Adapting the Psychological Capital Questionnaire (PCQ) for Kazakh-Speaking University Students in Kazakhstan: Translation, Validation, and a Three-Factor Model

2.1. Procedure and Research Methods

The research design was reviewed and approved by the Ethics Committee of Al-Farabi Kazakh National University in 2025 (Protocol No. IRB-A1667, Registration No. 119). Data collection was conducted online from February 20 to April 11, 2025, using the Google Forms platform. The questionnaire was distributed via convenience sampling through digital channels, primarily WhatsApp and Instagram. Participants were allowed to share the survey link with their acquaintances, which partially corresponded to the principles of snowball sampling, a typical feature of online research.

Participation in the survey was voluntary and anonymous. Before completing the questionnaire, participants were provided with an information sheet describing the study objectives, guarantees of confidentiality, the estimated completion time (12–25 minutes), and the voluntary nature of participation. The questionnaire also contained the contact details of the research team. Each participant was able to complete the survey only once, ensuring the uniqueness and reliability of the collected data. Inclusion criteria required participants to be aged 17 years or older and to have adequate proficiency in the Kazakh language. The study involved students from Al-Farabi Kazakh National University, Abai Kazakh National Pedagogical University, Ahmet Yassawi University, and Alikhan Bokeikhan University. Participation was entirely voluntary, and no incentives or rewards were offered.

2.2. Sample Division and Justification of Sample Size

The sample size for validating the Kazakh version of the questionnaire was determined in accordance with international recommendations for factor analysis (Floyd & Widaman, 1995; Fokkema & Greiff, 2017) [22, 23]. According to these sources, the minimum ratio ranges from 5:1 to 10:1 (participants per estimated parameter). In the present study, a ratio of at least 10:1 was maintained, ensuring the statistical reliability and validity of the results. The total sample for the adaptation and translation of the English version of the Psychological Capital Questionnaire consisted of 656 participants (N = 656), including 570 females and 86 males aged 16 to 27 years (М = 18,5; SD = 1,22). This sample was randomly divided into two subsamples: N = 456 for conducting the Exploratory Factor Analysis (EFA) and N = 200 for the Confirmatory Factor Analysis (CFA).

Prior to the translation and adaptation of the methodology, we received written permission from one of the authors of the original 24-item PsyCap questionnaire, Professor F. Luthans, to translate, culturally adapt, and use the PsyCap-24 questionnaire as part of a grant-funded research project (IRN AR23488243 “Development of Psychodiagnostics in Education in Kazakhstan and Worldwide: Cross-Cultural Analysis and Development of a Validation Algorithm”). The original PCQ-24 instrument consists of 24 items designed to assess four core personal characteristics. These include self-efficacy, which reflects confidence in achieving goals and solving tasks; hope, understood as the ability to redirect pathways toward goal attainment; optimism, defined as a positive outlook on success; and resilience, which refers to the capacity to cope with difficulties. The questionnaire consists of 21 positively worded statements and 3 reverse-worded items. Students are asked to rate the extent to which each statement reflects their personal characteristics using a 6-point Likert scale from 1 (“strongly disagree”) to 6 (“strongly agree”).

The translation and cultural adaptation of the PCQ-24 into Kazakh was carried out in accordance with international guidelines and standards proposed by Sousa and Rojjanasrirat [24], Sperber [25], and the ITC Guidelines for Translating and Adapting Tests [26].

The first step involved translating the English version of the questionnaire into Kazakh by two independent professional translators. In the second step, a third translator, who was both bilingual and bicultural, compared the two versions and merged them into a single preliminary Kazakh version of the questionnaire. The third step involved back-translation, in which another translator translated the preliminary Kazakh version back into English. In the fourth step, a multidisciplinary committee was established, consisting of a bilingual practicing psychologist who was part of the research team, a native Kazakh-speaking researcher, and two translators who had previously been involved in translating the questionnaire. The committee carefully compared and evaluated the back-translated version against the original English questionnaire. All discrepancies were discussed, reviewed, and corrected.

After completing all procedures for translating and semantically aligning the concepts, we asked another Kazakh-English bilingual with a degree in philology to compare the final Kazakh version with the original English questionnaire. The translated version was assessed as accurate and appropriate.

The fifth step involved pilot-testing the Kazakh-language version of the questionnaire with a sample of 30 participants. This phase focused on evaluating face validity, particularly the clarity and cultural relevance of the questionnaire items. This was conducted with bilingual Kazakh-English students, which enabled us to identify any ambiguous or culturally inappropriate wording in the statements. Participants completed the translated Kazakh version of the scale and provided written comments on each item. Based on the feedback received, targeted editorial revisions were made to improve the linguistic accuracy and cultural appropriateness of the items. Students were also asked to evaluate the clarity and comprehensibility of all questions. Following this stage, the final Kazakh version of the PsyCap questionnaire, consisting of 24 items, was approved.

3.1. Primary Data Processing

All questionnaires were administered to participants in paper-based format. After completion, an initial screening of the responses was conducted, during which questionnaires with incorrect or incomplete data were excluded from the overall sample. Following this procedure, a total of 708 questionnaires remained. Of these, 52 questionnaires containing two or more missing items were excluded from further analysis. For 11 questionnaires with a single missing response, missing values were replaced using median substitution through the SPSS “Replace Missing Values” function. The dataset was not screened for statistical outliers. Duplicate entries were unlikely, as all questionnaires were completed manually in an offline format. Data-entry accuracy was verified using descriptive statistics, including frequency analyses for each item, which allowed potential input errors (e.g., implausible values such as “66” or “55”) to be identified and corrected.

Before data analysis, the reverse-coded items (item 13, item 20, and item 23) were recoded to ensure a consistent directionality across all scale items. This procedure allowed higher scores on reverse items to correspond to higher levels of the measured construct rather than the opposite. The calculation of total scale scores, internal consistency indices, factor analysis, and reliability and validity assessment was conducted only after the reverse items had been inverted.

In the first stage of the questionnaire adaptation process, EFA was conducted to identify latent factors without considering the structure of the original questionnaire. The sample for the EFA consisted of 456 participants, including 395 women and 61 men, with a mean age of 18.49 years. For exploratory factor analysis (EFA), the principal component analysis (PCA) method was used to identify factors. Varimax rotation with Kaiser normalization was applied as the rotation technique. The number of factors was determined based on eigenvalue analysis (eigenvalues > 1) and inspection of the scree plot generated using SPSS.

This procedure was conducted to identify potential cultural differences in the understanding of the psychological phenomena under study. The obtained Kaiser-Meyer-Olkin (KMO) coefficient was found to be 0.943, indicating a high level of sampling adequacy for factor analysis. The result of Bartlett’s test of sphericity was statistically significant (χ² (276) = 5522.33, p <0,001), confirming the presence of substantial correlations among the variables.

Based on the EFA results, four latent factors were identified, accounting for 59.29% of the total variance. All 24 items of the questionnaire showed factor loadings ˃ 0.4. After rotation, the contribution of each factor to the total variance was distributed as follows:

Factor 1 (Self-Efficacy) explained 20.75% of the variance;

Factor 2 (Resilience) explained 15.55% of the variance;

Factor 3 (Hope) explained 13.72% of the total variance;

Factor 4 (including the three reverse questions) explained 9.27% of the total variance. The distribution of factor loadings is presented in Table 1.

Thus, the EFA confirmed the presence of structural changes in the questionnaire. The original version of the questionnaire included four scales: efficacy, resilience, hope, and optimism. The EFA results showed that the items from the optimism scale were redistributed to the resilience scale. In our opinion, this redistribution can be considered theoretically justified. As shown in Table 1, two items (items 7 and 21) demonstrated cross-loadings, while three items (items 13, 20, and 23) showed negative loadings on an additional fourth factor. These results can be explained theoretically, as these items are reverse-worded and reflect opposite meanings. This factor does not represent an independent factor and will not be considered in further research.

The evaluation of the EFA results was conducted by relying on the original theoretical model. Thus, it was decided to include item 7 in Factor 3 (Hope), as this item originally belonged to this scale. The structure of the questionnaire was organized as follows:

Factor 1 (Efficacy): 1, 2, 3, 4, 5, 6, 9, 21

Factor 2 (Resilience): 14, 15, 16, 17, 18, 19, 22, 24

Factor 3 (Hope): 7, 8, 10, 11, 12

Factor 4 (Undefined): 13, 20, 23

Verification of the factor structure: CFA was conducted to assess the consistency of the obtained data with the original scale structure of the PCQ-24. For the second phase of the study, a new group of students who had not participated in the initial EFA was recruited. Conducting CFA on a new sample allowed us to avoid the risk of capitalizing on chance and exclude the likelihood of overfitting the model to the random features of a particular sample. The confirmatory factor analysis was performed on a sample of 200 participants, including 174 women and 26 men, with a mean age of 18.52 years.

The Lilliefors-corrected Kolmogorov-Smirnov test and the Shapiro-Wilk test were conducted to evaluate the normality of the data. Normality tests were conducted separately for each scale (efficacy, resilience, and hope), and the results are presented in Table 2.

Table 2 shows that non-normality of distribution was recorded for all three scales, as indicated by both the Kolmogorov-Smirnov test with Lilliefors correction and the Shapiro-Wilk test.

To further test and refine the model, CFA was conducted using the Diagonally Weighted Least Squares (DWLS) method. This method is considered more appropriate when working with ordinal data, such as the Likert scale. It provides more reliable estimates of model parameters when the data distribution deviates from normality.

The obtained modification indices enabled an improved model structure and the redistribution of items from factor 4. Items 21 and 22 were reallocated to factor 2 and factor 3, respectively. Items from factor 4 were reallocated to factor 2. Thus factor 1 (Efficacy) includes items 1, 2, 3, 4, 5, 6, 9 (7 questions); factor 2 (Resilience): 14, 15, 16, 17, 18, 19, 21, 24, 13, 20, 23 (11 questions); factor 3 (Hope): 7, 8, 10, 11, 12, 22 (6 questions). The final structure of the questionnaire is presented in Fig. (1).

As shown in Figure 1, the inter-factor correlations range from 0.80 to 0.85, indicating a significant relationship between the latent variables, while still allowing them to be considered separate constructs. All questionnaire items have high factor loadings, indicating that these items are highly representative of the factors they identify. The exceptions include items 13, 20, and 23, which are reverse items. Given this circumstance, we consider it acceptable that their loadings are in the range of 0.2 to 0.5.

CFA was applied to test the adequacy of the identified model. For the confirmatory factor analysis (CFA), the R software environment (version 4.4.3) was used with the lavaan package (version 0.6-19). The model was estimated using the diagonally weighted least squares (DWLS) estimator and the NLMINB optimization method. Model fit was assessed using the following fit indices: χ², Comparative Fit Index (CFI), Tucker-Lewis Index (TLI), Root Mean Square Error of Approximation (RMSEA), and Standardized Root Mean Square Residual (SRMR). The criteria for acceptable model fit were defined as CFI and TLI ≥ .90, RMSEA ≤ .08 and SRMR ≤ .08 [27]. The fit indices of the final model are presented in Table 3.

As shown in Table 3, the CFI and TLI ˃ 0,95, indicating excellent model fit. The RMSEA and SRMR < 0,08, suggesting an acceptable model fit. Thus, the data obtained indicate a sufficiently good fit of the model to the data. Due to high inter-factor correlations (≈ 0.80-0.85), alternative models were tested to examine a potentially better-fitting structure. In particular, a bifactor model was evaluated, including a general factor and group-specific factors. However, compared with the model presented above, the bifactor model demonstrated substantially weaker fit indices; therefore, its results were not included in the article.

After administering the Kazakh-language version of the questionnaire to the students and collecting the data, we determined the internal consistency of the methodology. To assess the internal consistency of all three scales, two reliability measures, Cronbach's alpha coefficient (α) and McDonald's omega coefficient (ω), were calculated, and these showed acceptable internal reliability. The Cronbach's (α) and McDonald's (ω) reliability coefficients ranged from 0.894 to 0.908. According to accepted psychometric standards, values above 0.80 indicate high coherence and values above 0.70 indicate satisfactory coherence. The internal reliability measures are reflected in Table 4 below.

As shown in Table 4, all scales show high internal consistency. The “Resilience” scale, which includes reverse-coded items, demonstrated a relatively low Cronbach’s alpha (α = 0.628). However, when reliability was re-examined with these items removed, the alpha coefficient increased substantially (α = 0.875), indicating strong internal consistency among the direct items of the scale. These findings may suggest potential issues with the wording of the reverse-coded items. To further examine this assumption, item-total correlations and Cronbach’s alpha values after item deletion were analyzed (see Table 5).

As shown in Table 5, the three reverse-coded items demonstrate negative correlations with the overall scale. Cronbach’s alpha increases when these items are removed from the scale structure, suggesting that the issue lies specifically with these items and, potentially, with their wording.

The inter-item correlation matrix (see Table 6) showed that the reverse-coded items were negatively correlated with the remaining items of the scale.

Thus, the issue may lie within the items themselves, which requires further examination and clarification in future studies.

The process of adapting a psychodiagnostic instrument is similar to the process of creating a new methodology, as the structure of the original questionnaire may undergo significant changes due to cultural influences. Based on the results of exploratory and confirmatory analyses, the structure of the questionnaire was modified (with one of the factors, optimism, excluded). Despite good model fit indices, it is necessary to test the reliability and validity of the final questionnaire to determine how well the instrument fulfills its psychodiagnostic functions.

To assess the reliability of the adapted psychodiagnostic instrument, the test-retest method was used. The sample for reliability testing included 112 individuals (N = 112). The normality of the data distribution was evaluated using the Lilliefors-corrected Kolmogorov-Smirnov test and the Shapiro-Wilk test. The statistical analysis indicated that several scales deviated from a normal distribution. Given these circumstances, Spearman’s rank correlation coefficient was applied to examine the relationship between the variables obtained from the first and second administrations of the questionnaire.

A period of 45 days elapsed between the first and second administration of the questionnaire. It is worth noting that, according to international standards, the interval should be no more than six months and no less than four weeks. The obtained results are shown in Table 7.

Thus, as shown in Table 7, the results of the first and second administrations across all scales, as well as the total score, are significantly correlated, indicating the stability of the measurements and the reliability of the methodology. The results obtained using this methodology are stable over time.

Verification of the convergent validity of the questionnaire. At present, in Kazakhstan, there are no validated instruments available in the Kazakh language to measure the psychological constructs examined in this study. This circumstance complicates the process of assessing the validity of the adapted psychodiagnostic instrument. To address this issue, we selected existing instruments that had previously been adapted for Russian-speaking populations and translated them into Kazakh, allowing bilingual Kazakh participants who are fluent in both Russian and Kazakh to respond to bilingual items. To assess the convergent validity of the psychodiagnostic instrument, the following standardized instruments were selected and translated into Kazakh:

1. The self-efficacy test developed by Maddux and Sheer, adapted by Boyarintseva under the supervision of Krichevsky.
2. The Russian version of the Brief Resilience Scale, adapted by Alfimova and Golimbet.
3. Diagnostics of dispositional optimism: validity and reliability of the Russian Version of LOT-R adapted by Gordeeva et al [28].

The study sample consisted of 277 participants (N = 277). Before conducting the validity assessment, the obtained data were tested for normality of distribution using the Kolmogorov - Smirnov test with Lilliefors correction and the Shapiro - Wilk test. The results are presented in Table 8.

As shown in Table 8, the efficacy and hope scales have a non-normal distribution, while the resilience scale shows only minor deviations from normality. In this regard, Spearman's rank correlation coefficient was used to analyse the relationship between the PCQ results and the instruments presented above. The obtained results are reflected in Table 9.

As shown in Table 9, there are significant positive correlations between the scales of the developed psychodiagnostic instrument and existing standardized measures, indicating that the presented scales effectively measure the stated construct domains. It is important to note that to assess hope, we used the methodology measuring the level of optimism, as the Kazakh language uses overlapping expressions for the concepts of hope (umіt, meaning “hope”) and optimism (omirge umittene qarau, meaning “to look at life with hope”), which may have common contexts of use. The Kazakh language does not have a direct lexical equivalent for the concept of “optimism.” In international literature, as well as in certain dictionaries and academic sources, optimism is often defined in terms of hope. Optimism (from the Latin optimus, meaning “the best”) is commonly understood as “hope for a positive future and confidence that difficulties will end favourably,” or as “a person’s hopefulness or confidence about the success of something or the future” [29]. In Kazakh cultural understanding, hope (“umit”) is viewed as a core element of a positive worldview [30], closely associated with belief (“senim”), goodness or well-being (“zhaksylyk”), and culturally reinforced through blessings (“bata”).

Optimism can be considered an active form of hope, reflecting not only the expectation of a positive outcome but also confidence in its realization. Hope is often associated with positive expectations of the future and belief in a better future. Optimism, in turn, is a broader concept, encompassing not only positive expectations but also a general outlook on life. These linguistic and cultural differences make it reasonable to use optimism as an indicator of hope, given that optimistic people tend to report higher levels of hope. Cross-cultural research in sociology and cultural studies shows that optimism and hope are key factors in shaping individual and social behaviour. Therefore, the use of optimism to assess hope in the Kazakh context can be justified both theoretically and empirically, given their proximity and interconnection in both lexical and psychological aspects. Theoretical justification for selecting optimism as an indicator for hope can thus serve as a confirmation of the interrelation of these concepts in the studied Kazakh culture and psychology.

The final stage in the adaptation of the PCQ involves the identification of test norms. Table 10 below reflects descriptive statistics on the scales.

Based on the empirical data, percentile ranks were calculated to reflect each respondent’s position within the sample according to their score. To simplify the categorisation of the data obtained by each student, quartile values (25th and 75th percentiles) were calculated, which were then used as starting points for the distribution of scores into low, medium, and high levels. No differences were found between men and women. The test norms are shown below in Table 11.

However, the predominance of female participants in the sample suggests that the findings may not be equally generalizable to both genders, as a clear gender imbalance is present. The overrepresentation of women raises the question of how reliable the questionnaire would be when applied to male participants. This limitation highlights the need for future research to test gender invariance and ensure that the instrument demonstrates comparable reliability for both men and women.

Thus, the Kazakh version of the PCQ appears to deviate structurally from the original four-factor model.

3.2. Psychological Capital Indicators in the Kazakhstani Sample

The students in the sample showed the following results across the three scales of Psychological Capital. On the Self-efficacy scale, the majority of participants (51.75%) demonstrated an average level, reflecting a moderate sense of their own competence and ability to achieve their goals. At the same time, 22.5% of students showed a high level, indicating a strong confidence in their abilities, while 25,75% had a low level of effectiveness, which may indicate doubts about their personal abilities and a tendency to self-deprecation.

On the Resilience scale, an average level also predominated (43,5%), while 30,5% of participants showed high scores, indicating a well-developed ability to overcome difficulties and maintain inner balance. At the same time, 26% of students demonstrated a low level of resilience, which may reflect difficulties in adapting to stressful situations and reduced stress tolerance.

On the Hope scale, 88,25% of participants demonstrated a high level, indicating strong goal orientation, the ability to find ways to achieve objectives, and confidence in a successful future. Only a small proportion of students (1,75%) showed a low level of hope, which can be considered an exception within this sample.

Foreign scholars are conducting studies to identify the connection or influence of psychological capital on life satisfaction [31, 32], depression [33], social support, and work engagement [34]. In future research, we also plan to examine the relationships between psychological capital and the specific features of students’ mental health.