Sustainable Gratitude Based on Theistic Spiritual Values Scale (SGBTSV-45): A Higher Education Measurement Instrument

1. Participants and sampling

The study involved a sample of 543 university students recruited from diverse higher education institutions across Indonesia. To ensure the sample remained aligned with the research objectives, a purposive-convenience sampling technique was employed. This approach specifically targeted emerging adults who operate within a theistic-cultural academic framework, focusing on individuals who were readily accessible and expressed a voluntary willingness to participate through digital platforms.

Data collection was executed via Google Forms, a digital administration method chosen to facilitate a broad geographical reach across the Indonesian archipelago while ensuring participant safety and convenience. Respondents provided their self-assessments through the 45-item SGBTSV-45 scale, evaluating their grateful dispositions across the eight previously defined dimensions.

Ethical integrity was a cornerstone of the procedure; all participants provided informed consent prior to their involvement. To foster honest and uninhibited self-reporting essential for psychometric accuracy all data were collected under a strictly anonymous protocol, adhering to the highest ethical standards in psychological research.

2. Instrument development and operationalization

The SGBTSV-45 was rigorously constructed based on a comprehensive multidimensional framework designed to operationalize sustainable gratitude through 45 distinct statement items. These items are systematically distributed across eight pivotal dimensions that capture the holistic synergy between theistic spiritual values and psychological fortitude: (1) transpersonal, (2) personal, (3) emotional, (4) moral, (5) prosocial, (6) behavioral action, (7) psychological well-being, and (8) inner strength.

To facilitate a nuanced quantification of respondents’ grateful dispositions, the scale utilizes a five-point Likert-type response format. This measurement approach is engineered to capture both the intensity and frequency of theistic-based gratitude, allowing participants to evaluate the degree to which each statement aligns with their lived experience. The scoring protocol adheres to a polytomous system, where each item response is assigned a numerical value reflecting its alignment with the latent construct, as systematically presented and elaborated in Table 1.

To further enhance psychometric rigor and minimize response bias, the scale incorporates both favorable (positively worded) and unfavorable (negatively worded) items. While favorable items follow a direct linear progression, unfavorable items undergo a reverse-scoring procedure (x’ = (k + 1)-x), where k represents the maximum scale value. This bidirectional scoring mechanism is essential for the subsequent Rasch Model calibration, as it ensures that all item measures are correctly oriented toward the same latent continuum before the estimation of person measures and item difficulty parameters.

3. Content validation (Expert judgment)

Prior to field testing, the initial item pool underwent a rigorous qualitative and quantitative evaluation conducted by a panel of five experts specializing in educational psychology, psychometrics, and spiritual education. This validation phase was instrumental in ensuring that each item remained theoretically grounded in theistic spiritual values while maintaining high psychometric standards. The experts scrutinized the instrument for content relevance, representativeness, and technical clarity. Their evaluative feedback facilitated the refinement of item phrasing and the elimination of conceptual ambiguities, thereby establishing robust content validity before empirical administration.

4. Readability and pilot testing

Once the experts reached a consensus on the validity of the items, a comprehensive readability test was performed with a group of university students who shared identical demographic and academic characteristics with the target research subjects. This stage served as a bridge between theoretical validation and practical application, ensuring that the population and sample remained conceptually aligned.

The primary objective of this assessment was to verify that the linguistic structure and sociocultural nuances of the SGBTSV-45 were easily comprehensible to the end-users. By gathering direct feedback from students, the researchers were able to confirm that the items were free from interpretative bias and ecologically valid for the Indonesian higher education context. This dual-layered validation process combining expert theoretical scrutiny with student-centered linguistic feedback guaranteed that the instrument was fully optimized for large-scale data collection and subsequent Rasch Model analysis.

5. Psychometric analysis (Rasch model)

The empirical data were analyzed using the Rasch Measurement Model (via Winsteps software, version 5.6.1) to transform ordinal raw data into linear interval measures. This probabilistic approach was specifically selected to achieve superior measurement precision, ensure objective calibration, and verify the instrument’s structural integrity. The psychometric evaluation of the SGBTSV-45 encompassed the following rigorous assessments:

1. Validity and reliability of content

To verify the structural integrity of the SGBTSV-45, a rating scale diagnostic (category functionality analysis) was performed. As suggested by Widhiarso and Sumintono (2016), this analysis examines the monotonic progression of the Observed Average and Category Measure to ensure that the five-point Likert response format is psychometrically sound and easily distinguishable by the respondents. The empirical results of this diagnostic evaluation are presented in Figure 1. The diagnostic results demonstrate a sequential movement of the Observed Average scores from 0.05 to 0.82, accompanied by a progression of Category Measure values from −1.75 to 1.77. This consistent movement from negative to positive values confirms that the instrument effectively captures the increasing intensity of sustainable gratitude with high-precision psychometric accuracy. Furthermore, the category probability curves demonstrated that each response category (1–5) maintained a distinct peak. This lack of threshold disordering indicates that no category was redundant or overlapping in the participants’ cognitive interpretations, affirming that university students could clearly differentiate between the five levels of theistic-based gratitude.

Figure 1. Category functionality and rating scale diagnostic results.

Category functionality and rating scale diagnostic result, in terms of scale agreement, can be seen from the observed average and category measure columns. The scale is considered appropriate if the values in both columns move sequentially from negative to positive.

During the instrument validation process, a critical phase of this validation involved synthesizing expert reviewer feedback to refine the instrument’s content validity. The detailed identities, academic backgrounds, and areas of expertise of the expert reviewers are presented in Table 4 in the supplementary materials. This process ensured that each item is operational, contextual, and accurately integrates theistic spiritual values with the sustainable nature of gratitude. By aligning these formulations with specific sub-indicators, the SGBTSV-45 provides a robust framework for measuring how spiritual foundations foster lasting well-being in academic settings. A key refinement involved shifting secular terminology to theistic-centered language. For instance, in the transpersonal dimension (Item 1), the initial phrase ‘I feel lucky’ was replaced with ‘recognizing that God has granted me the capability’ following expert recommendations to better reflect the theistic source of gratitude. In response to expert feedback on Item 45 (Self-Strength aspect), the wording was shifted from a general avoidance sentiment to a specific academic scenario. Rater 3 recommended contextualizing the item within classroom dynamics to enhance its face validity for university students. The revised statement “When encountering differences of opinion in group tasks, I prefer to withdraw rather than seek a collective solution” clarifies the behavioral indicators of theistic spiritual values when navigating collaborative conflict, thereby strengthening the instrument’s contextual relevance.

Regarding Item 29 within the Prosocial dimension, specifically for the Sustainable Acts of Kindness indicator, the initial draft stated: ‘Dialogue with friends builds collaborations, but the positive influence on the community remains largely unplanned.’ Rater 2 recommended transforming this into an active and intentional statement to better capture the essence of sustainable contribution. Consequently, the item was refined to: ‘Through dialogue with friends, I am able to build collaborations that bring a positive influence to the community.’ This revision was critical to shift the focus from a passive, incidental outcome to an intentional pro-social agency, which is a core component of gratitude rooted in theistic spiritual values. The complete thematic transformation of these items is detailed in Table 2.

Following expert consensus on item validity, a readability test was conducted with 10 third-semester university students representing various regions across Indonesia. This geographic diversity ensured that the instrument’s linguistic structure was culturally resonant and easily understood within the broad Indonesian higher education context. This stage served as a bridge between theoretical validation and practical application, ensuring that the population and sample remained conceptually aligned.

The primary objective was to verify that the linguistic structure and sociocultural nuances of the SGBTSV-45 were transparent to the target respondents. During this pilot phase, students provided critical feedback to further optimize the instrument’s clarity. The key recommendations received included: (1) simplifying complex academic jargon to enhance immediate comprehension, (2) adjusting local idioms to ensure national linguistic neutrality, and (3) refining sentence length to maintain respondent focus. These specific refinements, which address the students’ perspectives on item clarity, are detailed in Table 3. This dual-layered validation combining expert scrutiny with student-centered feedback guaranteed that the instrument was free from interpretative bias and fully optimized for large-scale data collection and subsequent Rasch Model analysis.

Beyond content refinement, the structural integrity and functional precision of the SGBTSV-45 rating scale are further corroborated by the visual evidence presented in the Category Probability Curves ( Figure 2). According to the psychometric criteria established by Boone (2017), a rating scale operates optimally when each response category demonstrates a distinct and prominent peak. As evidenced in Figure 2, each of the five categories (ranging from 1 to 5) successfully achieved an independent peak. This lack of threshold disordering indicates that no category was redundant or overlapping in the participants’ cognitive interpretations. Consequently, it can be concluded that the five-point Likert format employed in the SGBTSV-45 is clearly interpretable and psychologically representative of the university student population within this theistic-cultural framework. The detailed instrument blueprint supporting this structure is presented in Table 5.

Figure 2. Diagnostic rating scale test.

Another condition that explains the appropriateness of a scale can also be observed in Figure 2 (see legend). The criteria within a rating scale are considered comprehensible to respondents when each response category demonstrates a distinct peak within the probability distribution curve.

2. Construct validity: Unidimensionality analysis

The evaluation of unidimensionality, or construct validity, was conducted by examining the raw variance explained by measures and the unexplained variance across the first through fifth contrasts. According to the psychometric standards established by Boone (2017) an instrument demonstrates sufficient unidimensionality when the raw variance explained by measures reaches a minimum threshold of 20%, provided that the unexplained variance in the subsequent contrasts remains consistently below 15%. This ensures that the instrument predominantly captures a single underlying latent trait without significant interference from secondary dimensions.

The detailed outcomes of this Rasch-based principal component analysis of residuals, including the percentage of raw variance explained and the unexplained variance across contrasts one to five, are comprehensively presented in Figure 3 (Construct Validity of Research Instruments). Based on this analysis, the assessment of construct validity through unidimensionality testing revealed that the raw variance explained by measures reached 30.0%, thereby exceeding the Rasch model criterion of >20%. Furthermore, the unexplained variance across the first to fifth contrasts remained within acceptable thresholds, specifically 14.1%, 6.7%, 5.5%, 4.1%, and 3.8%, respectively. As the first contrast did not surpass the 15% cut-off point, these findings provide strong empirical support for the scale’s unidimensionality, confirming that the instrument measures a single dominant latent trait without substantial interference from secondary dimensions.

Figure 3. Construct validity of research instruments.

Unidimensionality analysis, as an examination of construct validity, evaluates the values of raw variance explained by measures and unexplained variance in the first to fifth contrasts.

3. Item fit order

Item fit analysis was conducted to determine whether each item was appropriately understood and functioned consistently within the measurement model. The evaluation criteria applied to assess item suitability were as follows: (a) Outfit Mean Square (MNSQ): 0.5 < MNSQ < 1.5; (b) Outfit Z-Standard (ZSTD): −2.0 < ZSTD < +2.0; and (c) Point Measure Correlation (Pt Mean Corr): 0.4 < Pt Corr < 0.85. An item is considered valid if it satisfies at least one of these three criteria. However, based on the Rasch measurement principles articulated by Bond and Fox (as cited in Hrnjicic & Alihodžic, 2024), fulfillment of the Outfit MNSQ criterion is regarded as sufficient evidence of item acceptability, thereby reducing reliance on the remaining two indices.

The comprehensive item fit statistics are documented in Figure 4 (Item Fit Levels of the Research Instrument), which is provided in the appendix.

Figure 4. Item fit levels of the research instrument.

Item fit analysis delineates the extent to which the content of each individual item is comprehensively understood by students. It evaluates whether each item functions consistently with the underlying construct and is interpreted meaningfully and coherently by respondents.

Referring to the results summarized in Figure 4, all items met the primary Outfit MNSQ criterion (0.5 < MNSQ <1.5). This indicates that every item functions within acceptable model expectations. Accordingly, all items in the research instrument are considered psychometrically valid and suitable for measurement purposes, irrespective of the remaining statistical indicators.

4. Item measure

Item difficulty levels, or item measures, were analyzed to determine the relative level of challenge each item posed to respondents within the Rasch measurement framework. The complete distribution of item difficulty is documented in Figure 5 (Distribution of Item Difficulty Levels), which is provided in the Appendix for detailed reference. Following the classification framework proposed by Boone et al. (2014), item difficulty was categorized into four tiers based on the integration of the logit mean and the standard deviation (SD): Very Difficult (> +1 SD), Difficult (0.00 logit to +1 SD), Easy (−1 SD to <0.00 logit), and Very Easy (< −1 SD). The statistical analysis produced a Standard Deviation (SD) value of 0.46, which served as the threshold for classification. Based on this criterion, the items were distributed as follows:

Figure 5. Distribution of item difficulty levels.

Item difficulty (Item Measure) reflects the extent to which an item is easy or difficult for respondents to answer. The classification of item difficulty is determined based on the standard deviation in conjunction with the mean logit value. Specifically, items are categorized as very difficult (greater than +1 SD), difficult (between 0.0 logit and + 1 SD), easy (between 0.0 logit and − 1 SD), and very easy (less than −1 SD).

Overall, the instrument demonstrates a well-balanced distribution of item difficulty levels. This variability indicates that the scale possesses adequate measurement sensitivity, enabling it to accurately assess respondents across a broad spectrum of latent trait levels, from low to high sustainable gratitude dispositions.

5. Wright map

The alignment between respondent ability and item difficulty was examined using the Wright Map (Person–Item Map), which is presented in Figure 6 in the Appendix. This map provides a visual representation of both person measures and item measures on a shared logit scale, enabling an evaluation of targeting accuracy within the Rasch framework.

Figure 6. Wright map of the instrument.

Based on the Wright Map, respondents’ abilities are distributed within an approximate range of −2 logits to +2 logits, with the majority of respondents clustered between −1 logit and + 1 logit. The mean person ability (person mean) is located around +0 logit, while the mean item difficulty (item mean) is also situated at approximately 0 logit. This alignment indicates that the average ability level of the respondents is relatively well matched to the average level of item difficulty.

The distribution of person measures spans approximately from −2.00 to +2.00 logits, with the majority of respondents clustered between −1.00 and + 1.00 logit. The mean person measure is located at +0.52 logit, positioned relatively close to the item mean, which is anchored at 0.00 logit. This proximity indicates a well-balanced calibration between respondents’ latent trait levels and the hierarchical difficulty of the items. Such alignment suggests that the SGBTSV-45 demonstrates strong measurement precision and is appropriately targeted to the university student population, with no substantial indication of floor or ceiling effects. Complementing the person distribution, the item difficulty hierarchy predominantly spans a range from −1.00 to +1.00 logit, partitioned into two primary strata:

The overall difficulty spectrum extends from approximately −1.00 logit (the most accessible item, N33) to slightly above +1.00 logit (the most challenging item, N9). The concentration of the majority of items within the −1.00 to +1.00 logit interval demonstrates a comprehensive spread of difficulty levels, ranging from easy to moderately difficult. This diverse array of item difficulty, coupled with the balanced distribution of respondent ability around the mean, underscores the instrument’s high sensitivity and robust capacity to differentiate various levels of sustainable gratitude among the target population.

6. Reliability and separation analysis

The reliability of an instrument is fundamental to evaluating the consistency of measurements across repeated administrations (Boone & Noltemeyer, 2017). Within the framework of the Rasch Measurement Model, the reliability of the SGBTSV-45 is comprehensively assessed through three primary indices: person reliability, item reliability, and Cronbach’s alpha $\left(\alpha \right)$ , supplemented by the separation index. These parameters provide a robust estimation of the internal consistency and the stability of the interaction between the university students and the 45-item scale.

The empirical results of the reliability testing for the SGBTSV-45 are summarized in Figure 7. The analysis yielded a person reliability coefficient of 0.83 and an item reliability of 0.99. According to the criteria established by Boone (2020), the consistency of respondent answers is categorized as “good,” while the technical quality of the items is “excellent.” Furthermore, a Cronbach’s alpha value of 0.85 confirms a high degree of internal consistency. Collectively, these findings demonstrate that the SGBTSV-45 is a stable tool, capable of producing consistent longitudinal data when administered to similar student populations.

Figure 7. Results of instrument reliability testing.

Instrument reliability is essential to determine the extent to which repeated measurements yield consistent and stable information. Reliability analysis using the Rasch Model produces three key indices: person reliability, item reliability, Cronbach’s alpha, and separation indices, which collectively indicate the consistency and discriminatory power of the instrument.

Beyond general reliability, the separation index was analyzed to determine the scale’s capacity to differentiate levels of person ability and item difficulty into distinct strata. A higher separation value signifies superior instrument quality and greater discriminatory power (Avinç & Doğan, 2024). The analysis revealed a person separation of 2.21 (fair) and an item separation of 10.66 (excellent). According to Storey et al. (2025), separation indices are classified as follows: < 2 (weak), 2–3 (fair), 3–4 (good), 4–5 (very good), and > 5 (excellent). The analysis revealed a person separation of 2.21 (fair) and an item separation of 10.66 (excellent). By applying the strata formula $\frac{H=\left(4\mathit{\text{xseparation}}\right)+1}{3}$ , the person separation value resulted in a strata index (H) of 3.28 (rounded to 3). This indicates that the SGBTSV-45 can effectively categorize university students into three distinct groups based on their intensity of sustainable gratitude: low, moderate, and high. More significantly, the item separation resulted in an H value of 14.54 (rounded to 14), signifying that the difficulty levels are distributed across 14 distinct hierarchical groups. This exceptionally wide spread of item difficulty, ranging from fundamental to complex theistic-spiritual indicators, underscores the SGBTSV-45’s superior sensitivity. It confirms the instrument’s utility as a high-resolution diagnostic tool capable of capturing the nuanced spectrum of sustainable gratitude in higher education settings.