Introduction

In mental health research, increasing attention has focused on protective factors, with resilience receiving predominant emphasis. However, a consensus on its operational definition remains lacking1,2. Within this context, the psychological immune system model3 has been introduced, which includes various cognitive, motivational and behavioral dimensions that promote mental well-being, optimal adaptation and frustration tolerance. The model systematizes numerous protective and resource constructs within a coherent theoretical framework, resulting in a multidimensional concept of psychological immunity. The psychological immune system integrates three subsystems comprising 16 factors.

The concept of psychological immunity as an indicator of resilience to stress and adversity has primarily been studied in Hungary. Previous research has shown its positive relationship with mental and physical health3,4,5 life satisfaction4,6 life aspirations5 emotional intelligence3 meaningfulness of life and experience of positive emotions7. By contrast, psychological immunity has been negatively associated with depression5 psychopathological symptoms3,8 and burnout syndrome4. In addition, extensive research has been conducted on individual components of the psychological immune system to date9,10,11,12,13,14,15,16.

The Psychological Immune Competence Inventory (PICI)3 was developed to measure psychological immunity, along with its 16 components and three second-order subsystems. As it facilitates the assessment of complex individual profiles of mental resilience characteristics, it can provide more detailed information than the widely used Brief Resilience Scale (BRS)17. However, unlike the BRS, the PICI requires more time to administer due to the larger number of items. The psychometric properties of the PICI inventory have been validated in Hungary3 and in a pilot study in Slovakia18. Validation data from other countries are still lacking, although translated versions are used internationally5,19.

The primary objective of this study was to examine the psychometric properties of the Slovak adaptation of the Psychological Immune Competence Inventory3,20. An original study by the author3 and a pilot validation study of the Slovak PICI inventory18 served as a background. The pilot study was conducted on a research sample of 213 medical students and confirmatory factor analysis showed an acceptable fit to the data. However, Cronbach’s alpha values indicated that there was insufficient reliability of the three factors in the Slovak version. This problem may be due to the fact that the English version3 was used for translation instead of the original Hungarian version21. Consequently, a translation was conducted from the original Hungarian version and subsequently compared with the existing Slovak translation to improve its psychometric parameters. This process led to the development of the revised Slovak version, which was used in the present study. It was expected that its factor structure would be consistent with its original version and that it would also have adequate psychometric parameters. In addition, secondary objectives included evaluating the invariance of the instrument across different gender, age and education groups, as well as determining its test–retest stability.

Methods

Research sample and data collection

The research sample consisted of 585 individuals from the general population, aged 18 to 50 years (M = 36.25; SD = 8.08). Of these, 370 were women and 214 were men. Participants were reached through an online invitation that contained a link to register for the study. They participated voluntarily, without financial compensation, although receiving their results upon completion may have served as motivation. Data were collected in person from January to June 2024 using computer-assisted self-interviewing (CASI) on tablets and computers. The test battery was hosted on a secure and licensed online platform, REDCap22,23. It consisted of multiple instruments and required between 40 and 90 min to complete. However, this study focused solely on data obtained from the demographic questionnaire and the Psychological Immune Competence Inventory (PICI). Although these instruments were placed at the beginning of the battery, the study also aimed to detect potential careless response style using three specific indicators (response time per item, maximum longstring index and Mahalanobis distance), as described in detail in the Statistical analysis section. After considering the results of these analyzes, no participant was excluded. Given that psychological immunity is conceptualized as a trait-like construct, the long-term stability of the instrument was assessed through a test–retest procedure. Participants were re-contacted 6 to 8 months after the initial assessment and asked to complete only this one questionnaire, hosted on the REDCap platform22,23. A total of 261 participants, including 181 women, participated in the online retest.

The study was performed in line with the principles of the Declaration of Helsinki. Approval was granted by multiple ethics committees: (a) the Ethics Council of the Faculty of Arts, Comenius University in Bratislava (ER/15/2023); (b) Ethical Committee of the Faculty of Medicine, Comenius University and University Hospital in Bratislava, Old Town Hospital (119/2023); (c) Ethical Committee of the Regional Public Health Authority with the seat in Banská Bystrica (approved on April 19, 2024); (d) Independent Ethical Committee of the Banská Bystrica Self-Governing Region (BBSK) for Biomedical Research (approved on May 21, 2024), and (e) the Ethics Committee of L. Pasteur University Hospital in Košice (approved on June 4, 2024). All participants were informed about the aims of the study, and written informed consent was obtained from each individual prior to participation.

Measures

A demographic questionnaire along with the Psychological Immune Competence Inventory (PICI) were used to meet the research objectives.

The psychological immune competence inventory

The PICI Inventory3 is an 80-item instrument designed to assess the global level of psychological immunity, but it also provides a detailed individual profile by examining the test scores in its three subsystems and 16 factors: (a) Approach-Belief Subsystem (Positive Thinking, Sense of Control, Sense of Coherence and Sense of Self-Growth), (b) Monitoring-Creating-Executing Subsystem (Creative Self Concept, Self-Efficacy, Goal Orientation, Problem Solving Capacity, Change and Challenge Orientation, Social Monitoring Capacity, Social Mobilizing Capacity and Social Creation Capacity) and (c) Self-Regulating Subsystem (Synchronicity, Impulse Control, Emotional Control and Irritability Control). Each item is answered by a four-point Likert scale, with 1 indicating “completely does not describe me” and 4 indicating “completely describes me”. The psychometric properties of the Slovak version of the PICI have to date been examined in a pilot study18using a version translated from English3. The pilot study indicated an acceptable fit of the theoretical model to the data. However, the three factors showed insufficient reliability (Sense of Control, Sense of Self- Growth and Impulse Control). Consequently, for the purpose of the present study, we decided to revise the Slovak version by comparing it with the Hungarian translation of the instrument. An expert in psychology, fluent in Hungarian and Slovak, performed the translation. Following this, four psychology specialists convened for a panel discussion to reach consensus on the final version. Subsequently, it was back-translated into English and forwarded to the inventory author for approval. In this study, Cronbach’s alphas for the test/retest are as follows: Global level of Psychological Immunity (0.95/0.96); Approach-Belief Subsystem (0.89/0.91); Monitoring-Creating-Executing Subsystem (0.91/0.93); Self-Regulating Subsystem (0.89/0.91); Positive Thinking (0.82/0.84), Sense of Control (0.60/0.69), Sense of Coherence (0.81/0.85), Sense of Self-Growth (0.73/0.78), Creative Self Concept (0.68/0.75), Self-Efficacy (0.69/0.75), Goal Orientation (0.74/0.80), Problem Solving Capacity (0.77/0.76), Change and Challenge Orientation (0.85/0.86), Social Monitoring Capacity (0.86/0.89), Social Mobilizing Capacity (0.76/0.78), Social Creation Capacity (0.84/0.85), Synchronicity (0.78/0.82), Impulse Control (0.74/0.79), Emotional Control (0.79/0.76) and Irritability Control (0.76/0.78).

Statistical analysis

Data analysis was performed using the statistical software JASP, version 0.18.324. To detect any potential careless response style, three indicators were employed: response time per item, the maximum longstring index and Mahalanobis distance. Thresholds were established at less than two seconds per item25 and a maximum of 20 consecutive identical responses in an 80-item questionnaire26,27.

Subsequently, the descriptive statistics and Cronbach’s alphas were obtained. In order to assess the normality of the data distribution, the Shapiro-Wilk test and Q–Q plot analysis were used. Furthermore, to gain a more comprehensive understanding of descriptive statistics, an analysis of gender differences was performed for each of the PICI factors, as well as a calculation of their correlations with age.

To verify the factor structure of the Slovak version, we replicated the research design used in the pilot study18. Confirmatory factor analysis was performed using structural equation modeling (SEM) and individual items were treated as ordinal variables. For parameter estimation, the diagonally weighted least squares method (DWLS) with robust standard error correction was used. To assess the fit of the model to the research data, the ratio of the chi-square to degrees of freedom was used, along with the following fit indices: RMSEA, SRMR, CFI, TLI and PNFI. Subsequently, the standardized factor loadings and the intercorrelations between the psychological immunity factors were examined. In assessing various aspects of the construct validity for each individual factor, the average variance extracted (AVE) and composite reliability (CR) analyzes were performed. To establish convergent validity, the AVE values have to be ≥ 0.5, while the CR must be equal to 0.7 or greater28.

The invariance testing aimed to determine whether observed differences in test scores could be meaningfully interpreted as reflecting true variations in psychological immunity levels, rather than measurement bias arising from differential functioning of the instrument across groups. The psychometric equivalence of the instrument across frequently examined sociodemographic groups (gender, age, and education) was evaluated using the four-step procedure recommended by Putnick and Bornstein29. This approach involves sequentially fitting models to evaluate configural, metric, scalar, and strict invariance. The process starts with configural invariance estimation using multi-group SEM analysis, with each subsequent model imposing additional equality constraints across the groups. The metric invariance is examined by constraining the factor loadings to be equal across the groups. For scalar invariance assessment, both factor loadings and intercepts are constrained to be equal. Finally, strict invariance measurement involves constraining factor loadings, intercepts, and residuals to be equal across the groups. As a foundational approach, structural equation modeling (SEM) was utilized using the DWLS method. Historically, the focus of the invariance evaluation was on the criterion of χ2 change significance. However, due to its very high sensitivity to even small model deviations in larger samples, some researchers have shifted to alternative fit indices like ΔCFI, ΔRMSEA, or ΔSRMR30,31. To evaluate model deterioration, Chen’s30 guidelines were followed; these recommend a maximum CFI decrease of 0.01, an RMSEA increase of 0.015, and an SRMR increase of 0.03 for metric invariance. For scalar and residual invariance, the same thresholds apply, except for the SRMR index, where the guideline indicates a maximum rise of 0.01.

To determine test–retest reliability, the stability and intraclass correlation coefficients between the first and second measurements were calculated. Due to the high attrition rate (55.38%), we also focused on detecting potential systematic error by using binary logistic regression. First, a dichotomous dependent variable was created based on participation in the retest, with 0 representing droppers and 1 representing stayers. Then, to predict attrition, independent variables were added to the regression model: age, gender, academic degree and test scores of each psychological immunity factor at the first measurement.

Results

Based on the analysis of careless response style indicators, no participants were excluded from the sample. On average, the questionnaire completion time was 12 min and 19 s (SD = 4 min and 50 s), with the mean maximum long string index being 5.37 (SD = 1.71). The shortest completion time recorded was 4 min and 1 s, indicating 3 s per item. The longest sequence of consecutive identical responses was 15 items, with no Mahalanobis distance p-value less than 0.001.

The results of the descriptive statistics, along with the reliability and normality data, are presented in Table 1. In the reliability analysis, the Cronbach alphas ranged from 0.89 to 0.91 for the second-order factors and from 0.60 to 0.86 for the first-order factors. Normality tests indicated a violation of the normal distribution of the data in most factors. However, based on the central limit theorem, the parametric Welch t-test was used to analyze gender differences. The results indicated that men generally scored higher than women in all psychological immunity factors, except for Social Monitoring Capacity. However, the effect sizes of these gender differences were minimal to negligible, with Cohen’s d values ranging from 0.01 (Social Mobilization Capacity) to 0.35 (Emotional Control). Regarding age-related associations, Spearman’s correlations revealed a positive association with all factors of psychological immunity, except for Social Monitoring Capacity and Impulse Control, for which a negative association was found. However, the correlations were also very weak, ranging from 0.01 to 0.24.

Table 1 Descriptive characteristics, alphas and normality for factors of the Slovak PICI.
Full size table

Regarding the results of the confirmatory factor analysis, the initial SEM produced ambiguous findings about the expected factor structure of the inventory (χ2 = 17965.742, df = 3061, χ2/df = 5.869, CFI = 0.912, TLI = 0.909, PNFI = 0.868, RMSEA = 0.091, SRMR = 0.093). While some of the goodness-of-fit indices suggest an adequate model fit to the data (CFI, TLI, PNFI), several other important indices indicate a suboptimal fit (χ2/df, RMSEA, SRMR). However, after applying reasonable modification indices, these parameters significantly improved and indicated an acceptable fit of the model to our data (χ2 = 12571.246, df = 3050, χ2/df = 4.122, CFI = 0.944, TLI = 0.942, PNFI = 0.895, RMSEA = 0.073, SRMR = 0.079, Δχ2 = 5394.496, Δdf = 11, p < 0.001). In the context of the applied modification indices, some covariances were considered: within the Monitoring-Creating-Executing Subsystem (between the Problem Solving Capacity and the Social Creation Capacity; between the Social Monitoring Capacity and the Social Creation Capacity; and between the Social Monitoring Capacity and the Problem Solving Capacity), between two factors within the Self-Regulating Subsystem (the Impulse Control and the Emotional Control), and between items 14 and 62, which both load on the Impulse Control factor and are similar in content, while not sharing conceptual meaning with the other items.

The standardized factor loadings, the average variance extracted (AVE) and the composite reliability (CR) are presented in Table 2 and the intercorrelations in Table 3.

Table 2 Standardized factor loadings and the convergent validity of factors.
Full size table
Table 3 Intercorrelations of factors in the Slovak version of the PICI inventory.
Full size table

The results of the invariance tests are presented in Table 4. Within the context of gender, the psychometric equivalence of the instrument was examined between a group of men (n = 214) and women (n = 370), excluding from the analyzes a single participant who did not identify with either group. Based on acceptable differences in the fit indices, the results support the gender invariance of the Slovak version of the instrument (ΔCFI ranging from − 0.004 to 0.000; ΔRMSEA = 0.001; and ΔSRMR = 0.000 to 0.002). Age invariance was evaluated in three age groups: 18–29 years (n = 123), 30–39 years (n = 254) and 40–50 years (n = 208). The most noticeable deterioration was observed in the CFI fit index when testing for metric invariance (ΔCFI = − 0.012). After rounding, however, this is still regarded as an acceptable indication of the index’s maximum deterioration (ΔCFI = 0.01). Consequently, the differences in the fit indices also support the instrument invariance with respect to age (ΔCFI ranging from − 0.012 to 0.000; ΔRMSEA = 0.000 to 0.007; and ΔSRMR = 0.000 to 0.005). The evaluation of invariance related to education was affected by the composition of the sample, predominantly composed of individuals with higher education levels. Most of the participants achieved an academic degree: Bachelor (n = 41); Master (n = 338) and Ph.D. or higher (n = 78). Of those who had not completed higher education, most had attained at least secondary education with a high school diploma (n = 124), and a very few had solely completed primary school (n = 4). For this reason, it was only possible to assess the psychometric equivalence between two education groups: individuals without a university degree (n = 128) and those who possess a university degree (n = 457). The results showed acceptable differences in the individual fit indices (ΔCFI = − 0.006 to 0.000; ΔRMSEA = 0.001 to 0.003; and ΔSRMR = 0.000 to 0.002), indicating psychometric equivalence between these groups.

Table 4 Invariance measurement results.
Full size table

The test–retest reliability of the instrument was evaluated through the stability correlation coefficients (r), along with the intraclass correlation coefficients (ICC) between the first and second measurements. The results are presented in Table 5, indicating a relatively high stability of the instrument over a period of 6 to 8 months. The stability correlation coefficient for the total score was 0.84, while the second-order factors exhibited a stability ranging from 0.81 to 0.84, and the stability of the first-order factors ranged from 0.65 to 0.83. Similarly, intraclass correlation coefficients showed comparable results, as the ICC of total psychological immunity was 0.84, the ICC coefficients for second-order factors ranged from 0.82 to 0.84, and for first-order factors from 0.63 to 0.81.

Table 5 Test–retest reliability of the instrument over a period of 6 to 8 months.
Full size table

To detect possible attrition bias, binary logistic regression was used. Several predictor variables were included to estimate the probability of attrition: first-measurement test scores on individual psychological immunity factors and selected demographic variables, such as gender, age and academic degree. The results indicated that attrition was not significantly predicted by age, academic degree or initial scores on 16 psychological immunity factors (χ2(2) = 20.997, p = 0.337), as illustrated in Table 6. Gender was the only significant variable identified, and women showed a higher probability of participating in the retest. This may represent a limitation in test-retest reliability. Otherwise, no significant systematic error due to attrition was observed.

Table 6 Results of binary logistic regression to identify attrition bias.
Full size table

Discussion

The primary objective was to conduct a psychometric evaluation on the Slovak adaptation of the PICI inventory. Emphasis was placed on verifying its factor structure, along with analyzing the convergent validity and internal consistency of its scales. In addition, an invariance was examined across groups differentiated by gender, age and education, and was supplemented by an assessment of its test-retest reliability.

In the present study, a revised Slovak version of the instrument20 was employed, developed through a comparison with the Hungarian original21. The revision appears to result in a considerable enhancement in the internal consistency of its scales, as all Cronbach’s alpha values are at moderate to very high levels, without any indication of redundancy. Furthermore, the findings indicate adequate convergent validity of the psychological immunity scales, as evidenced by their average variance extracted (AVE) and composite reliability (CR). Although some first-order factors did not meet the recommended AVE threshold of ≥ 0.5, this criterion is generally regarded as conservative. Convergent validity can still be regarded as sufficient when the factor achieves a CR value of ≥ 0.732,33. Given that all factors have satisfactory composite reliability, we consider the convergent validity of these factors to be sufficient.

The results of the confirmatory factor analysis are consistent with the findings of the pilot study18 and serve to support the original factor structure. Although the fit indices are slightly deteriorated in comparison with the pilot CFA, this can be readily corrected by adding a few reasonable modification indices, resulting in an acceptable model fit to our data. With regard to one of the reasons for misspecification, four items with weak factor loadings were repeatedly identified in this research sample as well as in the pilot study: 34 (“I never trust fate or luck to solve my problems.” in the Sense of Control factor), 14 (“I am the type of person that says the first thing that comes to my mind” in the Impulse Control factor), 77 (“I have often started a new project before I have finished an earlier one.” in the Goal Orientation factor) and 37 (“Over the last year, my personality has not changed as I wished at all.” in the Sense of Self-Growth factor). Our item-level factor loading results cannot be compared with studies from other countries due to the unavailability of reporting these data. However, it is possible that these four items may have culturally specific meanings. As this requires further investigation, caution is recommended in interpreting the test results of these four psychological immunity scales with respect to the fact that they contain an item with low factor loading.

Invariance measurement indicated that psychometric properties were similar across different gender, age and education groups. However, the results might be compromised due to smaller sample sizes within specific age and education categories. In particular, the group of people aged 18 to 29 and the group of those without an academic degree were under-represented. To the best of our knowledge, these results represent the first invariance test of the PICI instrument, hence it was not possible to compare the results with other studies. However, research investigating the invariance of instruments that measure a closely related concept of resilience suggests configural, metric, and scalar invariance across gender, education, and age groups34,35,36,37. These observations align with our outcomes. However, it should be noted that, since these instruments have a different factor structure and measure a slightly different construct, these findings are only indicative of agreement on the feasibility of making meaningful comparisons of psychological immunity between these groups.

The instrument demonstrates relatively high test-retest reliability after a period of 6 to 8 months. In the original study3,21 the instrument’s stability was examined after a two-week period, indicating a high degree of test-retest reliability. These findings align with the theoretical framework defining psychological immunity as a dispositional variable3,38,39. Conversely, longitudinal research with breast cancer patients receiving psychological treatment40 demonstrated that psychological immunity can also vary situationally. The authors of the study noted that scores on global psychological immunity and its five factors (Positive Thinking, Sense of Control, Sense of Coherence, Emotional Control, and Impulse Control) increased during treatment and at 1-year follow-up. This finding suggests the potential mobilization of specific internal resources in coping with a life-threatening illness and subsequent post-traumatic growth. However, it is important to note that these results could be influenced by biases arising from post-medication relief and the study’s relatively small sample size. The findings of the present study also revealed a statistically significant test-retest difference for a single subscale (Impulse Control), although the effect size was negligible. An analogous uncertainty surrounds the notion of resilience. It remains unclear whether this is a stable trait or a dynamic state. Some researchers argue that it is a trait, relatively stable across various contexts41. Nonetheless, some of them acknowledge that certain components of resilience are considered to be state-specific42. According to others, resilience is perceived as the outcome of actions intended to cope or bounce back from challenging circumstances, mainly relying on skills gained through experience43. However, a critique exists against the dichotomy of trait versus state44. Critics suggest that resilience should be understood as a transactional and dynamic interaction between individual and environmental factors. This perspective emphasizes the processual nature of resilience45. To gain a deeper understanding of the nature of psychological immunity, future research should also explore its development across different ontogenetic stages and situational contexts.

Limitations

The limitations of the research are largely due to the cross-sectional design reliant on self-evaluation and the composition of the research sample consisting of participants who joined voluntarily. The results might also be influenced by the higher education level of the participants, as well as by the under-representation of young people in the 18 to 29 age-group. Additionally, an observed limitation is the gender imbalance in the retest sample, which leans toward a predominance of women. However, given that the mean test scores show only negligible differences between the genders, we believe that gender bias did not have a major impact on the test-retest results. Furthermore, no other systematic bias was detected using binary logistic regression.

Conclusion

The Slovak adaptation of the Psychological Immune Competence Inventory (PICI) serves as an effective tool for gaining a comprehensive understanding of different dimensions of mental resilience and is equipped with appropriate psychometric properties. Although the model of the psychological immune system is complex, it is unlikely to encompass a complete set of protective and resource-related personal competences. However, it provides the groundwork for a deeper investigation of the processes and systemic interactions that underpin mental resilience, as well as its links to other variables. The self-report format of the PICI suggests potential areas for improvement. For example, it may be beneficial to develop a version of the rating scale for use by external evaluators in the future. Nonetheless, the Slovak PICI appears to be applicable in a wide range of contexts, providing a comprehensive insight into the resource competencies of an individual.