The German emotion beliefs questionnaire: factorial validity, measurement invariance, and descriptive reference values in a population-representative sample.
BMC psychology · 2026
Abstract originale
Abstract Background Beliefs about emotions are considered a central mechanism of emotion regulation and mental health [1]. The 16-item Emotion Beliefs Questionnaire (EBQ; [2]) assesses beliefs about the usefulness and controllability of positive and negative emotions. While the original study supported a three-factor model, subsequent population-based studies favored the theoretically proposed four-factor structure. Although two German-language adaptations exist, neither has been validated in a population-representative sample, and descriptive reference values, evidence for measurement invariance across key sociodemographic groups, and systematic investigations of sociodemographic correlates of emotion beliefs are still lacking. Methods The German version of the EBQ [3] was administered to a representative sample of the German general population ( N = 2368; mean age = 50.3; SD = 17.13; range: 18–93; 44.7% men, 55.2% women). Confirmatory factor analyses (MLR) were conducted to compare competing three- and four-factor models. Measurement invariance across gender, age, and educational level was examined and descriptive reference values were calculated. Results Although the four-factor model fit better than the three-factor solution on all indices examined, the two controllability factors were very highly correlated ( r = .90) and showed limited discriminant validity, indicating a largely valence-independent controllability dimension. Bifactor analyses supported interpretation of the total score (ω_H = 0.792, ECV = 0.603), and the EBQ further demonstrated good psychometric properties. Measurement invariance was supported up to the strict level across age and gender, and approximately at the scalar level across education. Educational attainment showed the most consistent associations with emotion beliefs across the total score and all subscales. Descriptive percentile reference values are provided by age group and gender. Conclusions The German EBQ demonstrated satisfactory psychometric properties in a representative adult sample. Population-based descriptive reference values may facilitate research comparisons and preliminary clinical contextualization of individual scores. The findings support a largely valence-independent controllability dimension and highlight educational attainment as a consistent sociodemographic correlate of emotion beliefs.
Parole chiave
Leggi il testo integrale originale
Testo estratto dalla pubblicazione: sono esclusi immagini, media, supplementi e parti soggette a diritti distinti.
The German emotion beliefs questionnaire: factorial validity, measurement invariance, and descriptive reference values in a population-representative sample
Background
Beliefs about emotions are considered a central mechanism of emotion regulation and mental health [1]. The 16-item Emotion Beliefs Questionnaire (EBQ; [2]) assesses beliefs about the usefulness and controllability of positive and negative emotions. While the original study supported a three-factor model, subsequent population-based studies favored the theoretically proposed four-factor structure. Although two German-language adaptations exist, neither has been validated in a population-representative sample, and descriptive reference values, evidence for measurement invariance across key sociodemographic groups, and systematic investigations of sociodemographic correlates of emotion beliefs are still lacking.
Methods
The German version of the EBQ [3] was administered to a representative sample of the German general population (N = 2368; mean age = 50.3; SD = 17.13; range: 18–93; 44.7% men, 55.2% women). Confirmatory factor analyses (MLR) were conducted to compare competing three- and four-factor models. Measurement invariance across gender, age, and educational level was examined and descriptive reference values were calculated.
Results
Although the four-factor model fit better than the three-factor solution on all indices examined, the two controllability factors were very highly correlated (r = .90) and showed limited discriminant validity, indicating a largely valence-independent controllability dimension. Bifactor analyses supported interpretation of the total score (ω_H = 0.792, ECV = 0.603), and the EBQ further demonstrated good psychometric properties. Measurement invariance was supported up to the strict level across age and gender, and approximately at the scalar level across education. Educational attainment showed the most consistent associations with emotion beliefs across the total score and all subscales. Descriptive percentile reference values are provided by age group and gender.
Conclusions
The German EBQ demonstrated satisfactory psychometric properties in a representative adult sample. Population-based descriptive reference values may facilitate research comparisons and preliminary clinical contextualization of individual scores. The findings support a largely valence-independent controllability dimension and highlight educational attainment as a consistent sociodemographic correlate of emotion beliefs.
Supplementary Information
The online version contains supplementary material available at https://doi.org/10.1186/s40359-026-05837-z .
Background
Emotional experiences constitute a fundamental and pervasive component of human functioning, with evidence indicating that individuals report being in an emotional state up to 90% of the time [4]. Given this prevalence, it is hardly surprising that people develop varied beliefs on what emotions are, what purpose they serve, and to what extent they can be controlled [5, 6]. According to Ford and Gross [1], these beliefs can be conceptualized along two primary dimensions: the perceived goodness of emotions (i.e., to what extent emotions are considered valuable or helpful) and their perceived controllability (i.e., to what extent emotions can be consciously altered). From a clinical perspective, beliefs about emotions are highly relevant, as they shape both the motivation and implementation of emotion regulation processes [1]. Beliefs about the goodness (later operationalized as usefulness by [2]) of emotions shape the appraisal of emotional states as desirable or undesirable and thus influence which regulation strategies individuals are motivated to employ, for instance, whether negative emotions are accepted or suppressed. Research shows that perceiving emotions as undesirable promotes avoidance-based strategies [7]. Beliefs about controllability, in turn, determine whether regulation is initiated, how much effort is invested, and whether regulation attempts are sustained or abandoned. Empirically, beliefs regarding the uncontrollability of emotions are associated with less frequent use of adaptive strategies such as cognitive reappraisal and lower self-efficacy in emotion regulation [8, 9]. This influence on emotion regulation represents a central mechanism through which emotional beliefs are linked to mental health. Targeting dysfunctional emotion beliefs represents a promising avenue for clinical interventions, as modifying these beliefs may facilitate the acquisition and effective use of adaptive regulation strategies and, in turn, improve mental health [10, 11].
To provide a psychometric assessment of the belief dimensions proposed by Ford and Gross [1], Becerra et al., [2] developed the Emotion Beliefs Questionnaire (EBQ). Earlier scales, for example the Implicit Theories of Emotions Scale (ITES; [12]), the Beliefs about Emotions Scale (BES; [13]), or the Emotion and Regulation Beliefs Scale (ERBS; [14]) did not simultaneously meet all three key criteria outlined by the authors: (1) the distinction between beliefs about controllability and usefulness, (2) assessment at the superordinate level (i.e., beliefs about emotions in general rather than about one’s own emotion regulation self-efficacy), and (3) differentiation between positive and negative emotions. The EBQ addresses these gaps by assessing beliefs about the controllability and usefulness of positive and negative emotions across four subscales, with higher scores reflecting stronger endorsement of maladaptive beliefs. However, in the original study using an Australian sample (N = 161), the four-factor model showed signs of empirical underidentification, as the two controllability subscales were perfectly correlated, rendering a three-factor solution the best-fitting viable model. Accordingly, the results indicated a three-factor structure consisting of General Controllability, Negative Usefulness, and Positive Usefulness, with an overarching general factor [2]. In contrast to these initial findings, international validation studies in American [15], Iranian [16], Italian [17] and Polish [18] samples provide support for the originally proposed four-factor model based on fit indices. The EBQ has been shown to demonstrate acceptable to good internal consistency [2, 15–18], satisfactory test-retest reliability [16] as well as measurement invariance across age, education and gender categories [15–18]. EBQ scores show the expected associations with difficulties in emotion regulation [17], emotional reactivity [18] and lesser use of adaptive emotion regulation strategies [19]. Across studies, higher EBQ scores are consistently associated with elevated symptoms of depression, anxiety, and stress [15, 17, 18, 20]. Mazidi et al., [21] used the EBQ to identify six emotion belief profiles in a clinical sample, with the two most maladaptive profiles exhibiting the greatest psychopathology and emotion regulation difficulties.
Two independent German-language adaptations of the Emotion Beliefs Questionnaire (EBQ) have been developed using different translation approaches. The first adaptation by Biel et al., [3] employed a forward-backward translation procedure conducted by native German speakers. In a sample of psychotherapy trainees (N = 104), the authors replicated the original three-factor structure using exploratory factor analysis. Internal consistency was good for the total scale (α = 0.87) and the General Controllability subscale (α = 0.85), and acceptable for Negative Usefulness (α = 0.76) and Positive Usefulness (α = 0.70). Evidence for concurrent validity was provided by associations with emotion acceptance and symptoms of depression, anxiety, and stress. Beyond its initial validation, the Biel et al., version has already demonstrated clinical relevance, supporting its conceptualization, from a psychodynamic perspective, as a measure of affective mentalization: In a prospective panel study, baseline EBQ scores predicted somatic symptom-related distress at follow-up beyond autoregressive effects, with findings replicated in psychosomatic outpatient samples [22]. Further evidence comes from a multi-study investigation linking EBQ-assessed affective mentalization to depression. Higher EBQ scores were associated with greater depression severity in a large psychosomatic sample, and longitudinal analyses in a general population sample indicated that baseline EBQ scores predicted subsequent depressive symptoms. Additionally, an experimental study showed that a targeted intervention improved affective mentalization and reduced induced depressive states [23]. Together, these findings support the clinical sensitivity and predictive validity of the Biel et al., adaptation.
A second German validation by Gutzweiler and Grüning [24], based on an independent translation, examined a larger sample of students and young adults (N = 348) using confirmatory factor analysis. Results favored a four-factor model, which outperformed competing models, including the three-factor model. Internal consistency ranged from acceptable to good (α = 0.68-0.79; ω = 0.70-0.80), and construct validity was supported by expected associations with emotion regulation, emotional reactivity, self-efficacy, and psychopathology. The measure demonstrated largely strict measurement invariance across gender as well as across levels of emotional reactivity and self-efficacy [24]. Taken together, the two German adaptations reflect the broader pattern of cross-national discrepancies in EBQ factor structure. While the exploratory three-factor solution reported by Biel et al., [3] aligns with the original Australian findings, the confirmatory four-factor solution identified by Gutzweiler and Grüning [24] is consistent with subsequent validations in diverse cultural contexts.
Several limitations remain with respect to the Biel et al., version, despite its promising clinical applications. The initial validation relied on a small, young, and predominantly female convenience sample, limiting generalizability to the broader German population. Crucially, population-based reference values are currently lacking, which restricts the interpretability of individual scores. Reference values are essential for the meaningful (clinical) application of self-report instruments, as they allow individual scores to be interpreted relative to a defined reference population. While such norms have been established for the English [15] and Polish [18] versions, no population-based reference values are currently available for the German EBQ. In addition, measurement invariance across key sociodemographic variables remains insufficiently examined in German samples as German validation studies have not systematically tested invariance across age or educational levels. This gap is theoretically relevant, as emotion-related beliefs and regulation strategies are known to vary across the lifespan. According to socioemotional selectivity theory, age-associated shifts in motivational priorities influence emotion regulation goals and preferences [25], potentially affecting how EBQ items are interpreted across age groups. Furthermore, the role of sociodemographic factors such as education and socioeconomic status (SES) in shaping emotion beliefs remains largely unexplored. Existing evidence suggests that lower SES is associated with greater anxiety and emotional distress, and that SES moderates the effectiveness of adaptive emotion regulation strategies such as cognitive reappraisal [26]. Given the close link between emotion beliefs, appraisal processes, and perceived regulatory utility [1], it is plausible that SES-related differences in regulatory outcomes are partly reflected in differences in emotion beliefs. However, this question has not yet been systematically addressed within the EBQ framework.
Against this background, the present study aims to further validate the German EBQ [3] in a large, population-representative sample of German-speaking adults. Specifically, the study seeks to (1) examine the factorial validity by comparing three- and four-factor models using confirmatory factor analysis, (2) test measurement invariance across key sociodemographic variables, including gender, age, and educational background, (3) investigate associations between emotion beliefs and sociodemographic characteristics, and (4) provide population-based reference values to support both research and clinical application by enabling the interpretation of individual scores relative to the general population.
Methods
Subjects
Data were acquired between March and June 2025 by a survey research institute (USUMA, Berlin, Germany) as part of a broader German household survey. A multistage random sampling procedure was used: proportionally to the distribution of private households, Germany was divided into 258 sampling areas covering all regions. Within these areas, households were selected using a random-route procedure, and target persons within households were randomly selected. In total, N = 6663 households were contacted, approaching a representative sample of the German general population. N = 2556 subjects aged ≥ 16 years gave written informed consent to participate, resulting in N = 2522 analyzable interviews (effective response rate: 37.9%). The survey consisted of an interviewer-administered section designed to collect sociodemographic data and a self-administered questionnaire. In our analysis, participants < 18 years (N = 37) were excluded. Furthermore, subjects with missing items in the EBQ (N = 117) were also excluded, resulting in a final sample of N = 2368.
Measure
The Emotion Beliefs Questionnaire [2] is a 16-item self-report questionnaire assessing beliefs about the controllability (e.g., whether emotions can be regulated) and usefulness (e.g., whether emotions are helpful or harmful) of positive and negative emotions across four subscales. Items are rated on a seven-point Likert scale (1 = strongly disagree to 7 = strongly agree). Higher scores indicate stronger endorsement of maladaptive beliefs regarding the uncontrollability and lack of usefulness of emotions. Subscale scores and a total score are calculated by summing the corresponding items, resulting in a total EBQ score ranging from 16 to 112. The items of the EBQ and their corresponding subscales are presented in Table 2. For the present study, the German version of the EBQ validated by Biel et al., [3] was used. The translation followed the European Social Survey’s Translation Guidelines, involving multiple independent translations. The German EBQ items are provided in Supplement 1.
Statistical analysis
Statistical analysis was carried out with R version 4.5.1. (R Core Team [27]), primarily using lavaan, semTools, psych, car, lmtest, and sandwich. To examine the factor structure of the Emotion Beliefs Questionnaire, confirmatory factor analyses (CFA) using a robust maximum likelihood estimator (MLR) were conducted, in which a one-factor, a three-factor, and a four-factor model were tested. The four-factor model comprised the dimensions Negative Controllability, Positive Controllability, Negative Usefulness, and Positive Usefulness, each with four items per factor. In the three-factor model, the two controllability dimensions were combined into a single factor, while the one-factor model assumed a general factor across all items. Model fit was assessed using scaled Chi-square (χ²), robust Root Mean Square Error of Approximation (RMSEA), robust Comparative Fit Index (CFI), robust Tucker-Lewis Index (TLI), Standardized Root Mean Square Residual (SRMR) and Bayesian information criterion (BIC). Model fit was evaluated according to established cut-off criteria [28]. Nested models were compared using the Satorra-Bentler scaled difference test [29]. Internal consistency was assessed using Cronbach’s alpha (α) and McDonald’s omega (ω_total). To evaluate the interpretability of the EBQ total score despite the multidimensional factor structure, we additionally fitted a bifactor model comprising one general factor loading on all 16 items and three orthogonal group factors (Controllability, Negative Usefulness, Positive Usefulness) using MLR. Omega hierarchical (ωH), explained common variance (ECV), percentage of uncontaminated correlations (PUC), and factor-score determinacy (FD) were computed from the standardized solution to assess the strength and interpretability of the general factor. Omega hierarchical subscale (ωHS) was calculated to quantify reliable subscale-specific variance beyond the general factor [30]. The corresponding ωH and ωHS coefficients were additionally calculated from the unstandardized bifactor solution for the unweighted raw-item sum scores used in the descriptive reference values and regression analyses. As a robustness check, we additionally fitted a bifactor model with four orthogonal group factors, separating Negative and Positive Controllability. Following established recommendations, ωH values of 0.70-0.80 or higher [31] and FD ≥ 0.90 were considered indicative of a well-defined general factor. ECV and PUC were interpreted jointly rather than against fixed cut-offs: Reise et al., [32] proposed that when PUC falls below 0.80, ECV > 0.60 in combination with ωH > 0.70 indicates that the degree of multidimensionality present does not preclude interpreting the instrument as primarily reflecting a single common dimension.
Discriminant validity of the EBQ dimensions was assessed using latent factor intercorrelations, the Fornell-Larcker criterion, and the heterotrait-monotrait ratio (HTMT; [33]). The Fornell-Larcker criterion compares the average variance extracted (AVE, the mean proportion of item variance accounted for by its factor) with the correlations between factors. Discriminant validity is supported when the square root of the AVE for each factor exceeds its correlations with all other factors, indicating that a factor shares more variance with its own indicators than with any other construct. The HTMT estimates the correlation two constructs would show if both were measured without error, by relating the average correlation of items across constructs to the average correlation of items within constructs. Lower values indicate greater discriminant validity, values below 0.85 were considered indicative of adequate discriminant validity [33].
Descriptive reference values for the EBQ total score were derived using the raw sum scores. Percentiles were calculated for the total sample and stratified by gender and age group. Measurement invariance across gender, age, and education was examined sequentially by fitting configural, metric, scalar, and strict models using MLR. Robust CFI and RMSEA and the SRMR were used to evaluate model fit. Measurement invariance was primarily evaluated using changes in CFI and RMSEA, with decreases in CFI of no more than 0.010 and increases in RMSEA of no more than 0.015 considered indicative of invariance [34, 35]. Because education-group sizes were highly unequal and two groups were small, Chen’s [34] more stringent criteria (ΔCFI ≤ 0.005 in combination with ΔRMSEA ≤ 0.010 or ΔSRMR ≤ 0.010) were additionally considered as a conservative sensitivity benchmark. For education, we further examined model fit separately within each group and conducted item-level score-test diagnostics of the equality constraints.
Because several Positive Usefulness items were strongly skewed and showed pronounced floor effects, sensitivity analyses were conducted. The CFA models were re-estimated treating the seven-point EBQ items as ordered categorical indicators using WLSMV. Measurement invariance analyses were likewise repeated under the ordinal specification through the scalar level, testing equality of thresholds, factor loadings, and intercepts across groups following Wu and Estabrook [36]. Where necessary, adjacent upper response categories were collapsed consistently across groups to address unobserved or sparsely populated categories.
Multiple linear regression analyses examined associations of age, gender, educational attainment, household income, and marital status with the EBQ total score and the three subscale scores (Controllability, Negative Usefulness, and Positive Usefulness). All four models were estimated in the same complete-case sample (N = 2335). Age was mean-centered at 50.42 years. Men, basic school graduation, household income below €1250, and married and living with partner served as reference categories. For each outcome, we report unstandardized coefficients (B) with 95% confidence intervals and standardized coefficients. For age, coefficients were fully standardized as B × SDx/SDy. For categorical predictors, coefficients were standardized with respect to the outcome only (B/SDy) and therefore represent outcome-standardized dummy-coded contrasts. The dummy-coded predictors themselves remained unstandardized. Model fit was summarized using R² and adjusted R². Omnibus tests were conducted for each complete predictor block, with partial R² and ΔR² quantifying its unique contribution after adjustment for the remaining predictors. Benjamini-Hochberg false discovery rate correction was applied separately within each EBQ outcome, with the 14 individual non-intercept coefficients forming one correction family and the five omnibus block tests a second. Outcome-specific correction was used because the EBQ total score is composed of the three subscale scores and the corresponding analyses therefore represent closely related, structurally overlapping sets of tests. As a more conservative sensitivity analysis, the Benjamini-Hochberg correction was additionally applied across all four outcomes jointly (56 coefficient tests and 20 omnibus tests). Bivariate associations were examined to distinguish unadjusted from mutually adjusted associations: Pearson correlations were calculated for age, and separate unadjusted linear models with omnibus F tests were fitted for the categorical predictors. Regression assumptions were evaluated using residual diagnostic plots, generalized variance inflation factors, the Breusch-Pagan test, models including a quadratic age term, and influence diagnostics based on standardized residuals, leverage, and Cook’s distance. As a further sensitivity analysis, HC3 heteroskedasticity-consistent inference was used for coefficient and omnibus tests.
Results
Sample Characteristics
Table 1 shows the sociodemographic characteristics of the analyzed sample. The sample consisted of 44.7% men, 55.2% women, and 0.1% diverse gendered individuals. The mean age was 50.32 years (SD = 17.13; range: 18–93 years).
Table 1Sample characteristics
Total (N = 2368) Men (N = 1059) Women (N = 1306)
Age, M (SD) 50.32 (17.13) 50.44 (17.09) 50.26 (17.17)
Age groups N (%) N (%) N (%)
< 30 years 356 (15.0) 163 (15.4) 193 (14.8)
30–39 years 367 (15.5) 158 (14.9) 207 (15.8)
40–49 years 399 (16.8) 163 (15.4) 235 (18.0)
50–59 years 454 (19.2) 211 (19.9) 243 (18.6)
60–69 years 452 (19.1) 216 (20.4) 236 (18.1)
≥ 70 years 340 (14.4) 148 (14.0) 192 (14.7)
Marital status
Married (living with partner) 1086 (46.1) 498 (47.2) 588 (45.2)
Married (living without partner) 51 (2.2) 18 (1.7) 33 (2.5)
Unmarried (in a relationship) 268 (11.4) 117 (11.1) 150 (11.5)
Unmarried (not in a relationship) 439 (18.6) 235 (22.3) 203 (15.6)
Divorced 316 (13.4) 136 (12.9) 179 (13.8)
Widowed 197 (8.4) 50 (4.7) 147 (11.3)
Education
Without graduation 60 (2.5) 30 (2.8) 30 (2.3)
Basic school graduation 549 (23.2) 267 (25.3) 281 (21.5)
Middle maturity 989 (41.8) 378 (35.8) 610 (46.7)
Technical school 99 (4.2) 46 (4.4) 53 (4.1)
A-level 357 (15.1) 186 (17.6) 170 (13.0)
University graduation 307 (13.0) 150 (14.2) 157 (12.0)
Currently in school 5 (0.2) 0 (0.0) 5 (0.4)
Occupational status
Regularly employed 1431 (60.5) 668 (63.2) 760 (58.3)
Currently unemployed/short-time work/ marginally employed 144 (6.1) 55 (5.2) 89 (6.8)
Schooling/professional training 103 (4.4) 46 (4.4) 57 (4.4)
Retired 612 (25.9) 279 (26.4) 333 (25.5)
Long-term non-employed/stay-at-home 55 (2.3) 9 (0.9) 46 (3.5)
Volunteer work/parental leave/other leave 19 (0.8) 0 (0.0) 19 (1.5)
Household income
< 1250 € 222 (9.4) 86 (8.2) 136 (10.5)
1250–2500 € 756 (32.1) 327 (31.0) 427 (32.9)
> 2500 € 1378 (58.5) 641 (60.8) 736 (56.7)
Note. M = mean, SD = standard deviation; n = 8 subjects did not report marital status; n = 3 unmarried subjects did not report relationship status: n = 2 subjects did not report education; n = 4 subjects did not report occupational status; n = 12 subjects did not report household income; n = 3 participants with diverse gender were included in the total sample but not shown separately; Percentages are calculated excluding missing values per variable
Descriptive statistics and internal consistency
Item statistics are presented in Table 2. The highest mean item scores were observed for item 11 (“Negative emotions are harmful“; M = 4.59, SD = 1.69) and item 3 (“There is very little use for negative emotions”; M = 4.27, SD = 1.79), while the lowest scores were found for item 16 (“Positive emotions are harmful”, M = 1.89, SD = 1.39) and item 4 (“Positive emotions are very unhelpful to people”, M = 1.91, SD = 1.34). The items generally exhibit low to moderate skewness and largely unremarkable kurtosis, although some items (particularly items regarding the usefulness of positive emotions) show greater skewness and increased kurtosis, suggesting possible floor effects.
Corrected item-total correlations ranged from r = .36 to r = .68. Corrected item-subscale correlations ranged from r = .53 to r = .71 and were mostly higher than the corresponding item-total correlations. The EBQ subscales demonstrated acceptable to good internal consistency (α = 0.79-0.88; ω = 0.79-0.88). The unweighted raw-item sum score also showed high internal consistency (α = 0.88; ω_total = 0.91), with ω_total estimated from the correlated three-factor CFA model. Reliability and descriptive statistics for EBQ total and subscale scores are provided in the supplementary materials (Tables S1-S2; Figure S1).
Table 2Item characteristics of the German EBQ
Item Subscale M (SD) Skew Kurtosis ra rb
1 Once people are experiencing negative emotions, there is nothing they can do about modifying them 1 3.11 (1.66) 0.41 -0.73 0.63 0.63
2 People cannot control their positive emotions 2 3.15 (1.6) 0.29 -0.70 0.55 0.55
3 There is very little use for negative emotions 3 4.27 (1.79) -0.23 -0.89 0.40 0.53
4 Positive emotions are very unhelpful to people 4 1.91 (1.34) 1.55 1.80 0.50 0.70
5 It doesn’t matter how hard people try, they cannot change their negative emotions 1 3.08 (1.55) 0.40 -0.52 0.68 0.71
6 People cannot learn techniques to effectively control their positive emotions 2 3.14 (1.65) 0.38 -0.63 0.62 0.60
7 People don’t need their negative emotions 3 3.89 (1.85) -0.04 -1.00 0.46 0.59
8 There is very little use for positive emotions 4 2.2 (1.41) 1.10 0.37 0.51 0.70
9 People cannot control their negative emotions 1 3.39 (1.55) 0.16 -0.64 0.61 0.61
10 It doesn’t matter how hard people try, they cannot change their positive emotions 2 3.2 (1.56) 0.28 -0.61 0.62 0.64
11 Negative emotions are harmful 3 4.59 (1.69) -0.50 -0.51 0.36 0.57
12 People don’t need their positive emotions 4 2.04 (1.52) 1.53 1.55 0.42 0.65
13 People cannot learn techniques to effectively control their negative emotions 1 3.03 (1.63) 0.44 -0.63 0.65 0.58
14 Once people are experiencing positive emotions, there is nothing they can do about modifying them 2 3.36 (1.57) 0.15 -0.63 0.58 0.58
15 The presence of negative emotions is a bad thing for people 3 4.25 (1.81) -0.27 -0.92 0.43 0.68
16 Positive emotions are harmful 4 1.89 (1.39) 1.54 1.53 0.44 0.68
Note. Item wording corresponds to the original English version of the EBQ. The German translation administered in the present study is provided in the Supplementary Material. Subscale 1 = Negative Controllability, Subscale 2 = Positive Controllability, Subscale 3 = Negative Usefulness, Subscale 4 = Positive Usefulness; M = mean, SD = standard deviation; ra = corrected item-total correlation (item with total score excluding the item); rb = corrected item-subscale correlation (item with subscale score excluding the item); Item’s range 1–7
Confirmatory factor analyses and model comparison
Table 3 presents the global fit indices. The one-factor model showed poor fit with fit indices clearly below recommended thresholds [28]. The three-factor model, in which Positive and Negative Controllability were combined into a single factor, demonstrated acceptable fit (see Table 3), with a BIC difference of 116 points in favor of the four-factor solution. As depicted in Fig. 1, the latent correlation between the two controllability factors was high, r = .90, 95% CI [0.87, 0.93]; HTMT = 0.910. The ordinal sensitivity analysis yielded the same overall pattern of results: The three-factor model showed substantially better fit than the one-factor model, while the four-factor model again showed somewhat better fit than the three-factor model (Table S9). The latent correlation between Negative and Positive Controllability remained very high, r = .92, 95% CI [0.91, 0.93].
Table 3Global fit indices of the 1-, 3-, and 4-factor models
Scaled χ² (df) CFI TLI RMSEA (90% CI) SRMR BIC
1-factor model 3514.4 (104) 0.680 0.630 0.144 [0.140, 0.148] 0.110 132,289
3-factor model 681.2 (101) 0.949 0.939 0.058 [0.054, 0.063] 0.039 128,043
4-factor model 586.4 (98) 0.957 0.947 0.054 [0.050, 0.058] 0.037 127,927
Note. χ² = chi-square; df = degrees of freedom; CFI = Comparative Fit Index; TLI = Tucker-Lewis Index; RMSEA = Root Mean Square Error of Approximation; CI = confidence interval; SRMR = Standardized Root Mean Square Residual; BIC = Bayesian Information Criterion. χ², CFI, TLI, and RMSEA are reported using robust corrections for MLR estimation. 90% CI refers to the robust RMSEA
Together, these findings indicate limited discriminant validity between Negative and Positive Controllability, whereas discriminant validity was acceptable for the remaining factors. Overall, the results support the factorial validity of the EBQ. However, the two controllability facets showed limited discriminant validity, raising questions about the practical utility of distinguishing them in scoring. In contrast, the three-factor model showed low to moderate latent factor intercorrelations (r = .10-0.60, see Fig. 1). The three-factor model was therefore retained for scoring purposes, based on parsimony and clearer discriminant validity. Detailed results for the three- and four-factor models, including latent factor correlations and factor loadings with standard errors, p values, and 95% confidence intervals, are provided in Tables S3-S6. Additional indices of discriminant validity, including AVE, HTMT, and the Fornell-Larcker criterion, are reported in Tables S2, S7 and S8.
Sensitivity Analyses of Factor Structure and Score Interpretation
To examine whether the substantial overlap between Negative and Positive Controllability was attributable to their parallel item wording, we estimated a four-factor correlated-uniqueness model allowing residual covariances between the four corresponding negative/positive item pairs (see Table S10). Although model fit improved relative to the baseline four-factor model, as indicated by a Satorra-Bentler scaled chi-square difference test [29], Δχ²(4) = 38.73, p < .001, the latent correlation between Negative and Positive Controllability remained virtually unchanged (r = .902 vs. 0.898). Thus, the substantial overlap between the two controllability facets cannot be explained by parallel item wording, supporting their combination into a single Controllability factor for scoring purposes while acknowledging the better item-level fit of the four-factor model.
To evaluate the appropriateness of the EBQ total score considering its multidimensional structure, we fitted a bifactor model with one general factor loading on all 16 items and three orthogonal group factors (Controllability, Negative Usefulness, Positive Usefulness, Table S11). The model converged without Heywood cases or inadmissible estimates and showed good overall fit, robust CFI = 0.963, robust TLI = 0.949, robust RMSEA = 0.053, SRMR = 0.044, BIC = 127,904. Standardized general-factor loadings were generally higher for the Controllability items (0.62–0.79) than for the Positive Usefulness (0.39–0.49) and Negative Usefulness items (0.28–0.38), which in turn retained substantial loadings on their respective specific factors (Table S11, Panel B). Based on the standardized solution, the general factor showed an omega hierarchical of ωH = 0.792. The corresponding coefficient for the unweighted raw-item total score was highly similar (ωH = 0.788), indicating that the interpretation of the general factor was essentially unchanged for the score used in the descriptive reference values and regression analyses. Factor-score determinacy was high (0.949), indicating that the general factor was well represented by its factor-score estimates. Explained common variance (ECV = 0.603) and the percentage of uncontaminated correlations (PUC = 0.667) further indicated a substantial, although not exhaustive, general factor. Reise et al., [32] suggested that when PUC is below 0.80, the strength of the general factor should be considered jointly with ECV and ωH. In this situation, ECV values above 0.60 together with ωH values above 0.70 indicate that the multidimensionality may still be sufficiently limited to permit interpretation of a total score. The present values (PUC = 0.667, ECV = 0.603, ωH = 0.792) are consistent with this guideline, although ECV exceeds the suggested value only marginally. Thus, the findings support interpretation of the EBQ total score.
At the subscale level, Negative Usefulness (ωHS = 0.604) and Positive Usefulness (ωHS = 0.563) retained meaningful reliable variance beyond the general factor. In contrast, Controllability retained virtually no specific reliable variance (ωHS = 0.003; ECV = 0.036) and showed lower factor-score determinacy (0.641), indicating that its reliable variance was largely shared with the general factor. The corresponding ωHS coefficients for the unweighted raw-item subscale scores were virtually identical (Controllability = 0.003, Negative Usefulness = 0.604, Positive Usefulness = 0.562). This does not imply that the observed Controllability subscale is unreliable in the conventional sense (see Table S2), but that it provides little reliable variance beyond the general EBQ factor. As a robustness check, we additionally fitted a four-group-factor bifactor model separating Negative and Positive Controllability. This model showed virtually identical fit, robust CFI = 0.962, robust TLI = 0.949, robust RMSEA = 0.054, SRMR = 0.044, BIC = 127,908. Standardized general-factor loadings ranged from 0.281 to 0.791 (see Table S11, Panel D). General-factor indices were highly similar (ωH = 0.780, ECV = 0.580, PUC = 0.800, determinacy = 0.941), and both controllability-related group factors showed comparably low specific reliable variance (ωHS = 0.007 and 0.154, respectively), indicating that the limited incremental value of Controllability beyond the general factor is not an artifact of combining negative and positive controllability items. Taken together, these findings support retaining the EBQ total score, while qualifying its interpretation as a broad general index that more strongly reflects controllability beliefs than usefulness beliefs rather than as a strictly unidimensional measure.
Descriptive reference values
Descriptive percentile reference values for the EBQ total score stratified by gender and age group are presented in Table 4. Percentile values showed slight variation across age groups and between men and women (see also Fig. 2). Supplementary Tables S12-S14 provide corresponding descriptive percentile reference values for the EBQ subscales stratified by gender and age group. Overall, subgroup differences were small. For the Positive Usefulness subscale, several lower percentile ranks corresponded to the minimum attainable score, suggesting limited variability at the lower end of the distribution. Supplementary raw-score conversion tables providing empirical percentile ranks for each possible raw score of the EBQ total scale and the three subscales are presented in Tables S15-S16.
Table 4Descriptive percentile reference values for the EBQ total score by gender and age group
EBQ percentile Total Men Women
18-93y 18-29y 30-39y 40-49y 50-59y 60-69y ≥ 70y 18-29y 30-39y 40-49y 50-59y 60-69y ≥ 70y
1% 16 22 16 16 16 20 18 16 16 16 16 16 21
5% 25 26 25 24 26 28 30 20 24 23 24 19 31
10% 30 32 31 31 31 33 37 26 28 28 32 26 38
25% 40 40 39 40 40 41 47 35 39 37 40 40 47
50% 51 49 52 50 50 52 56 45 51 48 51 51 56
75% 61 60 63 59 62 62 64 56 60 60 59 60 64
90% 69 68 69 65 71 70 69 67 68 68 67 66 70
95% 74 77 74 76 76 75 74 73 72 74 72 72 77
99% 88 81 82 88 91 98 85 87 86 87 91 85 91
Note. y = years. Participants with diverse gender are included in the total sample but are not reported separately because of the small subgroup size (n = 3). Response scale: 1–7; theoretical raw sum score range: 16–112; 50% = median. The percentile values are descriptive reference values derived from the present population-based sample and are intended for research comparisons and preliminary clinical contextualization
Measurement invariance
Measurement invariance was supported across gender and age up to the strict level, with only small changes in fit across increasingly restrictive models (Table 5). For education, the robust configural model showed CFI = 0.942, RMSEA = 0.061, and SRMR = 0.048, while the scalar model showed CFI = 0.933, RMSEA = 0.060, and SRMR = 0.054 (ΔCFI = − 0.0071, ΔRMSEA = 0.0006, ΔSRMR = 0.0019). Although all changes remained within the primary criteria (ΔCFI ≤ 0.010, ΔRMSEA ≤ 0.015; [35]; [34]), the decreases in CFI at the scalar and strict levels were somewhat larger under the more conservative sensitivity benchmark discussed by Chen [34] for small and unequal groups, including ΔCFI = − 0.0054 from scalar to strict. Together with the weaker group-specific fit in the two smallest education groups (Table S17), we interpret these findings as providing approximate support for scalar invariance across education groups under the continuous-item specification. Evidence for strict invariance was less conclusive. Local score-test diagnostics did not identify any loading, intercept, or residual-variance constraint that remained significant after correction for multiple testing (Table S18). Because scalar invariance is sufficient for latent mean comparisons [37], the approximate support for scalar invariance across education groups permits cautious interpretation of the substantive group comparisons reported in the manuscript. Strict invariance is not required for such comparisons.
Table 5Measurement invariance across age, gender, and education for the three-factor EBQ model
Model Scaled χ² df CFI RMSEA [90% CI] SRMR ΔCFI ΔRMSEA ΔSRMR
Age Configural 1264.5 606 0.946 0.060 [0.055, 0.065] 0.049 - - -
Metric 1365.8 671 0.944 0.058 [0.054, 0.063] 0.054 − 0.0019 − 0.0020 0.0050
Scalar 1482.2 736 0.941 0.057 [0.053, 0.061] 0.055 − 0.0029 − 0.0012 0.0014
Strict 1551.3 816 0.937 0.056 [0.052, 0.060] 0.057 − 0.0037 − 0.0012 0.0012
Gender Configural 770.5 202 0.950 0.058 [0.053, 0.062] 0.041 - - -
Metric 795.5 215 0.949 0.056 [0.052, 0.060] 0.042 − 0.0002 − 0.0016 0.0014
Scalar 824.4 228 0.949 0.055 [0.051, 0.059] 0.042 − 0.0006 − 0.0013 0.0004
Strict 807.1 244 0.950 0.053 [0.049, 0.057] 0.043 0.0009 − 0.0023 0.0001
Education Configural 1326.9 606 0.942 0.061 [0.057, 0.066] 0.048 - - -
Metric 1423.3 671 0.940 0.059 [0.055, 0.064] 0.053 − 0.0017 − 0.0022 0.0050
Scalar 1593.3 736 0.933 0.060 [0.056, 0.064] 0.054 − 0.0071 0.0006 0.0019
Strict 1672.4 816 0.928 0.059 [0.055, 0.063] 0.056 − 0.0054 − 0.0008 0.0015
Note. Group sizes: Age: < 30 years n = 356; 30–39 years n = 367; 40–49 years n = 399; 50–59 years n = 454; 60–69 years n = 452; ≥ 70 years n = 340 (N = 2368). Gender: men n = 1059; women n = 1306 (N = 2365, three participants with diverse gender were excluded). Education: basic school graduation n = 549; without graduation n = 60; middle maturity n = 989; technical school n = 99; A-level n = 357; university graduation n = 307 (N = 2361). Models were estimated using MLR with FIML. χ² denotes the scaled test statistic; df = degrees of freedom; CFI = Comparative Fit Index; RMSEA = Root Mean Square Error of Approximation; CI = confidence interval; SRMR = Standardized Root Mean Square Residual. Robust CFI and RMSEA values are reported, with 90% confidence intervals for RMSEA. Change values are reported to four decimal places because rounding to three would obscure comparisons with the 0.010 criterion. Criteria for evaluating invariance are described in the text
Results for the four-factor model were comparable (Tables S19-S20). Ordinal sensitivity analyses using WLSMV likewise supported threshold, loading, and intercept invariance across gender, age, and education, with only small changes in fit across successive invariance levels (|ΔCFI| ≤ 0.005; |ΔRMSEA| ≤ 0.013). These findings indicate that the substantive conclusions were robust to treating the EBQ items as ordered categorical indicators (Tables S21-S22).
Impact of sociodemographic variables on EBQ scores
Multiple linear regression analyses examined associations between sociodemographic variables and EBQ total and subscale scores. All four models were statistically significant (all p < .001), although explained variance was modest (R² = 0.037-0.060). For the EBQ total score, the model explained 6.0% of the variance, F(14, 2320) = 10.53, p < .001, R² = 0.060, adjusted R² = 0.054. Educational attainment showed the largest unique association, F(5, 2320) = 16.25, p_FDR < 0.001, partial R² = 0.034, ΔR² = 0.033 (Table 6 and S25). As depicted in Fig. 3, relative to basic school graduation, lower scores were observed for middle maturity (standardized coefficient = − 0.21, 95% CI [− 0.32, − 0.11]), A-level (− 0.49 [− 0.63, − 0.35]), and university graduation (− 0.56 [− 0.70, − 0.42]; all p_FDR < 0.001). The technical-school contrast did not remain significant after FDR correction (− 0.23 [− 0.45, − 0.02], p_FDR = 0.077), and the estimate for participants without graduation was imprecise (− 0.01 [− 0.28, 0.25]). Age showed a small positive association (0.08 [0.02, 0.13], p_FDR = 0.013), whereas women reported slightly lower scores than men (− 0.12 [− 0.20, − 0.04], p_FDR = 0.013). Although income and marital status were associated with EBQ total scores in bivariate analyses (see S27), their adjusted omnibus associations were not significant after FDR correction.
Table 6Sociodemographic associations with EBQ total and subscale scores
Predictor EBQ Total Controllability Negative Usefulness Positive Usefulness
Age (centered) 0.08 [0.02, 0.13] 0.06 [0.01, 0.11] 0.05 [-0.00, 0.10] 0.07 [0.02, 0.13]
Gender: Women (ref. Men) -0.12 [-0.20, -0.04] -0.12 [-0.21, -0.04] -0.05 [-0.13, 0.03] -0.07 [-0.15, 0.01]
Education: Without graduation -0.01 [-0.28, 0.25] -0.08 [-0.35, 0.19] -0.20 [-0.47, 0.06] 0.36 [0.09, 0.62]
Education: Middle maturity -0.21 [-0.32, -0.11] -0.21 [-0.32, -0.11] -0.10 [-0.21, 0.01] -0.15 [-0.26, -0.04]
Education: Technical school -0.23 [-0.45, -0.02] -0.28 [-0.50, -0.06] -0.12 [-0.34, 0.09] -0.05 [-0.27, 0.16]
Education: A-level -0.49 [-0.63, -0.35] -0.42 [-0.56, -0.28] -0.34 [-0.48, -0.20] -0.37 [-0.51, -0.23]
Education: University graduation -0.56 [-0.70, -0.42] -0.45 [-0.60, -0.31] -0.46 [-0.60, -0.31] -0.38 [-0.53, -0.24]
Income: €1250-<€2500 -0.11 [-0.26, 0.04] -0.16 [-0.31, -0.01] 0.13 [-0.02, 0.28] -0.18 [-0.33, -0.03]
Income: >€2500 -0.14 [-0.30, 0.01] -0.20 [-0.36, -0.04] 0.06 [-0.09, 0.22] -0.14 [-0.30, 0.01]
Marital status: Married (living without partner) 0.03 [-0.25, 0.30] 0.04 [-0.24, 0.32] -0.13 [-0.41, 0.15] 0.16 [-0.12, 0.44]
Marital status: Unmarried (in a relationship) 0.05 [-0.10, 0.20] 0.09 [-0.06, 0.23] -0.12 [-0.27, 0.03] 0.14 [-0.01, 0.28]
Marital status: Unmarried (not in a relationship) -0.01 [-0.15, 0.12] -0.02 [-0.15, 0.12] -0.08 [-0.21, 0.06] 0.08 [-0.05, 0.22]
Marital status: Divorced -0.08 [-0.21, 0.05] -0.06 [-0.19, 0.07] -0.05 [-0.18, 0.09] -0.08 [-0.22, 0.05]
Marital status: Widowed 0.09 [-0.08, 0.26] 0.11 [-0.06, 0.28] -0.10 [-0.27, 0.07] 0.20 [0.03, 0.37]
R² 0.060 0.044 0.037 0.045
Adjusted R² 0.054 0.039 0.031 0.039
Note. Cells show standardized coefficients with 95% confidence intervals. For age, coefficients are fully standardized (B × SDX / SDY); for categorical predictors, coefficients are outcome-standardized dummy-coded contrasts (B / SDY). Bold coefficients remained significant after Benjamini-Hochberg FDR correction across the 14 non-intercept coefficients within the respective outcome. Reference categories were men, basic school graduation, household income < €1250, and married and living with partner. Complete-case analysis; N = 2335
Subscale analyses showed a more differentiated pattern (Tables S23-S25). Educational attainment was the only predictor block significantly associated with all three subscales and showed the largest unique contribution in each model (ΔR² = 0.021-0.024, all p_FDR < 0.001). For Controllability, university graduation was associated with lower scores (− 0.45 [− 0.60, − 0.31]), as were female gender (− 0.12 [− 0.21, − 0.04]) and household income above €2500 (− 0.20 [− 0.36, − 0.04]). For Negative Usefulness, education was the only predictor block remaining significant after FDR correction, with the clearest contrasts for A-level (− 0.34 [− 0.48, − 0.20]) and university graduation (− 0.46 [− 0.60, − 0.31]). For Positive Usefulness, A-level (− 0.37 [− 0.51, − 0.23]) and university graduation (− 0.38 [− 0.53, − 0.24]) were associated with lower scores, whereas participants without a school-leaving qualification showed higher scores (0.36 [0.09, 0.62]); given the small size of this group (n = 59), this contrast should be interpreted cautiously. Marital status also showed a significant omnibus association with Positive Usefulness, F(5, 2320) = 2.98, p_FDR = 0.018, and age was positively associated with scores (0.07 [0.02, 0.13], p_FDR = 0.022). Some additional coefficient-level associations with income and marital status were observed but were less stable across sensitivity analyses (Tables S23-S25 and S29).
Regression diagnostics did not indicate substantial multicollinearity (adjusted GVIFs ≤ 1.31), and adding a quadratic age term did not improve model fit (all p ≥ .076, ΔR² ≤ 0.0013). Cook’s distances were small overall (maximum = 0.010). The Breusch-Pagan test indicated heteroskedasticity only for Negative Usefulness (p < .001). Sensitivity analyses using HC3 inference and FDR correction pooled across all four outcomes agreed with the primary analyses in 144 of 152 significance decisions. All eight discrepancies involved adjusted p values close to 0.05 (0.044-0.084; Table S29). Overall, educational attainment showed the largest and most consistent sociodemographic associations with EBQ scores, whereas associations with age, gender, income, and marital status were smaller and more outcome-specific. Given the modest explained variance (3.7%-6.0%), these associations should not be overinterpreted.
Discussion
This study examined the psychometric properties of the German EBQ [3] in a representative sample of the German general population, including tests of measurement invariance across sociodemographic groups, descriptive reference values and sociodemographic correlates of emotion beliefs.
A key finding of the present study concerns the factor structure of the German EBQ. Confirmatory factor analyses showed better fit for the four-factor model than for the three-factor model on all indices examined (ΔCFI = 0.008, ΔTLI = 0.008, ΔRMSEA = 0.004, ΔBIC = 116). At the same time, the two controllability factors were very highly correlated, and neither the Fornell-Larcker criterion nor the HTMT criterion met recommended thresholds. These two lines of evidence point in different directions: the four-factor model captures item-level covariance more accurately, whereas the controllability facets it separates are barely distinguishable from one another in practice. This tension is not specific to the present sample. Comparably high intercorrelations between the controllability factors have been reported across linguistic and cultural contexts: r = .93 in a Polish [18], r = .97 in a U.S. general population sample [15], r = .87 in an Italian sample [17], and correlations between 0.94 and 0.95 across three samples including an Iranian one [16]. Previous validation studies retained and interpreted the four-factor model, generally without reporting formal tests of discriminant validity between Positive and Negative Controllability. Our sensitivity analyses further suggest that the high correlation is not primarily attributable to parallel item wording: freeing residual covariances between parallel-worded item pairs left the correlation between the two facets essentially unchanged. Complementary bifactor analyses further showed that Controllability retained virtually no reliable specific variance beyond the general factor. We therefore adopted the three-factor solution as a scoring decision grounded in parsimony and interpretability. The initial validation study by Becerra et al., [2] already favored a three-factor solution with a combined controllability factor and a general factor, based on conceptual considerations similar to those in the present study. Similarly, the Norwegian adaptation by Raanes et al., (2024) reported a high intercorrelation between the controllability factors (r = .90), favored the three-factor solution, and concluded that the EBQ “appears unable to effectively distinguish controllability beliefs by valence” (p. 290). Further support for the three-factor solution was recently reported in an Iranian clinical sample of adults undergoing psychotherapy (N = 385; [21]). Taken together, the consistently high intercorrelations across studies support the interpretation that beliefs about the controllability of positive and negative emotions, while statistically separable, overlap to a degree that limits their practical distinctness in the general population, and may largely reflect a broader, valence-independent belief in emotional controllability. Individuals who believe they can control positive emotions typically hold the same conviction regarding negative emotions and vice versa. Such a generalized sense of lacking emotional control has been linked to an increased risk of mental health difficulties across a range of psychopathological conditions, including depressive symptoms [11].
The present study provides population-based descriptive percentile reference values for the German EBQ. Such reference values allow individual scores to be contextualized against an appropriate reference population and may therefore facilitate the use of the EBQ in German-speaking research and preliminary clinical contextualization. From a clinical perspective, such reference values may be relevant because maladaptive beliefs about emotions may limit the extent to which individuals benefit from emotion regulation interventions. Although many psychotherapeutic approaches aim to strengthen adaptive regulation strategies, such strategies may be less readily adopted or applied when individuals view emotions as inherently uncontrollable or useless [9, 15]. Consistent with this, maladaptive emotion beliefs have been associated with psychological distress, maladaptive regulation strategies, and lower use of cognitive reappraisal [8, 11]. In a clinical sample, Mazidi et al., [21] identified six distinct emotion-belief profiles that differed meaningfully in their associations with emotion regulation difficulties and psychological distress. Notably, one profile was characterized by specific maladaptive beliefs about the uselessness of negative emotions despite largely adaptive beliefs about positive emotions, suggesting that clinically relevant emotion beliefs may differ not only in overall adaptiveness but also in their content and valence specificity. These findings indicate that interventions targeting maladaptive emotion beliefs may benefit from considering individuals’ specific belief profiles rather than conceptualizing maladaptive emotion beliefs as a uniform construct. Accordingly, future studies should examine whether the EBQ can support individualized intervention planning by identifying clinically meaningful patterns in beliefs about controllability, usefulness, or the adaptive functions of negative emotions. The reference values provided in the present study offer a basis for evaluating this potential clinical application, but such use requires further validation in clinical and prospective studies.
Establishing measurement invariance is essential to ensure that observed group differences reflect underlying differences in emotion beliefs rather than measurement artifacts [38]. The present study systematically examined measurement invariance of the German EBQ across age, gender, and educational level, with support up to the strict level for age and gender and approximate support at the scalar level for education. Previous studies have supported EBQ measurement invariance across gender and, in some cases, age or cultural groups (Larionow et al., 2023; [16, 17]). To date, Johnston et al., [15] appears to be the only study to have additionally examined invariance across educational groups. A German adaptation examined invariance only across gender [24], leaving age and education untested. This study’s results are particularly relevant in the light of the present findings on education. Educational attainment emerged as the most consistent sociodemographic correlate of EBQ scores, showing associations with the total score and all subscales after controlling for age, gender, income, and marital status (ΔR² = 0.021-0.033, p_FDR < 0.001). Although these effects were modest and the regression models explained only a small proportion of variance overall, their consistency suggests that educational attainment may nevertheless represent a relevant correlate of emotion beliefs. Higher educational attainment was associated with more adaptive emotion beliefs, a pattern not previously documented in the EBQ literature. The approximate scalar invariance observed across educational groups further supports interpreting these differences as substantive variation in emotion beliefs rather than as an artifact of differential measurement functioning.
Prior research has shown that lower SES is associated with greater emotional distress and may moderate the effectiveness of adaptive regulation strategies such as cognitive reappraisal [26]. One possible explanation is that educational attainment is associated with broader social, emotional, and language-based competencies in adulthood [39], which may support more differentiated representations of emotional experience. Emotional granularity - the ability to differentiate and label emotional states - has been linked to adaptive emotion regulation and psychological wellbeing [40, 41]. Individuals with higher educational attainment may therefore have greater resources for interpreting emotional experiences as understandable, manageable, and potentially useful. In line with Ford and Gross’s [1] model, such beliefs may constitute one pathway linking education to regulatory outcomes, consistent with evidence that lower educational attainment is associated with greater reliance on maladaptive strategies such as suppression and avoidance coping [42]. Notably, individuals with lower educational attainment are less likely to access mental health services and more likely to rely on pharmacological rather than psychological interventions, including antidepressants and sedatives [43]. This suggests that structural and psychological barriers to engaging with psychotherapy may be particularly relevant in this group. The present findings raise the possibility that less adaptive emotion beliefs constitute an additional belief-level barrier, by reducing the perceived usefulness, acceptability, or feasibility of engaging with emotion-focused psychological interventions and regulation strategies. Because emotion beliefs are thought to shape whether and how individuals engage in emotion regulation [1], education-sensitive interventions may benefit from targeting not only specific regulation strategies, but also beliefs about the controllability and usefulness of emotions. Such components could be integrated into psychoeducational or emotion regulation interventions using accessible language, concrete examples, and everyday emotion-related scenarios. Future research should examine whether education-sensitive adaptations enhance intervention efficacy and psychotherapy engagement among individuals with lower educational attainment, and whether changes in emotion beliefs mediate improvements in both engagement and emotion regulation outcomes across educational groups.
Biel et al., [3] noted a low factor loading for Item 16 (“It is harmful to feel positive emotions”; factor loading = 0.26) in their initial validation and recommended revising the item wording. In the present population-representative sample, however, the item showed no psychometric irregularities and did not compromise the scale. This finding is methodologically relevant, as the weak item performance reported by Biel et al., [3] may have been related to their small and demographically unbalanced convenience sample, which consisted predominantly of female and highly educated participants. The robust performance of Item 16 in the present study suggests that the item works adequately in representative samples and that revision of its wording does not appear necessary at this stage.
Several limitations should be noted. First, the cross-sectional design precludes causal conclusions. It remains unclear whether higher educational attainment contributes to more adaptive emotion beliefs or whether third variables, such as cognitive ability or stable personality traits, account for this association. Moreover, the regression models explained only a small proportion of variance in EBQ scores, indicating that sociodemographic characteristics capture only a limited part of individual differences in emotion beliefs. Second, as a self-report measure, the EBQ captures consciously held, trait-level emotion beliefs only. It does not assess implicit or situationally activated beliefs, which may be more closely linked to actual regulatory behavior [6]. Third, educational attainment serves as a broad proxy for the proposed mechanisms, such as emotional granularity or access to cognitive-reflective resources. Because these mechanisms were not directly assessed in the present study, the interpretation of the observed education effects remains speculative, and future studies should operationalize these constructs more explicitly to test the proposed pathways. Fourth, the near-zero specific reliable variance of the Controllability subscale (ωHS = 0.003) limits its interpretation as a distinct dimension. Although the subscale showed good internal consistency (Table S2), its reliable variance was almost entirely attributable to the general emotion-belief factor. Researchers using the EBQ should therefore exercise caution when interpreting Controllability as a distinct subscale and may wish to prioritize the Negative and Positive Usefulness subscales when subscale-specific interpretation is of interest.
The present findings point to several areas for further research. First, validation studies in clinical samples are necessary to derive clinically meaningful cut-off scores based on the current descriptive reference values and to examine the predictive validity of the EBQ for treatment response and relapse trajectories. Second, given the demands of clinical screening settings and large-scale population studies, developing an economical short-form version of the EBQ would be a sensible next step. Item response theory approaches could facilitate the selection of highly informative items while preserving measurement precision across the relevant range of emotion beliefs. Third, the range of language versions of the EBQ now available (including English, German, Polish, Turkish, Persian, Italian, and Norwegian) provides an important foundation for cross-cultural studies that can clarify whether the observed education-related associations are replicable across cultures. Finally, future longitudinal and intervention studies should investigate the extent to which changes in emotion beliefs precede and mediate changes in emotion regulation strategies over time.
Conclusion
In conclusion, this study provides a comprehensive psychometric evaluation of the German version of the Emotion Beliefs Questionnaire in a large, population-representative adult sample. Extending prior work on the EBQ, the findings support a three-factor model comprising Controllability, Negative-Usefulness, and Positive-Usefulness as a parsimonious scoring solution. The study further provides strong evidence of measurement invariance across age and gender and approximate support for scalar invariance across educational groups, supporting meaningful group comparisons with some caution for educational level. In addition, the study provides the first population-based descriptive reference values for a German EBQ version, facilitating the interpretation of individual scores relative to the general adult population in research and preliminary clinical contextualization. Importantly, educational attainment emerged as the most consistent sociodemographic correlate of emotion beliefs, highlighting a potentially relevant sociocognitive factor that may contribute to inequalities in emotion regulation resources and mental health outcomes. Taken together, the findings support the EBQ as a psychometrically robust instrument for assessing emotion beliefs and underscore the relevance of these beliefs as a transdiagnostic and potentially modifiable construct. Further research should extend these findings in clinical, longitudinal, and cross-cultural contexts and examine how the EBQ can best be used for efficient and clinically informative assessment.
Supplementary Information
Below is the link to the electronic supplementary material.
References
1.Ford BQ Gross JJ Why Beliefs About Emotion Matter: An Emotion-Regulation Perspective Curr Dir Psychol Sci 2019 28 1 74 81 10.1177/0963721418806697 Ford BQ, Gross JJ. Why Beliefs About Emotion Matter: An Emotion-Regulation Perspective. Curr Dir Psychol Sci. 2019;28(1):74–81. 10.1177/0963721418806697 .
2.Becerra R Preece DA Gross JJ Assessing beliefs about emotions: Development and validation of the Emotion Beliefs Questionnaire PLoS ONE 2020 15 4 e0231395 10.1371/journal.pone.0231395 32287328 Becerra R, Preece DA, Gross JJ. Assessing beliefs about emotions: Development and validation of the Emotion Beliefs Questionnaire. PLoS ONE. 2020;15(4):e0231395. 10.1371/journal.pone.0231395 .PMC7156043 32287328
3.Biel HM Löwe B Briken P Hennig T Krott NR Jungclaußen I Liebherz S Weigel A Reininger KM Affektive Mentalisierung als wahrgenommene Kontrollierbarkeit und Nützlichkeit von Emotionen: Validierung einer deutschsprachigen Version des Emotion Beliefs Questionnaire Die Psychother 2023 68 2 128 36 10.1007/s00278-022-00633-7 Biel HM, Löwe B, Briken P, Hennig T, Krott NR, Jungclaußen I, Liebherz S, Weigel A, Reininger KM. Affektive Mentalisierung als wahrgenommene Kontrollierbarkeit und Nützlichkeit von Emotionen: Validierung einer deutschsprachigen Version des Emotion Beliefs Questionnaire. Die Psychother. 2023;68(2):128–36. 10.1007/s00278-022-00633-7 .
4.Trampe D Quoidbach J Taquet M Emotions in Everyday Life PLoS ONE 2015 10 12 e0145450 10.1371/journal.pone.0145450 26698124 Trampe D, Quoidbach J, Taquet M. Emotions in Everyday Life. PLoS ONE. 2015;10(12):e0145450. 10.1371/journal.pone.0145450 .PMC4689475 26698124
5.Kisley MA Shulkin J Meza-Whitlatch MV Pedler RB Emotion beliefs: Conceptual review and compendium Front Psychol 2024 14 1271135 10.3389/fpsyg.2023.1271135 38239475 Kisley MA, Shulkin J, Meza-Whitlatch MV, Pedler RB. Emotion beliefs: Conceptual review and compendium. Front Psychol. 2024;14:1271135. 10.3389/fpsyg.2023.1271135 .PMC10794336 38239475
6.Peter S Oliver BR Kabo HR Raynaud AV Wiggers M Somerville MP Measuring emotion beliefs: A systematic review Cogn Emot 2026 40 3 517 47 10.1080/02699931.2025.2526676 40644681 Peter S, Oliver BR, Kabo HR, Raynaud AV, Wiggers M, Somerville MP. Measuring emotion beliefs: A systematic review. Cogn Emot. 2026;40(3):517–47. 10.1080/02699931.2025.2526676 .40644681
7.Trincas R Bilotta E Mancini F Specific Beliefs about Emotions Are Associated with Different Emotion-Regulation Strategies Psychology 2016 07 13 1682 99 10.4236/psych.2016.713159 Trincas R, Bilotta E, Mancini F. Specific Beliefs about Emotions Are Associated with Different Emotion-Regulation Strategies. Psychology. 2016;07(13):1682–99. 10.4236/psych.2016.713159 .
8.De Castella K Platow MJ Tamir M Gross JJ Beliefs about emotion: Implications for avoidance-based emotion regulation and psychological health Cogn Emot 2018 32 4 773 95 10.1080/02699931.2017.1353485 28737108 De Castella K, Platow MJ, Tamir M, Gross JJ. Beliefs about emotion: Implications for avoidance-based emotion regulation and psychological health. Cogn Emot. 2018;32(4):773–95. 10.1080/02699931.2017.1353485 .28737108
9.Deplancke C Somerville MP Harrison A Vuillier L It’s all about beliefs: Believing emotions are uncontrollable is linked to symptoms of anxiety and depression through cognitive reappraisal and expressive suppression Curr Psychol 2023 42 25 22004 12 10.1007/s12144-022-03252-2 Deplancke C, Somerville MP, Harrison A, Vuillier L. It’s all about beliefs: Believing emotions are uncontrollable is linked to symptoms of anxiety and depression through cognitive reappraisal and expressive suppression. Curr Psychol. 2023;42(25):22004–12. 10.1007/s12144-022-03252-2 .
10.Hong E-J Kangas M The Relationship between Beliefs about Emotions and Emotion Regulation: A Systematic Review Behav Change 2022 39 4 205 34 10.1017/bec.2021.23 Hong E-J, Kangas M. The Relationship between Beliefs about Emotions and Emotion Regulation: A Systematic Review. Behav Change. 2022;39(4):205–34. 10.1017/bec.2021.23 .
11.Kneeland ET Dovidio JF Joormann J Clark MS Emotion malleability beliefs, emotion regulation, and psychopathology: Integrating affective and clinical science Clin Psychol Rev 2016 45 81 8 10.1016/j.cpr.2016.03.008 27086086 Kneeland ET, Dovidio JF, Joormann J, Clark MS. Emotion malleability beliefs, emotion regulation, and psychopathology: Integrating affective and clinical science. Clin Psychol Rev. 2016;45:81–8. 10.1016/j.cpr.2016.03.008 .27086086
12.Tamir M John OP Srivastava S Gross JJ Implicit theories of emotion: Affective and social outcomes across a major life transition J Personal Soc Psychol 2007 92 4 731 44 10.1037/0022-3514.92.4.731 17469955 Tamir M, John OP, Srivastava S, Gross JJ. Implicit theories of emotion: Affective and social outcomes across a major life transition. J Personal Soc Psychol. 2007;92(4):731–44. 10.1037/0022-3514.92.4.731 .17469955
13.Rimes KA Chalder T The Beliefs about Emotions Scale: Validity, reliability and sensitivity to change J Psychosom Res 2010 68 3 285 92 10.1016/j.jpsychores.2009.09.014 20159215 Rimes KA, Chalder T. The Beliefs about Emotions Scale: Validity, reliability and sensitivity to change. J Psychosom Res. 2010;68(3):285–92. 10.1016/j.jpsychores.2009.09.014 .20159215
14.Veilleux JC Salomaa AC Shaver JA Zielinski MJ Pollert GA Multidimensional Assessment of Beliefs About Emotion: Development and Validation of the Emotion and Regulation Beliefs Scale Assessment 2015 22 1 86 100 10.1177/1073191114534883 24835246 Veilleux JC, Salomaa AC, Shaver JA, Zielinski MJ, Pollert GA. Multidimensional Assessment of Beliefs About Emotion: Development and Validation of the Emotion and Regulation Beliefs Scale. Assessment. 2015;22(1):86–100. 10.1177/1073191114534883 .24835246
15.Johnston TE McEvoy PM Gross JJ Becerra R Preece DA The Emotion Beliefs Questionnaire: Psychometric properties, norms, and links to affective outcomes J Affect Disord 2024 356 577 85 10.1016/j.jad.2024.04.002 38574870 Johnston TE, McEvoy PM, Gross JJ, Becerra R, Preece DA. The Emotion Beliefs Questionnaire: Psychometric properties, norms, and links to affective outcomes. J Affect Disord. 2024;356:577–85. 10.1016/j.jad.2024.04.002 .38574870
16.Ranjbar S Mazidi M Gross JJ Preece D Zarei M Azizi A Mirshafiei M Becerra R Examining the Cross Cultural Validity and Measurement Invariance of the Emotion Beliefs Questionnaire (EBQ) in Iran and the USA J Psychopathol Behav Assess 2023 45 3 755 66 10.1007/s10862-023-10068-2 Ranjbar S, Mazidi M, Gross JJ, Preece D, Zarei M, Azizi A, Mirshafiei M, Becerra R. Examining the Cross Cultural Validity and Measurement Invariance of the Emotion Beliefs Questionnaire (EBQ) in Iran and the USA. J Psychopathol Behav Assess. 2023;45(3):755–66. 10.1007/s10862-023-10068-2 .
17.Rogier G Cavalli RG Maggiolo C Velotti P Factorial Structure of the Emotional Beliefs Questionnaire: Testing Measurement Invariance and Competitive Models J Psychopathol Behav Assess 2023 45 2 558 71 10.1007/s10862-023-10038-8 Rogier G, Cavalli RG, Maggiolo C, Velotti P. Factorial Structure of the Emotional Beliefs Questionnaire: Testing Measurement Invariance and Competitive Models. J Psychopathol Behav Assess. 2023;45(2):558–71. 10.1007/s10862-023-10038-8 .
18.Larionow P Preece DA Mudło-Głagolska K Assessing emotion beliefs with the Polish version of the Emotion Beliefs Questionnaire (EBQ): Psychometric properties, norms, and links with emotional reactivity and psychopathology Curr Psychol 2024 43 17 15939 51 10.1007/s12144-023-05544-7 Larionow P, Preece DA, Mudło-Głagolska K. Assessing emotion beliefs with the Polish version of the Emotion Beliefs Questionnaire (EBQ): Psychometric properties, norms, and links with emotional reactivity and psychopathology. Curr Psychol. 2024;43(17):15939–51. 10.1007/s12144-023-05544-7 .
19.Johnston TE Gross JJ Chen W McEvoy P Becerra R Preece DA Emotion beliefs and emotion regulation strategy use Pers Indiv Differ 2025 240 113066 10.1016/j.paid.2025.113066 Johnston TE, Gross JJ, Chen W, McEvoy P, Becerra R, Preece DA. Emotion beliefs and emotion regulation strategy use. Pers Indiv Differ. 2025;240:113066. 10.1016/j.paid.2025.113066 .
20.Becerra R Gainey K Murray K Preece DA Intolerance of uncertainty and anxiety: The role of beliefs about emotions J Affect Disord 2023 324 349 53 10.1016/j.jad.2022.12.064 36586604 Becerra R, Gainey K, Murray K, Preece DA. Intolerance of uncertainty and anxiety: The role of beliefs about emotions. J Affect Disord. 2023;324:349–53. 10.1016/j.jad.2022.12.064 .36586604
21.Mazidi M Brett JD Azizi A Moloodi R Shaban M Gross JJ Becerra R Profiles of beliefs about emotions in treatment-seeking adults: A person‐centred approach using the emotion beliefs questionnaire Br J Clin Psychol 2026 65 1 218 35 10.1111/bjc.70014 40999600 PMC12889208 Mazidi M, Brett JD, Azizi A, Moloodi R, Shaban M, Gross JJ, Becerra R. Profiles of beliefs about emotions in treatment-seeking adults: A person‐centred approach using the emotion beliefs questionnaire. Br J Clin Psychol. 2026;65(1):218–35. 10.1111/bjc.70014 .PMC12889208 40999600
22.Reininger KM Biel HM Hennig T Zitzmann S Weigel A Spitzer C Toussaint A Löwe B Beliefs about emotions predict psychological stress related to somatic symptoms Br J Clin Psychol 2023 62 4 699 716 10.1111/bjc.12438 37706588 Reininger KM, Biel HM, Hennig T, Zitzmann S, Weigel A, Spitzer C, Toussaint A, Löwe B. Beliefs about emotions predict psychological stress related to somatic symptoms. Br J Clin Psychol. 2023;62(4):699–716. 10.1111/bjc.12438 .37706588
23.Reininger KM Löwe B Hennig T Toussaint A-K Weigel A Maehder K Briken P Moritz S Biel HM On the relationship between affective mentalization and depression severity: Cross-sectional, longitudinal, and experimental evidence Psychoanal Psychother 2025 39 2 95 119 10.1080/02668734.2025.2485051 Reininger KM, Löwe B, Hennig T, Toussaint A-K, Weigel A, Maehder K, Briken P, Moritz S, Biel HM. On the relationship between affective mentalization and depression severity: Cross-sectional, longitudinal, and experimental evidence. Psychoanal Psychother. 2025;39(2):95–119. 10.1080/02668734.2025.2485051 .
24.Gutzweiler R Grüning DJ Measuring four facets of emotion beliefs in Germany: A German-language adaptation of the EBQ and its comparability across gender and different emotion abilities PLoS ONE 2025 20 1 e0316007 10.1371/journal.pone.0316007 39746019 Gutzweiler R, Grüning DJ. Measuring four facets of emotion beliefs in Germany: A German-language adaptation of the EBQ and its comparability across gender and different emotion abilities. PLoS ONE. 2025;20(1):e0316007. 10.1371/journal.pone.0316007 .PMC11694981 39746019
25.Carstensen LL Socioemotional Selectivity Theory: The Role of Perceived Endings in Human Motivation Gerontologist 2021 61 8 1188 96 10.1093/geront/gnab116 34718558 Carstensen LL. Socioemotional Selectivity Theory: The Role of Perceived Endings in Human Motivation. Gerontologist. 2021;61(8):1188–96.10.1093/geront/gnab116 PMC8599276 34718558
26.Hittner EF Rim KL Haase CM Socioeconomic status as a moderator of the link between reappraisal and anxiety: Laboratory-based and longitudinal evidence Emotion 2019 19 8 1478 89 10.1037/emo0000539 30556700 PMC10448892 Hittner EF, Rim KL, Haase CM. Socioeconomic status as a moderator of the link between reappraisal and anxiety: Laboratory-based and longitudinal evidence. Emotion. 2019;19(8):1478–89. 10.1037/emo0000539 .PMC10448892 30556700
27.R Core Team. (2025). R: A language and environment for statistical computing.https://www.R-project.org/
28.Hu L Bentler PM Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives Struct Equation Modeling: Multidisciplinary J 1999 6 1 1 55 10.1080/10705519909540118 Hu L, Bentler PM. Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Struct Equation Modeling: Multidisciplinary J. 1999;6(1):1–55. 10.1080/10705519909540118 .
29.Satorra A Bentler PM Ensuring Positiveness of the Scaled Difference Chi-square Test Statistic Psychometrika 2010 75 2 243 8 10.1007/s11336-009-9135-y 20640194 Satorra A, Bentler PM. Ensuring Positiveness of the Scaled Difference Chi-square Test Statistic. Psychometrika. 2010;75(2):243–8. 10.1007/s11336-009-9135-y .PMC2905175 20640194
30.Rodriguez A Reise SP Haviland MG Evaluating bifactor models: Calculating and interpreting statistical indices Psychol Methods 2016 21 2 137 50 10.1037/met0000045 26523435 Rodriguez A, Reise SP, Haviland MG. Evaluating bifactor models: Calculating and interpreting statistical indices. Psychol Methods. 2016;21(2):137–50. 10.1037/met0000045 .26523435
31.Reise SP The Rediscovery of Bifactor Measurement Models Multivar Behav Res 2012 47 5 667 96 10.1080/00273171.2012.715555 PMC3773879 24049214 Reise SP. The Rediscovery of Bifactor Measurement Models. Multivar Behav Res. 2012;47(5):667–96. 10.1080/00273171.2012.715555 .PMC3773879 24049214
32.Reise SP Scheines R Widaman KF Haviland MG Multidimensionality and Structural Coefficient Bias in Structural Equation Modeling: A Bifactor Perspective Educ Psychol Meas 2013 73 1 5 26 10.1177/0013164412449831 Reise SP, Scheines R, Widaman KF, Haviland MG. Multidimensionality and Structural Coefficient Bias in Structural Equation Modeling: A Bifactor Perspective. Educ Psychol Meas. 2013;73(1):5–26. 10.1177/0013164412449831 .
33.Henseler J Ringle CM Sarstedt M A new criterion for assessing discriminant validity in variance-based structural equation modeling J Acad Mark Sci 2015 43 1 115 35 10.1007/s11747-014-0403-8 Henseler J, Ringle CM, Sarstedt M. A new criterion for assessing discriminant validity in variance-based structural equation modeling. J Acad Mark Sci. 2015;43(1):115–35. 10.1007/s11747-014-0403-8 .
34.Chen FF Sensitivity of Goodness of Fit Indexes to Lack of Measurement Invariance Struct Equation Modeling: Multidisciplinary J 2007 14 3 464 504 10.1080/10705510701301834 Chen FF. Sensitivity of Goodness of Fit Indexes to Lack of Measurement Invariance. Struct Equation Modeling: Multidisciplinary J. 2007;14(3):464–504. 10.1080/10705510701301834 .
35.Cheung GW Rensvold RB Evaluating Goodness-of-Fit Indexes for Testing Measurement Invariance Struct Equation Modeling: Multidisciplinary J 2002 9 2 233 55 10.1207/S15328007SEM0902_5 Cheung GW, Rensvold RB. Evaluating Goodness-of-Fit Indexes for Testing Measurement Invariance. Struct Equation Modeling: Multidisciplinary J. 2002;9(2):233–55. 10.1207/S15328007SEM0902_5 .
36.Wu H Estabrook R Identification of Confirmatory Factor Analysis Models of Different Levels of Invariance for Ordered Categorical Outcomes Psychometrika 2016 81 4 1014 45 10.1007/s11336-016-9506-0 27402166 Wu H, Estabrook R. Identification of Confirmatory Factor Analysis Models of Different Levels of Invariance for Ordered Categorical Outcomes. Psychometrika. 2016;81(4):1014–45. 10.1007/s11336-016-9506-0 .PMC5458787 27402166
37.Putnick DL Bornstein MH Measurement invariance conventions and reporting: The state of the art and future directions for psychological research Dev Rev 2016 41 71 90 10.1016/j.dr.2016.06.004 27942093 Putnick DL, Bornstein MH. Measurement invariance conventions and reporting: The state of the art and future directions for psychological research. Dev Rev. 2016;41:71–90. 10.1016/j.dr.2016.06.004 .PMC5145197 27942093
38.Vandenberg RJ Lance CE A Review and Synthesis of the Measurement Invariance Literature: Suggestions, Practices, and Recommendations for Organizational Research Organizational Res Methods 2000 3 1 4 70 10.1177/109442810031002 Vandenberg RJ, Lance CE. A Review and Synthesis of the Measurement Invariance Literature: Suggestions, Practices, and Recommendations for Organizational Research. Organizational Res Methods. 2000;3(1):4–70. 10.1177/109442810031002 .
39.OECD. Skills that Matter for Success and Well-being in Adulthood: Evidence on Adults’ Social and Emotional Skills from the 2023 Survey of Adult Skills. OECD Publishing; 2025. 10.1787/6e318286-en .
40.Ip KI Yu K Gendron M Emotion Granularity, Regulation, and Their Implications in Health: Broadening the Scope from a Cultural and Developmental Perspective Emot Rev 2024 16 4 224 37 10.1177/17540739231214564 Ip KI, Yu K, Gendron M. Emotion Granularity, Regulation, and Their Implications in Health: Broadening the Scope from a Cultural and Developmental Perspective. Emot Rev. 2024;16(4):224–37. 10.1177/17540739231214564 .
41.Kalokerinos EK Erbas Y Ceulemans E Kuppens P Differentiate to Regulate: Low Negative Emotion Differentiation Is Associated With Ineffective Use but Not Selection of Emotion-Regulation Strategies Psychol Sci 2019 30 6 863 79 10.1177/0956797619838763 30990768 Kalokerinos EK, Erbas Y, Ceulemans E, Kuppens P. Differentiate to Regulate: Low Negative Emotion Differentiation Is Associated With Ineffective Use but Not Selection of Emotion-Regulation Strategies. Psychol Sci. 2019;30(6):863–79. 10.1177/0956797619838763 .30990768
42.Smrtnik Vitulić H, Prosen. Coping and Emotion Regulation Strategies in Adulthood: Specificities Regarding Age, Gender and Level of Education. Drustvena Istrazivanja. 2016;43–62. 10.5559/di.25.1.03 .
43.Halme M Rautava P Sillanmäki L Sumanen M Suominen S Vahtera J Virtanen P Salo P Educational level and the use of mental health services, psychotropic medication and psychotherapy among adults with a history of physician diagnosed mental disorders Int J Soc Psychiatry 2023 69 2 493 502 10.1177/00207640221111091 35819228 PMC9983050 Halme M, Rautava P, Sillanmäki L, Sumanen M, Suominen S, Vahtera J, Virtanen P, Salo P. Educational level and the use of mental health services, psychotropic medication and psychotherapy among adults with a history of physician diagnosed mental disorders. Int J Soc Psychiatry. 2023;69(2):493–502. 10.1177/00207640221111091 .PMC9983050 35819228
Come citare questo studio
Lammers JR, Brähler E, Koechel A, Conrad R, Ramm M. (2026). The German emotion beliefs questionnaire: factorial validity, measurement invariance, and descriptive reference values in a population-representative sample.. BMC psychology. DOI: 10.1186/s40359-026-05837-z
Consulta la pubblicazione originale ↗