Abstract
Accurate self-assessment of Big Five personality traits in middle childhood remains challenging. To address this gap, this study developed and validated the Chinese Pictorial Big Five Personality Inventory for Children (CPBFI-C), a novel self-report tool designed for children as young as 6 years old. The CPBFI-C was developed based on the facet-informed structure of the Big Five Inventory-2, with items sampled to represent diverse facet content within each domain and combined with pictorial stimuli tailored to Chinese cultural contexts to enhance comprehension and engagement. In Study 1, based on a sample of 2,484 primary school students (Sample 1, aged 6–14), the CPBFI-C demonstrated acceptable internal consistency, good test–retest reliability, and excellent structural validity via exploratory structural equation modeling. Crucially, the scale achieved strict measurement invariance across age groups (6–9 vs. 10–14 years) and partial strict invariance across sex. Furthermore, partial scalar invariance was established between the picture-based version and the text-only version administered to Sample 2, which comprised 2,282 students in Grades 4–7. The alignment method was employed for sex and test format and confirmed approximate measurement invariance, thus allowing for meaningful group comparisons. These results support the scale's robust psychometric properties across diverse demographics and formats. In Study 2, utilizing three independent subsamples for external validation, the CPBFI-C showed statistically significant and meaningful self–other agreement with parent-reported personality. Furthermore, it exhibited strong external validity through theoretically consistent correlations with children's self-reported depressive symptoms and teacher-evaluated academic performance. Alignment-with-covariates analysis further showed broadly comparable associations across sex groups. In summary, the CPBFI-C is a reliable, valid, and engaging instrument, and provides researchers with an effective Big Five self-assessment tool for measuring children as young as 6 years old.
Introduction
Many interindividual differences in characteristics of thoughts, feelings, and actions can be summarized by the Big Five personality model (McCrae & John, 1992). The five factors are extraversion, agreeableness, conscientiousness, neuroticism (alternatively labeled emotional stability), and openness (alternatively labeled open-mindedness, intellect, or imagination), which are ordered at different levels of breadth (Costa & McCrae, 2014; DeYoung et al., 2007; Soto & John, 2017a). The Big Five personality model has been widely used for personality research and found in different cultural backgrounds (Hofstee et al., 1997; Trull & Geary, 1997).
Given the vast range of empirical evidence, the Big Five model can be used as an overarching taxonomy for both adults’ and children's personality traits, and, at least by middle childhood, children's personality traits share the same Big Five structure as adult traits (Herzhoff et al., 2017; Shiner & Caspi, 2003). However, for the evaluation of children's personality in middle childhood, most existing approaches rely on reports from parents and teachers, and there remains a lack of instruments that adequately capture the breadth of personality-relevant behaviors within each domain. During this period, children can already make consistent assessments of their own personality (Measelle et al., 2005). Given that pictures can stimulate children’s interest and help focus their attention (Valla et al., 2000), culturally appropriate pictures reflecting Chinese social situations may be beneficial. This study is guided by the facet-informed structure of the Big Five Inventory-2 (BFI-2) and combines pictorial and textual formats to develop a Big Five personality scale that is suitable for self-assessment in middle childhood.
Evaluation of Children's Big Five Personality Traits
Researchers have found the Big Five personality structure among children through teacher or parent evaluations. Kohnstamm et al. (1998) asked parents from seven countries (Belgium, The Netherlands, Germany, Greece, China, the USA, and Poland) to freely describe their 2- to 13-year-old children. Most of the phrases used by the parents in every country could be classified as Big Five personality factors. Digman and Inouye (1986) conducted factor analysis on teacher ratings of 499 children using 43 unipolar adjective scales and found the five-factor model in sixth-grade children. Another study asked 224 teachers to rate 10 randomly selected children, aged 4 to 12, on the Flemish Big-Five Bipolar Rating Scales (Mervielde et al., 1995). The complete five-factor structure emerged in children in primary school based on the teachers’ evaluations. Asendorpf and Van Aken (2003) carried out a 9-year longitudinal study on 230 children, starting at age 4 and ending at age 12. For measuring children's personality, they used inventories such as the California Child Q-Set and bipolar adjective Big Five scales, as well as multiple informants like teachers, parents, and peers. Their results showed that, throughout childhood, Big Five traits related to judgments and behavioral observations of inhibition and aggressiveness. These traits were also linked to school achievement outcomes.
The Big Five personality structure has also been verified in Chinese children. Zhang et al. (2002) used a 14-category coding system, partly based on the Big Five model, to code Chinese and Dutch parents’ free descriptions of their 3- to 14-year-old children's personalities. Of the personality descriptions, 86% from the Chinese parents and 77% from the Dutch parents could be classified into the Big Five adult personality categories.
As mentioned above, many researchers have measured children’s Big Five personality traits using informant reports, such as parent or teacher ratings. However, children's personalities should not be measured solely by the assessments of others. Adult observers’ descriptions of children's personalities may be influenced by their own cognitive categories and personalities (Maćkiewicz & Cieciuch, 2016). These descriptions may also be shaped by the specific situations they witness (Achenbach et al., 1987). Children know their own behavior across different conditions. They are the best informants about their internal states, which cannot always be seen by parents and teachers (Measelle et al., 2005). A meta-analysis by Achenbach et al. (1987) showed that the test–retest reliability of children's self-ratings of behavioral or emotional problems is much higher than the average correlation between children's self-ratings and those from parents, teachers, and mental-health workers. Therefore, it is necessary and valuable to adopt appropriate methods that allow children to evaluate their own personalities.
Cognitive Development of Children for Personality Self-Evaluation
The methods for studying the structure and development of children's personalities should be appropriate for their age and development of cognitive and language skills. According to cognitive development theory, egocentrism decreases between the ages of 4 and 7. By this age, children can better identify things based on features like size, shape, and color (Shaffer, 2009). Children between the ages of 6 and 12 fall into middle childhood (National Research Council, 1984). According to Piaget’s theory of cognitive development, this period corresponds to the concrete operational stage (Piaget & Inhelder, 1969). They start to use psychological constructs and traits to describe these regularities (Shaffer, 2009).
Children in middle childhood may be more reliable at reporting their personality compared to younger children (Clark et al., 2017). Measelle et al. (2005) studied self-reports of Big Five traits in children aged 5, 6, and 7 using the Berkeley Puppet Interview. Two puppets made opposite statements, and the children then described themselves. The researchers rated the children's responses based on which puppet they matched. The results showed that even 6-year-olds could give coherent, distinct reports on their Big Five traits.
Development and Characteristics of the BFI-2
Among the various measurements developed under the Big Five model, the series of scales based on the BFI has been widely used to assess personality in different countries and cultures (Rammstedt et al., 2020; Shchebetenko et al., 2020; Soto & John, 2017b; Vedel et al., 2021; Zhang et al., 2022). Some researchers have also examined the factor structure of the BFI and BFI-2 in adolescents aged 14 years and above (Fossati et al., 2011; Ober et al., 2021).
The BFI uses short phrases to capture the core features of the Big Five. The items combine adjectives that express the dimensions’ key meanings with synonyms or specific explanations. This approach gives the BFI the simplicity and brevity of adjectives but avoids problems with ambiguous or multi-meaning adjectives (John & Srivastava, 1999; Soto & John, 2017a).
On the basis of the BFI, the BFI-2 (Soto & John, 2017a) not only retains the BFI's characteristics of relatively short and easy-to-understand phrases but also tries to achieve bandwidth–fidelity tradeoff by adopting domain–facet hierarchical assessment. Guided by previous research examining hierarchical personality structure and measurement, Soto and John (2017a) defined three facets for each BFI-2 domain: a factor-pure facet, which is identified as central to its own domain and independent from the other four domains, and two complementary facets, which are prominent in the personality literature and represented in the original BFI's item content. In this way, the BFI-2 is not only effective and extensive but also accurate by balancing descriptive breadth at the domain level and specificity at the facet level. Therefore, this study draws on the BFI-2 framework to construct a Big Five personality assessment tool for Chinese children, with items selected to represent diverse facet-related content within each domain.
Combination of Pictures and Text
Some researchers believe that the combination of visual stimulation and verbal stimulation can improve the reliability and validity of children's responses (Frick et al., 2010), since pictures are more concrete and can focus children's attention, stimulate their interest, and improve their participation (Valla et al., 2000). Therefore, some researchers have developed interviews and scales that not only rely on the form of questions and answers but also combine pictorial stimuli, such as Dominic-R (Valla et al., 2000) and the Pictorial Instrument for Children and Adolescents (Ernst et al., 2000). They believe that measuring children's internalized symptoms by pictures is more suitable for younger children (Dubi & Schneider, 2009).
Further, researchers have attempted to use a similar approach to develop a personality scale that is suitable for children's self-assessments. Maćkiewicz and Cieciuch (2016) constructed the Pictorial Personality Traits Questionnaire for Children (PPTQ-C) to measure the Big Five personality traits of children between the ages of 7 and 13. The PPTQ-C is a reliable and valid instrument that creatively combines pictures and words to compensate for the inadequacies of the self-report inventory in evaluating the Big Five personality traits of younger children. The scale contains 15 items, and each dimension contains 3 items. Each item comprises two pictures, showing the low level and high level of the same feature, respectively. Two different scales were used for children of different ages, with a 3-point response scale for younger children (6–9 years old) and a 5-point response scale for older children (10–12 years old). Although the PPTQ-C has expanded the age limit for the self-assessment of children's Big Five personality traits, it has some deficiencies. The PPTQ-C only measures the five dimensions of the Big Five personality traits and does not specifically consider the facets. This may result in the measurement of one dimension focusing on one or two key facets. In addition, some pictures are unsuitable for Chinese social-life situations when applying this scale to Chinese children. For example, there is a set of pictures that ask children “What do you do when someone in class needs … ?” The child in one picture is holding a sandwich, but Chinese children may not be able to tell what it is because not all of them will be familiar with this type of food.
Development of the Chinese Pictorial Big Five Personality Inventory for Children (CPBFI-C)
The specific objective of this study was to develop a short pictorial self-report tool for measuring the Big Five personality traits of Chinese children aged 6–14, referring to the BFI-2 framework and the combination of verbal and visual methods of the PPTQ-C. The importance and originality of this study lies in addressing the limitations of previous inventories by enhancing the breadth of domain-level assessment through the inclusion of diverse facet-related content, as well as by incorporating pictorial stimuli that are adapted to the everyday life contexts of Chinese children.
The items were selected on the basis of the BFI-2 item pool. Compared with the Chinese revision of the BFI-2 by Zhang et al. (2022), the text expression of the items was modified to a certain extent so that the vocabulary could be understood by Chinese children in the middle of childhood. Each facet contains two to three items that are more suitable for children's life. Then, a professional graphic designer illustrated the items in line with Chinese children's life scenes. The preliminary item pool of the CPBFI-C was assessed with a sample of children aged 6–14. Next, we constructed the 15-item CPBFI-C using a combination of empirical and rational criteria. Each item was selected to represent one of the BFI-2 facets, ensuring that the final scale captured diverse facet content within each domain. The selection criteria included confirmatory factor analysis (CFA) factor loadings, the exclusion of items with cross-loadings, theoretical considerations, and reliability.
In this study, we conducted a series of examinations to test the psychometric properties of the CPBFI-C. In Study 1, we examined the CPBFI-C's internal consistency, test–retest reliability, structural validity, and measurement invariance across sex and age. Between approximately 10 and 12 years of age, children enter preadolescence—a transitional period marked by rapid biological, cognitive, and socioemotional changes. During this stage, accelerations in growth, emerging secondary sexual characteristics, heightened emotional reactivity, and expanding cognitive capacities introduce new internal and social challenges (Gilmore & Meersand, 2014). These developmental transformations may influence how personality traits are perceived, expressed, and measured. Therefore, the participants were divided into two age groups (6–9 years and 10 years and above) to examine whether the measurement structure of the CPBFI-C remained comparable across developmental stages. The internal correlations among the dimensions and items were calculated to assess convergent and discriminant validity, and served as a supplementary index of internal consistency. To further evaluate the utility of pictorial stimuli in assessing the Big Five traits in children, we examined the measurement invariance between the original picture-based version and a text-only version of the scale.
In Study 2, we examined the similarity between self- and other-described personalities, which is referred to as self–other agreement (Funder & West, 1993). We also tested the CPBFI-C's external validity by computing the correlations of the CPBFI-C scales with the depression level and academic performance of the children. Studies have shown a strong correlation between children's Big Five personality traits and their academic performance and levels of depression. For example, a meta-analysis showed that the extraversion, agreeableness, conscientiousness, and openness of elementary/middle-school students were significantly positively correlated with their academic performance (Mammadov, 2022). Moreover, Carrasco Ortiz and del Barrio Gándara (2007) found that children and adolescents with high levels of neuroticism or low levels of extraversion, openness, agreeableness, and conscientiousness showed more depressive symptomatology.
Study 1
Method
Data Collection and Participants
A total of 2,519 primary school students were recruited and asked to fill in the initial item pool of the CPBFI-C. The data was collected in an online evaluation system or offline paper-and-pencil tests. The online participants were invited by a county school in Shanxi Province, China, and participated in the questionnaire survey by clicking on a WeChat link to enter the evaluation system. The offline paper-and-pencil tests were conducted in three primary schools in Yunnan Province, Hunan Province, and Beijing. Cases with 20% or more missing values were excluded. For the participants who completed the test–retest assessment, additional screening was conducted based on the test–retest discrepancy scores. Given that children's self-reported personality ratings may be affected by the testing context and socially desirable response styles, and that agreeableness and conscientiousness are among the Big Five dimensions most closely associated with impression management (Connelly & Chang, 2015), the participants whose discrepancy scores on either dimension fell outside the corresponding mean discrepancy score ± a 2.5 standard deviation interval were excluded. This procedure was intended to reduce the influence of potentially context-induced response instability on estimates of trait stability. The remaining 2,484 children comprised Sample 1 and reported an average age of 10.31 years (SD = 1.49), ranging from 6 to 14. In Sample 1, 1,231 children (49.56%) were boys. Their education levels ranged from Grade 1 to Grade 6. For details of the demographic characteristics, see Table 1. Sample 1 was used to examine the CPBFI-C's internal consistency, internal correlations, structural validity, and measurement invariance across sex and age.
Description of the samples.
Subsample 1 was used to assess the test–retest reliability. A total of 191 primary school students from Yunnan Province completed the CPBFI-C twice with a three-week interval. After data screening, the final test–retest sample included 177 children (82 boys, 46.33%). The average age of the children was 9.53 years (SD = 1.53, 7–12 years old), and the education level was between Grade 1 and Grade 5.
In addition, a new group of participants (Sample 2) was used to further validate the CPBFI-C. A total of 2,282 students were recruited from two primary schools and six middle schools in Sichuan Province. In this sample, the students completed the text-only version of the CPBFI-C, which did not include any pictorial elements. This version was used to examine the measurement invariance between the picture-based and text-only versions of the CPBFI-C. There were 1,076 boys (47.15%). The average age of the children was 12.36 years (SD = 0.74, 7–15 years old). Their grade levels ranged from Grade 4 to Grade 7.
The study was approved by the Institutional Review Board of the Institute of Psychology at the Chinese Academy of Sciences. The survey was conducted with the consent of the children’s guardians or the permission and supervision of their school. Before the survey portion of the study, we verbally informed all the participants of the voluntary nature of their participation, as well as their right to withdraw at any time.
Measurement.
The CPBFI-C
The CPBFI-C scale contains 15 items. Each item corresponds to a facet of the BFI-2 and contains a short phrase describing the measured facet and a corresponding picture. The children were asked to carefully observe the similarity between the behavior of themselves and the children with a scarf in the pictures, and then rated it on a 5-point Likert-type scale (1 = very much like me, 2 = kind of like me, 3 = uncertain, 4 = some are not like me, 5 = very unlike me). The mean score of the three items within each domain was calculated (after reverse coding) as the indicator for that domain. The higher the mean score, the higher the level of the corresponding Big Five personality trait.
Data Analysis
All of the data analysis was conducted using SPSS 26.0, Mplus 8.0, and R 4.3.1. Missing data on the CPBFI-C items was minimal (Sample 1: M = 0.22%; Subsample 1 at Time 2: M = 0.23%) and was replaced using the expectation–maximization algorithm.
We examined the internal consistency reliability by first using Cronbach's α and McDonald's ω. In addition to the commonly reported alpha coefficient, we calculated McDonald's total omega (ωt) via the omega() function in the psych package in R, which has been recommended as a more appropriate index of internal consistency when item loadings are not assumed to be equal (Dunn et al., 2014). The test–retest reliability was estimated using the correlation between the Time 1 and Time 2 scores in Subsample 1. Additionally, average inter-item correlations were computed as a supplementary indicator of internal consistency (Agbo, 2010). To further examine whether these internal association patterns varied across developmental stages, we conducted supplementary grade-specific analyses of the intercorrelations among the CPBFI-C domains.
The structural validity of the CPBFI-C was evaluated using the exploratory structural equation modeling (ESEM) framework (Marsh et al., 2010), which integrates the flexibility of exploratory factor analysis with the confirmatory rigor of CFA. The models were estimated with maximum likelihood and oblique geomin rotation (the default in Mplus). The model fit was evaluated using the Comparative Fit Index (CFI), Tucker–Lewis Index (TLI), root-mean-square error of approximation (RMSEA), and standardized root-mean-square residual (SRMR). According to conventional guidelines (Hu & Bentler, 1999; Wen et al., 2004), an acceptable model fit was indicated by CFI and TLI ≥ .90, RMSEA ≤ .08, and SRMR ≤ .08.
Measurement invariance was examined across sex, age groups, and formats (picture vs. text-only version) using CFA models. Tests of measurement invariance were conducted by applying increasingly more stringent equality constraints on the measurement parameters than the prior model. The four levels of invariance tested, from least to most strict, were configural, metric, scalar, and strict. Invariance was supported when ΔCFI < .10, ΔRMSEA < .15, and ΔSRMR < .30 (or < .10 for scalar/strict levels; Chen, 2007). The first model tested configural invariance, ensuring the same items and number of factors across groups. Using the marker–indicator approach (Brown & Moore, 2012; Millsap & Olivera-Aguilar, 2012), the first loading of each factor was fixed to 1 in both groups. The metric invariance model then constrained the factor loadings to equality, indicating that the items contributed similarly to the latent factors across groups. The scalar invariance model further constrained the item intercepts, and the strict invariance model added equality constraints on the residual variances. Subsequently, we also implemented invariance testing using ESEM models, given their advantages in accounting for cross-loadings.
We also employed the alignment approach within the CFA framework to estimate the latent mean differences across sex and test formats (Asparouhov & Muthén, 2014, 2023; Wen et al., 2019). Alignment is a multiple-group latent-variable technique that starts from the configural model and minimizes a total loss function to locate approximate measurement invariance instead of imposing exact equality constraints. With this approach, we tested and reported between-group differences in the five domains without requiring full scalar invariance.
Transparency and Openness
The data set analyzed and the code used during this study are available on the Science Data Bank website. The CPBFI-C scale can be found in the supplementary material.
Results
Internal Consistency and Inter-Item Correlations
The means and standard deviations of the CPBFI-C domains and items are shown in Table 2.
Means, standard deviations, and test–retest reliability of the CPBFI-C domains and items..
Note. E = extraversion; A = agreeableness; C = conscientiousness; N = neuroticism; O = openness; E1 = sociability; E2 = assertiveness; E3 = energy level; A1 = compassion; A2 = respectfulness; A3 = trust; C1 = organization; C2 = productiveness; C3 = responsibility; N1 = anxiety; N2 = depression; N3 = emotional volatility; O1 = aesthetic sensitivity; O2 = creative imagination; O3 = intellectual curiosity.***p < .001.
The Cronbach's α coefficients of the five CPBFI-C domains (see Table 3) for Sample 1 ranged from .43 to .63, with an average of .56, and the McDonald's ω reliability coefficients ranged from .43 to .64, with an average of .56. For Sample 2, the Cronbach's α reliability coefficients ranged from .64 to .84, with an average of .70, and the McDonald's ω reliability coefficients ranged from .64 to .71, with an average of .61. Supplementary grade-level analyses for the picture-based version (Sample 1) using both Cronbach's α and McDonald's ω showed a broadly similar pattern across grades, although some fluctuation was observed across traits and grade levels (see Table S1 in the supplementary material).
Cronbach's α and McDonald's ω for the original picture-based version and the text-only version.
The test–retest reliability was conducted with Subsample 1 and was .64 for extraversion, .65 for agreeableness, .60 for conscientiousness, .70 for neuroticism, and .61 for openness. The average test–retest reliability was .64, indicating moderate temporal stability. Table 2 presents the test–retest reliability of the CPBFI-C domains and individual items.
Table 4 shows the intercorrelations among the CPBFI-C domains and the items in Sample 1. Absolute correlations averaged .26 (total range .13–.40) between the domain scales. Absolute intercorrelations among items within the same domain averaged .30 (.20 for extraversion, .37 for agreeableness, .32 for conscientiousness, .33 for neuroticism, and .28 for openness), compared with only .14 for items across different domains. As an additional analysis, grade-specific intercorrelation patterns were examined (see Tables S2–S8 in the supplementary material). The overall pattern was broadly similar across grades, with average absolute domain correlations ranging from .24 to .31 and same-domain item correlations generally exceeding different-domain correlations. Although some variability was observed, these fluctuations appeared to be trait-specific (e.g., extraversion) rather than systematically grade-related. Overall, these results indicate that the discriminant validity of the CPBFI-C is satisfactory in the total sample and remains broadly stable across grade levels.
Intercorrelations of the CPBFI-C domains and items.
Note. N = 2,484. E = extraversion; A = agreeableness; C = conscientiousness; N = neuroticism; O = openness; E1 = sociability; E2 = assertiveness; E3 = energy level; A1 = compassion; A2 = respectfulness; A3 = trust; C1 = organization; C2 = productiveness; C3 = responsibility; N1 = anxiety; N2 = depression; N3 = emotional volatility; O1 = aesthetic sensitivity; O2 = creative imagination; O3 = intellectual curiosity.
***p < .001. **p < .01. *p < .05.
Structural Validity: ESEM
The tested five-factor ESEM models were excellently fitted to the data in Sample 1, χ2(40) = 80.77, p < .001, CFI = .992, TLI = .980, RMSEA = .020 [90% confidence interval = .014–0.027], SRMR = .010, and it is reasonable that χ2 is almost always significant when the sample size is large (Chen, 2007). The standardized factor loadings of the models are presented in Table 5.
Standardized factor loadings of the five-factor ESEM model.
Note. N = 2,484. E1 = sociability; E2 = assertiveness; E3 = energy level; A1 = compassion; A2 = respectfulness; A3 = trust; C1 = organization; C2 = productiveness; C3 = responsibility; N1 = anxiety; N2 = depression; N3 = emotional volatility; O1 = aesthetic sensitivity; O2 = creative imagination; O3 = intellectual curiosity. Loadings greater than 0.30 in the ESEM model are in bold.
Measurement Invariance Across Sex
Table 6 shows the results of the measurement invariance tests of the CPBFI-C across sex (for the factor loadings and inter-factor correlations of the CFA models, see Tables S9–S12 in the supplementary material). The configural invariance model results showed a good model fit, χ2(160) = 459.87, p < .001, CFI = .941, TLI = .922, RMSEA = .039, SRMR = .035.
Fit indices from the CFA models for the measurement invariance tests of the CPBFI-C across sex.
Note. C3 = responsibility; O1 = aesthetic sensitivity.
***p < .001.
The results demonstrated support for metric invariance, with the constraints of the item factor loading equivalence only producing a minimal change in the model fit, ΔRMSEA = −.001, ΔCFI = −.001, ΔSRMR = .001.
The scalar invariance model was not supported because the changes in indices compared with the metric model (Model 2) were more than the indicator of non-invariance, ΔRMSEA = −.011, ΔCFI = −.045, ΔSRMR = .010. The modification indices indicated that the intercept of Item O1 (“Values art and beauty”) was non-invariant between males and females. When the constraint of this item’s intercept was released, the changes in indices between the partial scalar model and the metric model provided evidence for the partial scalar invariance, ΔRMSEA = .001, ΔCFI = −.004, ΔSRMR = .002. The estimated intercept of O1's reversed score was 3.21 and 4.00 for the male and female group, respectively.
The overall model fit was worse in the strict invariance model compared to the partial scalar invariance model, ΔRMSEA = .005, ΔCFI = −.027, ΔSRMR = .023. Indicated by modification indices, after freely estimating the residuals of O1 (“Values art and beauty”) and C3 (“Sometimes behaves irresponsibly”) across groups sequentially, the partial strict invariance of the CPBFI-C was supported across sex, ΔRMSEA = .001, ΔCFI = −.010, ΔSRMR = .007. The residuals were generally higher for the male group (1.03 for C3; 1.72 for reversed O1) than for the female group (0.71 for C3; 1.04 for reversed O1).
To further examine the robustness of these findings, measurement invariance was also tested using the ESEM framework. The results indicated that the CPBFI-C achieved scalar invariance and partial strict invariance across sex (for the factor loadings and inter-factor correlations of the ESEM models, see Tables S13–S17 in the supplementary material).
Given that full scalar invariance was not achieved under the CFA framework, we further applied the alignment method to compare latent means across groups. The alignment results indicated a good model fit, χ2(160) = 459.87, p < .001, CFI = .941, TLI = .922, RMSEA = .039, SRMR = .035, and all loadings were invariant after alignment. The intercepts of two items (13.33%) were found to be non-invariant, well below the critical threshold of 20% proposed by Asparouhov and Muthén (2014), confirming approximate measurement invariance and allowing for comparison of latent means across sex. The comparison of latent means showed that boys scored significantly lower than girls on agreeableness and conscientiousness, while no significant differences were observed in the other three domains.
Measurement Invariance Across Age Groups
We divided Sample 1 into two subgroups: those aged 6 to 9 years (n = 668, Mage = 8.40, SD = 0.81) and those aged 10 to 14 years (n = 1,816, Mage = 11.01, SD = 0.98). The four models were excellently fitted to the data at the configural, metric, scalar, and strict levels, and the changes in indices (RMSEA, CFI, and SRMR) compared with the prior model were all less than the indicator of non-invariance (see Table 7 and Tables S18–S21 in the supplementary material). The results suggested that the CPBFI-C has a strict measurement invariance across age groups. The results of the measurement invariance testing across age groups using the ESEM model indicated that the CPBFI-C achieved scalar invariance and partial strict invariance (for details see Tables S22–S26 in the supplementary material).
Fit indices from the CFA models for the measurement invariance tests of the CPBFI-C across age groups.
***p < .001.
Measurement Invariance Between the Original Version and the Text-Only Version
The measurement invariance between the original picture-based version and the text-only version of the CPBFI-C was examined using both CFA and ESEM approaches.
The CFA results indicated that the configural and metric invariance models demonstrated an acceptable model fit (for the factor loadings and inter-factor correlations of the CFA models, see Table 8 and Tables S27–S29 in the supplementary material). However, the scalar invariance model did not hold, as the model fit deteriorated substantially after constraining the intercepts to be equal across the versions, ΔRMSEA = .010, ΔCFI = −.034, ΔSRMR = .006. After freeing the non-invariant intercepts of Item 2 and Item 9, partial scalar invariance was achieved, ΔRMSEA = .000, ΔCFI = −.004, ΔSRMR = .000. The strict invariance model was not supported, as it required releasing a large number of residual constraints to reach an acceptable fit.
Fit indices from the CFA models for the measurement invariance tests of the CPBFI-C across the original picture-based version and the text-only version.
Note. C3 = responsibility; E2 = assertiveness.
***p < .001.
In contrast, the ESEM-based invariance testing showed an unacceptable change in model fit at the metric level (for details see Tables S30–S31 in the supplementary material). Therefore, the alignment method was applied to further examine approximate measurement invariance. The results suggested that a “FIXED” alignment model was most appropriate, χ2(160) = 890.528, p < .001, CFI = .947, TLI = .930, RMSEA = .044, SRMR = .036. All factor loadings and most item intercepts were invariant across the two versions, except for Items 2, 3, and 9, whose intercepts showed notable non-invariance.
The aligned factor mean comparisons revealed that the children scored significantly higher on extraversion, agreeableness, conscientiousness, and openness, and significantly lower on neuroticism when responding to the picture-based version rather than the text-only version. Because the latent means derived from the alignment method are expressed relative to a reference group (with the latent means fixed to zero for identification), we further conducted t-tests on the observed scores to provide a more direct comparison between the pictorial and text-only versions (see Table 2). The results were consistent with the alignment results, and the effect sizes at the domain level were small in magnitude (Cohen’s d = 0.09–0.28).
Discussion
In Study 1, we first examined the reliability of the CPBFI-C, which was designed as a brief and developmentally appropriate measure for Chinese children in middle childhood. The average Cronbach's α and McDonald's ω coefficients were both .56, indicating that the internal consistency was modest. This level of internal consistency is comparable to other brief Big Five inventories characterized by similar construct heterogeneity. For instance, the 15-item BFI-2-XS showed average α coefficients of .59 (Soto & John, 2017b), .60 (Gallardo-Pujol et al., 2022), and .62 (Rammstedt et al., 2021) across three validation samples. The internal consistency of the CPBFI-C was acceptable given its brevity and multidimensional design. Each domain was represented by only three items, and these items were intentionally selected to capture diverse facet-related content within each trait domain. Such breadth of content inevitably increases construct heterogeneity and tends to reduce internal consistency, particularly Cronbach's α. The small number of items also leads to a decrease in reliability. Cortina (1993) has shown that a smaller number of items, more dimensions, and lower average item intercorrelation all lead to a decrease in Cronbach's α. Moreover, the grade-specific α and ω estimates showed no substantial decline across grades, and the test–retest coefficients were all above .60, providing additional evidence that the scale yields reasonably stable assessments over time. Although the text-only version showed somewhat higher internal consistency, this likely reflects, at least in part, the older age and more advanced verbal comprehension of the children in that sample. Therefore, the relatively higher reliability of the text-only version does not negate the usefulness of the picture-based format; rather, the latter appears to offer a reasonably reliable and developmentally appropriate brief measure for the younger age group it was intended to assess.
The average absolute correlation between the domain scales was .26, which was similar to that of the BFI-2 (.20 and .24 in two samples, respectively; Soto & John, 2017a). The within-domain correlation, indexed by the average interitem correlation among the items within each domain, was .30 and lower than that reported for the BFI-2 (Soto & John, 2017a). However, as recommended by Clark and Watson (1995), the average interitem correlation was a supplementary index for internal consistency and should fall within .15 to .50, especially for the broad higher-order construct. Thus, the result of the present study was acceptable. The average correlation of items between different domains was .14 and as small as that of the BFI-2, which was .16 in an Internet validation sample and .19 in a student validation sample (Soto & John, 2017a). These results supported the adequate reliability of the CPBFI-C and indicated satisfactory discrimination between the CPBFI-C domains as well as convergence among the items within each domain.
The test results of the structural validity by ESEM showed that the model fit index was good and all items were loaded on the corresponding dimension, indicating the good structural validity of the CPBFI-C and confirming the emergence of five clear personality dimensions of Chinese children in middle childhood.
We further examined the measurement invariance of the CPBFI-C across different sex and age groups using CFA and ESEM. The measurement invariance testing across sex demonstrated that the CPBFI-C maintained comparable factor structures and factor loadings for boys and girls, supporting its cross-sex applicability. The scalar invariance across sex, however, was only partially supported. The intercept constraint of one item (“Values art and beauty”) representing aesthetic sensitivity was released. The partial metric invariance model showed that girls scored higher on aesthetic sensitivity than boys. This result may be due to boys and girls having different perceptions of art. Istók et al. (2009) found that, when participants were asked to list adjectives describing the aesthetic value of a musical piece without listening to any music during the task, female participants more frequently listed adjectives such as “beautiful,” “touching,” and “feelings evoking,” whereas male participants more frequently listed “ugly” and “drab.” Strict invariance across sex was also partially established. The residual of two items (“Values art and beauty” for aesthetic sensitivity and “Sometimes behaves irresponsibly” for responsibility) was non-invariant and generally higher for the male group, which may be due to the influence of social desirability. Although full scalar invariance was not achieved, the alignment results confirmed approximate measurement invariance, with only 13.33% of the intercepts showing non-invariance. These findings indicate that the latent mean comparisons across sex are meaningful. The observed lower latent means for boys in agreeableness and conscientiousness align with prior developmental literature suggesting that girls tend to exhibit higher prosociality and self-regulation during middle childhood (Else-Quest et al., 2006).
The CPBFI-C demonstrated strict invariance across the younger (6–9 years) and older (10–14 years) age groups, indicating that its measurement properties remain stable across the transition from late childhood to early adolescence. This developmental period, often characterized by rapid cognitive, emotional, and social changes (Gilmore & Meersand, 2014), could potentially disrupt the consistency of self-perceived traits. However, the invariance results suggest that children as young as 6 can reliably understand and evaluate personality-related behaviors. This supports the feasibility of using self-report measures to capture the Big Five traits even during preadolescence, a stage often marked by heightened variability in self-concept and affective regulation.
The measurement invariance analyses confirmed that the picture-based CPBFI-C, as the core version of the instrument, demonstrated solid psychometric equivalence when compared with its text-only counterpart. Although the CFA and ESEM results did not support strict invariance, the alignment method showed that all factor loadings and most item intercepts were approximately invariant across the two formats, with an acceptable proportion of intercept non-invariance observed for Items 2, 3, and 9. These findings indicate that the inclusion of pictorial elements did not alter the underlying construct representation of the Big Five traits. Importantly, the picture-based format was designed to enhance comprehension and engagement among younger children, and the present results confirmed that this adaptation did not compromise the measurement validity. Moreover, despite the picture-based group being younger (Mage = 10.31) than the text-only group (Mage = 12.36), both formats exhibited comparable model structures and reliability. In this context, the observed mean differences between the picture-based and text-only versions are more likely to reflect developmental differences rather than format-related measurement bias. Consistent with the disruption hypothesis (Soto & Tackett, 2015), children tend to report higher levels of socially desirable traits, whereas these traits show a temporary decline during early adolescence. Accordingly, the relatively higher scores in the picture-based group are likely attributable to age-related normative trends rather than the influence of pictorial versus text-based item presentation.
Study 2
Method
Participants
Study 2 tested the self–other agreement and external validity of the CPBFI-C using three subsamples from Sample 1. Subsample 2 comprised 573 children who completed the questionnaire through the online evaluation system. One of their parents was also invited by the schoolteacher to evaluate the corresponding child's Big Five personality traits through the online evaluation system. The questionnaires of the parents and children were matched by WeChat ID. The data was used for the test of self–other agreement. Subsample 2 was independent of Subsample 1. In Subsample 2, there were 284 boys (49.56%); the average age of the children was 11.51 years old (SD = 1.04); the age range was 6–14 years old; and the education level was between Grade 1 and Grade 6. The average age of the parents was 37.23 years old (SD = 4.24). Four hundred and fifty-five questionnaires (79.41%) were completed by mothers.
Subsample 3 and Subsample 4 were used for the test of external validity. Subsample 3 was independent of Subsample 1 and Subsample 2. The primary school students from Hunan Province were asked to evaluate their depression level with the offline paper-and-pencil test. After excluding non-responses and data with no less than 20% missing values on the corresponding scale, 1,094 students comprised Subsample 3. Among them, 536 (48.99%) were boys; the age range was 9–14 years old (M = 10.13, SD = 0.90); and the education level was between Grade 4 and Grade 6.
Subsample 4 was independent of Subsample 1 and Subsample 2, but included some participants from Subsample 3. Of the participants who completed the questionnaire through the paper-and-pencil test, 1,281 students’ corresponding class teachers evaluated the students’ academic performance. In Subsample 4, 1,018 students were from Hunan Province (79.47%) and 263 students were from Beijing. There were 650 boys (50.74%); the average age was 9.71 years old (SD = 1.28); the age range was 7–14 years old; and the education level was between Grade 2 and Grade 6.
Measurement
The CPBFI-C.
The BFI-2-S
The parents in Subsample 2 were asked to assess their corresponding child's Big Five personality traits with the short form of the BFI-2 (BFI-2-S; Soto & John, 2017b), which contains six 5-point Likert items (1 = strongly disagree to 5 = strongly agree) for each domain. The facets in the BFI-2-S are the same as the BFI-2, and each facet contains two items. In this sample, the Cronbach's α for the BFI-2-S domains was .33 for extraversion, .68 for agreeableness, .64 for conscientiousness, .58 for neuroticism, and .66 for openness.
Depression
The children in Subsample 3 were asked to complete the Children’s Depression Inventory (CDI) (Kovacs, 1992) based on their experiences during the past week. The Inventory includes 27 items. Each item has three options describing different levels of depressive symptoms, which are scored from 0 to 2, respectively. The least severe degree of depression is denoted by 0 and the most severe degree of depression by 2. The depression score is the sum of the 27 items. Higher scores indicate more severe depressive symptoms. The Chinese version showed good psychometric properties among Chinese children and adolescents (Wu et al., 2010). The Cronbach's α was .84 in this study.
Academic Performance
The corresponding class teachers were asked to subjectively assess each child's academic performance in Subsample 4 by answering the question “How is the child's academic performance?” on a 5-point Likert scale (1 = outstanding to 5 = dreadful). The academic performance scores were scored in reverse when analyzing the data.
Data Analysis
The data analyses were carried out using SPSS version 26. The missing data from the Children's Depression Inventory was replaced using the expectation–maximization algorithm as the number of missing values was minimal (for the 27 items of the Inventory in Subsample 3, M = 1.80%). The self–other agreement was examined by computing the correlations between the children's self-rated and parent-rated Big Five personality traits in Subsample 2 at the domain level as well as at the item level. Finally, to test the external validity, we calculated the correlation between the level of self-rated depressive symptoms of the children in Subsample 3 and the academic performance of the children evaluated by the teachers in Subsample 4 and the CPBFI-C at the domain and item levels. Because the measurement model showed only approximate rather than exact scalar invariance across sex in Study 1, we further applied the alignment-with-covariates framework using the aligned SEM model (Asparouhov & Muthén, 2023) to estimate the associations between the Big Five traits and external variables across groups under approximate measurement invariance. To address potential developmental differences, supplementary grade-specific analyses were conducted for self–other agreement and criterion-related validity.
Results
Self–Other Agreement
The Pearson correlations of the children's self-rating Big Five personality traits and their parents’ other rating are shown in Table 9. The results showed that the self–other agreement of all the domains and items was significant. The average correlation between the domains was .31 and the average correlation between the items was .22. The domain with the highest correlation between self and other evaluation was openness (r = .34). At the item level, the two items with the highest self–other agreement were C1 (organization, r = .31) and N3 (emotional volatility, r = .29).
Correlation between the children's self-rating Big Five personality and their parent-rating Big Five personality, self-rating depression level, and teacher-rating academic performance.
Note. E = extraversion; A = agreeableness; C = conscientiousness; N = neuroticism; O = openness; E1 = sociability; E2 = assertiveness; E3 = energy level; A1 = compassion; A2 = respectfulness; A3 = trust; C1 = organization; C2 = productiveness; C3 = responsibility; N1 = anxiety; N2 = depression; N3 = emotional volatility; O1 = aesthetic sensitivity; O2 = creative imagination; O3 = intellectual curiosity.***p < .001. **p < .01. *p < .05.
Given the very small sample sizes in Grades 1–3 (n < 30), grade-specific analyses were interpreted primarily for Grades 4–6. As shown in Table S32 in the supplementary material, the average domain-level self–other correlations showed a generally increasing trend across these grades, from .28 in Grade 4 to .30 in Grade 5 and .47 in Grade 6, indicating a tendency toward stronger self–other agreement in the higher grades.
External Validity
The results of the correlations are presented in Table 9. The CPBFI-C domains were significantly correlated with self-reported depression level and teacher-reported academic performance. The domain with the highest correlation with self-rated depressive symptoms was neuroticism (r = .41), and the corresponding item with the highest correlation was N2 (depression, r = .44). The domain with the highest correlation with teacher-reported academic performance was openness (r = .26), with the highest item-level correlation observed for O3 (intellectual curiosity, r = .20).
To further examine whether the associations between the children's Big Five personality traits and external criteria (depressive symptoms and academic performance) differed across sex, we applied the aligned SEM model within the alignment-with-covariates framework (Asparouhov & Muthén, 2023) in Mplus. Using the fixed alignment approach, the models achieved approximate measurement invariance across sex—depressive symptoms: χ2(180) = 391.315, p < .001, CFI = .925, TLI = .900, RMSEA = .047, SRMR = .043; academic performance: χ2(180) = 385.740, p < .001, CFI = .929, TLI = .906, RMSEA = .043, SRMR = .040—and showed that the majority of loadings and intercepts were comparable between boys and girls. Moreover, no significant group differences were found in the intercepts of the external variables, suggesting that the mean levels of depressive symptoms and academic performance did not differ significantly between boys and girls.
Based on the results of the aligned SEM model, the overall pattern of relationships between the Big Five domains and the external criteria appeared to be broadly similar across sex groups. After controlling for all five traits simultaneously, neuroticism positively predicted depressive symptoms (for male: β = .30, p = .045; for female: β = .62, p < .001), whereas openness negatively predicted depressive symptoms (for male: β = −.12, p = .229; for female: β = −.19, p = .008) and positively predicted academic performance (for male: β = .29, p = .008; for female: β = .27, p = .003). These results suggest a broadly comparable pattern of associations across sex groups, although the strength of some specific coefficients differed.
Additional grade-specific alignment-with-covariates analyses were conducted to examine whether the relations between the Big Five domains and the external criteria varied across grade. For depressive symptoms, the comparison was conducted across Grades 4, 5, and 6. For academic performance, because the Grade 3 sample was relatively small (n = 64), the comparison was limited to Grades 2, 4, 5, and 6. In both analyses, no significant between-grade differences were found in either the intercepts of the external variables or the regression relations between the Big Five domains and the external criteria.
Discussion
The results on self–other agreement showed that the correlations between the children's self-reported Big Five traits and their parents’ ratings were significant, although modest in magnitude. This relatively low level of agreement may partly reflect the differences in assessment format between the child and parent versions, as well as the relatively young age of the children in the present study. Notably, the supplementary grade-specific analyses, although exploratory in nature, suggested a tendency toward stronger self–other agreement in the higher grades. This pattern is broadly consistent with Barbaranelli et al.’s (2003) study, which found that older children's self-reports tended to converge more strongly with ratings provided by other informants, especially parents. Moreover, also in line with Barbaranelli et al. (2003), openness was the factor on which children and parents converged most in the present study.
The external validity of the CPBFI-C was supported by significant correlations between the criteria and both the domain scores and item-level indicators. The alignment results supported the robustness of the CPBFI-C structure across sex and indicated that the observed associations between the Big Five traits and external criteria were not biased by measurement non-invariance. In other words, the relationships between the children's personality traits and both depressive symptoms and academic performance can be meaningfully compared across sex groups. The small but significant correlation between Neuroticism and academic performance is consistent with the findings of Andersen et al. (2020), who replicated this association in two large samples of students from Grades 4 to 8, each comprising more than 100,000 participants. Among the five traits, openness—especially its item reflecting intellectual curiosity (O3)—had the strongest correlation with academic performance, which echoes Mammadov’s (2022) research on primary school students. The alignment-with-covariates analysis also revealed that openness was the most salient predictor of academic performance when all five traits were considered simultaneously. A possible explanation is that openness is a personality trait that reflects cognitive exploration (DeYoung, 2015). It refers to the ability and tendency to discover, understand, and apply complex patterns of both sensory and abstract information. According to Mammadov (2022), the significant correlations between extraversion/agreeableness and academic performance among primary school students may be attributed to two factors. First, primary school learning often emphasizes group activities and classroom interactions, where extraverted and agreeable students tend to participate more actively and cooperate effectively. Second, since teachers rated students’ academic performance in this study, their evaluations likely reflected not only academic outcomes but also students’ engagement and social adaptability, which are central to extraversion and agreeableness. When it comes to secondary or post-secondary education stages, which emphasize independent learning and objective assessment, such correlations become weaker or vanish. The results also reflect those of Carrasco Ortiz and del Barrio Gándara (2007). They used the same inventory to measure children's depression level and also found that children with lower levels of neuroticism or higher levels of extraversion, agreeableness, conscientiousness, and openness showed lower levels of depressive symptoms.
General Discussion
The present research aimed to develop and validate a culturally appropriate self-report measure of the Big Five personality traits for Chinese children in middle childhood. Given the limitations of parent and teacher evaluations in capturing children's internal characteristics (Beitchman & Corradini, 1988), it is important to provide children with a self-report tool that is appropriate to their cognitive and developmental level. Compared with younger children, children in middle childhood may be better able to provide meaningful self-reports of their personality traits (Clark et al., 2017). To support children's comprehension and engagement, the present study adopted a combined pictorial and verbal format, following prior work suggesting the usefulness of this approach in child personality assessment (Maćkiewicz & Cieciuch, 2016). At the same time, the facet-informed structure of the BFI-2 provided a useful framework (Soto & John, 2017a), taking into account both breadth and conceptual accuracy. Guided by these considerations, the present study developed the CPBFI-C as a pictorial self-report measure for Chinese children aged 6 to 14, with items selected to capture diverse facet-related content within each domain.
Taken together, the findings from Study 1 and Study 2 provide robust evidence for the psychometric soundness and practical utility of the CPBFI-C. The good psychometric properties of the CPBFI-C across the two studies indicate that the five-factor personality model was satisfactorily replicated among Chinese children in middle childhood (Herzhoff et al., 2017; Shiner & Caspi, 2003). Moreover, the results suggest that the patterns of item responses reflecting different facet-related content are meaningfully organized within each domain, which is consistent with the structure proposed in the BFI-2 (Soto & John, 2017a).
More specifically, the ESEM results from Study 1 supported a clear five-factor structure, and the measurement invariance analyses demonstrated that this structure is largely equivalent across sex and strictly invariant across age groups. This stability is noteworthy, given the rapid developmental changes during this period, and underscores the robustness of the Big Five framework in childhood (Gilmore & Meersand, 2014). The comparison between the picture-based and text-only versions (Study 1) revealed approximate measurement invariance, indicating that the pictorial elements aid comprehension without fundamentally altering the constructs being measured. Although the pictorial version did not show clear psychometric superiority over the text-only version, its value lies in making personality self-reporting more accessible and developmentally appropriate for Chinese children. Because children's self-reports can capture internal experiences that are not fully observable to parents or teachers (Measelle et al., 2005), an age-appropriate self-report format remains important even when the two versions are largely equivalent psychometrically.
Study 2 extended these findings by linking the CPBFI-C to parent ratings and external criteria. The significant self–other agreements, although modest, are comparable to those found in other child personality studies and provide initial evidence for the convergent validity of children's self-perceptions (Barbaranelli et al., 2003). In addition, the associations of personality traits with depressive symptoms and academic performance supported the external validity of the CPBFI-C. The alignment-with-covariates analysis revealed theoretically coherent and sex-invariant associations between personality traits and key outcomes.
These findings also highlight the practical significance of the CPBFI-C for child personality assessment. By providing a developmentally appropriate format for younger children's Big Five self-reporting, the CPBFI-C may help address the need for age-appropriate personality measures in childhood (Vicentini et al., 2025) and facilitate research on personality development earlier in life, particularly in the Chinese cultural context.
Limitations
The current study has several limitations. First, the main character in the pictures of the CPBFI-C is a boy, which may lead to different feelings of involvement between boys and girls when completing the questionnaire. From the perspective of social learning theory, research on sex differences in children's imitation learning suggests that children may exhibit a preference to imitate same-sex patterns in some situations (Perry & Bussey, 1979). In future picture drawings and scale construction, it might be possible to distinguish the sex of the main character.
Second, each facet is represented by a single item, which limits the ability to obtain robust reliability estimates at the facet level. Accordingly, we recommend focusing on the five domain scores in practical applications. Some researchers have proposed that nuances, a meaningful level of the trait hierarchy below facets, may help to better describe individual differences (Mõttus et al., 2017). A nuance corresponds roughly to a single item (or groups of very similar items) in a facet scale (McCrae, 2015). Future research could further incorporate the concept of nuances and expand the number of items representing each facet, thereby enabling more comprehensive and reliable facet-level measurement.
Third, further work is required to provide more evidence of external validity. Although the teacher-rated academic performance used in this study is more reliable than self-reports, future research should incorporate more objective academic records (e.g., standardized test scores) for validation. Additionally, an examination of the relationships with other behavioral criteria, such as aggressive behavior and other externalizing problems, would further strengthen the evidence of the scale's validity. Future research should also include comparable external criteria for both the pictorial and text-only versions, which would allow for a more direct evaluation of criterion-related validity across formats.
Conclusions
In conclusion, we have demonstrated that the CPBFI-C is a valid and reliable self-report measurement of Chinese children's Big Five personality traits, as evidenced by its structural validity, measurement invariance across key demographics, significant self–other agreement, and theoretically consistent relationships with external criteria. The successful application of this tool in a diverse sample from various Chinese cities underscores its broad applicability. Chinese children in middle childhood can indeed self-evaluate their Big Five personality traits in an interesting and psychometrically sound way. Most importantly, we have also provided more evidence for a similar Big Five personality structure in children compared with adolescents and adults at the domain level, while demonstrating that a brief domain-level measure can capture diverse facet-related behavioral content, thereby helping to bridge the personality assessment gap between middle childhood and later developmental stages.
Supplemental Material
sj-docx-1-pac-10.1177_18344909261466376 - Supplemental material for Development and Validation of the Chinese Pictorial Big Five Personality Inventory for Children (CPBFI-C): Evidence of Reliability, Validity, Self–Other Agreement, and Measurement Invariance
Supplemental material, sj-docx-1-pac-10.1177_18344909261466376 for Development and Validation of the Chinese Pictorial Big Five Personality Inventory for Children (CPBFI-C): Evidence of Reliability, Validity, Self–Other Agreement, and Measurement Invariance by Weiqi Mu, Fugui Li, Long Zhang, Xue Li, Siying Li, Ninghao Zhu, Hui Cao and Mingjie Zhou in Journal of Pacific Rim Psychology
Footnotes
Ethical Considerations
This study was approved by the Institutional Review Board of the Institute of Psychology at the Chinese Academy of Sciences (protocol number: H20019). The survey was conducted with the consent of the children’s guardians or the permission and supervision of their schools. Before the survey portion of the study, we verbally informed all the participants of the voluntary nature of their participation, as well as their right to withdraw at any time.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the 13th Five-Year Plan Project of Beijing Education Sciences (grant number: BEAA19043), and the CAS Key Laboratory of Mental Health, Institute of Psychology, Chinese Academy of Sciences (grant number: KLMH2018GJ03).
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Data Availability Statement
Supplemental Material
Supplemental material for this article is available online.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
