Psychometric Properties of the Generalized Pliance Questionnaire - Children (GPQ-C)

Context and Objectives

Empirical research on pliance, a functional class of rule-governed behavior, has been limited in children. Pliance is a type of verbally governed behavior in which an individual follows a rule provided by another person or by themselves, primarily motivated by social approval. The Generalized Pliance Questionnaire (GPQ) was recently developed for adults showing excellent psychometric properties. However, no validated child version existed to measure this important behavioral construct. This study aimed to develop and validate a child version of the GPQ adapted for Colombian children (GPQ-C), examining its psychometric properties, factor structure, and relationships with relevant constructs such as emotional symptoms, psychological inflexibility, and repetitive negative thinking.

Method

Participants: The sample consisted of 797 participants (60.2% female) with ages ranging from 8 to 13 years (M = 9.57, SD = 1.10), distributed across fourth to ninth grade (equivalent to USA). 62.3% came from private schools and 37.7% from public schools. All participants were Colombian.

Primary Instrument - Generalized Pliance Questionnaire for Children (GPQ-C): The GPQ-C consists of 8 items rated on a 5-point scale (5 = always true, 1 = never true). It was derived from the original adult GPQ through item reduction based on content relevant to children's life. Seven items from the original GPQ were eliminated for not being relevant (items 3, 6, 7, 10, 13, 15, and 16). The remaining 8 items were adapted to children's language. The items primarily assess children's tendency to seek social approval and follow rules established by others.

Additional Instruments:

• Acceptance and Fusion Questionnaire for Youth (AFQ-Y): Consists of 17 items on a 5-point Likert scale measuring psychological inflexibility. The Spanish validated version was used.
• Depression, Anxiety, and Stress Scale for Youth (DASS-Y): A 24-item scale containing three subscales (Depression, Anxiety, Stress) with good internal consistency properties. Acceptable alpha values (.78, .79, and .69).
• Penn State Worry Questionnaire for Children (PSWQ-C): Consists of 14 items measuring worry and repetitive negative thinking. Demonstrated good psychometric properties (alpha of .88) with single-factor structure.
• Persistent Negative Thinking Questionnaire for Children (PTQ-C): Measures persistent negative thinking in 15 items with 5-point Likert scale. Showed excellent psychometric properties (alpha of .92) and single-factor structure.

Procedure: The study was approved by the institutional ethics committee. Participants were recruited from public and private schools in Bogotá (Colombia) and surrounding areas. Informed consent was obtained from parents. Children completed the questionnaire package in regular class under the supervision of a trained psychologist, anonymously. Administration time was approximately 15-20 minutes. The order of questionnaires was: DASS-Y, PSWQ-C, PTQ-C, AFQ-Y, and GPQ-C.

Data Analysis: A cross-validation study was conducted by randomly dividing the sample into two subsamples of approximately equal size using SPSS 19. In the first subsample (N = 394), exploratory factor analysis (EFA) was performed with unweighted least squares (ULS) extraction and polychoric Direct Oblimin rotation. Unidimensionality was assessed using UniCo, ECV, and MIREAL. Cronbach's alpha was calculated for internal consistency. In the second subsample (N = 403), confirmatory factor analysis (CFA) was conducted to validate the structure found. Goodness-of-fit indices were used: RMSEA, CFI, NNFI, and SRMR. Metric and scalar invariance analyses were performed by gender and age group. Two-way ANOVA was conducted to analyze differences in GPQ-C scores by gender and age. Pearson correlations were calculated between GPQ-C and other scales to evaluate convergent validity.

Results

Factor Analysis: Table 1 shows the 8 items of the GPQ-C, their English translations, descriptive data, and corrected item-total correlations. All items showed good discrimination, with corrected item-total correlations ranging from .41 (Item 8) to .59 (Item 7). Cronbach's alpha was .81.

In the first subsample (N = 394), EFA showed that the Kaiser-Meyer-Olkin (KMO) test was acceptable (.87) and Bartlett's test result was statistically significant (776.91(28), p < .001). EFA suggested a single-factor structure explaining 48.9% of variance (eigenvalue = 3.92). Table 1 also shows that factor loadings were high for all items, ranging from .49 (Item 8) to .71 (Items 3 and 6). Unidimensionality indices (UniCo, ECV, and MIREAL with values of 24) strongly supported unidimensionality of the GPQ-C.

In the second subsample (N = 403), CFA was conducted to analyze the fit of the single-factor model. Overall fit of the single-factor model in GPQ-C was very good: χ²(20) = 33.84, p < .05; RMSEA = .042, 90% CI [.014, .065]; CFI = .98; NNFI = .99; SRMR = .04. These values indicate a well-fitting model to the data.

Metric and Scalar Invariance: Table 2 presents results of metric and scalar invariance analyses by gender and age group. Parameter invariance was supported at both levels (metric and scalar) across gender and age group (8-9 years vs. 10-13 years) because changes in RMSEA, CFI, and NNFI were lower than .01. This means that subgroups of children responded similarly to the GPQ-C, allowing their scores to be compared directly.

Descriptive Statistics: Table 3 presents descriptive data for the GPQ-C by sex and age group. Boys obtained lower scores (M = 20.22, SD = 8.07) in the 8-9 years group and (M = 17.81, SD = 7.43) in the 10-13 years group. Girls obtained higher scores (M = 21.35, SD = 8.04) in the 8-9 years group and (M = 20.35, SD = 7.84) in the 10-13 years group. Overall total score was M = 20.30, SD = 7.83.

Two-way ANOVA showed that differences in GPQ-C scores were statistically significant for both sex (F = 9.85, p = .002, η² = .014) and age group (F = 8.65, p = .003, η² = .013). There was no significant interaction between gender and age (F = 0.82, p = .37, η² = .001). Girls obtained higher scores than boys on the GPQ-C. Scores appeared to decrease with age, with highest scores in the 8-9 years group.

Convergent Validity: Table 4 presents Pearson correlations between the GPQ-C and other measures of relevant constructs. The GPQ-C showed very strong positive correlations with experiential avoidance and cognitive fusion (AFQ-Y: r = .72****), which is consistent with previous theory. The GPQ-C showed strong positive correlations with emotional symptoms: DASS-Depression (r = .42****), DASS-Anxiety (r = .46****), DASS-Stress (r = .44****). PSWQ-C showed a strong correlation with GPQ-C (r = .58****). PTQ-C showed the strongest correlation (r = .62****). All correlations were significant at p < .001 level.

Discussion and Conclusions

The results provide solid evidence that the GPQ-C is a psychometrically valid and reliable measure of generalized pliance in children. Exploratory factor analysis showed clear single-factor structure, and confirmatory factor analysis in the second subsample supported this finding. Goodness-of-fit indices were excellent, confirming that the structure of a single latent variable fits the data well.

Metric and scalar invariance was maintained across gender and age group, meaning that the GPQ-C can be used to compare scores between boys and girls of different ages without confusion due to differences in how they respond to the scale. This is an important finding demonstrating that the GPQ-C functions equivalently in these demographic subgroups.

The correlations found with other constructs provided important evidence of convergent validity. The GPQ-C showed very strong positive correlations with psychological inflexibility (AFQ-Y), which is theoretically coherent given that both pliance and psychological inflexibility relate to control by external rules. Furthermore, correlations with emotional symptoms (depression, anxiety, stress) were significant, suggesting that children with higher pliance tend to experience more emotional symptoms. This association seems logical because a greater need for social approval might generate more concern about how others perceive them, which in turn generates more rumination about social issues that may be unpredictable.

The strongest correlations were found with repetitive negative thinking (PSWQ-C and PTQ-C), indicating that children with higher pliance tend to engage more frequently in rumination and worry. This seems coherent because a greater need for social approval might trigger more worry about how others perceive them.

An important finding was that GPQ-C scores appeared to decrease with age, being higher in the 8-9 years group. This is consistent with the hypothesis that generalized pliance decreases with age because pliance is the initial type of rule-governed behavior that develops in childhood, and it is relatively easy to generalize to some extent in early childhood. However, with development, children typically develop greater skills in relational framing, allowing them more behavioral flexibility in response to environmental contingencies.

Another finding was that girls obtained higher pliance scores than boys. This is consistent with previous research data suggesting that in Latin America parents place more emphasis on social interactions in girls, which could emphasize the importance of following social rules and being accepted by others.

The study had some limitations worth discussing. First, there is limited experimental research clearly distinguishing between pliance and tracking as distinct functional classes of rule-governed behavior. The reason for this state of evidence is unclear, but could be due to difficulty designing instructions resembling pliance versus tracking, probably because participants' personal history influences their performance more than experimental rules. A recent exception has been the study by Kissi, Hughes, De Schryver, De Houwer, and Crombez, which showed that the insensitivity effect was clearer in the pliance condition compared to tracking and no-instruction conditions. Similarly, although the distinction between pliance and tracking has been largely emphasized in ACT, there are few attempts to measure these classes of rule-governed behavior in psychopathology and clinical studies. This contrasts significantly with attempts to measure other ACT mid-level terms such as experiential avoidance, cognitive fusion, or values. Second, the GPQ-C was correlated only with other self-report measures, which may have inflated the correlations found. Future studies should explore criterion validity of the GPQ-C against a behavioral task measuring insensitivity to contingencies. Third, the psychometric properties found in this study are exclusive to the Colombian population. Future studies should analyze the psychometric properties and validity of the GPQ-C in other Spanish-speaking countries and other languages.

Significance and contribution

This study contributes to the field of child psychological assessment by presenting a valid and reliable adaptation of the Generalized Pliance Questionnaire for children. The GPQ-C demonstrated clear factor structure, metric and scalar invariance across gender and age groups, and theoretically coherent correlations with related constructs, providing an instrument for assessing rule-governed behavior in research and clinical practice with Spanish-speaking populations.

---