Facilitating International Research on Career Indecision: Developing Career Indecision Profile-Short-5 in China and the U.S.

Abstract

While five-factor model of career indecision proposes five cross-culturally important indecision factors, a psychologically sound brief measure of career indecision that corresponds to the five-factor model is lacking. Thus, using a Chinese college sample (n = 588) and an U.S. college sample (n = 762), this study developed and validated the Career Indecision Profile-Short-5 factor (CIP-Short-5) in China and the U.S. Applying item response theory, the CIP-Short-5 consists of five items each for the five overarching domains of career indecision. It showed desirable option occurrence patterns and minimal gender-oriented differential item functioning in both China and the U.S. Additionally, the scale supported the five-factor model over an alternative, four-factor model of career indecision in both China and the U.S. The convergent and discriminant patterns of the CIP-Short-5 with criteria were largely supported in China and the U.S. Last, the results supported the configural and metric invariance of the CIP-Short-5 but did not fully support the scalar invariance of the CIP-Short-5 across China and the U.S. Together, the results offer psychometric evidence for the CIP-Short-5, which has important implications for research and practice on career indecision in the international context in general and in China in specific.

Keywords

career indecision five-factor model item response theory indecision measurement measurement invariance

Career indecision, which denotes a state of being unable to select an educational or career direction, has been one of the key topics in career counseling and development (Kulcsar et al., 2020; Xu & Bhang, 2019). Although decidedness itself is not the ultimate goal of career development, developing a sense of career direction benefits individuals in career preparation and transition (Krumboltz, 2009; Savickas, 2012). When examining career indecision, research has often focused on a key process aspect of career indecision, career decision-making difficulty, which describes factors impeding the career decision-making process. Centering on career decision-making difficulty, research has devoted great effort to developing the taxonomy and measurement of career indecision (Kulcsar et al., 2020; Xu & Bhang, 2019). In general, there have been two strategies in this research line: theory-oriented and data-oriented (Osipow, 1999; Xu & Bhang, 2019). Recently, Brown and colleagues empirically developed the 65-item Career Indecision Profile-65 (CIP-65) based on a meta-analytic analysis of existing indecision data (Brown et al., 2012; Hacker et al., 2013). Comparing the data-driven CIP model with theory-oriented models (e.g., Gati et al., 1996; Saka et al., 2008), Xu and Bhang (2019) concluded that a five-factor model could comprehensively and parsimoniously summarize all major indecision factors. This model is particularly favored over the original CIP model in East Asia, such as China and South Korea (Carr et al., 2014; Roche et al., 2017). However, a psychologically sound brief measure of career indecision that corresponds to the five-factor model is lacking in China. Additionally, a brief indecision measure that operates consistently on the five-factor model in the international context is lacking, which limits cross-culturally examination of career indecision. Thus, the focus of the present study was to develop the Career Indecision Profile-Short-5 factor (CIP-Short-5) in China and the U.S.

Taxonomy and Measurement of Career Indecision

Roughly, there have been four generations of indecision models (Osipow, 1999; Xu & Bhang, 2019). The first generation, which is represented by the Career Decision Scale (Osipow et al., 1976), does not involve an a priori structural model of career indecision. Rather, it developed items based on clinical observations, and its validity essentially rests upon its empirical ability to differentiate decided and undecided individuals. The second generation, which is represented by the Career Factors Inventory (Chartrand et al., 1990), developed items based on a predetermined theoretical model of indecision. Although this theory-oriented approach enhanced the theoretical meaningfulness of identified factors, the model itself (involving two major categories of difficulties) is not systematically rooted in decision-making science and the career decision-making process. Therefore, the third generation, which is represented by the Career Decision Difficulties Questionnaire (CDDQ; Gati et al., 1996) and the Emotional and Personality Career Difficulties Scale (EPCD; (Saka et al., 2008), adopted more sophisticated theoretical models that apply decision-making science in the context of career decision-making. However, residual issues existed regarding limited content representativeness and structural inconsistency of various theory-driven models. Thus, the fourth generation attempted to empirically extract and summarize factors from existing research to pursue a comprehensive and yet parsimonious model of career indecision.

The pursuit of a comprehensive model started with Kelly and Lee’s (2002) model, evolved through the Career Decision Profile (CIP; Brown et al., 2012; Brown & Rector, 2008), and currently arrived at Xu and Bhang’s (2019) integrative five-factor model. Kelly and Lee (2002) innovatively conducted factor analysis of existing measures related to career indecision and revealed a six-factor structure in the U.S. Brown and colleagues (Brown et al., 2012; Brown & Rector, 2008) used more powerful meta-analytic strategies and revealed a four-factor structure based on existing matrices related to career indecision in the U.S. Their four-factor model consists of neuroticism/negative affectivity (NNA), choice/commitment anxiety (CCA), lack of readiness (LR), and interpersonal conflicts (IC). However, research has shown a cross-cultural variation of the CIP model across Western and Eastern contexts in that while the four-factor holds in Western contexts (e.g., Italy and Iceland; (Abrams et al., 2013; Carr et al., 2014), a revised five-factor CIP model, which retains NNA, LR, and IC but divides original CCA into need for information (NI) and CCA, is more valid in Eastern contexts (e.g., South Korea and China; Abrams et al., 2014; Roche et al., 2017). Recently, Xu and Bhang (2019) summarized the literature pertaining to career indecision and particularly compared the CIP, the CDDQ, and the EPCD models. They concluded based on their conceptual exploration and empirical reanalysis that the five-factor CIP model could serve as a normative structure of career indecision, while the original four-factor CIP model represents a parsimonious variant of the five-factor model specific to Western contexts.

Notably, although the four-factor model might be an empirically parsimonious representation of the five-factor model in Western contexts, it does not necessarily indicate that choice/commitment anxiety and need for information are conceptually overlapping with each other. The dual-process theory of career decision-making (DTC; Xu, 2021a, 2021b) may be able to explain the difference and association of CCA and NI. The DTC was developed to address the challenges that Parsons’ (1909) seminal three-step model faces in the contemporary psychosocial context of career decision-making (Blustein et al., 2019; Krieshok et al., 2009; Lent & Brown, 2020; Savickas, 2015). It highlights that decision uncertainty regarding choosing the “right” educational or occupational direction cannot be eliminated during a given window of career decision-making. Accordingly, the DTC differentiates the management of uncertainty that can be removed (termed confusion) and that will likely persist (termed ambiguity). Viewed through the predictive model of the DTC (see Figure 1), need for information likely represents a barrier in confusion management, while choice/commitment anxiety likely represents an intermediate outcome of ambiguity management. However, these two constructs are also interrelated because information deficit and anxiety in commitment can give rise to each other. Thus, the five-factor model might be a theoretically more tenable model or a model that provides more nuanced guidance regarding intervention, even in Western contexts.

Figure 1.

Predictive Model of Dual-Process Theory. Note. NA = Neuroticism/negative affectivity; CCA = Choice/commitment anxiety; NI = Need for Information; LR = Lack of Readiness; IC= Interpersonal conflicts. Adapted from Xu, H. (2021). Career decision-making from a dual-process perspective: Looking back, looking forward. Journal of Vocational Behavior, 126, 103,556. https://doi.org/10.1016/j.jvb.2021.103556.

To measure the five factors of career indecision, there have been two primary measures: the 65-item CIP-65 (Hacker et al., 2013) and the 20-item CIP-Short (Xu & Tracey, 2017). While the CIP-65 was developed originally to measure the four CIP factors, it can be reconfigured to measure the five factors of career indecision, as shown in its Korean and Chinese versions. However, its length creates a thorny barrier for its use in indecision research, which often involves measures of other constructs and can rarely afford a 65-item measure for a single construct. Additionally, the length of the CIP-65 discourages practitioners from using the scale as a routine progress marker during intervention. Given these limitations, Xu and Tracey (2017) used item response theory (IRT) to develop the CIP-Short in the U.S. based on the CIP-65. They evaluated items in terms of ordered option occurrence and minimal gender-oriented differential item functioning (DIF) and retained the five best items each for the four CIP factors. While this tool could serve as a psychometrically sound brief measure of career indecision in the U.S. (Xu, 2020), obviously it does not strictly align with the five-factor model, resulting in a barrier in using it in China and South Korea. Given the complementary strength and weakness of the CIP-Short and its parental CIP-65, an important measurement need arises: to develop a brief indecision measure that corresponds to the integrative five-factor model of indecision. Such a measure is expected not only to facilitate indecision research and practice in East Asia, such as China, but also to pave the measurement foundation for cross-cultural research on career indecision.

Developing the CIP-Short-5 is particularly important in China. As China’s education system increasingly emphasizes the importance of career development during high school and college, the challenge of career indecision has been increasingly recognized (Xu et al., 2014). Although research has examined the cross-cultural validity of the CDDQ, EPCD, and CIP-65 in China (e.g., (Hou et al., 2015; Roche et al., 2017), a brief measure that builds on the integrative five-factor model is lacking. The integrative five-factor model is suitable to Chinese students because as Xu and Bhang (2019) summarized, it is empirically a better representation than the original CIP model in China and fits the context of career exploration and decision-making in China. Traditionally, career decision-making in China has never been an independent process because Confucianism emphasizes parental approval and family interests in this process (Leung et al., 2011). This collectivistic orientation often entails interpersonal conflicts in career decision-making. Economically, China has been experiencing rapid economic growth and transition, which creates difficulty in committing to a single choice based on present occupational information. Together, the cultural and economic contexts could elevate ambiguity in Chinese students’ career decision-making and consequently attenuate the connection between NI and CCA (see Figure 1) that has been often observed in developed Western countries. Given the importance and lack of a brief indecision measure corresponding to the five-factor model in the international context in general and in China in particular, the purpose of the present study was to develop the Career Indecision Profile-Short-5 factor (CIP-Short-5) in China and the U.S.

Development Strategy of the CIP-Short-5

Following the IRT strategy used in the development of the CIP-Short (Xu & Tracey, 2017), we first used IRT to shorten the CCA and NI subscales of the CIP-65 and examined the NNA, LR, and IC items identified by the CIP-Short in China, then validated the resultant CIP-Short-5 in the U.S., and finally examined the measurement invariance of the CIP-Short-5 (Vandenberg & Lance, 2000; Xu & Traceya, 2017) across China and U.S. There were three sets of considerations behind this strategy. First, the development process was driven mainly by applying IRT to CCA and NI items in Chinese college students because no brief measure of indecision exists in China and the five-factor model is particularly applicable in the Chinese context. However, we also retained IRT-derived NNA, LR, and IC items in the CIP-Short to pursue a consistent form of indecision measurement across China and the U.S. This consistent form of indecision measurement can enable cross-cultural comparison. Although alternatively we could develop the CIP-Short-5 first in the U.S. by applying the five-factor structure and then validate the scale in China, this strategy could risk retaining items that do not conform to the five-factor structure in China. The reason is that the structural separation of CCA and NI is more pressing in China than in the U.S. and CCA and NI items derived in the U.S. might not fully reflect this structural feature in China, which is a primary goal of this study.

Second, we selected IRT as the testing framework because while both IRT and classic test theory (CTT) describe the relation between a latent construct and test responses, IRT has advantages over CTT. More specifically, IRT enables an examination of item performance in relation to the different levels of the latent construct, whereas CTT offers only an omnibus indicator of item quality that speaks to the entire span of the latent construct (Whittaker & Worthington, 2016; Wirth & Edwards, 2007). In other words, IRT provides latent-construct-variant descriptions of item performance, while CTT provides latent-construct-invariant descriptions of item performance. Thus, once CTT has established the factor structure of a construct, IRT could reveal more nuanced information regarding item performance, which is desirable for selecting the best items and abbreviating a measure. One important criterion in our IRT analyses was gender fairness, which addresses whether the same level of latent constructs was associated with the same item score across man and woman. Gender difference has been a common area when investigating the variance of the CIP scores across demographic groups (Hacker et al., 2013; Xu & Tracey, 2017). More importantly, research has shown that women tend to score higher than men on NNA (Hacker et al., 2013). This result could reflect gender-biased NNA items and biasedly direct intervention to focusing on NNA for women. Additionally, gender is one of the most visible and important demographic constructs in China. Thus, the criterion of minimal gender difference in item functioning is important for the current development of the CIP-Short-5.

Last, we validated the CIP-Short-5 in China by examining its adherence to the five-factor model and its convergent and discriminant correlational pattern with related indecision measures. Following Brown et al.’s (2012) factor analytic findings and Xu and Bhang’s (2019) analysis, we chose self and identity of the EPCD, anxiety of the EPCD, lack of information of the CDDQ, career decision-making self-efficacy, inconsistent information of the CDDQ as the criteria for NNA, CCA, NI, LR, and IC of the CIP-Short-5, respectively. These criteria have been used as criteria in the development research of the CIP-Short (Xu & Tracey, 2017). The validation of the CIP-Short-5 in the U.S. was subjected to the same structural examination. Because research has supported the construct validity of the CIP-65 in the U.S. (Hacker et al., 2013; Xu & Tracey, 2017), we used the CIP-65 as a convenient criterion and examined the convergent and discriminant correlational pattern of the CIP-Short-5 with the parental CIP-65.

Summary of the Present Study

While the integrative five-factor model of career indecision (Xu & Bhang, 2019) has great potential to serve as a cross-cultural taxonomy of indecision, a brief indecision measure that corresponds to this model is lacking not only in the Chinese context but also in the international context. To facilitate cross-cultural indecision research in general and indecision research in China in specific, we drew on the five-factor model and used IRT to develop the CIP-Short-5. We anticipated that the CIP-Short-5 would show an adequate fit to the five-factor model and NNA, CCA, NI, LR, and IC of the CIP-Short-5 would be correlated mainly with self and identity of the EPCD, anxiety of the EPCD, lack of information of the CDDQ, career decision-making self-efficacy, inconsistent information of the CDDQ, respectively, in China. We also anticipated that the CIP-Short-5 would show an adequate fit to the five-factor model and NNA, CCA, NI, LR, and IC of the CIP-Short-5 would be correlated mainly with their corresponding subscale of the CIP-65 in the U.S.

Method

Participants and Procedures

The study participants consisted of 588 college students primarily from six universities across China with a mean age of 20.58 years (SD = 2.51). Of the participants, 29.8% (n = 175) identified as man, 69.4% (n = 408) identified as woman, and .9% (n = 5) identified as gender nonbinary. They reported a variety of academic majors that are related to education and psychology, information technology, language, business, law, and engineering. Of the sample, 29.9% (n = 176) identified as freshman, 4.3% (n = 25) identified as sophomore, 54.6% (n = 321) identified as junior, and 4.1% (n = 24) identified as senior. In terms of social economic status (SES), the participants reported a mean of 4.81 (SD = 1.78) on a ladder representing SES in China, which ranges from 1 (worst off) to 10 (best off). After obtaining institutional review board approval, we disseminated an online survey that included demographic questions and research questionnaires through social media and professional networking in China. Once potential participants read the cover letter of this study, they could decide if they wanted to participate voluntarily. All participants’ responses remained confidential and anonymous throughout the study. The Chinese dataset did not involve missingness.

To examine the psychometric performance of the CIP-Short-5 in the U.S., we reanalyzed the dataset that was used to develop the CIP-Short in the U.S. (Xu & Tracey, 2017). The dataset consisted of 762 undergraduate students recruited from a southwest state university in the U.S. They ranged in age from 17 to 47 (M = 19.41, SD = 2.85). Of the participants, 37.0% identified as man (n = 282) and 62.1% identified as woman (n = 473). In terms of race/ethnicity, 8.1% (n = 62) identified as African American/Black, 8.1% (n = 62) identified as Asian/Asian American, 22.2% (n = 169) identified as Latino (a)/Hispanic, 52.8% (n = 402) identified as White, 1.8% (n = 14) identified as Native American, and 3.9% (n = 30) identified as Multiracial. In terms of major, 71.6% (n = 546) were enrolled in a major and career exploratory program, and 27.7% (n = 211) had declared a major. The missingness rate of the American dataset was minimal across variables (<1.5%), and we used the Full Information Maximum Likelihood estimation (FIML) to handle missingness in structural examination (Schlomer et al., 2010).

Measures

Among Chinese Participants

The Career Indecision Profile-65 (CIP-65). We used the Chinese version of the CIP-65 (Roche et al., 2017) to measure the five aspects of career indecision: neuroticism/negative affectivity (NNA), need for information (NI), choice/commitment anxiety (CCA), lack of readiness (LR), and interpersonal conflicts (IC). The Chinese version of the CIP-65 was developed through a translation-back-translation procedure on the English, 65-item version of the CIP (Hacker et al., 2013). Sample items (in an abbreviated form) were “Often feel ashamed,” “Rather keep myself open than committing,” “Need a better idea of my talents,” “Confident I’ll achieve goals,” and “Important people discourage plans” for NNA, CCA, NI, LR, and IC, respectively. Participants were asked to rate each item on a 6-point Likert scale ranging from 1 (completely disagree) to 6 (strongly agree). Higher scores indicated greater difficulty. Roche et al. (2017) reported alpha coefficients ranging from .79 (IC) to .91 (CCA) in a Chinese sample and found support for the structural validity of the scale in the Chinese context. The current study revealed alpha coefficients of .93, .89, .87, .89, and .87 for NNA, NI, CCA, LR, and IC, respectively.

The Career Decision Difficulties Questionnaire (CDDQ). We used the Chinese version of the CDDQ (Hou, 2010) to measure three cognitive aspects of career indecision, consisting of lack of readiness, lack of information, and inconsistent information. The Chinese version of the CDDQ was translated from the English 44-item CDDQ (Gati et al., 1996) using the standard translation-back-translation procedure. Two sample items were “I find it difficult to make a career decision because I do not know what steps I have to take” and “I find it difficult to make a career decision because I’m equally attracted by a number of careers and it is difficult for me to choose among them.” Participants were instructed to rate each item on a 9-point Likert scale ranging from 1 (does not describe me) to 9 (describes me well). Higher subscale scores indicated greater difficulty in the corresponding area. Research has reported alpha coefficients of .62, .93, and .91 for the three subscales and supported the validity of the Chinese CDDQ in its associations with career exploration, cultural-value conflicts, and parental expectations (Leung et al., 2011; Xu et al., 2014). The current study found alpha coefficients of .72, .96, and .94 for the three subscales of the CDDQ.

The Emotional and Personality Career Difficulties Scale (EPCD). We used the Chinese version of the EPCD (Hou et al., 2015) to measure the three emotional and personality-related aspects of career indecision, consisting of pessimistic views, anxiety, and self-concept and identity. The Chinese version of the EPCD was translated and validated based on the original English version of the EPCD (Saka et al., 2008). Two sample items were “I often feel inferior to others” and “I am already considering a certain career, but am afraid it might not suit my personality.” Participants rated each item on a 9-point Likert-type scale ranging from 1 (does not describe me at all) to 9 (describes me well). Higher ratings indicated greater emotional and personality-related career indecision. Hou et al. (2015) revealed alpha coefficients of 0.78, 0.85, and 0.93 for the three subscales and found support for the validity of the Chinese EPCD in its associations with trait anxiety and vocational identity. The current study found alpha coefficients of 0.71, 0.90, and 0.84 for pessimistic views, anxiety, and self-concept and identity, respectively.

The Career Decision Self-Efficacy-Short Form (CDSE-SF). We used the Chinese version of the Career Decision Self-Efficacy Scale-Short (CDSE-SF; Long, 2003) to measure career decision-making self-efficacy. The Chinese version of the CDSE-SF was developed based on the 25-item English version of the CDSE-SF (Betz et al., 1996). The CDSE-SF measures self-efficacy in five important skill domains of career decision-making (Betz et al., 1996). Two sample items were “Accurately assess your abilities” and “Decide what you value most in an occupation.” Participants responded to each item on a 5-point Likert scale ranging from 1 (no competence at all) to 5 (complete competence). Higher total scores indicated greater career decision-making self-efficacy. Research has reported an alpha coefficient of .94 for the Chinese CDSE-SF and supported the validity of the scale in its associations with career adaptability and parental support (Guan et al., 2016; Long, 2003). The current study revealed an alpha coefficient of .94.

Among American Participants

The Career Indecision Profile-65 (CIP-65). The English version of the CIP-65 (Hacker et al., 2013) measures career indecision due to NNA, CCA, NI, LR, and IC. The CIP-65 was originally designed to measure the four-factor CIP model (Brown et al., 2012). However, based on CIP research in China and South Korea (Abrams et al., 2014; Roche et al., 2017) and Xu and Bhang’s (2019) integrative five-factor model, the CIP-65 could also be reconfigured to assess the five factors of career indecision. Participants rated each item on a 6-point Likert scale ranging from 1 (completely disagree) to 6 (strongly agree). After reverse coding for reverse items, higher subscale scores indicated greater indecision due to the corresponding area. Hacker et al. (2013) supported the validity of the CIP-65 in its differential scores across students enrolled in career planning vs. other undergraduate courses. They also found that the CIP-65 scores were negatively correlated with career decidedness. The current study revealed alpha coefficients of 0.94, 0.93, 0.91, 0.91, and 0.87 for NNA, CCA, NI, LR, and IC, respectively.

Analysis

Development and Validation of the CIP-Short-5 in China

Following the development procedure of the CIP-Short in the U.S. (Xu & Tracey, 2017), we used nonparametric IRT in the software of jMetrik to identify the five best items each for CCA and NI in China. We also used the same strategy to examine the performance of NNA, LR, and IC items that were originally identified by the CIP-Short in the U.S. to ensure that they are psychometrically adequate in China (i.e., without major deviations from the standards detailed below). Nonparametric IRT does not fit data to an a priori model (such as logistic functions; (Embretson & Reise, 2010) and thus does not hold a restrictive prerequisite while revealing the relationship between latent constructs and item/option responses (Sijtsma & Molenaar, 2002). Such flexibility is desirable when the purpose of IRT pertains to differentiating item performance (like the current study) as opposed to quantifying population parameters. Simulation research has shown that for parametric IRT on multidimensional polytomous responses, a sample size of 500 is adequate for most scenarios (Jiang et al., 2016). Additionally, nonparametric IRT requires a smaller sample size than parametric IRT (Stout, 2001). Thus, the current sample size in both countries (>500) was adequate. Before evaluating item performance, we first conducted exploratory factor analysis (EFA; principal axis factoring with direct oblimin rotation) on CCA and NI items to ensure that the unidimensionality assumption of IRT for each subscale was met (Ansley & Forsyth, 1985).

We used two criteria to evaluate item performance in the IRT framework: ordered option occurrence and minimal gender-oriented differential item functioning (DIF). First, we examined the option characteristic curves (OCCs) of each item, which delineates the relation between a latent construct and the probability of selecting an option (see Figure 2 for an example). The goal was to evaluate whether the occurrence of each option followed a bell-shaped pattern (i.e., occurrence probability increases first and then decreases) and whether the OCCs of each item progressed in an ordered fashion (i.e., as the latent construct increases, the occurrence probability of larger options increases and eventually exceeds the occurrence probability of smaller options). The criterion of ordered OCCs addresses whether item options are effective, which has been commonly used to shorten long scales. Second, we examined the item characteristic curve (ICC) of each item, which delineates the relation between a latent construct and the expected item score (see Figure 3 for an example). The purpose was to evaluate whether the same level of latent constructs was associated with the same item score across man and woman. The criterion of minimal gender DIF addresses whether scores can be fairly compared across man and woman.

Figure 2.

Option Characteristic Curves for Effective (above) and Less Effective (below) Items.

Figure 3.

Item Characteristic Curves for Items with Small (above) and Large (below) Gender DIF.

We used the two main criteria holistically in that we first rated each item in terms of the two criteria separately on a scale ranging from 1 (poor) to 3 (perfect). Items were rated lower when they demonstrated greater deviation from the perfect OCC and ICC patterns. Then, we used aggregate scores across the two criteria to rank items and selected the best five items. We selected five items for each dimension because this length (per dimension) matches the CIP-Short and has been a common choice for abbreviated measures in the literature (Xu & Tracey, 2017). There was only one exception to the ranking-oriented procedure. We did not retain two CCA items that tap into the changeability of interests (although they were ranked high) because the CIP50 already addresses changeability in relation to commitment. To broaden content representation, we selected CIP49 instead.

To validate the CIP-Short-5 in China, we examined the structural, convergent, and discriminant validities of the scale. Regarding its structural validity, we conducted confirmatory factor analysis to comparatively examine the four-factor and five-factor CIP models of career indecision in China. Following Hu and Bentler’s (1999)and Kline’s (2015) recommendations, we employed the following criteria holistically to evaluate the model-data fit of the structural alternatives: robust χ², comparative fit index (CFI) > 0.90, root mean square error of approximation (RMSEA) <.08, and standardized root mean square residual (SRMR) < 0.08. To enhance the robustness of results to the potential nonnormality of data, we adopted robust maximum likelihood (ML) parameter estimation in Mplus 8. Additionally, we examined the correlations of the CIP-Short-5 with the CDDQ, EPCD, and CDSE-SF to shed light on its convergent and discriminant validity.

Validation of the CIP-Short-5 in the U.S.

Following the same nonparametric IRT procedure abovementioned, we examined the item performance of the CCA and NI subscales in the U.S. to ensure that their psychometric performance is adequate in the U.S. (i.e., without major deviations from desirable OCCs and ICCs). Additionally, we examined the structural validity of the CIP-Short-5 in the U.S. by comparatively examining the four-factor and five-factor CIP models of career indecision. Using the CIP-65 as the criterion, we examined the correlations of the CIP-Short-5 with the CIP-65 to evaluate the extent to which the CIP-Short-5 can retain information captured by the parental CIP-65.

Measurement Invariance across China and the U.S.

We examined the measurement invariance of the CIP-Short-5 between China and the U.S. using the classic confirmatory factor analysis approach to measurement invariance (Vandenberg & Lance, 2000). We progressively examined configural, metric, and scalar models that corresponded to three levels of measurement invariance, configural, metric, and scalar invariance. These three invariance conditions imposed increasing structural constraints in that a configural model constrained the pattern of factor composition to be equivalent, a metric model further constrained factor loadings to be equivalent, and a scalar model further constrained intercepts to be equivalent. The satisfaction of all the three invariance conditions can ensure that the psychological meanings of the latent CIP-Short-5 subscale scores are equivalent across China and the U.S. and consequently justify the comparison of latent means. Statistically, we not only examined the model-data fit of all three models individually, using the criteria mentioned above, but also compared the three models using a nested-model comparison strategy. In each pairwise comparison (configural vs. metric models and metric vs. scalar models), we used the Santorra-Bentler scaled chi-square difference test (Muthén & Muthén, 2012) and the ΔCFI (Cheung & Rensvold, 2002; Meade et al., 2008) to evaluate which model fit the data better. Notably, because the chi-square difference test is sensitive to sample sizes and is often significant when the sample size is large than 200, research has recommended ΔCFI as a preferred method for detecting practically meaningful measurement variance (Cheung & Rensvold, 2002; Meade et al., 2008). Given the adequate sample sizes in the present study, we selected a ΔCFI <0.01 as the primary criterion and used ΔRMSEA <0.015 and ΔSRMR <0.01 as supplementary criteria (Chen, 2007; Cheung & Rensvold, 2002).

Results

Development and Validation of the CIP-Short-5 in China

Based on the Chinese version of the CIP-65, we first conducted EFA on CCA and NI items and revealed a three-factor structure: CCA, NI, and a combination of CCA and NI. The emergence of a combined factor was not particularly surprising given the interconnection between CCA and NI. To ensure unidimensionality in the subsequent IRT analysis, we excluded items in the combined factor and focused on items loading clearly on CCA and NI. Notably, we tried developing CCA and NI items first in the U.S. based on the five-factor model and then validating them in China; however, this version did not conform to the five-factor model in China, echoing our concern that developing separate CCA and NI items in the U.S. might not fully capture the five-factor structure. Following the EFA results in China, we used the mean scores of CCA and NI to indicate the levels of latent CCA and NI and delineated the OCCs and ICC for each CCA and NI item. Each item was first ranked in terms of ordered option occurrence. Figure 2 shows example items of different option effectiveness. As can be seen, the more effective item showed more separated and less chaotic OCCs. We than ranked each item in terms of minimal gender DIF. Figure 3 shows example items of different gender DIF. As can be seen, the more gender-fair item showed smaller gaps in the ICC. Considering both criteria along with content representativeness, we selected five best items each for CCA and NI (see Supplementary Materials for the OCCs and ICCS of all the items). Subsequently, based on the same two criteria, we examined the NNA, LR, and IC items of the CIP-Short that were originally derived in the U.S. Results did not detect major deviations of their OCCs and ICCS from desirable OCCs and ICCs, supporting the adequacy of their item-level performance in China. Thus, Table 1 presents the items of the final CIP-Short-5. We found alpha coefficients of 0.81, 0.84, 0.81, 0.81, and 0.87 for the NNA, CCA, NI, LR, and IC subscales, respectively.

Table 1.

Final Items of the CIP-Short-5 in the Metric Model.

Scale	Items	Number in CIP-65	Loadings	Intercepts
Scale	Items	Number in CIP-65	Loadings	China	U.S.
NNA	Takes a long time to feel good after setbacks	1	1.00	3.35	3.25
	Easily embarrassed	8*	1.14	4.13	3.36
	Hard to make decisions without help	10*	1.02	3.01	3.19
	Often feel fearful and anxious	39	1.44	3.59	3.55
	Often feel insecure	44	1.50	3.69	3.36
CCA	Conflict a number or careers are appealing	21	1.00	3.68	3.52
	Difficult because like so many things	24*	1.00	4.09	3.70
	Rather keep myself open than committing	28	0.75	3.90	3.42
	Can’t commit, don’t know other options	49*	1.03	4.11	3.21
	Concerned goals might change	50	1.04	4.18	3.86
NI	Need to learn more about what I want	11	1.00	4.54	3.67
	Need a better idea of my talents	20	1.05	4.93	3.89
	Need to learn more about my interests	42	1.18	4.77	3.84
	Need more information on successful options	52	1.14	4.71	3.99
	Need to learn decision-making	58	1.09	4.67	3.73
LR	Confident can overcome obstacles R	48*	1.00	3.05	4.76
	Try to excel at everything R	51*	0.78	2.67	5.09
	Will be able to find a career R	54	0.97	2.66	4.95
	Always work productively R	55	0.86	2.89	4.93
	Confident find career perform well in R	59	1.06	2.90	4.98
IC	Going against wishes of others	3	1.00	2.97	2.38
	Contradictory information from others	29	0.85	3.41	2.94
	Important people don’t support plans	33	1.27	2.59	2.10
	Important people disagree with plans	43*	1.32	2.46	2.20
	Important people discourage plans	53	1.24	2.53	2.11

NNA = Negative affectivity; CCA = Commitment anxiety; NI = Need for Information; LR = Lack of Readiness; IC = Interpersonal conflicts. R = reverse items. * Different intercepts.

Table 2 summarizes the results of structural examination. As shown by the values of χ²/df (724.76/265 and 1188.44/269), RMSEA (0.054 and 0.076), SRMR (0.06, and 0.06), and CFI (0.91 and 0.82), the five-factor model clearly outperformed the four-factor model and represented the data well. Thus, the structural validity of the CIP-Short-5 was supported in China. Table 3 shows the means, standard deviations, and bivariate correlations of the five factors of the CIP-Short-5 with the indecision domains of the CDDQ and EPCD and career decision-making self-efficacy. Results showed that NNA was positively associated with self and identity of the EPCD (r = 0.69, p < .05), CCA was positively associated with anxiety of the EPCD (r = 0.49, p < .05), NI was positively associated with lack of information of the CDDQ (r = 0.25, p < .05), LR was negatively associated with career decision-making self-efficacy (r = 0.60, p < .05), and IC was positively associated with inconsistent information of the CDDQ (r = 0.52, p < .05). Notably, all the five factors except NI were mainly associated with theoretically anticipated primary criterion. Thus, the results largely supported the convergent and discriminant validity of the CIP-Short-5 in China.

Table 2.

Summary of Model Comparison.

				RMSEA
	χ²	df	CFI	Estimate	90% C. I	SRMR
China (n = 588)
CIP-S: 5-Factor model	724.76	265	0.91	0.054	[0.050, 0.059]	0.06
CIP-S: 4-Factor model	1188.44	269	0.82	0.076	[0.072, 0.081]	0.09
U.S. (n = 762)
CIP-S: 5-Factor model	928.89	265	0.92	0.058	[0.054, 0.062]	0.06
CIP-S: 4-Factor model	1106.50	269	0.89	0.064	[0.060, 0.068]	0.06
China (n = 588) vs. U.S. (n = 762) on CIP-S-5
Model 1. Configural invariance	1651.96	530	0.91	0.056	[0.053, 0.059]	0.06
Model 2. Metric invariance	1680.38	550	0.91	0.055	[0.052, 0.058]	0.06
Model 3. Scalar invariance	2062.26	570	0.88	0.062	[0.060, 0.065]	0.07

Table 3.

Means, Standard Deviations, and Correlations of Variables in China.

Variable	Mean	SD	NNA	CCA	NI	LR	IC	CDSE	LR	LI	II	PESS	ANX
NNA	3.55	1.04	1.00
CCA	3.99	.94	.37**	1.00
NI	4.72	.78	.21**	.47**	1.00
LR	2.83	.83	.20**	-.10*	-.38**	1.00
IC	2.79	1.11	.36**	.31**	.07	.03	1.00
CDSE	3.30	.59	-.34**	-.13**	.08	-.60**	-.09*	1.00
LR	5.29	1.16	.51**	.44**	.23**	.03	.28**	-.09*	1.00
LI	5.16	1.71	.49**	.48**	.25**	.26**	.28**	-.40**	.55**	1.00
II	4.60	1.50	.44**	.48**	.13**	.15**	.52**	-.26**	.54**	.74**	1.00
PESS	5.23	1.13	0.41**	0.40**	0.20**	0.13**	0.27**	-0.14**	0.46**	0.60**	0.57**	1.00
ANX	5.68	1.60	.48**	.49**	.33**	.14**	.20**	-.25**	.47**	.63**	0.56**	0.62**	1.00
IDEN	5.20	1.55	0.69**	0.45**	0.26**	0.24**	0.29**	-0.28**	0.56**	0.65**	0.59**	0.61**	0.68**

Note. N = 588. NNA = Neuroticism/negative affectivity; CCA = Choice/commitment anxiety; NI = Need for Information; LR = Lack of Readiness; IC = Interpersonal conflicts; LR = CDDQ-Lack of Readiness; LI = CDDQ-Lack of Information; II = CDDQ-Inconsistent Information; PESS = EPCD-Pessimistic views; ANX = EPCD-Anxiety; IDEN = EPCD- Self-concept and identity. * p < .05, ** p < .01.

Validation of the CIP-Short-5 in the U.S.

We first used the same IRT procedure to examine the item-level performance of CCA and NI items that were derived in China. Results did not detect major deviations of their OCCs and ICCS from desirable OCCs and ICCs, supporting the adequacy of their item-level performance in the U.S. We found alpha coefficients of 0.83, 0.85, 0.88, 0.86, and 0.87 for the NNA, CCA, NI, LR, and IC subscales, respectively. Table 2 shows the results of our structural examination. As shown by the values of χ²/df (928.89/265 and 1106.50/269), RMSEA (.058 and .064), SRMR (.06, and .06), and CFI (0.91 and 0.89), the five-factor model still outperformed the four-factor model and represented the data well in the U.S., but its advantage was smaller compared to its advantage in China. This result showed that although CCA and NI were highly correlated to each other in the U.S., they technically represented two distinct concepts. Table 4 shows the means, standard deviations, and correlations of the CIP-Short-5 and CIP-65. Notably, NNA, CCA, NI, LR, and IC in the two versions of the CIP were positively correlated at 0.94, 0.92, 0.96, 0.91, and 1.00, respectively. This result indicated that the CIP-Short-5 retained at least 81% of the information captured by the parental CIP-65 in the U.S., which lent support to the construct validity of the CIP-Short-5.

Table 4.

Means, Standard Deviations, and Correlations of Variables in the U.S.

Variable	Mean	SD	NNA	CCA	NI	LR	IC	NNA-S	CCA-S	NI-S	LR-S
NNA	3.53	0.98	1.00
CCA	3.39	1.11	0.55**	1.00
NI	3.53	1.16	0.51**	0.85**	1.00
LR	4.80	0.74	-0.25**	-0.23**	-0.20**	1.00
IC	2.35	1.20	0.40**	0.49**	0.43**	-0.30**	1.00
NNA-S	3.34	1.14	0.94**	0.52**	0.48**	-0.27**	0.37**	1.00
CCA-S	3.54	1.15	0.39**	0.92**	0.79**	-0.11**	0.42**	0.36**	1.00
NI-S	3.82	1.19	0.46**	0.79**	0.96**	-0.10**	0.36**	0.43**	0.72**	1.00
LR-S	4.94	.88	-0.26**	-0.25**	-0.23**	0.91**	-0.33**	-0.29**	-0.13**	-0.14**	1.00
IC-S	2.35	1.20	0.40**	0.49**	0.43**	-0.30**	1.00**	0.37**	0.42**	0.36**	-0.33**

Note. N = 762. NNA = CIP-65-Neuroticism/negative affectivity; CCA = CIP-65-Choice/commitment anxiety; NI = CIP-65-Need for Information; LR = CIP-65-Lack of Readiness; IC = CIP-65-Interpersonal conflicts; NNA-S = CIP-S-Neuroticism/negative affectivity; CCA-S = CIP-S-Choice/commitment anxiety; NI-S = CIP-S-Need for Information; LR-S = CIP-S-Lack of Readiness; IC-S = CIP-S-Interpersonal conflicts. ** p < .01.

Measurement Invariance across China and the U.S.

To evaluate the utility of the CIP-Short-5 for cross-cultural research on career indecision, we progressively examined configural, metric, and scalar invariance. Table 2 presents information about the model-data fit of configural, metric, and scalar models. As shown by the values of χ²/df (1651.96/530), RMSEA (0.056), SRMR (0.06), and CFI (0.91), the configural model fit the data well, indicating that a free estimated five-factor model was an adequate representation of data for Chinese and U.S. colleague students together. We then examined the metric model, which constrained factor loadings to be equivalent across the two samples. As shown by the values of χ²/df (1680.38/550), RMSEA (0.055), SRMR (0.06), and CFI (0.91), the metric model fit the data well. Additionally, when comparing the metric and configural models, we found that Santorra-Bentler scaled Δχ² (20, n = 1341) = 27.18 (p > .05), ΔCFI was smaller than .01, and ΔRMSEA and ΔSRMR were smaller than 0.01. Therefore, the results indicated that the metric model was equivalent to the configural model in terms of model-data fit, supporting the invariance of factor loadings across China and the U.S. college students.

Based on the metric model, we additionally constrained intercepts and examined the scalar model. As shown by the values of χ²/df (2062.26/570), RMSEA (0.062), SRMR (0.07), and CFI (0.88), the scalar model fit the data mediocrely. Additionally, when comparing the scalar and metric models, we found that Santorra-Bentler scaled Δχ² (20, n = 1341) = 409.96 (p < .05) and ΔCFI was larger than 0.01. Although ΔRMSEA and ΔSRMR were smaller than .01, the results together appeared to indicate that the scalar model was a worse representation of the data than the metric model, which did not support the invariance of intercepts across China and the U.S. college students. The result of scalar non-invariance was not particularly surprising because scale invariance in general is an ideal situation that is often unachievable in applied settings (Asparouhov & Muthén, 2014) and particularly has not received consistent support in CIP measure research (Carr et al., 2014; Roche et al., 2017). We further used multi-group factor analysis alignment (Asparouhov & Muthén, 2014) to identify individual intercepts that varied across the two countries. The alignment approach is essentially akin to rotation in factor analysis, both of which maintain global model-data fit while rotating the solution to enhance the interpretability of individual parameters (e.g., intercepts in invariance research and loadings in factor analysis). Results showed that seven intercepts were statistically different across China and U.S. (see Table 1). The implication of this non-invariance pattern was subjected to discussion.

Discussion

To develop a brief measure that corresponds to the integrative five-factor model of career indecision (Xu & Bhang, 2019), the current study developed and validated the CIP-Short-5 in China and the U.S. Using IRT to identify the best items of the CIP-65, the CIP-Short-5 consists of five items each for NNA, CCA, NI, LR, and IC. It showed desirable option occurrence patterns and minimal gender DIF in both China and the U.S. Additionally, the scale supported the five-factor model over the four-factor model of career indecision in both China and the U.S. The convergent and discriminant patterns of the CIP-Short-5 with criteria were largely supported in China and the U.S. Last, the results supported the configural and metric invariance of the CIP-Short-5 but did not fully support the scalar invariance of the CIP-Short-5 across China and the U.S. Together, the results offer psychometric evidence for the CIP-Short-5, which has important implications for the measurement of career indecision in the international context in general and in China in specific.

Psychometric Performance and Utility of the CIP-Short-5

The current results shed light on four aspects of psychometric performance of the CIP-Short-5. First, the IRT results in China and the U.S. speak to the effectiveness of the item options of the CIP-Short-5 and the scale’s ability to provide gender-fair assessment of career indecision in both countries. These two appealing features have not been seen or examined in other indecision measures such as the CIP-65 and the EPCD. Second, the structural results support the ability of the CIP-Short-5 in measuring the five cross-culturally important indecision factors. Given the empirical and theoretical underpinnings of the five-factor model (Xu, 2021a; Xu & Bhang, 2019), the CIP-Short-5 outperforms the CDDQ and the EPCD in terms of domain representativeness. Together, the brevity and content coverage of the CIP-Short-5 make the scale a comprehensive and parsimonious measurement tool of career indecision. Following the item-level and structural performance, the third strength of the CIP-Short-5 lies in its convergent and discriminant correlations with important criteria in China and the U.S. The results further support the construct validity of the scale in light of measuring the five-factor model. Last, the results support the weak invariance (i.e., configural and metric invariance) of the CIP-Short-5 but do not support the strong invariance (i.e., scalar invariance) of the scale. There might be two strategies to handle the lack of scalar invariance. First, since complete scalar invariance is an ideal situation that rarely exists in applied settings, future research can explore the practical significance of the currently identified intercept differences. This step could complement the current alignment approach because the alignment approach is based on tests of statistical significance, which could be subject to sample size and idiosyncratic information in the dataset (Asparouhov & Muthén, 2014). Alternatively, future research can simply drop scalar non-invariant items identified in this study and try to develop scalar invariant items. This strategy will likely require several rounds of trial-and-error but could provide a scale with clear measurement invariance.

Given its psychometric advantages, the CIP-Short-5 adds to the CIP family and the fourth generation of indecision measurement and has promising utility in at least four areas. First, the CIP-Short-5 can be used as a research instrument. Its brevity enables researchers to add the scale to a research battery that measures a set of constructs. In fact, although it measures a comprehensive set of indecision factors, the 25-item CIP-Short-5 is still shorter than the CDDQ and the EPCD. This advantageous efficiency can help reduce testing fatigue and enhance response quality, which is particularly desirable for research that focuses on complex structural questions. Second, the CIP-Short-5 can be used as a screening/assessment tool to help identify at-risk individuals and formulate intervention direction(s) through between-person and within-person comparison, respectively. The population-referenced utility, however, relies on a test norm of career indecision, which has been partially addressed by the CIP-Short but needs to be further developed. It should be noted that the CDDQ and the EPCD might be able to provide more nuanced information about facets within each indecision domain than the CIP-Short-5. However, given the high cooccurrence rate of these facets (see the high internal consistency), whether differentiating facets can lead to meaningful differential intervention remains open to debate. Third, the CIP-Short-5 can be used as a progress tracker in intervention. Based on the feedback-informed approach to counseling (Anker et al., 2009), practitioners can use the scale to monitor the effectiveness of intervention and adjust it when appropriate. To facilitate this use, we encourage future research to examine the scale’s sensitivity to meaningful changes and stability over background noise. Last, although the measurement invariance of the CIP-short-5 can be improved, its commensurate factor structure and consistent items across China and the U.S. (and potentially other countries) make it a promising indecision measure for cross-cultural comparison regarding the antecedents and outcomes of career indecision.

Limitations and Suggestions for Future Research

Several limitations need to be noted when interpreting the current results. First, the current study used only college student samples in China and the U.S. and thus the generalizability of the results to other populations, such as high school students and early or middle career workers, remains unclear. Although college students have been the common focus of indecision research, we encourage future research to expand the psychometric evidence for the CIP-Short-5. Xu’s (2020) study on the CIP-Short in employees could be an example. Second, the CIP-Short-5 was derived from the CIP-65 in both China and the U.S., and thus, technically the responses to the CIP-Short-5 were a subset of responses to the CIP-65. We argue that the other items of the CIP-65 unlikely interfered with the CIP-Short-5 in a systematic fashion because all the CIP-65 items were arranged randomly. However, it is necessary for future research to use the CIP-Short-5 as a stand-alone measure and further validate it. Last, as shown by the predictive model of the DTC (see Figure 1), while the five-factor model of indecision addresses important aspects in the psychosocial background, confusion management, and ambiguity management of career decision-making, it falls short of capturing barriers in other important areas, such as decision ambiguity and maladaptive information processing strategies. Thus, it would be interesting for future research to continue refining the indecision taxonomy based on the DTC and refining indecision measurement accordingly.

Conclusions

The CIP-Short-5 builds on the empirical foundation of the CIP family (Brown et al., 2012; Hacker et al., 2013; Xu & Tracey, 2017) and the theoretical underpinnings of the DTC (Xu, 2021a, 2021b). Overall, the current study suggests that the CIP-Short-5 is a valid measure for the five-factor model of career indecision in both China and the U.S. Its brevity, option effectiveness, minimal gender DIF, and promising measurement invariance all support its potential as a psychometrically sound brief measure of career indecision in the international context in general and in China in specific. To conclude, the current study makes an important contribution to the field by offering a brief and yet comprehensive measure of career indecision, which could facilitate international research on career indecision.

Footnotes

Acknowledgments

We also thank Li Zhu for her assistance in collecting data.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Hui Xu

References

Abrams

M. D.

Lee

I. H.

Brown

S. D.

Carr

(2014). The career indecision profile measurement equivalence in the United States and South Korea. Journal of Career Assessment, 23(2), 1–11. http://doi.org/10.1177/1069072714535028

Abrams

M. D.

Ómarsdóttir

A. Ó.

Björnsdóttir

M. D.

Einarsdóttir

Martin

Carr

Brown

S. D.

Rector

(2013). Measurement invariance of the career indecision profile: United States and Iceland. Journal of Career Assessment, 21(3), 469–482. https://doi.org/10.1177/1069072712475181

Anker

M. G.

Duncan

B. L.

Sparks

J. A.

(2009). Using client feedback to improve couple therapy outcomes: A randomized clinical trial in a naturalistic setting. Journal of Consulting and Clinical Psychology, 77(4), 693–704. https://doi.org/10.1037/a0016062

Ansley

T. N.

Forsyth

R. A.

(1985). An examination of the characteristics of unidimensional IRT parameter estimates derived from two-dimensional data. Applied Psychological Measurement, 9(1), 37–48. https://doi.org/10.1177/014662168500900104

Asparouhov

Muthén

(2014). Multiple-group factor analysis alignment. Structural Equation Modeling: A Multidisciplinary Journal, 21(4), 495–508. https://doi.org/10.1080/10705511.2014.919210

Betz

Klein

Taylor

(1996). Evaluation of a short form of the career decision-making self-efficacy scale. Journal of Career Assessment, 4(1), 47–57. http://doi.org/10.1177/106907279600400103.

Blustein

D. L.

Ali

S. R.

Flores

L. Y.

(2019). Vocational psychology: Expanding the vision and enhancing the impact. The Counseling Psychologist, 47(2), 166–221. https://doi.org/10.1177/0011000019861213

Brown

S. D.

Hacker

Abrams

Carr

Rector

Lamp

Telander

Siena

(2012). Validation of a four-factor model of career indecision. Journal of Career Assessment, 20(1), 3–21. http://doi.org/10.1177/1069072711417154

Brown

S. D.

Rector

C. C.

(2008). Conceptualizing and diagnosing problems in vocational decision making. In Brown

S. D.

Lent

R. W.

(Eds.), Handbook of counseling psychology (pp. 392–407). Wiley.

10.

Carr

Rossier

Rosselet

J. G.

Massoudi

Bernaud

J.-L.

Ferrari

Nota

Soresi

Rowe-Johnson

Brown

S. D.

(2014). The Career Indecision Profile Measurement Equivalence in Two International Samples. Journal of Career Assessment, 22(1), 123–137. https://doi.org/10.1177/1069072713492930

11.

Chartrand

J. M.

Robbins

S. B.

Morrill

W. H.

Boggs

(1990). Development and validation of the career factors inventory. Journal of Counseling Psychology, 37(4), 491–501. https://doi.org/10.1037/0022-0167.37.4.491

12.

Chen

F. F.

(2007). Sensitivity of goodness of fit indexes to lack of measurement invariance. Structural Equation Modeling, 14(3), 464–504. https://doi.org/10.1080/10705510701301834

13.

Cheung

G. W.

Rensvold

R. B.

(2002). Evaluating goodness-of-fit indexes for testing measurement invariance. Structural Equation Modeling, 9(2), 233–255. https://doi.org/10.1207/S15328007SEM0902_5

14.

Embretson

S. E.

Reise

S. P.

(2010). Item response theory for psychologists. Psychology Press.

15.

Gati

Krausz

Osipow

S. H.

(1996). A taxonomy of difficulties in career decision making. Journal of Counseling Psychology, 43(4), 510–526. http://doi.org/10.1037/0022-0167.43.4.510

16.

Guan

Capezio

Restubog

S. L. D.

Read

Lajom

J. A. L.

(2016). The role of traditionality in the relationships among parental support, career decision-making self-efficacy and career adaptability. Journal of Vocational Behavior, 94, 114-123. https://doi.org/10.1016/j.jvb.2016.02.018.

17.

Hacker

Carr

Abrams

Brown

S. D.

(2013). Development of the career indecision profile factor structure, reliability, and validity. Journal of Career Assessment, 21(1), 32–41. https://doi.org/10.1177/1069072712453832

18.

Hou

(2010). Chinese translation of the career decision-making difficulty questionnaire. Beijing Normal University.

19.

Hou

Z.-J.

Liu

Y.-L.

Gati

(2015). The emotional and personality-related career decision-making difficulties questionnaire—validation of the Chinese version. Journal of Career Assessment, 24(2), 366–379. http://doi.org/10.1177/1069072715580565.

20.

Bentler

P. M.

(1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling: A Multidisciplinary Journal, 6(1), 1–55. http://doi.org/10.1080/10705519909540118

21.

Jiang

Wang

Weiss

D. J.

(2016). Sample size requirements for estimation of item parameters in the multidimensional graded response model [original research]. Frontiers in Psychology, 7, 109. https://doi.org/10.3389/fpsyg.2016.00109.

22.

Kelly

K. R.

Lee

W. C.

(2002). Mapping the domain of career decision problems. Journal of Vocational Behavior, 61(2), 302–326. https://doi.org/10.1006/jvbe.2001.1858

23.

Kline

R. B.

(2015). Principles and practice of structural equation modeling. Guilford publications.

24.

Krieshok

T. S.

Black

M. D.

McKay

R. A.

(2009). Career decision making: The limits of rationality and the abundance of non-conscious processes. Journal of Vocational Behavior, 75(3), 275–290. https://doi.org/10.1016/j.jvb.2009.04.006

25.

Krumboltz

J. D.

(2009). The happenstance learning theory. Journal of Career Assessment, 17(2), 135–154. http://doi.org/10.1177/1069072708328861

26.

Kulcsar

Dobrean

Gati

(2020). Challenges and difficulties in career decision making: Their causes, and their effects on the process and the decision. Journal of Vocational Behavior, 116, 103346. https://doi.org/10.1016/j.jvb.2019.103346.

27.

Lent

R. W.

Brown

S. D.

(2020). Career decision making, fast and slow: Toward an integrative model of intervention for sustainable career choice. Journal of Vocational Behavior, 120, 103448. https://doi.org/10.1016/j.jvb.2020.103448.

28.

Leung

Hou

Gati

(2011). Effects of parental expectations and cultural-values orientation on career decision-making difficulties of Chinese University students. Journal of Vocational Behavior, 78(1), 11–20. https://doi.org/10.1016/j.jvb.2010.08.004

29.

Long

(2003). The study on career decision-making self-efficacy of college students. Unpublished master’s thesis, Shanghai Normal University.

30.

Meade

A. W.

Johnson

E. C.

Braddy

P. W.

(2008). Power and sensitivity of alternative fit indices in tests of measurement invariance. Journal of Applied Psychology, 93(3), 568–592. http://doi.org/10.1037/0021-9010.93.3.568

31.

Muthén

L. K.

Muthén

B. O.

(2012). Chi-square difference testing using the Satorra-Bentler scaled chi-square. MPlus. http://www.statmodel.com/chidiff.shtml

32.

Osipow

S. H.

(1999). Assessing career indecision. Journal of Vocational Behavior, 55(1), 147–154. https://doi.org/10.1006/jvbe.1999.1704

33.

Osipow

S. H.

Carney

C. G.

Barak

(1976). A scale of educational-vocational undecidedness: A typological approach. Journal of Vocational Behavior, 9(2), 233–243. https://doi.org/10.1016/0001-8791(76)90081-6

34.

Parsons

(1909). Choosing a vocation. Houghton Mifflin Company.

35.

Roche

M. K.

Carr

A. L.

Lee

I. H.

Wen

J. H.

Brown

S. D.

(2017). Career indecision in China: Measurement equivalence With the United States and South Korea. Journal of Career Assessment, 25(3), 526–536. https://doi.org/10.1177/1069072716651623

36.

Saka

Gati

Kelly

K. R.

(2008). Emotional and personality-related aspects of career-decision-making difficulties. Journal of Career Assessment, 16(4), 403–424. https://doi.org/10.1177/1069072708318900

37.

Savickas

M. L.

(2012). Life design: A paradigm for career intervention in the 21st century. Journal of Counseling & Development, 90(1), 13–19. https://doi.org/10.1111/j.1556-6676.2012.00002.x

38.

Savickas

M. L.

(2015). Career counseling paradigms guiding, developing and designing. In Hartung

P. J.

Savickas

M. L.

Walsh

W. B.

(Eds.), APA handbooks in psychology. APA handbook of career intervention(Vol. 1, pp. 129-143), American Psychological Association. http://doi.org/10.1037/14438-000

39.

Schlomer

G. L.

Bauman

Card

N. A.

(2010). Best practices for missing data management in counseling psychology. Journal of Counseling Psychology, 57(1), 1–10. http://doi.org/10.1037/a0018082

40.

Sijtsma

Molenaar

I. W.

(2002). Introduction to nonparametric item response theory Vol. 5. Sage.

41.

Stout

(2001). Nonparametric item response theory: A maturing and applicable measurement modeling approach. Applied Psychological Measurement, 25(3), 300–306. https://doi.org/10.1177/01466210122032109

42.

Vandenberg

R. J.

Lance

C. E.

(2000). A review and synthesis of the measurement invariance literature: Suggestions, practices, and recommendations for organizational research. Organizational Research Methods, 3(1), 4–70. http://doi.org/10.1177/109442810031002

43.

Whittaker

T. A.

Worthington

R. L.

(2016). Item response theory in scale development research: A critical analysis. The Counseling Psychologist, 44(2), 216–225. http://doi.org/10.1177/0011000015626273

44.

Wirth

R. J.

Edwards

M. C.

(2007). Item factor analysis: Current approaches and future directions. Psychological Methods, 12(1), 58–79. http://doi.org/10.1037/1082-989x.12.1.58

45.

(2020). Career indecision profile-short: Reliability and validity among employees and measurement invariance across students and employees. Journal of Career Assessment, 28(1), 91–108. https://doi.org/10.1177/1069072719831975

46.

(2021a). Career decision-making from a dual-process perspective: Looking back, looking forward. Journal of Vocational Behavior, 126, 103556. https://doi.org/10.1016/j.jvb.2021.103556.

47.

(2021b). Career decision-making in an uncertain world: A dual-process framework. Current Psychology. http://doi.org/10.1007/s12144-021-01746-z.

48.

Bhang

C. H.

(2019). The structure and measurement of career indecision: A critical review. The Career Development Quarterly, 67(1), 2–20. https://dx-doi-org.web.bisu.edu.cn/10.1002/cdq.12159

49.

Hou

Z.-J.

Tracey

T. J. G.

(2014). Relation of environmental and self-career exploration with career decision-making difficulties in Chinese students. Journal of Career Assessment, 22(4), 654–665. http://doi.org/10.1177/1069072713515628

50.

Tracey

T. J. G.

(2017). Development of an abbreviated career indecision profile-65 using item response theory: The cip-short. Journal of Counseling Psychology, 64(2), 222–232. http://doi.org/10.1037/cou0000182

51.

Traceya

T. J.

(2017). Use of multi-group confirmatory factor Analysis in examining measurement invariance in counseling psychology research. The European Journal of Counselling Psychology, 6(1), 75-82. http://doi.org/10.5964/ejcop.v6i1.120