Alcohol consumption outcome expectancy questionnaire: psychometric validation in Ecuadorian university students

Introduction

Excessive alcohol consumption represents a serious global public health problem, with significant consequences for society. According to the Global Burden of Disease Study (GBD) of 2020, approximately 1.78 million alcohol-related deaths were recorded, and its intake is associated with deterioration of health and some chronic diseases.¹ This problem is not foreign to Ecuador, where 25,140 admissions to treatment for alcohol abuse were reported in 2021, 75% of which corresponded to men.² The high prevalence of alcohol consumption among adolescents and young adults in the region^3–5 has led to the need to develop psychometric tools, such as the Questionnaire of Alcohol Consumption Outcome Expectations (CERCA), which proposes three latent factors: 1) Stress reduction; 2) Improvement of social skills; and 3) Social facilitation.⁶ However, models obtained through exploratory factor analyses require validation to ensure their accuracy and applicability in different cultural and social contexts.

Alcohol consumption expectancies play a fundamental role in the motivation and behavior of individuals. Specifically, it has been found that there is a significant relationship between positive expectancies and the likelihood of alcohol consumption.^7–9

To better understand the underlying factors that influence these expectancies, it is essential to rely on sound psychological theories to analyze the complexity of this phenomenon. In this regard, three theories stand out for their relevance in the study of drinking expectancies: Albert Bandura’s Social Learning Theory,¹⁰ which postulates that people acquire new knowledge and skills through observation and imitation of others; Atkinson’s Expectation-Value Theory,¹¹ which suggests that behavioral decisions are based on outcome expectations and the value assigned to those outcomes; and Oliver’s Expectation Confirmation Theory,¹² which emphasizes that satisfaction occurs when actual expectations equal or exceed previous expectations.

Social Learning Theory provides a framework for understanding how people develop expectations about the consequences of their behavior from previous experiences and social reinforcement. In the case of CERCA, the latent factors of social interaction are grounded in this principle, as people develop social skills by observing models and receiving positive reinforcement, which encourages their participation in group activities. From this understanding, it is crucial to consider how outcome expectations influence decision-making, which relates to the Expectancy-Value Theory. This theory holds that behavioral decisions are based on outcome expectations and the value assigned to them. In this context, the variables measured in the CERCA model can be viewed as elements that affect an individual’s expectations about social interactions and their emotional consequences.

Complementing this perspective, The Expectancy Confirmation Theory offers an additional approach by highlighting that positive expectations can significantly influence the likelihood of repeating behaviors, such as alcohol consumption. If social experiences fulfill expectations of decreased stress, this not only encourages the repetition of social behaviors; but also contributes to the perception of an overall positive emotional state.

Together, these theories provide a comprehensive theoretical framework that not only explains how drinking expectancies are formed through social observation and outcome appraisal; but also highlights the relevance of satisfaction derived from social experiences, which is relevant for understanding and addressing drinking behaviors in different cultural contexts.^13–15

On the other hand, variables related to what is defined in the CERCA questionnaire as decreased stress (feeling relaxed, calm, happy, at ease, with well-being) form a construct that not only measures the absence of discomfort; but also captures the richness of positive experiences, thus grounding the notion of emotional well-being in the framework of subjective well-being.^16–18

Previous studies have shown that expectancies related to alcohol intake may be mediated by a variety of factors. Cooke and coworkers¹⁹ conducted research with first-year university students and found that, among others, these expectations are influenced by both emotional and social performance factors. For their part, Londoño and Carrasco²⁰ compared beliefs about alcohol consumption between young Colombians and Chileans, and found that Colombians believe that alcoholic beverages increase social disinhibition and the ability to interact in groups, in addition to providing relaxation and improving mood.

Based on this background, the main objective of this study is to confirm the validity of the Questionnaire of Alcohol Consumption Outcome Expectations (CERCA) through a confirmatory factor analysis with structural equations in students of a public university in Ecuador. Secondary objectives include: 1) To analyze the results of the exploratory factor analysis of the CERCA; 2) To calculate the fit indices needed to validate the confirmatory factor analysis; and 3) To validate the goodness of fit of the CERCA model as a psychometric test in a sample of university students. It is expected that the results of the confirmatory analysis conducted in this research, which reunifies in a two-factor structure (emotional well-being and social interaction) the CERCA questionnaire, will contribute to a better understanding and approach to alcohol-related expectancies, as well as to the implementation of more effective interventions in this area.

Methods

The study was a cross-sectional study with a quantitative approach, at an integrative level, corresponding to confirmatory research.²¹

The population (N = 1632) consisted of students in the seven education programs of an Ecuadorian public university. Given that the majority proportion of the population corresponds to women (68.8%), it was decided to carry out a stratified probability sampling by sex, with proportional allocation.²² Based on the population data, we took p = .688 (q = .312) where p is the probability of selecting an individual with the characteristic of interest, which in this case is the proportion of women in the population. For the calculations, because alcohol consumption represents a public health problem,^23,24 a confidence level of 99% (Z = 2.576) and an estimation error e = 3.5% was considered.

Since the population was finite, the formula used to calculate the sample was:

Data were collected using the Questionnaire of Alcohol Consumption Outcome Expectations Questionnaire, CERCA.⁶ The CERCA is an instrument designed to assess the expectations that individuals have about the effects of alcohol consumption in three main dimensions: 1) Decreased stress; 2) Improved social skills; 3) Social facilitation. The questionnaire consists of 12 items assessing these dimensions using a six-point Likert scale: 1 = Strongly Disagree; 2 = Strongly Disagree; 3 = Disagree; 4 = Agree; 5 = Strongly Agree; 6 = Strongly Agree. In addition to the CERCA, demographic data such as age, sex, ethnic self-recognition, type of cohabitation, and context of origin were collected.

Data collection was carried out in person at the university facilities. Participation was voluntary and written informed consent was obtained from all participants before applying the questionnaire. All participants were aged 18 years or older, and no minors were included in this study. Anonymity and confidentiality of responses were guaranteed. The application of the instrument was supervised by trained researchers to resolve any doubts during the process, following the ethical principles of the Declaration of Helsinki for research involving human subjects. The data collected were processed with R-Studio 2023.06.0 Build 421, SPSS v. 29, Amos Graphics v.24, and the macro for calculating McDonald’s in SPSS.²⁵

Throughout the study, the following analyses were performed: 1) Descriptive analysis of demographic variables and CERCA scores; 2) Exploratory Factor Analysis (EFA) to examine the factor structure of the CERCA in the Ecuadorian sample. The principal axis factorization extraction method with direct oblimin rotation (delta = 0) was used. The Kaiser-Meyer-Olkin index (KMO) and Bartlett’s test of sphericity were calculated to assess the adequacy of the data for the factor analysis, and reliability analysis was performed by calculating Cronbach’s alpha coefficient for each of the subscales identified and for the global scale, complemented with McDonald’s w; 3) Mann-Whitney U test to study the existence of statistically significant differences by sex and Kruskal-Wallis test for the variables ethnic self-recognition, living environment and context of origin; 4) Confirmatory Factor Analysis (CFA) by structural equation modeling to validate the factor structure obtained in the model proposed by the CERCA questionnaire, using AMOS v. 24. The estimates of the fit indices were based on the Asymptotic Distribution Free method (ADF), recommended when some categorical data and data do not fit the normal distribution.²⁶

In cases of large samples without multivariate normality, the literature recommends replacing the Chi-square with the alternative CMIN/DF index, which evaluates the discrepancy between the Chi-square and the degrees of freedom, avoiding the rejection of well-specified models showing statistically significant p-values due to the sensitivity of the Chi-square index to sample size.^27–31

Other indices considered were the comparative fit index, CFI, and the root-mean-square error of approximation, RMSEA. When a model reports a good fit in both, it is improbable that it does not adequately represent the data.³² The traditional cut-off points for these indices were taken, i.e. CFI > 0.95 and RMSEA < 0.05.³³ For the other indexes, such as the unstandardized Tucker Lewis fit index (TLI), the goodness-of-fit index (GFI) and the normed fit index (NFI), minimum required values of 0.95 were established, depending on the sample size. Likewise, for the root mean square residual (RMR), values lower than 0.05 were considered as an acceptance criterion.^34,35

Results

The characterization of the sample, n = 685, is presented in Table 1.

Comparatively, the CERCA model,²⁵ was conducted with a sample of 208 individuals confined in an addiction center, with 86.5% men. Still, the authors recommended considering how it works in other populations and acknowledged the low inclusion of women as a limitation of their study.

In the EFA used for the psychometric characterization of the CERCA, they performed an extraction using principal component analysis with varimax rotation. However, this popular approach known as “Little Jiffy” is considered problematic because it can lead to inadequate and unreliable factor solutions when applied to complex multidimensional data.^36,37 In the extraction, the principal component method should be used when the objective is to simplify the data without assuming an underlying theoretical model and the objective is to maximize the amount of total variance explained by the first components, and varimax rotation when the factors are not correlated with each other^38,39; however, this is not the case in the CERCA model,²⁵ nor do the authors theoretically justify the selection of the methods. In general, the indiscriminate use of the combination of Principal Components and varimax rotation carried out by SPSS by default is not a good practice.^40–42

Considering that the variables were measured on an ordinal scale, polychoric correlations were used.³⁹ All correlation values were greater than 0.61, indicating a strong relationship between the ordinal or categorical variables in terms of their latent association. The results, obtained using R-Studio 4.3.3 software, are presented in Table 2.

Confirmatory Factor Analysis (CFA)

The confirmatory factor analysis estimates for the three factors of the original CERCA model, obtained using AMOS v.24, are presented in Figure 1.

Figure 1. Standardized estimates for the confirmatory factor analysis of the CERCA model.

The factor loadings of the measured variables on their respective factors are all greater than .78, indicating a strong relationship between them. Similarly, the squared multiple correlations have values greater than .61, suggesting that a significant portion of the variance in the observed variables is explained by the latent factors.

Regarding the fit indices, the original model reports an Akaike Information Criterion (AIC)⁵⁶ of 192.2. It does not pass the Chi-square test and fails to meet the optimal thresholds for key fit indices, including the Comparative Fit Index (CFI), Root Mean Square Residual (RMR), Goodness of Fit Index (GFI), Normed Fit Index (NFI), Adjusted Goodness of Fit Index (AGFI), and Tucker-Lewis Index (TLI).^57,58 To improve the fit indices, the standardized covariance matrix suggests that removing the variables “coexisting,” “peaceful,” and “well-being”—which exhibit the highest covariances—could be a viable strategy. However, the high correlation (.99) between Factor 1 (Improvement of Social Skills) and Factor 3 (Social Facilitation) indicates that these two factors are highly interrelated and measure conceptually similar constructs.⁵⁹ This extreme correlation undermines the validity of the original three-factor CERCA model.

For the estimation of the modified CERCA model, Alcohol Consumption Expectancy Questionnaire (ACEQ), the variables identified as problematic in the three-factor model were removed. As a result, the Chi-square test was not significant, indicating that the specified model adequately fits the empirical data. The ACEQ yielded an AIC of 51.45, representing a substantial improvement over the original model proposed by Ref.⁶

Passing this chi-square test is uncommon in CFAs of expectations about alcohol consumption that have used relatively large samples, as seen in the studies by Ref.^60,61 where the models did not pass the inferential test.

Figure 2 presents the standardized estimates for the modified CERCA model (ACEQ), showing that all factor loadings are very high (above .86). Notably, in Factor 1, significantly higher values have been found compared to those reported in the unifactorial model by Ref.⁶² (.95, .94, .86 versus .77), which the researchers classified as having excellent fit. Compared to the CERCA model, a substantial increase has been achieved in the percentage of variance explained by the latent factor for each measured variable, with a minimum of 74%.

Figure 2. Standardized estimates for the confirmatory factor analysis of the modified CERCA model.

Table 4 presents a comparison of the Chi-square test results and the goodness-of-fit indices for both the CERCA and ACEQ models.^35,63–65

Table 4. Comparison of Chi-square test and goodness-of-fit indices for the CERCA and ACEQ models.

Fit index	Expected1	CERCA Model (3 factors)	ACEQ Model (2 factors)
		Observed
Chi-square $\left({\chi }^2\right)$	p > 0.05	c2 = 69.005; p = .000	c2 = 21.5 p = .064
Discrepancy between χ2 and degrees of freedom (CMIN/DF)	Between 1 y 3	2.875	1.65
Goodness-of-Fit Index (GFI)	.95-1	.952	.983
Adjusted Goodness-of-Fit Index (AGFI)	.90-1	.911	.964
Root Mean Square Residual (RMR)	< .05	.118	.042
Root Mean Square Error of Approximation (RMSEA)	< .05	0.052; p = 0.371	.031; p = .915
Comparative Fit Index (CFI)	.95 – 1.0	.951	.990
Normed Fit Index (NFI)	.95-1.0	.927	.976
Non-Normed Fit Index (NNFI or TLI, Tucker-Lewis Index)	.95-1	.926	.984

Table 5 presents the communalities and factor loadings for the modified two-factor CERCA model (ACEQ), using principal axis factorization with oblimin rotation as the extraction method. A KMO value of .903 and communalities greater than .67 were obtained. These two factors account for 89.5% of the variance.

Table 5. Communalities and factor loadings of the pattern matrix from the exploratory factor analysis of ACEQ.

		Factor
Communalities	Initial	1	2
Drinking alcohol makes me feel relaxed	.671	.018	.831
Drinking alcohol makes me feel happy	.783	.016	.911
Drinking alcohol makes me feel comfortable	.785	-.030	.952
Drinking alcohol helps me socialize	.877	.963	-.011
Drinking alcohol helps me have fun with my friends	.837	.938	-.013
Drinking alcohol helps me share	.856	.943	.000
Drinking alcohol helps me be sociable	.848	.909	.028

Discussion

All the descriptive measures presented in Table 4 met the cut-off points specified by Ref.^35,63,64 for establishing a good fit and align with the values reported in Ref.^65,66 The cut-off points for the CFI, GFI, and RMSEA indices were superior to those reported by Ref.⁶² (0.90 for CFI and GFI, and <.08 for RMSEA) and those presented in other studies validating models of motivations related to alcohol consumption, such as Ref.^60,61

The proposed two-factor model, ACEQ, consolidates social behavior into a single factor. Accordingly, Factor 1 is identified as “Emotional Well-being” (α = .927, ω = .928) and Factor 2 as “Social Interaction” (α = .968, ω = .968). The overall internal consistency yielded a Cronbach’s α of .952 and a McDonald’s ω of .948, both higher than those reported by Ref.⁶² (α = ω = .79) for the validation of the Alcohol Expectancy Questionnaire (AEQAB) in Portuguese university students. The high internal consistency of the subscales and the excellent goodness-of-fit indices of the modified model suggest that ACEQ is a robust instrument for assessing expectations regarding alcohol consumption in this population.

Although a three-factor structure was expected, in accordance with the original model by Ref.,⁶ confirmatory analyses indicated that a two-factor structure is more appropriate for the Ecuadorian population. This discrepancy, consistent with Fog’s study¹³ highlights the necessity of adapting theoretical models to specific cultural contexts, which may require eliminating variables that do not align well. The high correlation (.99) between the “Improvement of Social Skills” and “Social Facilitation” factors in the original CERCA model raises questions about the true distinction between these constructs within the Ecuadorian university student population. Future research is encouraged to further explore these models to confirm or refine the present conclusions.

In the CERCA model⁶ both the variables “socializing” and “coexisting” are included under the social skills factor. Depending on the context, both encompass the individual’s ability to relate to others and the specific behaviors within a social environment that facilitate interpersonal connections. Therefore, the ACEQ model not only offers greater parsimony and a better fit but also provides a more integrated conceptualization by unifying these elements under the label “social interaction.” This unification more accurately reflects the way social skills and activities intertwine in daily life and manifest in social practice.

Furthermore, the theoretical foundation supporting this social interaction factor is based on the triangulation of Bandura’s social learning theory,¹⁰ Atkinson’s achievement motivation theory,¹¹ and Oliver’s expectancy-value model.¹² This factor consists of interrelated components that mutually reinforce one another in the social learning process. It illustrates how expectations and values not only drive the acquisition of social skills but also encourage participation in social activities. Additionally, it integrates both expectations of social interaction and real-life experiences (e.g., having fun with friends) to form a comprehensive assessment of social engagement. In this way, unifying these variables allows for a more coherent representation of how expectations and experiences collectively shape satisfaction in social contexts.

The results provide empirical support for the validity of the Alcohol Consumption Expectancy Questionnaire (ACEQ) model in the Ecuadorian context. This validation lays the groundwork for the development of preventive interventions specifically targeting positive expectations related to alcohol consumption. These findings align with the predominance of positive expectations identified in the study by Ref.⁷

This study reveals that positive expectations regarding alcohol consumption are closely linked to emotional well-being and social interaction. These findings are consistent with those reported by Ref.⁵³ and Ref.⁵⁴ Additionally, they align with previous studies, such as that of Ref.,¹⁹ which found that alcohol consumption expectations among university students are strongly influenced by emotional and social factors. Similarly, the study by Londoño and Carrasco,²⁰ highlights that expectations surrounding alcohol consumption vary across cultural and social contexts, with disinhibition and social interaction emerging as the primary motives among young Colombians.

This study has significant implications for both academic research and public health policy development. Its diagnostic utility lies in its ability to elucidate the motivational factors underlying alcohol consumption. The identification of emotional well-being and social interaction as key determinants of alcohol consumption expectations aligns with the findings of Sánchez-Puertas et al.,⁶⁷ who emphasize that many alcohol prevention programs focus on promoting healthy alternatives for stress management and the development of social skills.

These findings hold both theoretical and practical relevance. On the one hand, they provide a solid foundation for the development of more effective and culturally tailored prevention and intervention programs. On the other hand, they pave the way for future research directions, enabling a deeper exploration of the dynamics of alcohol consumption expectations and their relationship with other psychosocial factors.

Conclusions

This study provides evidence of the validity and reliability of the Questionnaire on Expectations of Alcohol Consumption Outcome (CERCA) in its modified version (CERCAM) using a sample of Ecuadorian university students. The results from exploratory and confirmatory factor analyses indicate that a two-factor structure offers a superior fit compared to the original three-factor model proposed by Templos-Núñez et al.⁶ This finding suggests that alcohol consumption expectations in this population are best categorized into two core dimensions: emotional well-being and social interaction.

The high internal consistency of the subscales and the strong fit indices of the ACEQ model, reflected in high Cronbach’s alpha and McDonald’s ω values, support its reliability as a psychometric instrument. The model’s validity is further reinforced by the good fit indices obtained in the confirmatory factor analysis, particularly the comparative fit index (CFI), the normed fit index (NFI), and the root mean square error of approximation (RMSEA). Most notably, the Chi-square test yielded p > .05, indicating an adequate model fit.

The absence of significant gender differences in alcohol consumption expectations contrasts with some previous studies and warrants further investigation. However, this finding could reflect shifting social and cultural norms regarding alcohol consumption among Ecuadorian youth.

These results highlight the importance of addressing expectations related to stress reduction and social interaction, particularly among university students, to promote healthier alcohol consumption habits in academic settings.

Ethical approval

This study received ethical approval from the Ethics Committee for Research of the Faculty of Education, Science and Technology at Universidad Técnica del Norte (certificate No. UTN-FECYT-CEI-2024-0000001356-C), dated October 22, 2024. The research was also formally endorsed by the Governing Council of the Faculty (Resolution No. FECYT-CD-SE-40-0319-2024) and the university’s Research Council (Resolution No. UTN-CI-2024-219-R). The study was conducted in accordance with the ethical principles of the Declaration of Helsinki (2013 revision), the Ethical Guidelines for Educational Research of the British Educational Research Association (BERA), and the Code of Ethics of Universidad Técnica del Norte. Written informed consent was obtained from all participants involved in the study.