Adolescence is considered to be a critical transition period that shapes physical and mental health trajectories throughout life. However, extensive research has demonstrated an increased susceptibility to mental disorders in teenagers, adversely affecting developmental outcomes in personal, interpersonal, and educational domains (Castellanos-Ryan et al., 2016; Costello et al., 2011). Echoing this trend, recent research in China revealed that the prevalence rates of depression in adolescents ranged from 24.3 % to 36.6 %, while anxiety ranged from 36.2 % to 43.5 % (Chang et al., 2022; Wu et al., 2022). The COVID-19 outbreak has further aggravated the occurrence of mental health problems among Chinese youth (Chai et al., 2021; Ma et al., 2021). It is thus imperative and urgent to identify developmental antecedents contributing to psychopathology, as these would imply promising targets for intervention.

One influential body of research on vulnerability for psychopathology has emanated from the two-polarity model of personality development (Blatt, 2004), highlighting self-definition and interpersonal relatedness as two fundamental developmental dimensions. A healthy personality develops from a dialectically balanced interplay between these dimensions, whereas impairment in the development of one or both trajectories can lead to increased vulnerability to depression. For instance, impairment in the development of interpersonal relatedness can lead to compensatory dependency and anaclitic depression. On the other hand, impairment to self-definitional development can lead an individual to suffer from self-critical perfectionism and introjective depression. Specifically, dependency involves separation anxiety and a need to be close to loved ones, whereas self-criticism involves a preoccupation with failing to meet expectations (Blatt, 1995). A large volume of studies have associated both dependent and self-critical vulnerabilities with a wide range of psychopathology (Werner et al., 2019), such as externalizing problems (Vandenkerckhove et al., 2019), eating disorders (Duarte et al., 2014), anxiety disorders (Shahar et al., 2015), and suicidality (Sobrinho et al., 2016). Therefore, self-criticism and dependency have been identified as transdiagnostic personality vulnerabilities that contribute to the maintenance of various forms of psychopathology (e.g., Campbell et al., 2018).

The Depressive Experiences Questionnaire (DEQ; Blatt et al., 1976) has become a prevalent tool for measuring personality vulnerability (Stange et al., 2015). The 66-item DEQ measures three factors, including dependency, self-criticism, and the less commonly addressed factor of efficacy. Notably, to preserve each item's distinct contribution, Blatt et al. (1976) developed a unique factor weighting program, which used means, standard deviations, and factor score coefficients from a sample of American, non-clinical undergraduates to calculate standardized factor scores. However, researchers using the DEQ have complained that there is limited evidence for this instrument's psychometric properties. First, the original scoring method raises concerns about its applicability across diverse cultures and populations. For instance, studies in Belgium and Italy have shown that the original DEQ did not satisfy the criteria in their samples (Desmet et al., 2007; Falgares et al., 2018). Additionally, comparisons between genders are challenging, as the DEQ uses different scoring weights for males and females (Welkowitz et al., 1985). Moreover, the DEQ's length and complex scoring procedure have raised concerns among many researchers (e.g., Flett et al., 1995; Welkowitz et al., 1985). Lastly, the DEQ's orthogonality hypothesis (Blatt et al., 1976) has been questioned due to evidence of strong correlations between dependency and self-criticism (e.g., Desmet et al., 2007).

Researchers have since streamlined and condensed the DEQ in various ways. For example, Welkowitz et al. (1985) selected 43 items with high and differential loadings and used a unit-weighted scoring method to create the Revised DEQ (RevDEQ). Another attempt, the 19-item Reconstructed DEQ (RecDEQ; Bagby et al., 1994), demonstrated high and differential loadings across samples of non-clinical and clinical adults. Additionally, Viglione et al. (1995) created the Theoretical DEQ-21 (TDEQ-21) by selecting 21 items aligning with Blatt's concepts and loading over 0.40 in the original sample. They further refined it to the 12-item Theoretical DEQ-12 (TDEQ-12), with higher loadings in both community and clinical samples. To match the orthogonality of self-criticism and dependency, Santor et al. (1997) developed the McGill DEQ by selecting 48 items. Furthermore, a 20-item adolescent version of the DEQ was developed with modified language to suit teenagers' cognitive abilities (DEQ-A; Blatt et al., 1992; Fichman et al., 1994). To sum up, all revised DEQs have shortened the items list and used a unit-weighted scoring method.

Preliminary research comparing the psychometric properties of the revised DEQs has yielded differing results. In Desmet et al. (2007)’s sample of Belgian undergraduates, the RecDEQ had the best structural validity, good predictive validity, and adequately low intercorrelations between dependency and self-criticism. In contrast, Falgares et al. (2018) suggested that the TDEQ-21 and TDEQ-12 (TDEQ-21/12) were the most valid measures in samples of Italian undergraduates and clinical patients. Finally, Lin et al. (2021a) recommended the RecDEQ for use in China based on its construct validity and internal consistency in a sample of Chinese college students. However, the model fit failed to meet psychometric standards and predictive validity was unexamined.

While the DEQ-A was specifically revised for adolescents, other short versions have also been widely employed for this age group (e.g., Curran et al., 2017; Kopala-Sibley et al., 2017). However, research on the psychometric properties of the short DEQs in adolescents has been notably lacking. Personality vulnerability has been found to change over the course of adolescent development (Vandenkerckhove et al., 2020). Cognitive, social, emotional, and biological differences are evident between early, middle, and late adolescence, as well as emerging adulthood, suggesting a complex developmental process (Klimstra et al., 2009; Slobodskaya, 2021; Salmela-Aro, 2011). Hence, the lack of measurement invariance tests for the short DEQs across time and age groups weakens conclusions in studies of adolescents. Similarly, without cross-gender measurement invariance testing, significant teenaged gender differences in dependency and self-criticism (e.g., Campos et al., 2014; Falgares et al., 2018) are vulnerable to criticism. Furthermore, criterion-related validity among adolescents is needed, given that personality vulnerabilities might contribute to various forms of youth psychopathology (Campbell et al., 2018). However, previous studies have primarily used depression and interpersonal problems in adulthood for criterion-related validity, rather than considering a broader range of ill-being (Desmet et al., 2007; Falgares et al., 2018). A wider approach can further inform dependency and self-criticism as transdiagnostic processes underlying multiple mental disorders.

To address the need for psychometrically sound short measures of personality vulnerability among adolescents in China, the present study aims to investigate the psychometric properties of various short versions of the DEQ in a representative community youth sample. Furthermore, this study will allow a better understanding of how personality vulnerability evolves through adolescence. Based on previous research, the McGill DEQ has shown weak construct validity (Desmet et al., 2007; Falgares et al., 2018). Thus, the Rev-DEQ, RecDEQ, TEDQ-21/12, and DEQ-A were considered potential candidates to investigate in the present study. Additionally, the two-polarity model of personality development does not include an efficacy dimension, and prior research indicated that the three-factor model of the DEQ showed poor model fit (Blatt et al., 1992; Desmet et al., 2007; Falgares et al., 2018). Therefore, we mainly examined the reproducibility of the two-factor model of self-criticism and dependency among short DEQs. Moreover, we evaluated measurement invariances across gender, time, and age groups. To assess the transdiagnostic potential of self-criticism and dependency (e.g., Campbell et al., 2018), we used the Mental Health Inventory of Middle-School Students (MMHI-60; Wang et al., 1997) as the criterion. The MMHI-60 assesses ten areas of psychological disturbance in Chinese adolescents, such as depression, anxiety, obsessive-compulsiveness, and so on.