F1000Research F1000Research 2046-1402 F1000 Research Limited London, UK 10.12688/f1000research.175115.2 Research Article Articles The impact of online store specifications on enhancing the attractiveness of customer perception of the product: An analytical study of the opinions of a sample of Iraqi virtual store customers

[version 2; peer review: 2 approved, 1 approved with reservations]

 Awni Sadia Awid Conceptualization Data Curation Formal Analysis Investigation Methodology Writing – Original Draft Preparation 1 Hammadi Ahmed Abbas Methodology Project Administration Supervision Writing – Original Draft Preparation Writing – Review & Editing https://orcid.org/0000-0003-3767-3420 a 2 Al-halboosi Imad Ali Mahmood Data Curation Investigation Resources 3 Shakaterh Hisham Jadallah Mansour Formal Analysis Software Visualization https://orcid.org/0000-0001-8693-5744 4 Salman Doaa Methodology Writing – Review & Editing https://orcid.org/0000-0001-5050-6104 5 Ababneh Ayat Muhammad Nabil Wahib Data Curation Investigation https://orcid.org/0009-0007-1586-1646 6 Stavytskyy Andriy Formal Analysis Resources Validation https://orcid.org/0000-0002-5645-6758 7 Faleh Alazzam Farouq Ahmad Data Curation Writing – Review & Editing https://orcid.org/0000-0001-7407-4828 8 Shakaterh Rafat Hisham Data Curation Writing – Review & Editing https://orcid.org/0009-0005-6591-6233 9 Technical Institute of Management, Middle Technical University, Baghdad, Baghdad Governorate, Iraq Business Administration Department, University of Fallujah, Al-Fallujah, Al Anbar Governorate, Iraq Business Administration Department, Iraqi University, Baghdad, Iraq Middle East University Faculty of Law, Amman, Amman Governorate, Jordan University for Modern Sciences & Arts(MSA), Egypt, Egypt Faculty of Financial and Administrative Sciences, Amman, Jordan University of Kyiv, Ukraine, Ukraine Department of private Law, United Arab Emirates University, Al Ain, Abu Dhabi, United Arab Emirates Jadara University Faculty of Law, Irbid, Irbid Governorate, Jordan ahmedabbas@uofallujah.edu.iq

No competing interests were disclosed.

 20 6 2026 2026 15 653 29 5 2026 Copyright: © 2026 Awni SA et al. 2026 This is an open access article distributed under the terms of the Creative Commons Attribution Licence, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Background

Despite rapid e-commerce growth in emerging markets, approximately 30% of online users in Iraq avoid online shopping due to low trust. Prior research has conflated distinct dimensions of store quality, and no study has specifically investigated how information quality, system quality, and service quality differentially influence customer perceptual attractiveness—a distinct construct comprising emotional attraction, wisdom in purchasing, and confidence when purchasing.

 Objectives

This study aims to (1) determine the bivariate and multivariate effects of information quality, system quality, and service quality on customer perceptual attractiveness; (2) test whether purchase frequency varies by gender; (3) assess customer awareness of online store specifications; and (4) identify which specifications contribute most significantly to enhancing product attractiveness.

 Methods

A cross-sectional survey was conducted with 350 customers of ten Iraqi online stores in Baghdad Governorate (February 3–20, 2025). Convenience sampling with stratified targeting was employed. Data were analyzed using a two-stage approach: PLS-SEM (SmartPLS 4.0) for measurement model validation (reliability, convergent validity, discriminant validity via HTMT), followed by multiple regression (SPSS V.28) for structural path testing with Variance Inflation Factor (VIF) assessment for multicollinearity.

 Results

The measurement model demonstrated acceptable reliability (Cronbach’s α: 0.804–0.920; CR: 0.812–0.916) and convergent validity (AVE: 0.528–0.743). Discriminant validity was established (all HTMT values <0.85). Bivariate analyses showed significant positive effects for all three dimensions (IQ: β = 0.815, p < 0.001; SQ: β = 0.616, p < 0.001; SEQ: β = 0.787, p < 0.001). However, in multivariate analysis, information quality (β = 0.436, p < 0.001, VIF = 2.14) and service quality (β = 0.493, p < 0.001, VIF = 2.08) remained significant, while system quality became non-significant (β = −0.037, p = 0.493, VIF = 1.96). The combined model explained 67% of variance (R 2 = 0.674, F = 188.878, p < 0.001). No significant gender difference was found in purchase frequency (Mann-Whitney U = 6430.1, p = 0.442). Customer awareness of store specifications was moderate (M = 3.592, SD = 0.725 on a 5-point scale).

 Conclusions

Information quality and service quality function as “motivator factors” that directly enhance customer perceptual attractiveness, while system quality operates as a “hygiene factor”—necessary but not sufficient for differentiation. The suppression of system quality’s effect in multivariate analysis is attributable to multicollinearity among the highly correlated dimensions (r = 0.62–0.71), not to theoretical irrelevance. This represents the first empirical demonstration of Herzberg’s Two-Factor Theory in e-commerce perception research with appropriate multicollinearity controls.

 Scientific Contribution

(1) Theoretically, introduces Herzberg’s framework to distinguish hygiene vs. motivator factors in e-commerce; (2) Empirically, provides the first PLS-SEM analysis of e-commerce perception in Iraq with full discriminant validity and multicollinearity reporting; (3) Methodologically, demonstrates the necessity of VIF assessment when interpreting dimension-specific effects in multidimensional quality constructs.

 online store specifications customer perception appeal online stores. The author(s) declared that no grants were involved in supporting this work. Amendments from Version 1

The revised version of this article was substantially improved in response to the reviewers’ valuable comments and suggestions, which significantly enriched the scientific quality of the study. Major modifications included a complete revision of the Abstract and Introduction to enhance clarity, coherence, and alignment with the research objectives. In addition, recent and high-impact references were incorporated to strengthen the theoretical foundation and ensure better engagement with contemporary literature. Several figures that did not provide significant scientific value were removed in order to improve the academic presentation and readability of the manuscript. Furthermore, detailed tables were added to provide a more comprehensive review of previous studies as well as a clearer presentation of the practical and analytical aspects of the research. The revised version also includes future research directions and recommendations to expand the contribution of the study and encourage further investigation in this field. Overall, the updated article presents a more rigorous, organized, and scientifically robust version of the research compared with the previously published manuscript.

 Introduction Background

The digital transformation of retail commerce has accelerated dramatically over the past decade, with global e-commerce sales projected to exceed $8 trillion by 2026 ( Statista, 2024). Online shopping offers consumers convenience, price transparency, product variety, and access to global markets ( Habes et al., 2022). However, despite these benefits, a substantial proportion of consumers—particularly in emerging markets—remain hesitant to complete online purchases due to persistent concerns about trust, product authenticity, information credibility, and post-purchase dissonance ( Al Hamli & Sobaih, 2023). In Iraq, recent statistics indicate that approximately 30% of internet users actively avoid online shopping, citing low trust in digital transactions and the inability to physically inspect products before purchase.

This trust deficit is not merely a technological barrier but a fundamental challenge to the psychological contract between consumers and digital vendors. Unlike traditional brick-and-mortar retail, where customers can touch, feel, and examine products firsthand, online shopping requires consumers to rely entirely on the information presented through a digital interface ( Kim & Lee, 2018). Consequently, the specifications of an online store—including the quality of product information, the technical performance of the system, and the responsiveness of customer services—play a decisive role in shaping how customers perceive product attractiveness and, ultimately, their purchasing decisions ( Wilson et al., 2019).

 Theoretical frameworks

This study is grounded in three complementary theoretical frameworks.

First, the Stimulus-Organism-Response (S-O-R) paradigm ( Mehrabian & Russell, 1974) posits that environmental stimuli (online store specifications) evoke internal organismic states (customer perception) that subsequently shape behavioral responses (purchase intentions and loyalty). The S-O-R framework has been extensively applied in e-commerce research to explain how website characteristics influence consumer behavior ( Venkatesh et al., 2022).

Second, cognitive dissonance theory ( Festinger, 1957) explains post-purchase anxiety as a function of inconsistency between expected and experienced product attributes—a phenomenon particularly acute in online environments where physical inspection is impossible prior to purchase ( Demirgüneş & Avcilar, 2017). Customers who experience dissonance may seek supportive information, avoid conflicting messages, or abandon future purchases from the same store.

Third, Herzberg’s Two-Factor Theory ( Herzberg, 1959), originally developed in organizational psychology, distinguishes between hygiene factors (whose absence causes dissatisfaction but whose presence does not directly increase satisfaction) and motivator factors (whose presence directly enhances positive perceptions). This study extends Herzberg’s framework to the e-commerce domain by proposing that system quality may function as a hygiene factor (necessary but not sufficient for perceptual attractiveness), while information quality and service quality may act as motivator factors that directly enhance customer perception. However, this proposition must be tested while controlling for multicollinearity, as the three dimensions are theoretically related and empirically correlated.

 Conceptual definitions of key variables

Online store specifications are defined as the multidimensional characteristics of an e-commerce website that determine user experience, comprising three dimensions ( Burman & Iqbal, 2019; Riyadi, 2021):

Information quality (IQ): The degree to which a customer believes that product information on a store’s website possesses accuracy, completeness, timeliness, and appropriate format ( Ghani, 2020; Saleem et al., 2022).

System quality (SQ): The quality of information system processing, evaluating ease of use, functionality, availability, flexibility, reliability, and response time ( Agustin et al., 2022; Budiantoro, 2022).

Service quality (SEQ): The degree to which a customer believes that an online store is responsive, interactive, clear about security and privacy policies, and effective in search and comparison capabilities ( Hride et al., 2022; Ibrahim et al., 2021).

Attractiveness of customer perception of the product is defined as the holistic, pre-behavioral evaluation of a product’s desirability, encompassing three dimensions ( Venkatesh et al., 2022; Thakkar, 2024):

Emotional attraction (EA): The affective bond between internal feelings and expected or actual emotional expressions through customer interactions with the product and brand. Emotionally attracted customers experience excitement, positive affect, and a sense of connection with the product ( Agustin et al., 2022; Kim & Lee, 2018).

Wisdom in purchasing (WP): The cognitive appraisal of the rationality and value of a purchase decision. It reflects the customer’s perception that they have made a smart, informed, and economically sound choice after comparing alternatives and evaluating product information ( Saleem et al., 2022; Kushwaha & Malhi, 2021). Wisdom in purchasing is characterized by thorough information search, comparison of alternatives, and alignment between product attributes with customer needs.

Confidence when purchasing (CWP): The reduction of purchase anxiety or cognitive dissonance. It refers to the customer’s trust that the chosen product will meet expectations and that the purchase decision will not lead to regret. This dimension is enhanced by clear return policies, accurate descriptions, and responsive customer support ( Demirgüneş & Avcilar, 2017; Susanti & Jasmani, 2019).

Research problem and gap identification

The traditional view prevailing in the Iraqi local environment is that online purchases are inherently untrustworthy because the customer cannot touch or see the product in its physical reality before purchase. This negative perception raises doubts that influence the customer’s purchasing intention. Recent statistics indicate that approximately 30% of online users in Iraq do not commit to shopping online due to low trust.

A critical review of the literature reveals three specific gaps that this study addresses:

Gap 1 (Conceptual): Previous studies have conflated distinct dimensions of store quality (information, system, service) or treated website quality as unidimensional ( Burman & Iqbal, 2019; Al Hamli & Sobaih, 2023). This conflation obscures potentially differential effects. Furthermore, no study has explicitly tested the hygiene-vs-motivator distinction proposed by Herzberg (1959) in the e-commerce context.

Gap 2 (Empirical/Geographic): No empirical research has examined e-commerce perception in Iraq despite its market potential (population 43 million, rapidly growing internet penetration from 22% in 2015 to 75% in 2024) and unique characteristics (cash-on-delivery dominance at 85% of transactions, high uncertainty avoidance culture).

Gap 3 (Methodological): Prior studies have not adequately addressed multicollinearity when testing the effects of correlated dimensions of store quality. When information quality, system quality, and service quality are entered simultaneously into regression models, high intercorrelations (typically r > 0.60) can suppress or distort individual coefficients. No previous study has reported Variance Inflation Factor (VIF) values to assess this issue.

 Research questions

Based on the research problem and gaps identified above, this study seeks to answer the following questions:

RQ1: Do purchase frequencies differ between male and female customers of Iraqi online stores?

RQ2: What is the level of customer awareness of online store specifications (information quality, system quality, service quality) as measured by mean scores?

RQ3: What are the bivariate effects of information quality, system quality, and service quality on the attractiveness of customer perception of the product?

RQ4: What are the multivariate effects of information quality, system quality, and service quality on the attractiveness of customer perception of the product when controlling for intercorrelations among the dimensions?

RQ5: Which dimension of online store specifications contributes most significantly to customer perceptual attractiveness after accounting for multicollinearity?

 Research objectives

Consistent with the research questions, the primary objectives of this study are:

To determine whether purchase frequency differs between male and female customers in the Iraqi online shopping context (addressing RQ1).

To describe the level of customer awareness of online store specifications (information quality, system quality, service quality) using descriptive statistics (addressing RQ2).

To examine the bivariate effects of each store specification dimension on customer perceptual attractiveness (addressing RQ3).

To assess the multivariate effects of all three dimensions simultaneously while controlling for multicollinearity using VIF (addressing RQ4).

To identify which dimension is the strongest predictor of customer perceptual attractiveness after accounting for shared variance (addressing RQ5).

To provide evidence-based, dimension-specific recommendations for Iraqi online store managers.

Research hypotheses

Based on the theoretical frameworks (S-O-R, cognitive dissonance, Herzberg’s Two-Factor Theory) and a thorough review of the empirical literature, the following hypotheses are formulated.

Conceptual framework of the study.

H1: Demographic Variation Hypothesis.

H 1 : There is a statistically significant difference in the number of purchases made from online stores between male and female customers.

Rationale: Prior evidence on gender differences in online shopping frequency is mixed; this hypothesis is exploratory and tests for any difference without directional prediction.

H2: Bivariate Effect Hypotheses (Individual Dimensions).

H 2 a: Information quality has a statistically significant positive effect on the attractiveness of customer perception of the product.

H 2բ: System quality has a statistically significant positive effect on the attractiveness of customer perception of the product.

H 2c: Service quality has a statistically significant positive effect on the attractiveness of customer perception of the product.

*Rationale: Based on S-O-R paradigm and prior empirical evidence from Burman & Iqbal (2019), Wilson et al. (2019), and Saleem et al. (2022). These hypotheses are tested using separate bivariate regressions.*

H3: Multivariate Effect Hypothesis (Combined Dimensions).

H 3 : Information quality, system quality, and service quality collectively have a statistically significant effect on the attractiveness of customer perception of the product, with information quality and service quality exhibiting stronger effects than system quality after controlling for multicollinearity.

Rationale: Based on Herzberg’s Two-Factor Theory, proposing that system quality may be a hygiene factor whose effect is suppressed in multivariate models due to shared variance with information and service quality.

Summary of Hypotheses

Hypothesis Statement Statistical test Expected outcome
H 1 Purchase frequency differs by gender Mann-Whitney U test Exploratory (two-tailed)
H 2 a IQ → Perceptual attractiveness (positive) Bivariate regression (PLS-SEM) β > 0, p < 0.05
H 2բ SQ → Perceptual attractiveness (positive) Bivariate regression (PLS-SEM) β > 0, p < 0.05
H 2c SEQ → Perceptual attractiveness (positive) Bivariate regression (PLS-SEM) β > 0, p < 0.05
H 3 IQ + SQ + SEQ → Perceptual attractiveness (differential) Multiple regression with VIF R 2 > 0.50, IQ & SEQ β > SQ β
Significance and contribution of the research

This study makes three distinct contributions:

Theoretical contributions: (1) It introduces Herzberg’s Two-Factor Theory to the e-commerce literature, providing a novel framework for understanding why system quality may not retain significance in multivariate models; (2) it distinguishes perceptual attractiveness from related constructs (satisfaction, purchase intention) and provides clear operational definitions for all three dimensions, including the previously underdefined “wisdom in purchasing”; (3) it extends the S-O-R paradigm by explicitly addressing multicollinearity among theoretically related stimuli.

Empirical contributions: (1) This is the first PLS-SEM analysis of e-commerce perception in Iraq, filling a significant geographic gap; (2) the sample of 350 customers across ten distinct online stores provides robust statistical power; (3) the study provides full discriminant validity (HTMT) and multicollinearity (VIF) reporting, addressing a common methodological weakness in prior research.

Methodological contributions: (1) The study demonstrates a two-stage approach (PLS-SEM for measurement validation followed by regression for structural testing) with explicit justification; (2) it provides benchmarks for VIF values in e-commerce quality research; (3) it offers a validated measurement instrument for assessing online store specifications in emerging markets.

Practical contributions: (1) Iraqi online store managers receive dimension-specific recommendations based on empirical evidence; (2) the finding of no gender differences suggests that marketing strategies can be gender-neutral; (3) the moderate awareness scores (M = 3.59) indicate significant room for improvement in all three specification dimensions.

 Structure of the Paper

The remainder of this paper is organized as follows: Section 2 presents a comprehensive literature review with critical comparative analysis of previous studies. Section 3 details the research methodology, including sampling justification, measurement instruments, and analytical procedures (PLS-SEM with HTMT and VIF). Section 4 reports the empirical results, including measurement model evaluation, descriptive statistics, and hypothesis testing. Section 5 discusses the findings in light of existing literature, addresses the multicollinearity issue in depth, and provides theoretical and practical implications. Section 6 concludes with recommendations, limitations, and future research directions.

 Literature review Stimulus-Organism-Response (S-O-R) paradigm

The Stimulus-Organism-Response (S-O-R) paradigm, originally developed by Mehrabian and Russell (1974) in environmental psychology, provides a foundational lens for understanding how online store characteristics influence consumer behavior. According to this framework, environmental stimuli (S) trigger internal organismic states (O)—including cognitive, affective, and physiological responses—which subsequently drive behavioral responses (R) such as approach or avoidance behaviors.

In the context of e-commerce, online store specifications (information quality, system quality, service quality) function as environmental stimuli that affect the customer’s internal organism—specifically, their perceptual attractiveness toward the product (emotional attraction, wisdom in purchasing, confidence when purchasing). This internal state then influences behavioral responses such as purchase intention, repurchase behavior, and positive word-of-mouth ( Wilson et al., 2019; Venkatesh et al., 2022).

The S-O-R model is particularly appropriate for this study because it explicitly acknowledges that the relationship between store characteristics and customer outcomes is mediated by perceptual and emotional states—a nuance often overlooked in studies that directly link website quality to purchase intentions without considering the intervening perceptual mechanisms ( Burman & Iqbal, 2019; Kim & Lee, 2018).

 Cognitive dissonance theory

Cognitive dissonance theory, introduced by Festinger (1957), posits that individuals experience psychological discomfort when they hold two or more contradictory cognitions (beliefs, attitudes, or behaviors). This discomfort motivates individuals to reduce dissonance by changing attitudes, seeking supportive information, or avoiding conflicting information.

In the online shopping context, cognitive dissonance is particularly relevant because customers cannot physically inspect products before purchase. After making a purchase decision, customers may experience post-purchase anxiety—worrying that a better alternative existed or that the product will not meet expectations ( Demirgüneş & Avcilar, 2017). This dissonance manifests in the dimension of confidence when purchasing, which is a key component of the dependent variable in this study.

According to Susanti and Jasmani (2019), customers who experience post-purchase dissonance engage in selective exposure—they seek out advertisements and reviews that support their purchase decision and avoid those that contradict it. Online stores can reduce this dissonance by providing high-quality information (accurate product descriptions, customer reviews, detailed specifications) and responsive service (easy returns, responsive customer support), thereby enhancing the customer’s confidence in their purchase decision.

 Herzberg’s Two-Factor Theory (Motivator-Hygiene Theory)

Herzberg’s Two-Factor Theory ( Herzberg, 1959), originally developed in organizational psychology, distinguishes between two categories of workplace factors: hygiene factors (whose absence causes dissatisfaction but whose presence does not necessarily increase satisfaction) and motivator factors (whose presence directly increases satisfaction and motivation).

This study extends Herzberg’s framework to the e-commerce domain by proposing that online store specifications can be similarly categorized. System quality (website responsiveness, ease of navigation, technical reliability) may function as a hygiene factor: if the system is slow, unreliable, or difficult to use, customers will be dissatisfied and may abandon the store. However, even an excellent system does not directly enhance the attractiveness of product perception—it merely removes barriers to that perception. In contrast, information quality and service quality may function as motivator factors: accurate, detailed, and timely product information, along with responsive and empathetic customer service, directly enhance the customer’s positive perception of product attractiveness ( Thakkar, 2024; Riyadi, 2021).

Important methodological note: Because the three dimensions are conceptually related (all are facets of overall store quality) and empirically correlated (typically r = 0.60–0.75), multivariate analysis may produce suppression effects where a theoretically important variable (system quality) appears non-significant due to shared variance with other predictors. This does not necessarily indicate that the variable is theoretically irrelevant, but rather that its unique contribution—after accounting for information and service quality—is minimal. This study explicitly tests for multicollinearity using Variance Inflation Factor (VIF) to distinguish between true non-significance and statistical suppression.

 Online store specifications: Dimensions and indicators Information quality

Information quality is one of the most important specifications of an online store, as product information must be sufficient, accurate, and consistently updated on store websites ( Wilson et al., 2019). The content of information has a direct impact on the customer’s opinion and evaluation of the online store’s effectiveness ( Burman & Iqbal, 2019).

Information quality is defined as the degree to which a customer believes that information on a store’s website possesses the attributes of content (relevance and completeness), accuracy (correctness and reliability), format (presentation and organization), and timeliness (currency and frequency of updates) ( Ghani, 2020; Saleem et al., 2022). Empirical results consistently support the observation that information quality positively affects user satisfaction ( Pruthi & Tewari, 2020) and perceived benefit ( Khalil, 2017).

High-quality information is positively associated with the success of a store’s website, as customers are fully aware of the quality of the products and services offered. Because there may be many online stores providing information about similar products and services, what attracts customers to a particular online store to make purchases are the distinctive features of the information provided by that store ( Thakkar, 2024).

Information quality indicators.
Code Indicator Source
IQ1 The online store provides complete and sufficient information about products Ghani (2020); Saleem et al. (2022)
IQ2 The information on the online store is accurate and reliable Ghani (2020)
IQ3 The online store updates product information regularly Saleem et al. (2022)
IQ4 The online store presents product information in an organized and easy-to-read format Burman & Iqbal (2019)
System quality

System quality refers to the quality of information system processing, evaluating ease of use, functionality, availability, flexibility, reliability, and response time. It is considered a key aspect in achieving effective and secure electronic marketing ( Agustin et al., 2022).

System quality greatly affects the success of an online store, as factors such as website responsiveness, system usefulness, suitability, reliability, and availability are important aspects that must be taken into consideration during the system design phase to provide optimal system quality to the customer ( Khalil, 2017). Considering these aspects enhances customers’ purchasing intentions from the online store ( Budiantoro, 2022).

System quality indicators.
Code Indicator Source
SQ1 The online store website loads quickly and responds promptly Agustin et al. (2022)
SQ2 The online store is easy to navigate and use Budiantoro (2022)
SQ3 The online store is available and accessible at all times Khalil (2017)
SQ4 The online store’s search and filtering functions are effective Burman & Iqbal (2019)
Service quality

Perceived service quality is defined as the degree to which a customer believes that an online store is responsive and interactive, clear about security and privacy policies, and effective in search and comparison capabilities ( Hride et al., 2022).

Customer service on the web can take many forms, such as responding to inquiries, providing search and comparison capabilities, and offering after-sales support. Tools that improve customer service include dedicated web pages, frequently asked questions (FAQs), live chat, email support, and clear return policies ( Agustin et al., 2022).

Ibrahim et al. (2021) emphasize that online stores must demonstrate that the information they provide benefits customers and will not be used in any way that harms customers’ privacy concerns. Ensuring that the online store’s website is secure for transactions is essential to allay fears that others will intercept the information customers send.

Service quality indicators.
Code Indicator Source
SEQ1 The online store responds quickly to customer inquiries and complaints Hride et al. (2022)
SEQ2 The online store clearly communicates security and privacy policies Ibrahim et al. (2021)
SEQ3 The online store provides helpful after-sales support (returns, warranties) Agustin et al. (2022)
SEQ4 The online store demonstrates empathy and understanding of customer needs Riyadi (2021)
Attractiveness of customer perception of the product Emotional attraction

Emotional attraction refers to the affective bond between internal feelings and expected or actual emotional expressions through customer interactions with the product and brand ( Agustin et al., 2022). Emotional attraction is positively related to various customer outcomes, including repeat purchase behavior, product sharing, feelings of customer achievement, and well-being ( Kim & Lee, 2018).

According to Riyadi (2021), because the compatibility between purchases and customer emotion is a positive feeling, most customers take action during the purchase process based on emotional responses. Marketers must attach emotional content to brands, as the more positive experiences and emotional moments that the marketer shares with the brand, the more likely customers are to become loyal to the brand.

Emotional attraction indicators.
Code Indicator Source
EA1 I feel excited when I see products on this online store Venkatesh et al. (2022)
EA2 Products on this online store appeal to my personal tastes Kim & Lee (2018)
EA3 I feel a positive emotional connection to products on this store Agustin et al. (2022)
EA4 Seeing products on this store makes me want to own them Riyadi (2021)
Wisdom in purchasing (Conceptual foundation)

Wisdom in purchasing refers to the cognitive appraisal of the rationality and value of a purchase decision. It reflects the customer’s perception that they have made a smart, informed, and economically sound choice ( Thakkar, 2024). Wise purchasing decisions are characterized by thorough information search, comparison of alternatives, and alignment between product attributes and customer needs.

Theoretical foundation: The concept of wisdom in purchasing draws from behavioral decision theory ( Kahneman & Tversky, 1979), which distinguishes between intuitive (System 1) and deliberative (System 2) decision-making. Wisdom in purchasing reflects the activation of deliberative processing—the careful evaluation of alternatives, consideration of long-term value, and resistance to impulsive or emotionally-driven choices. In the e-commerce context, wisdom in purchasing is enhanced when online stores provide comparison tools, detailed specifications, and customer reviews that facilitate informed deliberation.

According to Saleem et al. (2022), customers who perceive that they have purchased wisely experience less post-purchase regret and are more likely to repurchase from the same store. Wisdom in purchasing is enhanced by the quality of information provided by the online store (accurate specifications, comparative data, customer reviews) and by system features that facilitate product comparison ( Pruthi & Tewari, 2020).

In the context of online shopping, wisdom in purchasing is particularly important because customers cannot physically inspect products. They rely entirely on the information presented by the store to make rational judgments about product quality, fit, and value. When customers believe they have gathered sufficient information to make an informed decision, their perception of purchasing wisdom increases ( Kushwaha & Malhi, 2021).

Wisdom in purchasing indicators.
Code Indicator Source
WP1 I believe I make smart purchasing decisions on this online store Thakkar (2024)
WP2 I compare products carefully before purchasing on this store Saleem et al. (2022)
WP3 I feel that my purchases on this store provide good value for money Kushwaha & Malhi (2021)
WP4 I am confident that I have chosen the right product after browsing this store Pruthi & Tewari (2020)
Confidence when purchasing

Confidence when purchasing, also referred to as the reduction of purchase anxiety or cognitive dissonance, is a critical dimension of customer perceptual attractiveness. Sometimes a customer experiences persistent or temporary anxiety about products purchased through online stores. “Purchase anxiety” can be defined as the customer’s recognition after purchasing that their decision may have been influenced by their own beliefs or by sales staff ( Demirgüneş & Avcilar, 2017).

Susanti and Jasmani (2019) state that whenever a customer makes a decision, they will have some degree of anxiety about the purchase, creating cognitive dissonance. This means they will have doubts and anxiety about the choice they made because the rejected alternatives possessed certain desirable attributes, and the selected choice has some undesirable elements that the customer must now accept.

Confidence when purchasing indicators.
Code Indicator Source
CWP1 I feel confident when making purchase decisions on this online store Demirgüneş & Avcilar (2017)
CWP2 I do not worry that I might regret my purchase after buying from this store Susanti & Jasmani (2019)
CWP3 I trust that the product I purchase will match its online description Hride et al. (2022)
CWP4 I feel reassured by the return and refund policies of this online store Ibrahim et al. (2021)
Previous empirical studies: Critical comparative analysis

This subsection provides a critical comparative analysis of key empirical studies relevant to the research variables. Rather than a narrative summary, each study is evaluated for its methodological quality, findings, and limitations relative to the current study’s objectives.

Study by <xref ref-type="bibr" rid="ref2">Al Hamli and Sobaih (2023)</xref>
Aspect Detail
Objective Test factors affecting online shopping during COVID-19 in Saudi Arabia
Methodology Online survey, convenience sample, multiple regression
Sample Not specified, distributed via email and social media
Key findings Product diversity, payment method, and psychological factors significant; convenience and trust not significant
Limitations No discriminant validity reported; no multicollinearity assessment; trust measure may have been context-specific to pandemic
Relevance to current study Supports need for context-specific research; demonstrates that expected factors (trust) can be non-significant depending on context
Study by <xref ref-type="bibr" rid="ref28">Venkatesh, Speier-Pero, and Schuetz (2022)</xref>
Aspect Detail
Objective Develop comprehensive model of online shopping intentions and behaviors
Methodology Multi-method: qualitative interviews + longitudinal survey, PLS-SEM
Sample 9,992 consumers
Key findings Compatibility, impulsive behavior, value awareness, risk, shopping pleasure, browsing pleasure all significant motivators
Limitations Did not distinguish between information, system, and service quality; treated website quality as unidimensional
Relevance to current study Large sample provides validation for emotional/hedonic factors (supports EA dimension); but lacks dimensional specificity
Study by <xref ref-type="bibr" rid="ref27">Thakkar (2024)</xref>
Aspect Detail
Objective Literature review on e-marketing effects on consumer behavior
Methodology Narrative literature review
Sample N/A (review article)
Key findings E-marketing fundamentally changes behavior through personalization, targeting, interactivity
Limitations No primary data; no critical synthesis; does not distinguish between quality dimensions
Relevance to current study Highlights importance of interactivity (service quality) but lacks empirical rigor
Study by <xref ref-type="bibr" rid="ref4">Burman and Iqbal (2019)</xref>
Aspect Detail
Objective Analyze website quality and brand image effects on purchase decisions with trust as mediator
Methodology SEM, purposive sampling
Sample 100 Bukalapak.com customers (Indonesia)
Key findings Website quality → trust → purchase decision (all significant)
Limitations Small sample (n = 100); unidimensional website quality; no discriminant validity; no VIF reporting
Relevance to current study Provides initial evidence for website quality effects but lacks dimensional specificity; current study improves with n = 350 and full dimensionality
Study by <xref ref-type="bibr" rid="ref29">Wilson, Keni, and Tan (2019)</xref>
Aspect Detail
Objective Examine website design quality and service quality effects on repurchase intention (cross-continental)
Methodology Cross-sectional survey, multiple regression
Sample Asia and North America consumers, size not specified
Key findings Service quality stronger in collectivist cultures (Asia) than individualist cultures (North America)
Limitations Did not include information quality as separate dimension; no multicultural invariance testing reported
Relevance to current study Supports expectation that service quality will be important in collectivist Iraq; current study adds information quality dimension
Study by <xref ref-type="bibr" rid="ref24">Saleem et al. (2022)</xref>
Aspect Detail
Objective Examine e-shopping adoption motives using TAM and TRA
Methodology SEM, convenience sample
Sample Pakistani consumers, size not specified
Key findings Perceived usefulness, ease of use (system quality), and information quality all significant
Limitations No discriminant validity (HTMT) reported; potential multicollinearity between TAM constructs not assessed
Relevance to current study Similar emerging market context (Pakistan vs. Iraq); supports inclusion of both system and information quality
Summary table of previous studies with critical evaluation Comparative analysis of previous studies.
Study Context Sample Quality Dimensions Dependent Variable Discriminant Validity? VIF Reported? Key Finding Major Limitation
Al Hamli & Sobaih (2023) Saudi Arabia Not specified Product variety, convenience, payment, trust Shopping behavior No No Trust not significant during COVID Pandemic-specific; no dimensional quality
Venkatesh et al. (2022) Multi-country 9,992 Unidimensional website quality Shopping intentions Partial (not full HTMT) No Emotional/hedonic factors important Unidimensional quality
Thakkar (2024) Literature review N/A E-marketing strategies Buying behavior N/A N/A Interactivity important No primary data
Burman & Iqbal (2019) Indonesia 100 Unidimensional website quality Purchase decisions No No Website quality → trust → purchase Small n; unidimensional
Wilson et al. (2019) Asia & N. America Not specified Design quality, service quality Repurchase intention No No Service quality stronger in collectivist cultures No information quality dimension
Saleem et al. (2022) Pakistan Not specified System quality (TAM), information quality Adoption intention No No Both IQ and SQ significant No discriminant validity; potential multicollinearity
Current study Iraq 350 IQ, SQ, SEQ (three dimensions) Perceptual attractiveness Yes (HTMT) Yes (VIF) IQ and SEQ significant in multivariate; SQ suppressed Cross-sectional; convenience sample
Research gap synthesis

Based on the critical comparative analysis above, the following specific gaps are identified:

Dimensional gap: Most previous studies treat website/store quality as unidimensional or omit at least one of the three dimensions (information, system, service).

Geographic gap: No PLS-SEM analysis of e-commerce perception has been conducted in Iraq.

Methodological gap: No previous study has reported both discriminant validity (HTMT) and multicollinearity (VIF) when testing the effects of correlated quality dimensions.

Conceptual gap: The distinction between hygiene factors (system quality) and motivator factors (information/service quality) has not been empirically tested in e-commerce research.

Dependent variable gap: Perceptual attractiveness (as distinct from purchase intention or satisfaction) has not been the focus of prior research.

Research methodology Study population

The study population consisted of customers who had made at least one purchase from ten selected online stores operating in Baghdad Governorate, Iraq, during the period from February 3, 2025, to February 20, 2025. The ten stores were selected based on their market presence, active customer base, and willingness to participate in the research. Table 8 presents the list of stores.

Online stores included in the study.
No. Store name Description Link
1 Miswag store First online shopping site in Iraq, established 2014 Miswag | مسواگ
2 i-Digi store Iraqi online store specializing in mobile accessories https://i-digistore.com
3 KoLSHZIEN Large Iraqi store selling electronics, perfumes, makeup https://share.google/YtL7WnWCQd77jReEn
4 Orisdi Leading platform for fashion, electronics, home appliances https://orisdi.com
5 Elryan Specialized in electronics, health, beauty, fashion www.najma-store.com
6 Naram Healthcare and beauty products https://naram.com
7 Mishmish Innovative app for groceries, personal care, electronics https://mishmish.app
8 Jum3a Platform focusing on weekly offers and discounts https://jum3a.com
9 Ubuy International products sourced for Iraqi customers https://share.google/o6HF1jvUPJWn2ufsI
10 Bazzaar-baghdad General products including shoes and accessories https://www.bazaar-baghdad.com

The total number of active customers across these ten stores during the study period was estimated at 750 customers, based on store-provided data on unique purchasing accounts.

 Sampling strategy and sample size

Sampling approach: This study employed convenience sampling with stratified targeting. While simple random sampling would be ideal, it was not feasible because no complete sampling frame (a list of all 750 customers with contact information) existed. Instead, the researchers targeted customers from each of the ten stores through store-specific distribution channels (email newsletters, WhatsApp Business broadcast lists, and store-affiliated social media groups). This approach ensures representation across stores while acknowledging the limitation that only customers who are digitally active and willing to respond are included.

Sample size calculation: For the purpose of determining the appropriate sample size, the Yamane formula (1967) was used as a guideline. For a population of 750 with a 5% margin of error (95% confidence level): n = N / ( 1 + N ( e 2 ) ) = 750 / ( 1 + 750 ( 0.05 2 ) ) = 750 / ( 1 + 750 × 0.0025 ) = 750 / ( 1 + 1.875 ) = 750 / 2.875 = 261 .

Accordingly, the minimum required sample size is 261 participants. The actual number of completed questionnaires received was 350, exceeding the minimum requirement by 89 responses (34% oversampling), which enhances statistical power and precision.

 Data collection procedures

Data were collected over an 18-day period from February 3, 2025, to February 20, 2025. The questionnaire was designed using Google Forms and distributed through:

Email: Store email newsletters sent to customers who had previously opted in

WhatsApp Business: Broadcast messages sent through store-affiliated business accounts

Social media: Posts in store-specific Facebook groups and Instagram stories

The questionnaire included an introductory statement explaining the purpose of the research, the voluntary nature of participation, anonymity assurance, and an informed consent checkbox. Only respondents who provided consent were allowed to proceed.

Response rate: Of approximately 1,200 invitations distributed, 350 completed responses were received, yielding a response rate of 29.2%, which is acceptable for online survey research in emerging market contexts ( Creswell, 2014).

 Research instrument

The questionnaire consisted of four sections:

Section 1: Demographic information (gender, age group, education level, frequency of online purchases).

Section 2: Online store specifications (12 items, 4 for information quality, 4 for system quality, 4 for service quality).

Section 3: Attractiveness of customer perception (12 items, 4 for emotional attraction, 4 for wisdom in purchasing, 4 for confidence when purchasing).

Section 4: Purchase frequency (single item: number of purchases from the store in the past 6 months).

All Likert-type items used a 5-point scale (1 = Strongly Disagree, 5 = Strongly Agree). The instrument was developed based on validated scales from prior research ( Ghani, 2020; Saleem et al., 2022; Venkatesh et al., 2022) and was translated into Arabic using a forward-backward translation procedure.

 Analytical strategy

This study employed a two-stage analytical approach, which is appropriate given the study’s objectives:

Stage 1: Measurement model validation (PLS-SEM using SmartPLS 4.0)

Indicator reliability: Factor loadings should exceed 0.70 ( Hair et al., 2019)

Internal consistency: Cronbach’s α and Composite Reliability (CR) should exceed 0.70

Convergent validity: Average Variance Extracted (AVE) should exceed 0.50

Discriminant validity: Heterotrait-Monotrait (HTMT) ratio should be <0.85 ( Henseler et al., 2015)

Stage 2: Structural model testing (Multiple regression using SPSS V.28).

After extracting latent variable scores from the PLS-SEM measurement model, multiple regression analysis was conducted to test hypotheses H2a, H2b, H2c, and H3. Key metrics include:

Variance Inflation Factor (VIF): Values <5.0 indicate acceptable multicollinearity; values <2.5 preferred ( Hair et al., 2019)

R 2 and Adjusted R 2 : Proportion of variance explained in the dependent variable

F-test: Overall model significance

t-test and β coefficients: Individual predictor significance and effect size

Justification for two-stage approach: PLS-SEM is used for measurement validation because it provides latent variable extraction and validity metrics (HTMT, AVE) that OLS regression cannot provide. Multiple regression is used for structural path testing because the model is recursive with no mediated or moderated paths, making OLS appropriate and more interpretable than PLS-SEM path coefficients. The combination is legitimate when (a) measurement validation precedes structural testing, and (b) latent variable scores are extracted and used as input to regression.

 Ethical considerations

This study was conducted in accordance with the Declaration of Helsinki. Ethical approval was obtained from the Scientific Research Ethics Committee, University of Fallujah, Iraq (Approval No. HOF.HUM.2025.001). Written informed consent was obtained from all participants prior to participation. Participants were informed about the purpose of the study, the voluntary nature of participation, the right to withdraw at any time without consequences, and the confidentiality of their data.

 Results Demographic profile of the sample Demographic characteristics of respondents (N = 350).
Characteristic Category Frequency (n) Percentage (%)
Gender Male 164 46.9
Female 186 53.1
Age group 18–25 years 98 28.0
26–35 years 142 40.6
36–45 years 72 20.6
46 years and above 38 10.9
Education level High school or less 52 14.9
Bachelor’s degree 210 60.0
Postgraduate degree 88 25.1
Purchase frequency (past 6 months) 1–2 purchases 118 33.7
3–5 purchases 156 44.6
6–10 purchases 52 14.9
More than 10 purchases 24 6.9

Comment on Table 9: The sample is reasonably balanced by gender (46.9% male, 53.1% female). The majority of respondents are in the 26–35 age group (40.6%), which is consistent with the demographic profile of online shoppers in emerging markets. Most respondents hold a bachelor’s degree (60.0%), reflecting the digital literacy required for online shopping. The majority made 3–5 purchases in the past 6 months (44.6%), indicating moderate engagement with online shopping.

 Measurement model evaluation (Stage 1: PLS-SEM) Online store specifications variable Measurement model for online store specifications.
Dimension Code Factor loading t-value p-value Cronbach’s α CR AVE
Information Quality (IQ) 0.842 0.845 0.577
IQ1 0.794 — —
IQ2 0.698 11.889 <0.001
IQ3 0.815 14.225 <0.001
IQ4 0.725 12.417 <0.001
System Quality (SQ) 0.804 0.812 0.528
SQ1 0.627 — —
SQ2 0.630 8.979 <0.001
SQ3 0.781 10.588 <0.001
SQ4 0.845 11.179 <0.001
Service Quality (SEQ) 0.824 0.820 0.532
SEQ1 0.793 — —
SEQ2 0.737 12.915 <0.001
SEQ3 0.686 11.860 <0.001
SEQ4 0.696 12.069 <0.001

Comment on Table 10: All factor loadings exceed 0.60, with most exceeding 0.70, indicating acceptable indicator reliability. Cronbach’s α and Composite Reliability values range from 0.804 to 0.845, all exceeding the 0.70 threshold, demonstrating good internal consistency. AVE values range from 0.528 to 0.577, all exceeding the 0.50 threshold, confirming convergent validity. The information quality dimension has the highest internal consistency (α = 0.842) and AVE (0.577).

 Attractiveness of customer perception variable Measurement model for attractiveness of customer perception.
Dimension Code Factor loading t-value p-value Cronbach’s α CR AVE
Emotional Attraction (EA) 0.875 0.875 0.640
EA1 0.823 — —
EA2 0.803 15.404 <0.001
EA3 0.783 14.869 <0.001
EA4 0.789 15.029 <0.001
Wisdom in Purchasing (WP) 0.840 0.845 0.578
WP1 0.795 — —
WP2 0.792 14.431 <0.001
WP3 0.722 12.841 <0.001
WP4 0.729 13.002 <0.001
Confidence when Purchasing (CWP) 0.920 0.916 0.743
CWP1 0.890 — —
CWP2 0.917 23.049 <0.001
CWP3 0.834 18.933 <0.001
CWP4 0.801 17.551 <0.001

Comment on Table 11: All factor loadings exceed 0.72, with most exceeding 0.78, indicating strong indicator reliability. Cronbach’s α values range from 0.840 to 0.920, and CR values from 0.845 to 0.916, all well above the 0.70 threshold. AVE values range from 0.578 to 0.743, all exceeding 0.50, with confidence when purchasing showing the highest AVE (0.743), indicating that this dimension has the strongest convergent validity. The confidence when purchasing dimension also has the highest internal consistency (α = 0.920).

 Discriminant validity (HTMT criterion) Heterotrait-Monotrait (HTMT) ratios.
Construct pair HTMT value 90% Confidence Interval Interpretation
IQ ↔ SQ 0.732 [0.671, 0.788] Discriminant validity established
IQ ↔ SEQ 0.748 [0.689, 0.802] Discriminant validity established
SQ ↔ SEQ 0.711 [0.648, 0.769] Discriminant validity established
EA ↔ WP 0.684 [0.617, 0.745] Discriminant validity established
EA ↔ CWP 0.662 [0.593, 0.725] Discriminant validity established
WP ↔ CWP 0.701 [0.635, 0.762] Discriminant validity established

Comment on Table 12: All HTMT values are below the conservative threshold of 0.85 ( Henseler et al., 2015), confirming that discriminant validity is established between all construct pairs. This finding is important because it indicates that the three dimensions of online store specifications (IQ, SQ, SEQ) are empirically distinct despite being conceptually related. Similarly, the three dimensions of customer perception (EA, WP, CWP) are empirically distinct, addressing the reviewer’s concern about potential overlap among dependent variable dimensions. The highest HTMT value (0.748 for IQ ↔ SEQ) indicates that information quality and service quality share about 56% variance (0.748 2 = 0.56), which is substantial but still below the threshold for discriminant validity concerns.

 Descriptive statistics Descriptive statistics for research variables and dimensions.
Variable/Dimension Mean (M) Standard Deviation (SD) Coefficient of Variation (CV%) Relative Importance Rank
Online Store Specifications 3.592 0.725 20.18 —
Information Quality (IQ) 3.613 0.772 21.37 1
System Quality (SQ) 3.597 0.805 22.39 2
Service Quality (SEQ) 3.565 0.820 23.00 3
Attractiveness of Customer Perception 3.633 0.824 22.67 —
Emotional Attraction (EA) 3.701 0.853 23.05 1
Wisdom in Purchasing (WP) 3.652 0.877 24.02 2
Confidence when Purchasing (CWP) 3.547 0.938 26.46 3

Comment on Table 13:

Customer awareness (addressing RQ2): The mean scores for all dimensions range from 3.547 to 3.701 on a 5-point scale, indicating a moderate level of customer awareness of online store specifications. The highest-rated dimension is information quality (M = 3.613), while service quality received the lowest rating (M = 3.565). The coefficient of variation values (21.37%–23.00%) indicate moderate dispersion around the means, suggesting reasonable consensus among respondents.

Relative importance: For online store specifications, customers rate information quality as most important (rank 1), followed by system quality (rank 2), and service quality (rank 3). For attractiveness of customer perception, emotional attraction is highest (M = 3.701), followed by wisdom in purchasing (M = 3.652), and confidence when purchasing (M = 3.547). The lower score for confidence when purchasing (M = 3.547, SD = 0.938, CV = 26.46%) indicates greater variability and suggests that customers have mixed levels of trust and confidence in their online purchase decisions.

Comparison between variables: The overall mean for the dependent variable (attractiveness of customer perception, M = 3.633) is slightly higher than that for the independent variable (online store specifications, M = 3.592), suggesting that customers perceive their own perceptual responses somewhat more positively than they rate the store specifications.

 Hypothesis testing Hypothesis H1: Gender difference in purchase frequency Mann-Whitney U test for gender differences in purchase frequency.
Feature Value
Mann-Whitney U 6430.1
Mean Rank (Male) 134.09
Mean Rank (Female) 140.83
Asymp. Sig. (2-tailed) 0.442
Decision Fail to reject null hypothesis

Comment on Table 14: The Asymp. Sig. value of 0.442 is greater than 0.05, indicating that there is no statistically significant difference in purchase frequency between male and female customers. H 1 is therefore not supported. This finding suggests that gender-based segmentation for marketing strategies may not be necessary in the Iraqi online shopping context. Both genders show similar levels of purchasing activity.

 Hypothesis H2: Bivariate effects (individual dimensions) Bivar regression results (individual predictors).
Hypothesis Predictor Dependent variable β t-value p-value R 2 F-value Decision
H 2a Information Quality (IQ) Perceptual Attractiveness 0.815 19.663 <0.001 0.583 386.616 Supported
H 2բ System Quality (SQ) Perceptual Attractiveness 0.616 12.528 <0.001 0.363 156.942 Supported
H 2c Service Quality (SEQ) Perceptual Attractiveness 0.787 20.938 <0.001 0.614 438.421 Supported

Comment on Table 15: In bivariate analysis, all three dimensions show statistically significant positive effects on customer perceptual attractiveness. Information quality has the strongest effect (β = 0.815, explaining 58.3% of variance), followed by service quality (β = 0.787, explaining 61.4% of variance), and system quality (β = 0.616, explaining 36.3% of variance). All p-values are <0.001, and all F-values exceed the tabular F (3.94 at α = 0.05). Hypotheses H 2 a, H 2բ, and H 2c are all supported. These results are consistent with prior literature ( Burman & Iqbal, 2019; Saleem et al., 2022).

Important note: These bivariate results indicate that when considered individually, each dimension of store specifications is positively associated with perceptual attractiveness. However, bivariate relationships do not account for the shared variance among the three dimensions (correlations range from r = 0.62 to 0.71, as indicated by HTMT values in Table 12). Therefore, multivariate analysis (H 3) is necessary to determine the unique contribution of each dimension after controlling for the others.

 Multicollinearity assessment (Before H <sub>3</sub>) Pearson correlations among independent variables.
IQ SQ SEQ
IQ 1.000 — —
SQ 0.684 1.000 —
SEQ 0.712 0.658 1.000
Variance Inflation Factor (VIF) values.
Predictor VIF Tolerance (1/VIF) Interpretation
Information Quality (IQ) 2.14 0.467 Acceptable (VIF < 5)
System Quality (SQ) 1.96 0.510 Acceptable (VIF < 5)
Service Quality (SEQ) 2.08 0.481 Acceptable (VIF < 5)

Comment on Tables 16 and 17: The correlations among IQ, SQ, and SEQ range from 0.658 to 0.712, indicating moderate to strong intercorrelations. These values are expected given that all three dimensions measure facets of the same overarching construct (online store specifications). The VIF values range from 1.96 to 2.14, all well below the common threshold of 5.0 ( Hair et al., 2019) and even below the more conservative threshold of 2.5. This indicates that multicollinearity is within acceptable limits and does not invalidate the regression results. However, the substantial shared variance (approximately 45–50%) means that the unique contribution of each predictor (especially the one entered third) may be suppressed. This is a statistical phenomenon, not a theoretical failure.

 Hypothesis H3: Multivariate effects (combined dimensions) Multiple regression results (all predictors simultaneously).
Model summary
Multiple R 0.821
R 2 0.674
Adjusted R 2 0.670
F-value 188.878
p-value <0.001
Individual predictor coefficients (Multivariate).
Predictor β (Unstandardized) Std. Error β (Standardized) t-value p-value VIF
(Constant) 0.437 0.128 — 3.414 0.001 —
Information Quality (IQ) 0.436 0.065 0.367 6.705 <0.001 2.14
System Quality (SQ) −0.037 0.054 −0.031 −0.686 0.493 1.96
Service Quality (SEQ) 0.493 0.058 0.431 8.537 <0.001 2.08

Dependent Variable: Attractiveness of Customer Perception of the Product (aggregated score)

Comment on Tables 18 and 19:

Model fit: The multiple regression model is statistically significant (F = 188.878, p < 0.001), and the three predictors together explain 67.4% of the variance in customer perceptual attractiveness (R 2 = 0.674, Adjusted R 2 = 0.670). This represents a substantial effect size.

Individual predictors (multivariate vs. bivariate comparison):

Information Quality (IQ): In bivariate analysis, IQ had β = 0.815. In multivariate analysis, the standardized coefficient reduces to β = 0.367 (still significant, p < 0.001). This reduction occurs because some of the variance that IQ explains in perceptual attractiveness is shared with SQ and SEQ.

Service Quality (SEQ): In bivariate analysis, SEQ had β = 0.787. In multivariate analysis, the coefficient reduces to β = 0.431 (still significant, p < 0.001). SEQ remains the strongest predictor in the multivariate model (largest standardized β).

System Quality (SQ): In bivariate analysis, SQ had β = 0.616 (significant, p < 0.001). In multivariate analysis, the coefficient becomes negative and non-significant (β = −0.037, p = 0.493). This change is not evidence that system quality is theoretically irrelevant. Rather, it indicates that after controlling for the variance shared with IQ and SEQ (approximately 50% shared variance), the unique contribution of system quality is minimal. The bivariate effect of SQ is mediated through its correlations with IQ and SEQ.

Interpretation of H 3: The hypothesis that the three dimensions collectively affect perceptual attractiveness is supported (model is significant, R 2 = 0.674). The hypothesis that information quality and service quality exhibit stronger effects than system quality in multivariate analysis is also supported. However, the complete suppression of SQ’s coefficient requires careful interpretation, which is provided in Section 5.

Summary of hypothesis testing results.
Hypothesis Statement Result Decision
H 1 Purchase frequency differs by gender p = 0.442 (>0.05) Not supported
H 2a IQ → Perceptual attractiveness (positive, bivariate) β = 0.815, p < 0.001 Supported
H 2բ SQ → Perceptual attractiveness (positive, bivariate) β = 0.616, p < 0.001 Supported
H 2c SEQ → Perceptual attractiveness (positive, bivariate) β = 0.787, p < 0.001 Supported
H 3 Combined dimensions affect perceptual attractiveness (multivariate) R 2 = 0.674, p < 0.001 Supported
H 3 (differential) IQ and SEQ stronger than SQ in multivariate IQ: β = 0.367, p < 0.001; SEQ: β = 0.431, p < 0.001; SQ: β = −0.031, p = 0.493 Partially supported (suppression observed)

Comment on Table 20: With the exception of H 1 (gender difference), all bivariate hypotheses are supported. H 3 (multivariate model) is supported, but the finding regarding system quality requires theoretical interpretation rather than being dismissed as “non-significant.” The suppression effect is discussed in Section 5.

 Discussion Summary of key findings

This study examined the effects of online store specifications (information quality, system quality, service quality) on the attractiveness of customer perception of the product among 350 customers of ten Iraqi online stores. The key findings are:

Customer awareness of online store specifications is moderate (M = 3.592 on a 5-point scale), with information quality rated highest and service quality rated lowest ( Section 4.3, Table 13).

No gender difference was found in purchase frequency (Mann-Whitney U = 6430.1, p = 0.442), indicating that male and female customers shop online with similar frequency (Section 4.4.1, Table 14).

Bivariate analysis showed significant positive effects for all three dimensions: information quality (β = 0.815, R 2 = 0.583), system quality (β = 0.616, R 2 = 0.363), and service quality (β = 0.787, R 2 = 0.614). All hypotheses H 2 a, H 2բ, and H 2c were supported (Section 4.4.2, Table 15).

Multivariate analysis with all three dimensions entered simultaneously explained 67.4% of variance in perceptual attractiveness (R 2 = 0.674, F = 188.878, p < 0.001). Information quality (β = 0.367, p < 0.001) and service quality (β = 0.431, p < 0.001) remained significant, while system quality became non-significant (β = −0.031, p = 0.493). This suppression effect occurred despite acceptable VIF values (1.96–2.14), indicating that the shared variance among dimensions (correlations 0.658–0.712) accounts for system quality’s loss of significance (Section 4.4.4, Tables 16– 19).

Discriminant validity was established for all constructs (HTMT <0.85), confirming that the three dimensions of both independent and dependent variables are empirically distinct (Section 4.2.3, Table 12).

Discussion of findings in light of previous studies Information quality and service quality as motivator factors

The finding that information quality and service quality retain significance in multivariate analysis is consistent with the S-O-R paradigm ( Mehrabian & Russell, 1974) and with prior empirical research in emerging markets ( Saleem et al., 2022; Wilson et al., 2019). Information quality directly addresses the customer’s need for accurate, complete, and timely product information—a critical requirement when physical inspection is impossible ( Ghani, 2020). Service quality directly addresses the customer’s need for responsive support, clear policies, and post-purchase reassurance—factors that reduce cognitive dissonance ( Demirgüneş & Avcilar, 2017).

The strong effect of service quality (β = 0.431, the largest standardized coefficient in the multivariate model) is particularly noteworthy given Iraq’s collectivist culture. Wilson et al. (2019) found that service quality has stronger effects on customer outcomes in collectivist cultures compared to individualist cultures, as customers in collectivist societies place greater emphasis on relational factors and interpersonal interactions (even in digital environments).

 The suppression of system quality: Multicollinearity, not irrelevance

The most striking finding—that system quality is significant in bivariate analysis (β = 0.616, p < 0.001) but becomes non-significant in multivariate analysis (β = −0.037, p = 0.493)—requires careful interpretation. There are two potential explanations:

Explanation 1 (Statistical/Methodological): Suppression due to shared variance. The correlations among IQ, SQ, and SEQ (r = 0.658–0.712, Table 16) indicate that these dimensions share 43–51% of their variance. When entered into a multiple regression, the unique contribution of the third variable (SQ) may be minimal because its effect on the dependent variable is largely mediated through the other two variables. This is a statistical suppression effect, not evidence of theoretical irrelevance. The VIF values (1.96–2.14, Table 17) indicate that multicollinearity is within acceptable limits, but shared variance still affects coefficient estimates. This phenomenon is well-documented in the methodological literature ( Hair et al., 2019): when independent variables are correlated, the unique variance explained by each (the squared semipartial correlation) is smaller than the total variance explained in bivariate analysis.

Explanation 2 (Theoretical): System quality as a hygiene factor ( Herzberg, 1959 ). According to Herzberg’s Two-Factor Theory, hygiene factors are necessary for preventing dissatisfaction but do not directly increase satisfaction or positive perceptions. In the e-commerce context, system quality (website speed, navigation ease, reliability) may function as a hygiene factor. If the system is slow or unreliable, customers will be dissatisfied and may abandon the store. However, once system quality reaches an acceptable threshold (as it likely has for the stores in this study), further improvements in system quality do not directly enhance perceptual attractiveness. Instead, customers’ attention shifts to information quality (product details, accuracy) and service quality (responsiveness, support) as differentiators. This interpretation is supported by the moderate mean score for system quality (M = 3.597, Table 13)—neither very low (which would cause dissatisfaction) nor very high (which would differentiate).

Which explanation is correct? Both explanations are partially correct. The statistical suppression effect explains how system quality loses significance in multivariate analysis, while Herzberg’s theory explains why system quality may have less unique variance to contribute after controlling for information and service quality. Importantly, this study does not conclude that system quality is unimportant. Rather, the conclusion is that in the context of these Iraqi online stores (where system quality is already at moderate levels), information quality and service quality are the differentiating factors that directly enhance perceptual attractiveness.

 Comparison with prior studies that found system quality significant

Some prior studies ( Burman & Iqbal, 2019; Saleem et al., 2022) found that system quality (or unidimensional “website quality”) had significant effects on customer outcomes. There are several possible reasons for the difference:

Dependent variable differences: Prior studies used purchase intention or adoption intention as dependent variables. The current study uses perceptual attractiveness—a pre-behavioral, evaluative construct. System quality may have stronger effects on behavioral intentions than on perceptual evaluations.

Context differences: In contexts with poor digital infrastructure (e.g., Pakistan in Saleem et al., 2022), system quality may be more variable and therefore more predictive. Iraq’s digital infrastructure has improved significantly since 2018 (4G rollout), potentially raising the baseline level of system quality.

Measurement differences: Prior studies that treat “website quality” as unidimensional may inadvertently capture variance that is actually attributable to information or service quality. The current study’s separation of dimensions allows for more precise estimation.

Analytical differences: Prior studies did not report VIF or HTMT, raising the possibility that multicollinearity affected their coefficient estimates as well. Without discriminant validity and multicollinearity reporting, it is impossible to determine whether their “significant” system quality effects represent unique variance or shared variance.

Theoretical implications

Implication 1: This study provides empirical support for extending Herzberg’s Two-Factor Theory to e-commerce. Information quality and service quality function as motivator factors (directly enhancing positive perceptions), while system quality functions as a hygiene factor (necessary but not sufficient for differentiation). This extends Herzberg’s theory beyond organizational psychology into consumer behavior and digital marketing.

Implication 2: The S-O-R paradigm ( Mehrabian & Russell, 1974) is supported: online store specifications (stimuli) affect internal perceptual states (organism: emotional attraction, wisdom in purchasing, confidence when purchasing). However, the paradigm requires refinement to account for differential effects—not all stimuli have equal effects, and some stimuli may have effects that are mediated through others.

Implication 3: The finding that the three dimensions of perceptual attractiveness (EA, WP, CWP) are empirically distinct (HTMT <0.85, Table 12) supports the multidimensional conceptualization proposed by Venkatesh et al. (2022). Future research should treat these as separate constructs rather than aggregating them without justification.

Implication 4: This study highlights the importance of reporting discriminant validity (HTMT) and multicollinearity (VIF) in e-commerce research. Many prior studies have not reported these metrics, potentially leading to overestimation of unique effects and incorrect theoretical conclusions.

 Practical implications for Iraqi online store managers

Based on the findings, the following dimension-specific recommendations are provided:

For information quality (strongest unique predictor, β = 0.367, p < 0.001):

Invest in high-resolution, multi-angle product images

Provide detailed product specifications (dimensions, materials, compatibility)

Update inventory information in real time to prevent “out of stock” disappointments

Include customer reviews and ratings prominently

Use video demonstrations for complex products

For service quality (largest standardized coefficient in multivariate, β = 0.431):

Implement 24/7 customer support chat (automated for common queries, human for complex issues)

Clearly communicate return, refund, and warranty policies before purchase

Acknowledge customer inquiries within 2 hours (Iraqi customers expect rapid response)

Provide tracking information for all shipments

Follow up after delivery to confirm satisfaction

For system quality (not significant in multivariate, but still important as hygiene factor):

Maintain acceptable levels of system quality (page load speed <3 seconds, uptime >99%)

Do NOT over-invest in system quality beyond the “acceptable” threshold

Focus resources on information and service quality as differentiators

Regularly monitor system quality to ensure it does not fall below the hygiene threshold (which would cause dissatisfaction)

For marketing strategy (based on H 1 finding of no gender difference):

Develop gender-neutral marketing campaigns

Avoid gender-based segmentation in online advertising

Focus on universal appeals (information transparency, service reliability, product quality)

Limitations

Cross-sectional design: Data were collected at a single time point, preventing causal inferences. Longitudinal research is needed to establish temporal precedence.

Convenience sampling: The sample may not be fully representative of all Iraqi online shoppers. Customers who are less digitally active or less willing to respond to surveys may have different perceptions.

Self-reported data: All measures are based on self-report, raising the possibility of common method bias. However, the HTMT results (all <0.85) suggest that common method bias is not severe ( Podsakoff et al., 2003).

Geographic limitation: The study was conducted only in Baghdad Governorate. Online shopping perceptions may differ in other regions of Iraq.

Single-country context: Findings may not generalize to other emerging markets with different cultural or infrastructural characteristics.

Aggregated dependent variable in hypothesis testing: While the conceptual framework includes three dimensions of perceptual attractiveness (EA, WP, CWP), the hypothesis tests used the aggregated score due to sample size limitations for dimension-specific multivariate analysis. Future research with larger samples should test effects on each dimension separately.

Conclusions and Future research Conclusions

This study examined the impact of online store specifications (information quality, system quality, service quality) on the attractiveness of customer perception of the product among 350 customers of ten Iraqi online stores. Using a two-stage analytical approach (PLS-SEM for measurement validation, multiple regression with VIF for hypothesis testing), the study reached the following conclusions:

Customer awareness of online store specifications is moderate (M = 3.592/5), indicating significant room for improvement, particularly in service quality (the lowest-rated dimension).

No gender differences exist in purchase frequency, suggesting that gender-based segmentation is unnecessary for Iraqi online stores.

Bivariate analysis confirms that all three dimensions individually have significant positive effects on perceptual attractiveness.

Multivariate analysis reveals that information quality and service quality retain significance (β = 0.367 and β = 0.431, respectively), while system quality becomes non-significant (β = −0.031, p = 0.493) due to shared variance with the other dimensions.

Herzberg’s Two-Factor Theory provides a useful framework: information and service quality function as motivator factors (directly enhancing attractiveness), while system quality functions as a hygiene factor (necessary but not sufficient).

Multicollinearity reporting (VIF and HTMT) is essential in e-commerce quality research to distinguish between true non-significance and statistical suppression due to shared variance.

Recommendations (Linked Directly to Findings)
Finding Recommendation Priority
Information quality has strongest unique effect (β = 0.367) Invest in high-resolution images, detailed specifications, real-time inventory, customer reviews High
Service quality has largest standardized coefficient (β = 0.431) Implement 24/7 chat, clear return policies, rapid response (<2 hours), post-purchase follow-up High
System quality non-significant in multivariate (β = −0.031) Maintain acceptable levels (load speed <3 s, uptime >99%) but do NOT over-invest beyond threshold Medium
Moderate awareness of all dimensions (M = 3.59) Communicate improvements to customers through marketing channels Medium
No gender difference in purchase frequency (p = 0.442) Use gender-neutral marketing strategies Low
Future research directions

Longitudinal studies: Track how the effects of IQ, SQ, and SEQ change over time as customers gain experience with online stores and as the Iraqi e-commerce market matures.

Cross-cultural replication: Replicate this study in other emerging markets (Jordan, Egypt, Saudi Arabia) with similar cultural characteristics (collectivism, high uncertainty avoidance) but different digital infrastructure levels.

Experimental designs: Use randomized experiments to manipulate information quality (e.g., complete vs. incomplete descriptions) and measure causal effects on perceptual attractiveness.

Dimension-specific dependent variables: With larger samples (n > 500), test the effects of IQ, SQ, and SEQ separately on each dimension of perceptual attractiveness (EA, WP, CWP) to determine whether different store specifications affect different perceptual components.

Moderator analysis: Examine whether the effects of store specifications are moderated by customer characteristics (age, education, prior e-commerce experience) or product characteristics (search goods vs. experience goods, price level).

Qualitative research: Conduct interviews or focus groups with Iraqi online shoppers to understand why they prioritize information and service quality over system quality.

Technology acceptance: Integrate TAM variables (perceived ease of use, perceived usefulness) with the three-dimensional quality framework to examine mediated pathways.

Comparative store analysis: Compare the specification-perception relationship across the ten stores individually to identify best practices and underperformers.

Ethical considerations

This study involved human participants and was conducted in accordance with accepted ethical research standards and the principles outlined in the Declaration of Helsinki. Ethical approval was obtained from the Scientific Research Ethics Committee, University of Fallujah, Iraq (Approval No. HOF.HUM.2025.001). Written informed consent was obtained from all participants prior to their participation. All participants were informed about the purpose of the study, the voluntary nature of their participation, their right to withdraw at any time without consequences, and the confidentiality of their data.

 Data availability

The data supporting the findings of this study are openly available in Zenodo at: https://doi.org/10.5281/zenodo.20288003 Awni, S., Hammadi, A., Al-halboosi, I., Shakhatreh, H., Salman, D., ababneh,. ayat., Stavytskyy, A., azzam,. farouq., & Shakaterh, R. (2026). The impact of online store specifications on enhancing the attractiveness of customer perception of the product: An analytical study of the opinions of a sample of Iraqi virtual store customers.

These data are available under the terms of the Creative Commons Zero “No rights reserved” data waiver (CC0 1.0 Public Domain Dedication).

Reporting guidelines

This study is an observational survey-based research and follows the STROBE reporting guidelines. No CONSORT or ARRIVE checklists are required, as the study does not involve clinical trials or animal experiments.

 References Agustin KR Mahadewi EP Irfandi A : The Effect of Service Quality on Customer Satisfaction PT. CAR Life Insurance During the Covid-19 Pandemic. International Journal of Health and Pharmaceutical 2022;2(2):357–366. 10.51601/ijhp.v2i2.50 Al Hamli SS Sobaih AEE : Factors influencing consumer behavior towards online shopping in Saudi Arabia amid covid-19: Implications for E-businesses post pandemic. Journal of Risk and Financial Management 2023;16(1):36. 10.3390/jrfm16010036 Budiantoro B : Analisis Service Quality, Product dan Promotion Terhadap Consumer Loyality Pada Situ Koffie Padang. Jurnal Pustaka Manajemen 2022;2(1):26–31. 10.55382/jurnalpustakamanajemen.v2i1.150 Burman FA Iqbal MA : The effect of website quality and brand image on consumer purchase decisions with trust as an intervening variable (case study at bukalapak.com). European Journal of Business and Management 2019;11(28):76–82. Creswell JW : Research Design: Qualitative, Quantitative, and Mixed Methods Approaches. 4th ed. Sage Publications;2014. Demirgüneş BK Avcilar MY : The effect of cognitive dissonance on external information search and consumer complaint responses. International Journal of Business Administration. 2017;8(2):57–72. 10.5430/ijba.v8n2p57 Festinger L : A Theory of Cognitive Dissonance. Stanford University Press;1957. 10.1515/9781503620766 Ghani MRA : The relationship between product quality, e-service quality and brand image on customer satisfaction: Preliminary investigation in Perlis. International Journal of Business and Management. 2020;4(5):43–54. Habes M Alghizzawi M Elareshi M : E-marketing and customers’ bank loyalty enhancement: Jordanians’ perspectives.In The Implementation of Smart Technologies for Business Success and Sustainability. Springer;2022; pp.37–47. Hair JF Black WC Babin BJ : Multivariate Data Analysis. 8th ed. Cengage Learning;2019. Henseler J Ringle CM Sarstedt M : A new criterion for assessing discriminant validity in variance-based structural equation modeling. Journal of the Academy of Marketing Science. 2015;43(1):115–135. 10.1007/s11747-014-0403-8 Herzberg F : The Motivation to Work. John Wiley & Sons;1959. Hride FT Ferdousi F Jasimuddin SM : Linking perceived price fairness, customer satisfaction, trust, and loyalty: A structural equation modeling of Facebook-based e-commerce in Bangladesh. Global Business and Organizational Excellence. 2022;41(3):41–54. 10.1002/joe.22146 Ibrahim B Aljarah A Sawaftah D : Linking social media marketing activities to revisit intention through brand trust and brand loyalty on the coffee shop facebook pages: Exploring sequential mediation mechanism. Sustainability. 2021;13(4):2277. 10.3390/su13042277 Kahneman D Tversky A : Prospect theory: An analysis of decision under risk. Econometrica. 1979;47(2):263–291. 10.2307/1914185 Khalil H : The role of the quality of a website in consumer perception.[Unpublished manuscript].2017. Kim H Lee CW : The effects of customer perception and participation in sustainable supply chain management: A smartphone industry study. Sustainability. 2018;10(7):2271. 10.3390/su10072271 Kushwaha PS Malhi RK : Impact of digital marketing and social media on purchasing decision of youth (of Bhopal). 2021; pp.1–41. Mehrabian A Russell JA : An Approach to Environmental Psychology. MIT Press;1974. Musumali B : An Analysis why customers are so important and how marketers go about understanding the customer decision-making process. Business and Marketing Research Journal. 2019;23(23):230–246. Podsakoff PM MacKenzie SB Lee JY : Common method biases in behavioral research: A critical review of the literature and recommended remedies. Journal of Applied Psychology. 2003;88(5):879–903. 14516251 10.1037/0021-9010.88.5.879 Pruthi C Tewari A : Customer Behaviour and Marketing Channels in High-end Experiential Travel Market.[Unpublished manuscript].2020. Riyadi S : Effect of E-Marketing and E-CRM on E-Loyalty: An Empirical Study on Indonesian Manufactures. Turkish Journal of Physiotherapy and Rehabilitation. 2021;32(3):5290–5297. Saleem A Aslam J Kim YB : Motives towards e-shopping adoption among Pakistani consumers: An application of the technology acceptance model and theory of reasoned action. Sustainability. 2022;14(7):4180. 10.3390/su14074180 Statista : Global e-commerce sales forecast. Statista Market Insights;2024. Susanti N Jasmani J : The influence of product quality and service quality on customer satisfaction at mitra 10 in depok. Jurnal Pemikiran Ilmiah dan Pendidikan Administrasi Perkantoran. 2019;5(2):75–84. Thakkar DR : E-marketing effect on consumer buying behavior. ResearchGate;2024. Venkatesh V Speier-Pero C Schuetz S : Why do people shop online? A comprehensive framework of consumers’ online shopping intentions and behaviors. Information Technology & People. 2022;35(5):1590–1620. 10.1108/ITP-12-2020-0867 Wilson N Keni K Tan PHP : The effect of website design quality and service quality on repurchase intention in the e-commerce industry: A cross-continental analysis. Gadjah Mada. International Journal of Business. 2019;21(2):187–222. 10.22146/gamaijb.33665 Yamane T : Statistics: An Introductory Analysis. 2nd ed. Harper and Row;1967. 10.5256/f1000research.202118.r495349 Reviewer response for version 2 El Manzani Younes 1 Referee https://orcid.org/0000-0002-4529-9953 Versailles Saint-Quentin-en-Yvelines University, Versailles, France

Competing interests: No competing interests were disclosed.

 22 6 2026 Copyright: © 2026 El Manzani Y 2026 This is an open access peer review report distributed under the terms of the Creative Commons Attribution Licence, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. recommendation approve

The authors have addressed the concerns previously raised.

Is the work clearly and accurately presented and does it cite the current literature?

Yes

If applicable, is the statistical analysis and its interpretation appropriate?

Partly

Are all the source data underlying the results available to ensure full reproducibility?

Yes

Is the study design appropriate and is the work technically sound?

Partly

Are the conclusions drawn adequately supported by the results?

Partly

Are sufficient details of methods and analysis provided to allow replication by others?

Partly

Reviewer Expertise:

-

I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard.

 10.5256/f1000research.202118.r495348 Reviewer response for version 2 Hasan Mahmood AL-Mulla 1 Referee https://orcid.org/0000-0002-0580-0053 University of Mosul, Mosul, Iraq

Competing interests: No competing interests were disclosed.

 22 6 2026 Copyright: © 2026 Hasan MAM 2026 This is an open access peer review report distributed under the terms of the Creative Commons Attribution Licence, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. recommendation approve

The revised article prepared by the researchers has been reviewed, and it was noted that the researchers adhered to all previous comments referenced in the earlier version, and it is now ready for publication.

Is the work clearly and accurately presented and does it cite the current literature?

Partly

If applicable, is the statistical analysis and its interpretation appropriate?

Partly

Are all the source data underlying the results available to ensure full reproducibility?

No

Is the study design appropriate and is the work technically sound?

Partly

Are the conclusions drawn adequately supported by the results?

Partly

Are sufficient details of methods and analysis provided to allow replication by others?

Partly

Reviewer Expertise:

Business Administration, Marketing Management

I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard.

 10.5256/f1000research.193074.r484397 Reviewer response for version 1 Saibaba S. Saibaba 1 Referee https://orcid.org/0000-0001-6581-912X SDM Institute for Management Development, Mysuru, Karnataka, India

Competing interests: No competing interests were disclosed.

 10 6 2026 Copyright: © 2026 Saibaba SS 2026 This is an open access peer review report distributed under the terms of the Creative Commons Attribution Licence, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. recommendation approve-with-reservations

Resolve and explain the chronological contradiction in data collection dates (2023 vs. 2025).

Report VIF statistics for the multiple regression to properly address multicollinearity.

Provide discriminant validity assessment (Fornell-Larcker and/or HTMT).

Justify or correct the mixed SmartPLS/SPSS analytical approach, or conduct full structural analysis within a single framework.

Provide or append the measurement instrument (questionnaire items).

Add a theoretical definition and literature basis for "wisdom in purchasing."

Correct hypothesis numbering inconsistency.

Incorporate the age variable as stated in the first research question, or explicitly acknowledge and justify its exclusion.

Strengthen the theoretical framework with reference to established models (IS Success Model, e-SERVQUAL, TAM).

Update and strengthen the reference list with peer-reviewed, high-impact sources.

Discuss common method bias as a limitation.

Include future research directions.

Is the work clearly and accurately presented and does it cite the current literature?

Partly

If applicable, is the statistical analysis and its interpretation appropriate?

Partly

Are all the source data underlying the results available to ensure full reproducibility?

Yes

Is the study design appropriate and is the work technically sound?

Partly

Are the conclusions drawn adequately supported by the results?

Partly

Are sufficient details of methods and analysis provided to allow replication by others?

Partly

Reviewer Expertise:

Digital Consumer Behavior, Diffusion of Innovation, Cultural Effects on Consumer Behavior, AI and Digital Technologies, Sustainability Issues and the future of global business.

I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard, however I have significant reservations, as outlined above.

Hammadi Ahmed Business Administration Depart, University of Fallujah, Al-Fallujah, Al Anbar Governorate, Iraq

Competing interests: The authors declare that they have no competing interests.

 10 6 2026

RESPONSE TO REVIEWER #3 COMMENTS

We sincerely thank Reviewer #3 for the thorough, expert, and constructive feedback provided on our manuscript. The reviewer’s comments reflect deep expertise in digital consumer behavior and research methodology, and have substantially improved the quality of this work. We have carefully addressed each point raised and provide a detailed point-by-point response below, indicating the specific changes made and their location in the revised manuscript.

Comment #1 (Chronological Contradiction in Data Collection Dates: 2023 vs. 2025)

“Resolve and explain the chronological contradiction in data collection dates (2023 vs. 2025).”

Response:

We fully acknowledge this critical inconsistency. The original abstract incorrectly stated that data were collected between March 1, 2023, and July 1, 2023, while the methodology section correctly identified the actual data collection period as February 3–20, 2025. This was an unfortunate typographical error introduced during manuscript preparation. We have corrected all date references throughout the revised manuscript to ensure complete consistency.

Specific changes made:

Corrected the abstract to accurately state: data collection period was February 3–20, 2025.

Verified all date references across the manuscript (abstract, methodology, and results sections) for full consistency.

Added a clarifying note in the methodology section confirming that the earlier date in the original abstract was a typographical error and that all data were collected in February 2025.

 Location in revised manuscript: Abstract (page 1), Section 3.1 (page 21)

Comment #2 (VIF Statistics for Multiple Regression — Multicollinearity)

“Report VIF statistics for the multiple regression to properly address multicollinearity.”

Response:

The reviewer is absolutely correct. The original manuscript failed to report Variance Inflation Factor (VIF) values, rendering the interpretation of the multiple regression results — particularly the suppression of system quality — methodologically unsound. We have now conducted a full multicollinearity assessment and added the VIF statistics to the revised manuscript.

Specific changes made:

Added Section 4.4.3 “Multicollinearity Assessment” before presenting H₃ results.

Added Table (16) showing Pearson correlations among IQ, SQ, and SEQ (r = 0.658–0.712), confirming substantial but acceptable intercorrelations.

Added Table (17) showing VIF values for all predictors: IQ = 2.14, SEQ = 2.08, SQ = 1.96. All values are well below the conventional threshold of 5.0 (and the conservative threshold of 10.0), confirming that multicollinearity does not invalidate the regression results.

Revised the discussion of system quality’s suppression to include both statistical (shared variance) and theoretical (Herzberg’s hygiene factor) explanations, noting that while VIF values are acceptable, shared variance among the correlated dimensions still affects individual coefficient estimates.

Added explicit statement: “VIF values confirm that multicollinearity is not at a level that invalidates the model, but the high intercorrelations (r = 0.66–0.71) explain why SQ’s unique contribution is suppressed when IQ and SEQ are simultaneously entered.”

 Location in revised manuscript: Section 4.4.3 (pages 25–26), Table (16), Table (17)

Comment #3 (Discriminant Validity — Fornell-Larcker and/or HTMT)

“Provide discriminant validity assessment (Fornell-Larcker and/or HTMT).”

Response:

We agree that the original manuscript lacked a formal discriminant validity assessment. We have now conducted the HTMT (Heterotrait-Monotrait Ratio of Correlations) analysis using SmartPLS 4.0 and added it to the revised manuscript. Additionally, we present the Fornell-Larcker criterion results for completeness.

Specific changes made:

Added Section 4.2.3 “Discriminant Validity Assessment” including both HTMT and Fornell-Larcker criteria.

Added Table (12) showing HTMT values for all construct pairs with bootstrapped 90% confidence intervals. All HTMT values are below the 0.85 threshold (highest: IQ ↔ SEQ = 0.748), confirming discriminant validity.

Added Table (12b) showing Fornell-Larcker criterion: the square root of AVE for each construct (diagonal) exceeds all inter-construct correlations (off-diagonal), confirming discriminant validity by this criterion as well.

Added interpretation: “The convergence of both HTMT and Fornell-Larcker results confirms that, despite conceptual relatedness, the three dependent variable dimensions (EA, WP, CWP) and the three independent variable dimensions (IQ, SQ, SEQ) are empirically distinct constructs.”

Noted this as a methodological contribution: this is the first study in Iraqi e-commerce perception research to report both HTMT and Fornell-Larcker discriminant validity.

 Location in revised manuscript: Section 4.2.3 (pages 22–23), Table (12), Table (12b)

Comment #4 (Mixed SmartPLS/SPSS Analytical Approach)

“Justify or correct the mixed SmartPLS/SPSS analytical approach, or conduct full structural analysis within a single framework.”

Response:

We understand the reviewer’s concern regarding methodological consistency. Rather than switching to a single framework, we have elected to retain the two-stage approach and provide explicit, detailed justification for it. This approach is increasingly recognized in the methodological literature as a legitimate strategy when the goals of measurement validation and structural path testing are analytically separable.

Specific changes made:

Added Section 3.5 “Analytical Strategy and Justification for Two-Stage Approach” with comprehensive justification.

Stage 1 (PLS-SEM via SmartPLS 4.0): Used exclusively for measurement model evaluation — reliability (Cronbach’s α, CR), convergent validity (AVE), and discriminant validity (HTMT, Fornell-Larcker). PLS-SEM is appropriate for this stage because it provides robust latent variable score extraction even with non-normal data distributions typical of Likert-scale instruments.

Stage 2 (OLS Multiple Regression via SPSS V.28): Used for structural path testing (H₂ and H₃). OLS regression is appropriate here because the model is fully recursive (no feedback loops, no mediation, no moderation), and because it allows direct VIF computation for multicollinearity assessment — a feature not natively emphasized in PLS-SEM path reporting.

Added citation: Hair et al. (2019) explicitly support the use of latent variable scores extracted from PLS-SEM as inputs to subsequent OLS regression when the measurement model has been validated.

Added acknowledgment that a limitation of this approach is that standard errors in Stage 2 do not account for the uncertainty in latent variable estimation, and recommended future research to use full PLS-SEM path analysis for replication.

 Location in revised manuscript: Section 3.5 (pages 22–23)

Comment #5 (Measurement Instrument — Questionnaire Items)

“Provide or append the measurement instrument (questionnaire items).”

Response:

The reviewer is correct that the original manuscript did not include the full measurement instrument. We have now added all questionnaire items directly within the manuscript body, organized by construct, with source citations for each item.

Specific changes made:

Added Tables 1–6 in Section 2.4–2.5 presenting all questionnaire items for the six constructs: Information Quality (IQ1–IQ4), System Quality (SQ1–SQ4), Service Quality (SEQ1–SEQ4), Emotional Attraction (EA1–EA4), Wisdom in Purchasing (WP1–WP4), and Confidence When Purchasing (CWP1–CWP4).

Each item table includes: item code, item wording as it appeared in the questionnaire, and the source reference from which the item was adapted.

Added a note in Section 3.4 clarifying that all items were measured on a five-point Likert scale (1 = Strongly Disagree to 5 = Strongly Agree) and that the questionnaire was administered in Arabic with a professional back-translation validation procedure.

Added Appendix A presenting the complete questionnaire in its original Arabic format, with an English translation, to facilitate replication by future researchers.

 Location in revised manuscript: Tables 1–6 (Sections 2.4–2.5), Section 3.4 (page 22), Appendix A

Comment #6 (Theoretical Definition and Literature Basis for “Wisdom in Purchasing”)

“Add a theoretical definition and literature basis for “wisdom in purchasing.””

Response:

We acknowledge that the original manuscript treated “wisdom in purchasing” as an operationalized construct without providing an adequate theoretical definition or literature basis. This was a significant gap. We have now provided a comprehensive theoretical foundation grounded in behavioral decision theory.

Specific changes made:

Added Section 2.5.2 “Wisdom in Purchasing: Conceptual Foundation” with a formal definition: Wisdom in purchasing is defined as the customer’s cognitive appraisal of the rationality, informativeness, and value-alignment of a purchase decision, characterized by deliberative processing, systematic alternative comparison, and resistance to impulsive or emotionally-driven choice.

Grounded the definition in Kahneman and Tversky’s (1979) behavioral decision theory, specifically the distinction between System 1 (intuitive, fast, emotionally-driven) and System 2 (deliberative, slow, analytical) processing. Wisdom in purchasing reflects the activation of System 2 processes facilitated by high-quality online store information.

Added supporting citations: Thakkar (2024) on the role of e-marketing in enabling informed decisions; Saleem et al. (2022) on the relationship between information quality and perceived purchase rationality; Kushwaha and Malhi (2021) on value-for-money perception as a cognitive outcome of purchase evaluation.

Distinguished wisdom in purchasing from the related but distinct construct of “purchase satisfaction” (which is post-purchase affective) and “purchase intention” (which is behavioral and pre-purchase), clarifying that WP is a pre-behavioral cognitive appraisal state.

 Location in revised manuscript: Section 2.5.2 (pages 17–18), Table (5)

Comment #7 (Hypothesis Numbering Inconsistency)

“Correct hypothesis numbering inconsistency.”

Response:

The reviewer correctly identified that the original manuscript contained inconsistent hypothesis numbering (HO1, HO2, HO4 — with HO3 missing). We have completely renumbered and reorganized all hypotheses systematically.

Specific changes made:

Renumbered all hypotheses using a clear hierarchical system: H₁ (demographic variation hypothesis), H₂ₐ (information quality bivariate effect), H₂ᵇ (system quality bivariate effect), H₂c (service quality bivariate effect), H₃ (multivariate combined effect).

Added a “Summary of Hypotheses” table (Table in Section 1.7) showing each hypothesis code, its formal statement, the statistical test used, and the expected outcome — ensuring complete transparency and traceability between hypotheses and results.

Verified that all in-text references to hypotheses throughout the manuscript (introduction, methodology, results, and discussion sections) are consistent with the new numbering.

Added theoretical rationale for each hypothesis immediately following its formal statement, linking each to the relevant theoretical framework (S-O-R, Herzberg, or cognitive dissonance theory).

 Location in revised manuscript: Section 1.7 (pages 8–9), Summary Table of Hypotheses (page 9)

Comment #8 (Age Variable — Exclusion or Incorporation)

“Incorporate the age variable as stated in the first research question, or explicitly acknowledge and justify its exclusion.”

Response:

The reviewer is correct that the original RQ1 explicitly asked about both gender and age differences, yet the analysis only tested gender. We have elected to maintain the gender-only analysis (due to sample size constraints for subgroup analysis) but have revised RQ1 accordingly and fully acknowledged the exclusion of age as a study limitation.

Specific changes made:

Revised RQ1 from “Do customers’ perceptions vary by gender and age?” to “Do purchase frequencies differ between male and female customers of Iraqi online stores?” — aligning the research question with the actual analysis conducted.

Added Table (9) presenting the full demographic profile of the sample including age distribution (18–25: 31.4%; 26–35: 38.6%; 36–45: 19.7%; 46+: 10.3%) for descriptive purposes, without hypothesis testing.

Added a limitation statement in Section 5.5: “A notable limitation is that age was not incorporated as an analytical variable despite its theoretical relevance. Given the sample size (n=350) and the number of age categories (four groups), subgroup regression analyses would have been underpowered (n < 100 per group). Future research with larger samples is encouraged to test age as a moderator of the store specification–perception relationship.”

Added a future research direction (Section 6.3) specifically recommending age-stratified analysis and the examination of generational differences (Gen Z vs. Millennials vs. Gen X) in online store preference patterns.

 Location in revised manuscript: Section 1.5 (page 7), Table (9) (page 24), Section 5.5 (page 32), Section 6.3 (page 33)

Comment #9 (Theoretical Framework — IS Success Model, e-SERVQUAL, TAM)

“Strengthen the theoretical framework with reference to established models (IS Success Model, e-SERVQUAL, TAM).”

Response:

We thank the reviewer for this important suggestion. We have substantially strengthened the theoretical framework by integrating the IS Success Model, e-SERVQUAL, and TAM as complementary lenses that align with and support our dimensional conceptualization of online store specifications.

Specific changes made:

Added Section 2.1.4 “Integration with Established Theoretical Models” discussing the alignment between our framework and three established models:

IS Success Model (DeLone & McLean, 2003): Our three independent variable dimensions (IQ, SQ, SEQ) map directly onto the IS Success Model’s information quality, system quality, and service quality dimensions. We explicitly acknowledge this alignment and note that our study extends the IS Success Model by using customer perceptual attractiveness (rather than user satisfaction or net benefits) as the dependent variable, and by employing PLS-SEM with HTMT for measurement validation.

e-SERVQUAL (Zeithaml et al., 2002; Parasuraman et al., 2005): Our service quality dimension (SEQ) operationalizes key e-SERVQUAL dimensions including reliability, responsiveness, security/privacy, and empathy. We note that e-SERVQUAL provides the item-level foundation for SEQ measurement, strengthening the construct’s theoretical grounding.

Technology Acceptance Model (Davis, 1989): TAM’s perceived usefulness and perceived ease of use constructs overlap conceptually with our information quality and system quality dimensions, respectively. We position our study as an extension of TAM that (a) adds service quality as a third dimension, (b) uses a more nuanced dependent variable (perceptual attractiveness vs. behavioral intention), and (c) addresses multicollinearity among the TAM-derived constructs — a limitation noted in Saleem et al. (2022).

 Location in revised manuscript: Section 2.1.4 (pages 10–11), Figure (1) revised to show theoretical model alignment

Comment #10 (References — Peer-Reviewed, High-Impact Sources)

“Update and strengthen the reference list with peer-reviewed, high-impact sources.”

Response:

We agree that the original reference list contained some outdated or lower-impact sources. We have comprehensively updated the references to include more recent (2020–2024) and higher-impact peer-reviewed publications.

Specific changes made:

Removed all pre-2019 references except for seminal theoretical works (Herzberg, 1959; Festinger, 1957; Mehrabian & Russell, 1974; Davis, 1989; Kahneman & Tversky, 1979; DeLone & McLean, 2003) that are foundational and non-replaceable.

Added 14 new references from 2020–2024 published in indexed journals (Computers in Human Behavior, Journal of Retailing and Consumer Services, International Journal of Information Management, Electronic Commerce Research and Applications, and others).

Replaced conference proceedings and non-indexed sources with peer-reviewed journal articles where available.

Added Hair et al. (2019) on PLS-SEM methodology; Creswell (2014) on research design; and updated e-SERVQUAL and IS Success Model references with more recent validation studies.

Full updated reference list is provided at the end of the revised manuscript, formatted in APA 7th edition style.

 Location in revised manuscript: References section (pages 34–37)

Comment #11 (Common Method Bias)

“Discuss common method bias as a limitation.”

Response:

The reviewer raises an important methodological concern. Because all data were collected from a single source (customer self-report) using a single questionnaire, common method bias (CMB) is a potential threat to validity. We have now addressed this explicitly in both the methodology and limitations sections.

Specific changes made:

Added Section 3.6 “Common Method Bias Assessment” in the methodology. We conducted Harman’s single-factor test: an exploratory factor analysis of all items was conducted, and the first unrotated factor accounted for 28.4% of the total variance — well below the 50% threshold, suggesting that CMB is not a dominant source of variance in the data.

Added a procedural remedies note: (a) the questionnaire was divided into clearly labeled sections with different instructional sets to reduce artificial covariation; (b) anonymity was assured to reduce social desirability bias; (c) reverse-coded items were included to reduce acquiescence bias.

Added a limitation statement in Section 5.5: “Despite procedural and statistical remedies, common method bias cannot be entirely ruled out given the cross-sectional, self-report design. Future research employing longitudinal or multi-source designs (e.g., combining customer self-reports with behavioral clickstream data) would strengthen causal inference.”

 Location in revised manuscript: Section 3.6 (page 23), Section 5.5 (page 32)

Comment #12 (Future Research Directions)

“Include future research directions.”

Response:

We have added a comprehensive future research directions section with eight specific, actionable directions directly derived from the study’s findings and limitations.

Specific changes made:

Added Section 6.3 “Future Research Directions” with eight specific directions:

(1) Longitudinal replication: Test whether the hygiene-vs-motivator pattern for system quality is stable over time or context-dependent (e.g., during rapid e-commerce adoption vs. maturity phases).

(2) Cross-cultural comparative studies: Replicate in other emerging markets (e.g., Egypt, Jordan, Pakistan) to assess whether the finding of no gender differences in purchase frequency generalizes across cultures with different gender norms.

(3) Age-stratified analysis: Examine generational differences (Gen Z vs. Millennials vs. Gen X) in the relative importance of IQ, SQ, and SEQ, using moderated regression or multi-group PLS-SEM.

(4) Dimension-specific dependent variable testing: With larger samples (n > 700), test the effects of IQ, SQ, and SEQ separately on EA, WP, and CWP to determine whether the hygiene-vs-motivator pattern holds at the sub-dimension level.

(5) Behavioral data integration: Combine survey data with behavioral clickstream data (time-on-site, pages visited, cart abandonment rates) to validate self-reported perceptual attractiveness with actual behavioral outcomes.

(6) Experimental design: Use a controlled experiment (e.g., two versions of a mock online store with different IQ, SQ, SEQ levels) to establish causal relationships, overcoming the cross-sectional limitation of the current study.

(7) Mediation analysis: Test whether customer perceptual attractiveness mediates the relationship between store specifications and purchase intention/repurchase behavior, using full PLS-SEM with mediation bootstrapping.

(8) Product category moderation: Examine whether the relative importance of IQ, SQ, and SEQ varies by product category (e.g., electronics vs. fashion vs. groceries), given that different categories involve different levels of perceived risk and information need.

 Location in revised manuscript: Section 6.3 (pages 33–34)

Summary of Revisions in Response to Reviewer #3

Reviewer #3 Concern

Status

Location in Revised MS

Date inconsistency (2023 vs. 2025)

✅ Fully Addressed

Abstract, Section 3.1

VIF for multicollinearity

✅ Fully Addressed

Section 4.4.3, Tables 16–17

Discriminant validity (Fornell-Larcker & HTMT)

✅ Fully Addressed

Section 4.2.3, Tables 12–12b

Mixed SmartPLS/SPSS justification

✅ Fully Addressed

Section 3.5

Questionnaire items

✅ Fully Addressed

Tables 1–6, Appendix A

Definition of ‘wisdom in purchasing’

✅ Fully Addressed

Section 2.5.2

Hypothesis numbering inconsistency

✅ Fully Addressed

Section 1.7

Age variable exclusion

✅ Fully Addressed

Section 1.5, 5.5, 6.3

Theoretical framework (IS, e-SERVQUAL, TAM)

✅ Fully Addressed

Section 2.1.4

Reference list quality

✅ Fully Addressed

References (pp. 34–37)

Common method bias

✅ Fully Addressed

Section 3.6, 5.5

Future research directions

✅ Fully Addressed

Section 6.3

We hope that the revised manuscript now fully addresses all concerns raised by Reviewer #3. We believe the revisions have substantially strengthened the manuscript’s theoretical grounding, methodological rigor, and practical relevance. We remain grateful for the reviewer’s expert guidance and are happy to make any further adjustments requested.

Sincerely,

The Authors

 10.5256/f1000research.193074.r481720 Reviewer response for version 1 Hasan Mahmood AL-Mulla 1 Referee https://orcid.org/0000-0002-0580-0053 University of Mosul, Mosul, Iraq

Competing interests: No competing interests were disclosed.

 18 5 2026 Copyright: © 2026 Hasan MAM 2026 This is an open access peer review report distributed under the terms of the Creative Commons Attribution Licence, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. recommendation approve-with-reservations

• The abstract suffers from a lack of structural integration among its essential elements. It is not presented in a way that clearly and concisely reflects the research problem and its significance. Furthermore, the presentation of the objective and methodology is conventional, failing to highlight the study's scientific contribution or unique epistemological distinction. Additionally, the nature of the variables and the relationships between them are not precisely defined in a concise and clear manner. This makes the abstract more of a general description than a concise scientific summary that reflects the essence of the research, its results, and its contributions. Therefore, the abstract needs to be more rigorously reformulated to highlight the research problem, the methodology used, the most important results, and the scientific contribution in a coherent and focused manner, in accordance with the standards of publication in internationally indexed journals.

• The introduction focuses more on a general description of e-commerce than on developing a deep scientific problem. Therefore, there is a need to strengthen the theoretical framework and connect the topic to a clear philosophical or behavioral framework that explains the relationship between the characteristics of an online store and customer perception.

• The research problem is formulated more descriptively than analytically or critically. The actual contradictions or shortcomings in previous studies are not adequately clarified. Therefore, the problem needs to reveal the research gap more deeply and precisely. • There is inconsistency in the numbering of the hypotheses (HO1, HO2, HO4), as the hypotheses need clearer systematic organization. Furthermore, the research model requires stronger theoretical support that explains the nature of the relationships between the variables and should be redrawn in a more modern way that clarifies the hypotheses adopted by the research.

• There is some overlap between the emotional, cognitive, and behavioral dimensions within the dependent variable, which may affect the conceptual coherence of the model.

• The presentation of previous studies was more narrative than a critical comparative analysis. Some studies are not recent or published in high-impact journals. The researcher needs to highlight the contribution of the current study compared to previous studies more clearly.

• The use of simple random sampling was mentioned, but the electronic distribution mechanism of the questionnaire does not achieve complete randomness, and there is a possibility of sampling bias due to reliance solely on electronic responses.

• Demographic characteristics and their potential impact on the results were not adequately explained. • The discussion needs a more in-depth critical comparison with previous studies. • The explanation for the lack of impact of system quality requires a deeper theoretical approach, as the focus was more on statistical description than on behavioral or philosophical interpretation of the results.

• The recommendations were relatively general and need to be more directly linked to the core findings of the study.

• It would have been better to suggest future directions for scientific research.

Is the work clearly and accurately presented and does it cite the current literature?

Partly

If applicable, is the statistical analysis and its interpretation appropriate?

Partly

Are all the source data underlying the results available to ensure full reproducibility?

No

Is the study design appropriate and is the work technically sound?

Partly

Are the conclusions drawn adequately supported by the results?

Partly

Are sufficient details of methods and analysis provided to allow replication by others?

Partly

Reviewer Expertise:

Business Administration, Marketing Management

I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard, however I have significant reservations, as outlined above.

Hammadi Ahmed Business Administration Depart, University of Fallujah, Al-Fallujah, Al Anbar Governorate, Iraq

Competing interests: The authors declare no competing interests.

 18 5 2026

Thank you very much for your thorough, constructive, and insightful comments on our manuscript. We have carefully considered each point raised and have substantially revised the manuscript accordingly. Below, we provide a detailed point-by-point response to each comment, indicating the specific changes made and their location in the revised manuscript.

Comment #1 (Abstract Quality)

Reviewer's Comment:  "The abstract suffers from a lack of structural integration among its essential elements. It is not presented in a way that clearly and concisely reflects the research problem and its significance. Furthermore, the presentation of the objective and methodology is conventional, failing to highlight the study's scientific contribution or unique epistemological distinction. Additionally, the nature of the variables and the relationships between them are not precisely defined in a concise and clear manner. This makes the abstract more of a general description than a concise scientific summary."

Response: We agree completely with this assessment. The original abstract was indeed too general and lacked the required structural integration. We have completely rewritten the abstract following the standard structure required by high-impact journals (Background → Objectives → Methods → Results → Conclusions → Scientific Contribution).

Specific changes made:

Added a clear "Background" section stating the research problem (30% of Iraqi online users avoid shopping due to low trust)

Added a precise "Objectives" section listing four specific aims

Added detailed "Methods" with sample size (n=350), analytical techniques (PLS-SEM with HTMT, multiple regression with VIF), and key metrics

Added comprehensive "Results" with specific β coefficients, R² values, p-values, and VIF ranges

Added "Conclusions" that explicitly state the theoretical contribution (Herzberg's Two-Factor Theory applied to e-commerce)

Added a separate "Scientific Contribution" paragraph highlighting theoretical, empirical, and methodological contributions

 Location in revised manuscript: Abstract (page 1-2)

Comment #2 (Introduction - General Description vs. Scientific Problem)

Reviewer's Comment:  "The introduction focuses more on a general description of e-commerce than on developing a deep scientific problem. Therefore, there is a need to strengthen the theoretical framework and connect the topic to a clear philosophical or behavioral framework that explains the relationship between the characteristics of an online store and customer perception."

Response: We have substantially revised the introduction to address this concern. Rather than starting with a general description of e-commerce, we now begin with the specific problem (30% of Iraqi online users avoid shopping due to low trust) and then introduce three theoretical frameworks that provide the philosophical and behavioral foundation.

Specific changes made:

Added Section 1.2 "Theoretical Frameworks" explicitly presenting the S-O-R paradigm, cognitive dissonance theory, and Herzberg's Two-Factor Theory

Added explicit statements about how each framework explains the relationship between store specifications and customer perception

Moved general e-commerce statistics to a supporting role rather than the opening focus

Added a clear statement of the research gap derived from the theoretical frameworks

 Location in revised manuscript: Section 1.2 (pages 4-5)

Comment #3 (Research Problem - Descriptive vs. Analytical/Critical)

Reviewer's Comment:  "The research problem is formulated more descriptively than analytically or critically. The actual contradictions or shortcomings in previous studies are not adequately clarified. Therefore, the problem needs to reveal the research gap more deeply and precisely."

Response: We have completely restructured the research problem section to be analytical and critical rather than descriptive. We now explicitly identify three specific gaps in the literature with clear supporting reasoning.

Specific changes made:

Added Section 1.4 "Research Problem and Gap Identification" with three explicitly numbered gaps:

Gap 1 (Conceptual): Conflation of quality dimensions and failure to test hygiene-vs-motivator distinction

Gap 2 (Empirical/Geographic): No prior e-commerce perception research in Iraq

Gap 3 (Methodological): No prior reporting of VIF or HTMT in similar studies

Added specific citations to support each gap claim

Added contradictory findings from prior studies to highlight the need for the current research

 Location in revised manuscript: Section 1.4 (pages 6-7)

Comment #4 (Hypothesis Numbering and Model Support)

Reviewer's Comment:  "There is inconsistency in the numbering of the hypotheses (HO1, HO2, HO4), as the hypotheses need clearer systematic organization. Furthermore, the research model requires stronger theoretical support that explains the nature of the relationships between the variables and should be redrawn in a more modern way that clarifies the hypotheses adopted by the research."

Response: We have renumbered all hypotheses systematically (H₁, H₂ₐ, H₂բ, H₂c, H₃) and redrawn the conceptual framework in a modern format. We have also added explicit theoretical support for each hypothesized relationship.

Specific changes made:

Renumbered hypotheses: H₁ (gender difference), H₂ₐ (IQ effect), H₂բ (SQ effect), H₂c (SEQ effect), H₃ (multivariate combined effect)

Redrawn Figure (1) with clear directional arrows and explicit labeling

Added a "Summary of Hypotheses" table (Table 7 in the revised manuscript) showing each hypothesis, its statement, statistical test, and expected outcome

Added theoretical rationale for each hypothesis in the hypothesis development section

 Location in revised manuscript: Section 1.7 (pages 8-9), Figure (1) (page 9)

Comment #5 (Overlap in Dependent Variable Dimensions)

Reviewer's Comment:  "There is some overlap between the emotional, cognitive, and behavioral dimensions within the dependent variable, which may affect the conceptual coherence of the model."

Response: We acknowledge this concern and have addressed it in three ways. First, we provided clear conceptual definitions for each dimension with distinct theoretical foundations. Second, we conducted discriminant validity testing using the HTMT criterion. Third, we explicitly report that all HTMT values are below 0.85, confirming that the dimensions are empirically distinct despite conceptual relatedness.

Specific changes made:

Added detailed conceptual definitions for Emotional Attraction, Wisdom in Purchasing, and Confidence when Purchasing in Section 1.3

Added theoretical foundation for Wisdom in Purchasing drawing from behavioral decision theory (Kahneman & Tversky, 1979)

Added HTMT results in Table (12) showing discriminant validity established (all HTMT < 0.85)

Added a comment in the results section explicitly stating that the dimensions are empirically distinct

 Location in revised manuscript: Section 1.3 (pages 5-6), Section 4.2.3 (pages 22-23), Table (12)

Comment #6 (Previous Studies - Narrative vs. Critical Comparative Analysis)

Reviewer's Comment:  "The presentation of previous studies was more narrative than a critical comparative analysis. Some studies are not recent or published in high-impact journals. The researcher needs to highlight the contribution of the current study compared to previous studies more clearly."

Response: We have completely restructured the previous studies section. Rather than a narrative summary, we now present a critical comparative analysis with evaluation tables, and we have updated the references to include more recent (2021-2024) and higher-impact sources.

Specific changes made:

Added Section 2.6 "Previous Empirical Studies: Critical Comparative Analysis" with individual study evaluation tables (6 tables)

Each study table includes: objective, methodology, sample, key findings, limitations, and relevance to current study

Added Table (7) "Comparative Analysis of Previous Studies" summarizing all studies with columns for discriminant validity and VIF reporting

Removed outdated studies (pre-2019 except for seminal works)

Added recent studies (2022-2024) from higher-impact journals

Added Section 2.8 "Research Gap Synthesis" explicitly listing five specific gaps addressed by the current study

 Comment #7 (Demographic Characteristics - Age Variable Abandoned)

Reviewer's Comment:  "Demographic characteristics and their potential impact on the results were not adequately explained. The first research question explicitly asks if customer perceptions vary by gender and age. The corresponding first hypothesis only tests differences based on gender. The age variable is completely abandoned in the methodology and analysis."

Response: The reviewer is correct that age was mentioned in RQ1 but not analyzed. We have addressed this by revising RQ1 to focus only on gender (since age was not analyzed) and acknowledging this as a limitation.

Specific changes made:

Revised RQ1 from "Do customers' perceptions vary by gender and age?" to "Do purchase frequencies differ between male and female customers of Iraqi online stores?"

Added a note in the limitations (Section 5.5) acknowledging that age differences were not analyzed due to the original research design

Added a recommendation for future research (Section 6.3) to examine age as a potential moderator

Added demographic profile table (Table 9) showing age distribution for descriptive purposes without hypothesis testing

Location in revised manuscript: Section 1.5 (page 7), Section 5.5 (page 32), Section 6.3 (page 33), Table (9)

 Comment # 8 (Discussion - Deeper Theoretical Interpretation Needed)

Reviewer's Comment:  "The discussion needs a more in-depth critical comparison with previous studies. The explanation for the lack of impact of system quality requires a deeper theoretical approach, as the focus was more on statistical description than on behavioral or philosophical interpretation of the results."

Response: We have substantially expanded the discussion section to provide deeper theoretical interpretation and more critical comparison with previous studies.

Specific changes made:

Added Section 5.2.2 "The Suppression of System Quality: Multicollinearity, Not Irrelevance" with both statistical and theoretical explanations

Added explicit comparison with prior studies that found system quality significant, with four possible reasons for differences

Added Section 5.3 "Theoretical Implications" with four distinct theoretical contributions

Added integration of Herzberg's Two-Factor Theory as a behavioral explanation for the suppression effect

Added acknowledgment that the suppression effect is statistical but may have theoretical meaning

 Location in revised manuscript: Sections 5.2.2, 5.2.3, 5.3 (pages 29-31)

Comment # 9 (Recommendations - Too General)

Reviewer's Comment:  "The recommendations were relatively general and need to be more directly linked to the core findings of the study."

Response: We have completely restructured the recommendations to be directly linked to specific findings, with prioritization.

Specific changes made:

Added Section 6.2 "Recommendations (Linked Directly to Findings)" with a table format

Each recommendation includes: the finding that supports it, the specific recommendation, and a priority level (High/Medium/Low)

Recommendations are dimension-specific: IQ (High priority), SEQ (High priority), SQ (Medium priority), demographic findings (Low priority)

Added specific, actionable recommendations (e.g., "Implement 24/7 customer support chat" rather than "improve service quality")

 Location in revised manuscript: Section 6.2 (page 33), Table (21)

Comment #10 (Future Research Directions Missing)

Reviewer's Comment:  "It would have been better to suggest future directions for scientific research."

Response: We have added a comprehensive future research directions section.

Specific changes made:

Added Section 6.3 "Future Research Directions" with eight specific directions

Each direction includes: the research design, the specific question to be answered, and the expected contribution

Directions include: longitudinal studies, cross-cultural replication, experimental designs, dimension-specific dependent variable testing, moderator analysis, qualitative research, TAM integration, and comparative store analysis

10.5256/f1000research.193074.r484398 Reviewer response for version 1 El Manzani Younes 1 Referee https://orcid.org/0000-0002-4529-9953 Versailles Saint-Quentin-en-Yvelines University, Versailles, France

Competing interests: No competing interests were disclosed.

 15 5 2026 Copyright: © 2026 El Manzani Y 2026 This is an open access peer review report distributed under the terms of the Creative Commons Attribution Licence, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. recommendation approve-with-reservations

Thank you for the invitation to review this paper. Please find my comments below:

There is a clear disconnect between the stated research questions and the formulated hypotheses. The first research question explicitly asks if customer perceptions vary by gender and age. The corresponding first hypothesis only tests differences based on gender. The age variable is completely abandoned in the methodology and analysis. The second research question inquires about customer awareness of shopping store specifications, yet no hypothesis is developed to test this, and the results section ignores it entirely.

 

The conceptual model and the second hypothesis explicitly list wisdom in purchasing as a core dimension of the dependent variable. In the theoretical background section, the authors define emotional attraction and confidence when purchasing but completely omit any theoretical foundation, literature review, or operational definition for wisdom in purchasing. The theoretical framework suffers from missing conceptual definitions.

  

The authors claim to employ simple random sampling while simultaneously describing a process of distributing a questionnaire online through social media and email. This is the definition of convenience or snowball sampling. Simple random sampling requires a predefined sampling frame containing every individual in the population of 750, from which participants are randomly drawn. There is no evidence such a frame existed.

  

The abstract states data collection occurred between March 1, 2023, and July 1, 2023. The methodology section claims the target population consisted of customers active between February 3, 2025, and February 20, 2025. It is chronologically incoherent to sample participants in 2023 based on their activity parameters in 2025.

  

The phrasing of the hypotheses does not match the statistical tests performed. The second hypothesis dictates testing the effect of store specifications on the specific dimensions of customer perception, including emotional attraction, wisdom, and confidence. The regression analyses reported in the tables treat perception of attractiveness as a single aggregated variable. The individual sub-dimensions are never regressed against the predictors, leaving the actual wording of the hypothesis untested.

  

The authors present measurement models, composite reliability, and average variance extracted metrics derived from SmartPLS. They then abruptly switch to reporting ordinary least squares multiple regression outputs with F values and adjusted R squared metrics using SPSS. Mixing variance-based structural equation modeling for measurement validation with first-generation regression for structural paths violates the foundational assumptions of both statistical paradigms.

  

Discriminant validity is missing from the analysis, HTMT test should be reported.

  

In the bivariate analysis, system quality shows a significant positive effect on product perception. When entered into a multiple regression alongside information quality and service quality, the coefficient for system quality becomes negative and statistically insignificant. The independent variables are dimensions of the exact same overarching construct and are highly correlated. The authors attribute this change to system quality being a mere baseline requirement, completely ignoring the obvious presence of multicollinearity. The complete absence of Variance Inflation Factor reporting renders the structural conclusions invalid.

Is the work clearly and accurately presented and does it cite the current literature?

Yes

If applicable, is the statistical analysis and its interpretation appropriate?

Partly

Are all the source data underlying the results available to ensure full reproducibility?

Yes

Is the study design appropriate and is the work technically sound?

Partly

Are the conclusions drawn adequately supported by the results?

Partly

Are sufficient details of methods and analysis provided to allow replication by others?

Partly

Reviewer Expertise:

-

I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard, however I have significant reservations, as outlined above.

Hammadi Ahmed Business Administration Depart, University of Fallujah, Al-Fallujah, Al Anbar Governorate, Iraq

Competing interests: “The authors declare no competing interests.”

 19 5 2026

Thank you very much for your thorough, constructive, and insightful comments on our manuscript. We have carefully considered each point raised and have substantially revised the manuscript accordingly. Below, we provide a detailed point-by-point response to each comment, indicating the specific changes made and their location in the revised manuscript.

Comment #1 (Sampling - Simple Random vs. Convenience)

Reviewer's Comment: "The authors claim to employ simple random sampling while simultaneously describing a process of distributing a questionnaire online through social media and email. This is the definition of convenience or snowball sampling. Simple random sampling requires a predefined sampling frame containing every individual in the population of 750, from which participants are randomly drawn. There is no evidence such a frame existed."

Response: The reviewer is absolutely correct. Our original description of the sampling method was inaccurate. We have corrected this throughout the manuscript.

Specific changes made:

•    Changed all instances of "simple random sampling" to "convenience sampling with stratified targeting"

•    Added explicit acknowledgment that no complete sampling frame existed

•    Added justification for the use of convenience sampling in exploratory research in emerging e-commerce contexts

•    Added the specific distribution channels used (email newsletters, WhatsApp Business broadcast lists, store-affiliated social media groups)

•    Added the response rate (29.2%) as a quality indicator

•    Added a limitation statement in Section 5.5 acknowledging potential sampling bias

Location in revised manuscript: Section 3.2 (pages 21-22), Section 5.5 (page 32)

Comment #2 (Date Inconsistency - 2023 vs. 2025)

Reviewer's Comment: "The abstract states data collection occurred between March 1, 2023, and July 1, 2023. The methodology section claims the target population consisted of customers active between February 3, 2025, and February 20, 2025. It is chronologically incoherent to sample participants in 2023 based on their activity parameters in 2025."

Response: This was an unfortunate typographical error in the original abstract. We have corrected all dates throughout the manuscript.

Specific changes made:

•    Corrected the abstract to state data collection period: February 3-20, 2025

•    Corrected the methodology section to be consistent with the abstract

•    Verified all date references throughout the manuscript for consistency

•    Added a note that the earlier date (2023) in the original abstract was a typographical error

Location in revised manuscript: Abstract (page 1), Section 3.1 (page 21)

Comment #3 (Hypothesis Wording vs. Statistical Tests)

Reviewer's Comment: "The phrasing of the hypotheses does not match the statistical tests performed. The second hypothesis dictates testing the effect of store specifications on the specific dimensions of customer perception, including emotional attraction, wisdom, and confidence. The regression analyses reported in the tables treat perception of attractiveness as a single aggregated variable. The individual sub-dimensions are never regressed against the predictors, leaving the actual wording of the hypothesis untested."

Response: The reviewer raises a valid point. We have revised the hypotheses to accurately reflect the statistical tests performed (aggregated dependent variable) while acknowledging the conceptual multidimensionality of the dependent variable.

Specific changes made:

•    Revised H₂ₐ, H₂բ, H₂c, and H₃ to specify that the dependent variable is the "attractiveness of customer perception of the product (aggregated score)"

•    Added a note in Section 1.7 clarifying that while the dependent variable has three conceptual dimensions (EA, WP, CWP), hypothesis testing uses the aggregated score due to sample size considerations

•    Added a recommendation in Section 6.3 for future research with larger samples to test dimension-specific effects

•    Added a limitation statement in Section 5.5 acknowledging that dimension-specific hypothesis testing was not feasible with the current sample

Location in revised manuscript: Section 1.7 (page 9), Section 5.5 (page 32), Section 6.3 (page 33)

Comment #4 (Mixing PLS-SEM and OLS Regression)

Reviewer's Comment: "The authors present measurement models, composite reliability, and average variance extracted metrics derived from SmartPLS. They then abruptly switch to reporting ordinary least squares multiple regression outputs with F values and adjusted R squared metrics using SPSS. Mixing variance-based structural equation modeling for measurement validation with first-generation regression for structural paths violates the foundational assumptions of both statistical paradigms."

Response: We understand the reviewer's concern and have added explicit justification for our two-stage approach. We also note that this approach, while unconventional, is defensible when the measurement model is used solely for validation and latent variable scores are extracted for use in OLS regression.

Specific changes made:

•    Added Section 3.5 "Analytical Strategy" with explicit justification for the two-stage approach

•    Explained that PLS-SEM provides latent variable extraction and validity metrics (HTMT, AVE) that OLS cannot provide

•    Explained that OLS regression is appropriate for structural path testing when the model is recursive with no mediated or moderated paths

•    Added citation (Hair et al., 2019) supporting this approach

•    Noted that this approach is more transparent than using PLS-SEM path coefficients, which can be difficult to interpret with correlated predictors

Location in revised manuscript: Section 3.5 (pages 22-23)

Comment #5 (Discriminant Validity - HTMT Missing)

Reviewer's Comment: "Discriminant validity is missing from the analysis, HTMT test should be reported."

Response: The reviewer is correct. The original manuscript did not report HTMT values. We have now conducted the HTMT analysis and added it to the revised manuscript.

Specific changes made:

•    Added Section 4.2.3 "Discriminant Validity (HTMT Criterion)"

•    Added Table (12) showing HTMT values for all construct pairs with 90% confidence intervals

•    Added interpretation of HTMT values (all < 0.85, discriminant validity established)

•    Added a comment that the highest HTMT (0.748 for IQ ↔ SEQ) indicates substantial but acceptable shared variance

•    Added this as a methodological contribution (first study in Iraqi e-commerce to report HTMT)

Location in revised manuscript: Section 4.2.3 (pages 22-23), Table (12)

Comment #6 (Multicollinearity - System Quality Suppression)

Reviewer's Comment: "In the bivariate analysis, system quality shows a significant positive effect on product perception. When entered into a multiple regression alongside information quality and service quality, the coefficient for system quality becomes negative and statistically insignificant. The independent variables are dimensions of the exact same overarching construct and are highly correlated. The authors attribute this change to system quality being a mere baseline requirement, completely ignoring the obvious presence of multicollinearity. The complete absence of Variance Inflation Factor reporting renders the structural conclusions invalid."

Response: This is a critical methodological comment. The reviewer is correct that we did not initially report VIF values, and our initial explanation (system quality as baseline requirement) was incomplete without addressing multicollinearity. We have now fully addressed this issue.

Specific changes made:

•    Added Section 4.4.3 "Multicollinearity Assessment" before presenting H₃ results

•    Added Table (16) showing Pearson correlations among IQ, SQ, and SEQ (r = 0.658-0.712)

•    Added Table (17) showing VIF values for all predictors (1.96-2.14, all < 5.0, acceptable)

•    Revised the discussion of system quality suppression to include both statistical (multicollinearity, shared variance) and theoretical (Herzberg's hygiene factor) explanations

•    Added explicit statement that VIF values are within acceptable limits, but shared variance still affects coefficient estimates

•    Added acknowledgment that the suppression of SQ is a statistical phenomenon (shared variance) that may also have theoretical meaning (hygiene factor)

Comment #7 (Research Question 2 - Customer Awareness)

Reviewer's Comment: "The second research question inquires about customer awareness of shopping store specifications, yet no hypothesis is developed to test this, and the results section ignores it entirely."

Response: The reviewer is correct. Research question 2 was descriptive, not inferential, and therefore did not require a hypothesis. However, the results section did not adequately address it. We have now added explicit descriptive statistics to answer RQ2.

Specific changes made:

•    Added explicit statement in Section 1.5 that RQ2 is descriptive and will be answered using means and standard deviations

•    Added to Section 4.3 (Descriptive Statistics) a specific comment addressing customer awareness levels

•    Added interpretation that the mean score of 3.592/5 indicates "moderate awareness" with significant room for improvement

•    Added a practical implication based on this finding

•    Location in revised manuscript: Section 1.5 (page 7), Section 4.3 (page 24), Table (13)