The following is a summary of the development of the Cross-cultural (Chinese) Personality Assessment Inventory (CPAI), and some related researches. For a more detailed summary, please refer to the book chapter by Cheung, Cheung, and Fan (2013).
The CPAI was developed in the 1990s. The construction of this personality inventory was initiated by Prof. Fanny Cheung in Hong Kong, and was carried out by a team of scholars coming from the Department of Psychology, the Chinese University of Hong Kong, and the Institute of Psychology, the Chinese Academy of Science. The CPAI was constructed by adopting a combined etic-emic approach – an approach that takes into account both universal and cultural-specific aspects of Chinese personality. The goal of the research team, when developing the CPAI, was to construct an inventory that is relevant to the Chinese cultural context, and would meet the scientific psychometric standard expected of established assessment instruments; in order to provide the Chinese society with a reliable, valid, and useful personality measure.
Historical Background
Most of the assessment tools in the 1980s and 1990s were imported from Western countries. Although this "imposed etic" method might have saved certain amount of research efforts, it also brought about some problems that could not be overlooked. Besides the sometimes-questionable qualities of translations and adaptations, problems also lay in how the tools were used and how the scores were interpreted. Specifically, most psychological assessment tools at that time were translated or adapted from Western measures based on Western personality theories, and were standardized in Western samples. Whether these scales could be applied to non-Western cultures need to be validated in the local context: "cross-cultural validity" and "cross-cultural equivalence". Methodologically, an adapted inventory is expected to demonstrate its cross-cultural validity and equivalence with the original measure at the item and the scale level, including linguistic equivalence, construct equivalence, psychometric equivalence, and psychological equivalence (Cheung, 2009; Fan, Cheung, Zhang, & Cheung, 2011; International Testing Commission, 2010). These standards are important but sometimes neglected. Researchers have noticed that, for example, when adopting a scale from one culture to another, its factor structure may not be perfectly replicated, and its reliability tends to reduce (Leong, Cheung, Zhang, Song, & Xie, 1998). Cheung and colleagues have found that the norms of the Minnesota Multiphasic Personality Inventory 2 (MMPI-2) may not be directly applicable to the Chinese population – if the American norms were to be applied without questions, a certain amount of normal Chinese individuals would be misclassified as having depression and schizophrenia (Cheung, Zhang, & Song, 2003). This shows that careless usage of adapted scales can potentially harm a person’s interest, or even his/her wellbeing.
Besides cross-cultural validity and equivalence, the cultural-relevance of the scales alto needs to be considered. In main-stream psychology, it has been assumed that personalities are stable, biologically-based, and cultural-independent characteristics. Cross-cultural research has identified cultural biases of some personality assessment inventories. Personality traits are formed in the interactions between the individual and his/her cultural environment, and could be viewed as ways in which the individual adapts to the environment (Cheung et al., 2013). The cultural perspective should be mainstreamed in Psychology (Cheung, 2012). This notion has received increased support from empirical studies, which found indigenous cultural-specific traits that were not represented in Western personality scales, especially in collectivistic cultures (e.g. Choi, Kim, & Choi, 1993; Verma, 1999; Yang, 1996).
In short, adapted scales could be used only when they have met the psychometric standards by having appropriate translations and revisions; however, even after this process of "fitting the shoes by chopping one's toes", these scales are still unable to comprehensively represent all personality aspects of non-Western populations. In the view of these problems, Prof. Fanny Cheung and Prof. Weizhen Song decided to collaborate with scholars in Hong Kong and Mainland China, to develop a personality inventory that is more culturally relevant to the Chinese population in 1990. Prof. Kwok Leung from the Chinese University of Hong Kong, and Prof. Jianxin Zhang from the Institute of Psychology, the Chinese Academy of Science, were also invited into the research team. After more than 20 years of development and revision, the CPAI is now an inventory that could be used for both normal personality testing and clinical assessment. It has an adolescent version (the CPAI-A; Cheung, Fan, Cheung, & Leung, 2008; Cheung, Leung, & Cheung, 2005) and an adult version (the CPAI-2; Cheung, Cheung, & Zhang, et al., 2008; Cheung, Leung, Song, & Zhang, 2001), and has been translated into English (Cheung, Cheung, Leung, Ward, & Leong, 2003), Japanese (Wada, Cheung, & Cheung, 2004), Korean (see Cheung et al., 2013), Vietnamese (Dang, Weiss, Tran, & To, 2010), Dutch (Born & Jooren, 2009), and Romanian (Iliescu & Ion, 2009). China contains one fifth of the world’s population, and has a long cultural history. It is hoped that the development of this inventory can help scholars to more comprehensively and accurately understand the personality characteristics of individuals living in this culture, and in turn contribute to the field of cross-cultural personality research.
Scale Development
The combined etic-emic approach has been adopted for the development of CPAI. Methodologically, it follows the practices of constructing a scientific measure that were well-established in Psychology. To independently create a personality inventory that is rooted in the local Chinese culture, the scale’s item content did not rely on adapting existing Western scales: instead, its development basically followed the "inductive – factor analysis – deductive" steps of scale development (Burisch, 1984). The specific method involved in the development of the CPAI modeled that of the development of MMPI-2 and MMPI-A (Butcher, Graham, Williams, & Ben-Porath, 1990), which included: determining personality aspects, assessing the item-subscale correlations, revising different aspects of scale content, and finally analyzing all subscales statistically and provide them with reasonable descriptions. Validity scales were also added into the Inventory at the later stages.
CPAI (for details, see Cheung et al., 2013; Cheung, Leung, Fan, Song, Zhang, & Zhang, 1996)
At the inductive stage of "determining personality aspects and generating items", as the research team wanted to obtain comprehensive and diverse personality descriptions – especially those that are more typical in the Chinese culture but are not reflected in Western personality models – they adopted a bottom-up strategy of information collection, and collected relevant personality descriptions from various sources, like popular contemporary Chinese novels, Chinese proverbs, informal street survey, and psychological literature (see Cheung et al., 1996, for details). For clinical scales, these sources included: (1) diagnostic criteria used in Hong Kong and Mainland China; and (2) applications of the Chinese MMPI and MMPI-2 clinical and content scales. By sorting and integrating the collected personality descriptions, 38 personality dimensions were extracted, giving the preliminary trial version of the CPAI 38 personality scales, including 26 normal personality scales and 12 clinical scales.
After the personality dimensions were determined, 20 -30 items were generated for each dimension. Besides these personality items, learning from the MMPI, the research team also gave the trial version of the CPAI a validity scale, i.e. the Infrequency Scale. A total of approximately 900 items were included in the CPAI trial version, which were screened for language difficulty level by junior high school Chinese language teachers; in addition, a small sample of respondents from Mainland China and Hong Kong with different educational and occupational backgrounds also reviewed the items for language fluency and content relevance. After deleting and revising certain items, the first edition of CPAI (i.e. the 1991 trial version) consisted of 906 items, with 589 items for the 26 normal personality scales, 280 items for the 12 clinical scales, and 37 items for the Infrequency Scale. A true-false response format was adopted.
1167 adults from different regions of Mainland China and 580 adults from Hong Kong were recruited to complete the 1991 trial version of the CPAI. Basing on the data collected, the research team deleted items with low item-total correlations or discriminability, or with overlapping meanings; personality dimensions that were highly correlated with each other were subsequently combined. As a result, the final 1991 version of the CPAI contained 22 normal personality scales, 12 clinical scales, and 1 validity scale; each scale carried at least 15 items, and the total number of items was 514
In 1992, the research team conducted a large scale testing of the 514 items to standardize the 1991 version of the CPAI; the normative sample was recruited via random household sampling in Hong Kong and quota sampling in Mainland China, which included adults with 18 - 65 years of age, and primary-or-higher education levels. This sample included minority ethnic group members. The final normative sample consisted of 1884 adults from Mainland China, and 441 adults from Hong Kong. By analyzing the data obtained from this standardization sample, number of scales were combined. Two additional validity scales, i.e. the Good Impression Scale, and the Response Consistency Index were developed with reference to the norms. Results of factor analysis extracted 4 normal personality dimensions: Social Potency, Dependability, Accommodation, and Interpersonal Relatedness, which accounted for 59% of the total variance; whereas the clinical dimensions extracted 2 factors: Emotional Problems, and Behavioral Problems, which accounted for 55.8% of the variance. In addition, with two additional testing sessions among college students, the research team found that the one-week and one-month test-retest reliability of the scales ranged from 0.94 to 0.56.
CPAI-2 (For details, see Cheung, Cheung, & Zhang, 2004a; Cheung, Cheung, & Zhang, et al., 2008)
Compared to the CPAI, the Cross-cultural (Chinese) Personality Assessment Inventory – 2 (CPAI-2) has two major revisions. 1. The items of the clinical scales were increased to cover more symptoms and widen the range of scores; correspondingly, the number of items in the normal personality scales was reduced. 2. Personality dimensions related "Openness", which were missing in the original CPAI, were added. Although Openness has not been a significant personality dimension even among other indigenous Chinese personality scales, the research team decided to incorporate Openness-related personality dimensions into the second version of the CPAI, in order to investigate whether the lack of Openness was the result of the unique Chinese personality structure. Keywords and behavioral descriptions representing the Openness characteristics were collected via a combined etic-emic approach, which involved literature review, unofficial interviews, and focus groups interviews. These keywords and behavioral descriptions were then used to produce personality dimensions; after which, items were generated respectively for each new dimension. After screening, piloting, and revision, 6 new scales related to Openness were added into the CPAI, forming the CPAI-2, and underwent a re-standardization process.
Re-standardization took place in 2001 and 2002, using the same sampling methods and procedures as was adopted when standardizing the CPAI in 1991. A total of 1575 valid questionnaires were returned from 6 major areas of Mainland China, and another 336 were returned from Hong Kong. Factor analysis of the re-standardization data suggested that, as in the original CPAI, the same 4 factors could be extracted from the normal personality scales, and another 2 factors from the clinical scales. The 6 newly added "Openness" scales were not extracted as one independent factor; instead, 4 of those scales loaded on the Social Potency factor (namely, "Novelty", "Diversity", "Divergent Thinking", and "Aesthetics") together with extraversion and leadership to characterize dynamism and expansiveness; the remaining 2 loaded either on the Accommodation factor or the Interpersonal Relatedness factor. This shows that Openness may not be an independent dimension in the Chinese personality taxonomy, but instead a complex one, as it involves not only the cognitive diversity and openness that are expressed in the Individualistic culture, but also social openness and acceptance that are treasured in the Collectivistic culture. After revision and re-standardization, the final version of the CPAI-2 has 28 normal personality scale, 12 clinical scales, and 3 validity scales, with a total of 541 items (details).
CPAI-A (For details, see Cheung, Fan, Cheung, & Leung, 2008; Cheung, Cheung, & Fan, 2013)
To provide a comprehensive personality assessment tool for the Chinese people, and to better understand the development of personality throughout the lifespan, an Cross-cultural (Chinese) Personality Assessment Inventory – Adolescent Version (CPAI-A) was developed in 2003. The CPAI-A was constructed using the CPAI-2 as the blueprint, with its development procedure largely following that of the latter. Certain personality dimensions that were more relevant to adolescents were added, such as "Sensation Seeking", "Life Goal", and "Eating Disorder". Some dimensions in the CPAI-2 that were deemed as relatively unimportant to the contemporary generation of adolescents, like "Thrift vs Extravagance", were deleted. CPAI-A was standardized in Hong Kong and Mainland China in 2005. The Hong Kong standardization sample was collected with a quota sampling method, data was collected from 2689 adolescents aged 12-18, including secondary school students and out-of-school adolescents (Cheung, Fan, Cheung, & Leung, 2008). The Mainland Chinese sample included 2744 adolescents from different regions throughout the country (see Cheung, Cheung, & Fan, 2013). The final version of the CPAI-A contained 25 normal personality scales, 14 clinical scales, and 3 validity scales, with a total of 509 items. Factor analysis of the standardization data suggested 4 factors for the normal personality dimensions: Social Potency/Expansiveness, Dependability, Emotional Stability, and Interpersonal Relatedness; and 2 factors for the clinical dimensions: Emotional Problems and Behavioral Problems (details).
Cross-Cultural Research on the CPAI
With both universal and culture-specific personality dimensions, the CPAI allows researchers to compare their own results with, and to make references to, published studies on universal personality dimensions (Cheung, 2006). In addition, the CPAI has been translated into different languages, which facilitated cross-cultural research on personality with particular reference to the relationship between universality and cultural-specificity of personality.
Church (2010) acknowledged that CPAI is one of the few indigenously-derived personality measures that have demonstrated incremental values in assessing personality beyond simply "manifestations" of a universal personality structure in a particular culture. The indigenous factor of the CPAI, Interpersonal Relatedness, is not covered in other Western personality scales.
Comparison of the CPAI to NEO-FFI
Joint factor analysis of the CPAI-2 and NEO-FFI produced 6 factors (Cheung, Cheung et al., 2008). None of the NEO-FFI scales loaded on the Interpersonal Relatedness factor, at the same time, "Diversity", "Divergent Thinking", and "Aesthetics" of the CPAI-2 all loaded on the "Openness" factor. In short, the 6 factors obtained were the five factors from the Big Five Model plus an Interpersonal Relatedness factor, suggesting that Interpersonal Relatedness is a culturally specific personality dimension derived from the Chinese culture.
Universality and Cultural-Specificity of the CPAI
In addition to studies comparing the CPAI to Western personality measures, other empirical studies have also shown that the cultural-specific CPAI factor or scales have incremental power in explaining various outcome variables in applied settings (for details, see below).
Results of various cross-cultural investigations suggested that, on one hand, the cultural-specific dimension of Interpersonal Relatedness exists not only in the Chinese culture, but also in other non-Chinese societies; on the other hand, this dimension may be less important to the daily lives of these non-Chinese populations, especially in individualistic cultures, and has therefore been receiving little attention from them (Fan et al., 2011).
The stability and replicability of the CPAI structure have been supported by various cross-cultural comparison studies (Cheung, Cheung, Howard, & Lim, 2006; Fan, Cheung, Zhang, & Zhang, 2011; Lin & Church, 2004). In one such study that employed a sample of Chinese, Chinese American, European American, Korean, and Japanese, a factor analysis with Procrustes Rotation (a method commonly utilized in Big Five researches; McCrae, Zonderman, Costa, Bond, & Paunonen, 1996 ), using the four factor structure of CPAI-2 normal personality as the target model, showed that the four factor structure fitted the data well in all five samples, with the consistency index of Social Potency and Dependability being 0.95 – 0.97, and that of Accommodation and Interpersonal Relatedness being 0.89 – 0.97. Good fits were also obtained in subsequent Confirmatory Factor Analysis ran with the five samples. In addition, Explorational Factor Analysis was run in two studies utilizing a Dutch college sample and a Romanian adult sample, respectively (Born & Jooren, 2009; Iliescu & Icon, 2009). Although the extraction of a four-factor structure was unsatisfactory in these two studies, the factor Interpersonal Relatedness had been retrieved as a unique factor in both studies, again demonstrating the existence of this indigenous factor in other cultures.
Applications of the CPAI
One important step in establishing a scale’s validity is to conduct a program of empirical studies to validate the utility of the scale in different settings and for different purposes.
In Industrial and Organizational Psychology, the universal CPAI factor, the Social Potency factor, had significant power in predicting work performances and self-rated leadership behaviors; the cultural-specific CPAI dimension, the Interpersonal Relatedness factor, was shown to have incremental value in cross-sectional analysis of MBA students’ job ranks, in evaluating hotel employees’ customer-orientation, and in predicting leadership behavior of high-ranked executives.
In terms of social relationship and behavior, the universal dimensions, such as "Optimism" and "Self-Acceptance", were significantly related to life satisfaction of Chinese people; cultural-specific dimensions, such as "Family Orientation", "Harmony", "Social Sensitivity", "Graciousness", "Relationship Orientation", and "Face", were also found to be closely related to variables such as filial piety, trust, interpersonal influence strategies, trust in society, style of interaction, civic behaviors, emotional intelligent, creativity, life satisfaction, and sexual identity. In addition, when explaining adolescents’ career development, while the universal dimensions of Social Potency and Dependability were significantly related to career exploration, the cultural-specific dimension of Interpersonal Relatedness also had sizable incremental value.
For details, please refer to Cheung & Cheung, 2002; Cheung, Fan, & To, 2008; Cheung, Fan, & Yao, 2012; Cheung, Zhang, & Cheung, 2010.
Clinical applications of the CPAI have been evaluated in a number of clinical studies involving psychiatric patients (Cheung, 2007; Cheung, Cheung, & Leung, 2008). The clinical scales of the CPAI were negatively correlated with life satisfaction measures in the standardization. Convergent validity of the CPAI with reference to the MMPI-2 was conducted. Significant correlations were obtained between the clinical scales of the CPAI and the Depression, Anxiety, Health Concern, and Distortion of Reality scales of the MMPI-2 (Cheung, Cheung, & Zhang, 2004b). However, certain cultural differences were observed. Specifically, the two cultures had different social norms on behaviors that may be symptomatic of mental disorders. For example, due to the modesty and self-suppression that are encouraged in the Chinese culture, Chinese people tended to score high on MMPI-2 items that suggested depression and anxiety (e.g. items from "Depression" and "Neurasthenia" on the MMPI-2). The normal personality scales especially those related to interpersonal relationship (Family Orientation, Face, Ah-Q Mentality) are particularly helpful in predicting social adjustment among patients.