A validation study of the Occupational Depression Inventory in Poland and Ukraine

This study examined the psychometric and structural properties of the Polish and Ukrainian versions of the Occupational Depression Inventory (ODI). We relied on two samples of Polish employees (NSample1 = 526, 47% female; NSample2 = 164, 64% female) and one sample of Ukrainian employees (NSample3 = 372, 73% female). In all samples, the ODI exhibited essential unidimensionality and high total-score reliability (e.g., McDonald’s omegas > 0.90). The homogeneity of the scale was strong (e.g., 0.59 ≤ scale-level Hs ≤ 0.68). The ODI’s total scores thus accurately ranked individuals on a latent occupational depression continuum. We found evidence of complete measurement invariance across our samples, a prerequisite for between-group comparisons involving observed scores. Looking into the criterion validity of the ODI, we found occupational depression to correlate, in the expected direction, with resilience and job-person fit in six areas of working life—workload, control, rewards, community, fairness, and values. The prevalence of occupational depression was estimated at 5% in Sample 1, 18% in Sample 2, and 3% in Sample 3. Our findings support the use of the ODI’s Polish and Ukrainian versions. This study adds to a growing corpus of research suggesting that the ODI is a robust instrument.


ODI
The ODI comprises nine core symptom items (rated from 0 for "never or almost never" to 3 for "nearly every day") and a supplementary question gauging turnover intention 5 .The nine core symptom items assess the nine diagnostic symptoms of major depressive disorder described in the DSM-5 9 .The symptoms are assessed within a two-week time window, consistent with the DSM-5.The ODI is intended to be used based on its total score (dimensional approach) and/or the diagnostic algorithm that accompanies the instrument (categorical approach).The diagnostic algorithm can be found in the form of an SPSS syntax in Supplemental Materials 1 and 2. The diagnostic algorithm is described in detail in the inaugural ODI paper 5 .The ODI was translated into Polish and Ukrainian using a back-translation method 27 .Native speakers first translated the original English version into Polish and Ukrainian.Then, different native speakers translated the Polish and Ukrainian versions back into English.We did not detect significant discrepancies between the original and back-translated English versions.The ODI's Polish and Ukrainian versions are displayed in Table 1 and additionally provided together with instructions to respondents in Supplemental Materials 1 (for Poland) and 2 (for Ukraine).Descriptive statistics are available in Table 2.The prevalence of occupational depression was 5% in Sample 1, 18% in Sample 2, and 3% in Sample 3.

Resilience scale
We used the Resilience Scale to assess resilience 28 .The Resilience Scale comprises 25 items rated from 0 for "definitely not" to 4 for "definitely." The Resilience Scale covers various aspects of resilience, such as persistence in action, openness to novelty, reliance on humor, tolerance to failure, or optimism.Resilience items include: "I easily adapt to new situations"; "I consider myself a strong person"; "My life has meaning." Cronbach's α was 0.93.Only Sample 2 completed the Resilience Scale.

Data analysis
Consistent with previous ODI studies, we examined the factorial structure of the ODI using ESEM bifactor analysis 30,31 .We considered two specific factors in addition to the general factor (Fig. 1), on account of the anhedonic-somatic and dysphoric symptom items populating the scale.In keeping with past research, we treated the items as ordinal, employed the weighted least squares-mean and variance adjusted-(WLSMV) estimator, and relied on a target rotation.We inspected the factor loadings on the general and specific factors and, as recommended, we computed the explained common variance (ECV) index to further estimate the importance of the general factor 31,32 .An ECV index exceeding 0.80 is suggestive of essential unidimensionality.The higher the ECV index, the greater the role of the general factor in accounting for the common variance extracted.We conducted our factor analyses in Mplus 8.7 33 .
We looked into the ODI's homogeneity using the Mokken package version 3.0.6 34in R version 4.2.0 35 .Homogeneity refers to the extent to which a scale's items hierarchically align on a single dimension.The hierarchy concerns item difficulty, which, in this context, refers to the probability that an item will be endorsed by respondents.Homogeneity is indexed by H coefficients.As per commonly applied standards 36  We investigated the total-score reliability of the ODI based on McDonald's omega, Cronbach's α, Guttman's lambda-2, and the Molenaar-Sijtsma statistic.We relied on the Pearson product-moment correlation coefficient to examine the relationships between the ODI and our other measures of interest and inquire into the ODI's criterion validity.
We examined the ODI's cross-sample measurement invariance focusing on configural invariance (equivalence in factor structures), weak invariance (equivalence in factor loadings), strong invariance (equivalence in item thresholds), and strict invariance (equivalence in item residuals).The equivalence constraints are cumulative.We relied on common standards for detecting deviations from measurement invariance.Invariance violations were signaled by increases in the RMSEA exceeding 0.015 and decreases in the CFI and TLI exceeding 0.010 37,38 .In a first analysis, we scrutinized measurement invariance across our three samples.In a second analysis, we scrutinized measurement invariance across our three samples and the sample used in the original validation study of the ODI 5 .Involving original validation samples in subsequent validation studies is recommended during scale development 39,40 .

Homogeneity and reliability
The results of the homogeneity analysis are summarized in Table 3.The ODI demonstrated strong homogeneity.The scale-level H coefficient reached 0.68 with a standard error of only 0.02.All item-level H coefficients largely exceeded the 0.30 threshold, and none of the pairwise H coefficients were low.The most frequently endorsed item was Item 4 (fatigue/loss of energy), and the least frequently endorsed item was Item 9 (suicidal ideation).

Homogeneity and reliability
The homogeneity of the ODI was strong (scale-level H = 0.59, standard error = 0.03), with no problematic H values at either item or pairwise levels (Table 3).Again, the least difficult item was Item 4 (fatigue/loss of energy), and the most difficult item was Item 9 (suicidal ideation).As was the case in Sample 1, the ODI showed high total-score reliability.McDonald's omega, Cronbach's α, Guttman's lambda-2, and the Molenaar-Sijtsma statistic were ≥ 0.90 (see Table 3).

Criterion validity
Occupational depression correlated, in the expected direction, with resilience and job-person fit in six areas of working life-workload, control, rewards, community, fairness, and values (Table 4).Moderate to large correlations were observed.Absolute rs ranged from 0.24 (for values) to 0.49 (for workload).

Homogeneity and reliability
The results of the homogeneity analysis are summarized in Table 3.The ODI demonstrated strong homogeneity.The scale-level H coefficient reached 0.60 with a standard error of only 0.02.All item-level H coefficients largely exceeded the 0.30 threshold, and none of the pairwise H coefficients were low.The most frequently endorsed item was Item 4 (fatigue/loss of energy), and the least frequently endorsed item was Item 9 (suicidal ideation).

Measurement invariance
As can be seen from Supplemental Material 3, we found measurement invariance to hold across our three samples considered separately as well as across our three samples and the original validation sample of the ODI.As we added constraints from configural invariance to strict invariance, none of the fit indices showed problematic alterations.As an illustration, CFI never decreased by more than 0.002, and TLI never decreased by more than 0.003.Complete invariance was thus reached.

Discussion
The ODI reflects a renewed approach to job-related distress and is increasingly used across countries and linguistic communities.The present study inquired into the psychometric and structural properties of the ODI's Polish and Ukrainian versions.Such an inquiry is important to ascertain whether Polish-and Ukrainian-speaking occupational health specialists can confidently use the instrument.We relied on up-to-date statistical techniques to accomplish our research goals.

Main findings
In all samples, the ODI met the requirements for essential unidimensionality.These results are consistent with the findings obtained in all previous ODI studies 5,6,[11][12][13]16,17 . Homogneity analysis indicated that the items of the ODI hierarchically align on a single dimension, in keeping with ODI studies conducted in other geographic and linguistic contexts 6,[11][12][13]16,17 .These findings suggest that the ODI's total score accurately ranks respondents on the latent continuum underlying the scale.We found the ODI to exhibit strong total-score reliability.This finding was documented based on four different indicators.It is of note that the ODI displays strong total-score reliability despite covering nine symptoms and showing no explicit redundancy in the content of its items.It is well-known that, in many scales, total-score reliability (typically indexed by Cronbach's α) is inflated by the repetition of nearly identical, virtually interchangeable questions 41,42 .The ODI's total-score reliability is unlikely to be biased upward by item synonymy because each of the ODI's items covers a specific symptom of major depression. More probably the unity of the item set reflects clinically meaningful bonds among the symptoms assessed.
Speaking to the criterion validity of the ODI, we found occupational depression to correlate negatively with resilience and job-person fit in six areas of working life-workload, control, rewards, community, fairness, and values.The present study is the first to examine the nomological network of occupational depression in relation to these variables.Our results are consistent with the findings pertaining to general depressive symptoms and disorders.Depression has been linked to a lack of resilience in past research 43 .Leiter and Maslach (2004)  reported small to large correlations between workload, control, rewards, community, fairness, and values as operationalized by the Areas of Worklife Scale and burnout-an entity known to overlap with (occupational) depression 7,44,45 .
The availability of uncomplicated, yet dependable, assessment tools is of paramount importance in identifying deteriorated health and impaired functioning in the workplace.The Polish and Ukrainian versions of the ODI can serve as both a research asset and a signaler to take preventive and interventional measures in organizations.

Study limitations and strengths
The present study has at least three limitations.First, we relied on a cross-sectional design, which prevented us from addressing properties such as test-retest reliability.Second, the alpha reliabilities of some of the subscales of the Areas of Worklife Scale were in the 0.70s, a range that is considered barely adequate in the context of basic research 46 .Our alpha reliabilities, however, comport with those documented by the creators of the Areas of Worklife Scale 20 .Third, we did not inquire into the ODI's discriminant validity vis-à-vis attribution-free measures of depression.Fortunately, the ODI's discriminant validity has been examined in several past ODI studies based on a variety of depression scales 5,16,17 .
As for its strengths, the present study involved three different samples, thus incorporating a replication component within its design.In addition, the study relied on advanced statistical techniques attached to classical test theory and item response theory.The techniques employed allowed us to conduct both synoptic and granular analyses of the data.Combined with the use of three different samples, this modus operandi increases the conclusiveness of the study findings.That the ODI showed complete measurement invariance across our samples suggests that differences in observed scores reflect genuine symptom variations rather than artifacts related to idiosyncratic utilization of the scale.Reaching complete measurement invariance is highly important for a measure.Strict invariance is indeed a prerequisite for between-group comparisons involving observed scores 37,38 .
We note that the Ukrainian data were collected in April 2023, i.e., in war times.It is likely that the war context influenced workers and organizations in the country.The nature of that influence on occupational depression is, however, unclear.On the one hand, the war context and its tragedies may have produced a "relativism effect" leading individuals to be less sensitive to work-related hassles and stressors.On the other hand, at least in some occupations, job demands may have increased dramatically, and job resources may have decreased considerably, setting a perfect floor for job strain and mental health issues.Unfortunately, our study does not allow us to clarify such questions.While it is tempting to call for further research, conducting further research is challenging in the current circumstances.

Conclusions
The Polish and Ukrainian versions of the ODI exhibit robust psychometric and structural properties.Given the estimated health and economic cost of job-related distress, it is crucial for occupational health specialists to assess the phenomenon reliably and validly.In view of its characteristics, and because it is available free of charge, the ODI has the potential to help researchers, practitioners, and public health decision-makers address job-related distress effectively.

Figure 1 .
Figure 1.Exploratory structural equation modeling bifactor structure under examination.Solid lines are indicative of target loadings.Two specific factors are considered on account of the anhedonic-somatic and dysphoric symptom items populating the Occupational Depression Inventory.OD general Occupational Depression factor, ANH-SOM Anhedonic-Somatic specific factor, DYS Dysphoric specific factor, ODI1 anhedonia, ODI2 depressed mood, ODI3 sleep alterations, ODI4 fatigue/loss of energy, ODI5 appetite alterations, ODI6 feelings of worthlessness, ODI7 cognitive impairment, ODI8 psychomotor alterations, ODI9 suicidal ideation.

Table 1 .
Polish and Ukrainian versions of the occupational depression inventory (ODI).The full ODI form ( , homogeneity is considered weak if 0.30 ≤ H < 0.40, moderate if 0.40 ≤ H < 0.50, and strong if H ≥ 0.50; a scale-level H coefficient below 0.30 suggests that the scale of interest cannot be regarded as unidimensional.Pairwise H coefficients should be > 0; item-level H coefficients should be > 0.30.

Table 3 .
Homogeneity analysis of the Occupational Depression Inventory.H scale-level H, H i item-level H, SE standard error, CI confidence interval.None of the pairwise H coefficients were low (≥ 0.54 in Sample 1; ≥ 0.42 in Sample 2; ≥ 0.32 in Sample 3).

Table 4 .
Criterion Ns vary from 140 to 164 due to missing values; there were no missing values for occupational depression (or any of the items of the Occupational Depression Inventory).CI confidence interval.