The Emotional and Personality-Related Career Decision-Making Difficulties Questionnaire—Validation of the Chinese Version

Review of Career Decision-Making Assessment

Several assessment tools have been developed to measure career indecision. The earliest widely used assessment is the Career Decision Scale (CDS; Osipow, Carney, Winer, Yanico, & Koschier, 1976). However, studies using CDS revealed that, in spite of its wide use, disputes exist: Some researchers discovered a multidimensional structure of CDS in factor analysis (Feldt et al., 2010; Shimizu, Vondracek, Schulenberg, & Hostetle, 1988), others suggested that CDS was unidimensional (Laplante, Coallier, Sabourin, & Martin, 1994). Osipow (1994) recommended using the total score of CDS to represent the level of career indecision, while Savickas and Jarjoura (1991) proposed CDS to be an indecision type indicator. These inconsistent results suggested potential validity issues of CDS and have motivated further research efforts to develop other measures for assessing career indecision or career decision-making difficulty.

In existing literature, a number of later measures can be found, such as the My Vocational Situation Scale (MVS; Holland, Daiger, & Power, 1980), the Vocational Decision Scale (Jones & Chenery, 1980), the Career Decision Diagnostic Assessment (Bansberg & Sklare, 1986; Larson, Busby, Wilson, Medora, & Allgood, 1994), and the Career Factor Inventory (Chartrand, Robbins, Morrill, & Boggs, 1990). However, none of those assessment tools is based on a systematic and theoretical construct of career indecision or decision-making difficulty, so there is a need for a “new framework, which combines a theoretical analysis and empirical tests” (Gati, Krausz, & Osipow, 1996, p. 511). To address this need, Gati, Krausz, and Osipow (1996) developed the Career Decision-Making Difficulties Questionnaire (CDDQ). CDDQ was based on the premise that the ideal decision maker should be a person who is aware of the need to make a decision, has motivation to do it, and has competence to make the right decision. Anyone who deviates from the ideal decision maker will probably have difficulty to make a decision. Gati et al. (1996) proposed a hierarchical model, which classified the career decision-making difficulties into three major clusters: lack of readiness, lack of information, and inconsistent information, each of which comprises 3–4 difficulty categories. Unlike the CDS and other previous assessment tools, CDDQ was developed by integrating a theoretical framework of career decision with empirical data. This improvement facilitated later research studies in this realm.

However, CDDQ focuses more on the cognition-related difficulties of career decision making, disregarding the emotional and personality-related aspects (Jin, Nam, Joo, & Yang, 2014). Scholars have found that both emotion and personality play important roles in career decision-making process and career choices, but such constructs received relatively little research attention (Emmerling & Cherniss, 2003; Kidd, 1998; Young, Paseluikho, & Valach, 1997; Young & Valach, 1996). Among relevant studies, Saunders, Peterson, Sampson, and Reardon (2000) observed both depression and dysfunctional career thoughts had significant relationships with indecision. With the Chinese population, Li, Hou, and Jia (2015) found that the emotion of regret mediated the influence of social comparison on career choice certainty. Lounsbury, Hutchens, and Loveland (2005) and Fabio, Palazzeschi, Levin, and Gati (2014) reported that career decision-making difficulties were related to the Big Five model of personality. Studies by Leong and Chervinko (1996) and Yu (2008) supported the association between perfectionistic personality and career indecision. Gati, Amir, and Landman (2010) found that counselors rated personal and emotional causes of clients’ career issues as more severe and as those that needed longer treatment than external and cognitive causes. The above studies lent evidence for the important role of emotional and personality factors in career indecision and decision-making difficulties.

To address this “missing part,” the EPCD (Saka et al., 2008) was developed to assess the emotional and personality-related aspects of career decision making. Like the CDDQ, EPCD was also a theoretically based instrument consisting of three major clusters: pessimistic views, anxiety, and self-concept and identity. Compared to other indecision measures, EPCD focuses on the emotional and identity-related aspects that are argued as important components in career decision-making process. The pessimistic views cluster refers to the negative thinking, feeling, and expectation toward oneself, the world of work, self-efficacy, and self-control, and comprises three difficulty categories: pessimistic views about the process, about the world of work, and about the individual’s control. People who have higher levels of pessimistic views will have negative perception of the work world, do not believe they have personal control over their decision, and have low career decision-making self-efficacy. Consequently, they tend to have undesirable feelings like depression and are also likely to take avoidance-oriented behaviors during the career decision-making process. The anxiety cluster refers to the negative emotional state during the whole career decision process. People with high level of anxiety tend to express low tolerance level of uncertainty, ambiguity, and get into an undecided state. The anxiety cluster in EPCD includes anxiety emotions before, during, and after career decision-making process. The anxiety cluster comprises four categories: anxiety about the process, about the involved uncertainty, about choosing, and about the outcomes. The third major cluster in EPCD is self-concept and identity. It comprises four difficulty categories: self-esteem, general anxiety, uncrystallized identity, and conflictual attachment and separation. This major cluster assesses the difficulties related to individuals’ ideas about who they are and where they want to go during the decision-making process.

A number of studies investigated and supported the convergent, divergent, and predictive validity of EPCD. Gati, Gadassi, et al. (2010) studied the association between EPCD and Big Five personality, perfectionism, locus of control, and career decision-making self-efficacy in three different samples. They found that EPCD had moderate negative correlation with the conscientiousness and extraversion factors, positive correlation with agreeableness, and relatively high and negative correlation with career decision-making self-efficacy. People who scored higher on EPCD also had significantly higher level of need for cognitive closure, perfectionism, and external locus of control. In a longitudinal study with 747 university students in Israel, Saka and Gati (2007) found that EPCD not only had a good structural validity but could specifically capture some stable and enduring components of career decision-making difficulties, given that the students who showed higher level of decision-making difficulties measured by EPCD had less progress and lower choice confidence after a 24-week time period. Finally, in a 3-year follow-up study, Gati, Peretz, and Fisher (2011) found that participants who reported having made a career decision after 3 years scores lower on EPCD at the first administration than did those who remained undecided after 3 years. Furthermore, whereas the decided group showed a decrease in the EPCD scores during the 3-year interval, the undecided group showed no significant decrease.

It is important to note, however, that the aforementioned studies supporting the reliability and validity of EPCD were all conducted in Israel, and only three studies were identified in which EPCD was applied to individuals in other countries or cultures. In one study, Saka, Gati, and Kelly (2008) tested its cross-cultural validity with a sample of American university students and found that the internal reliabilities of 11 difficulty categories of EPCD in American sample (from .66 to .94) were similar to it in the Israeli sample (from .70 to .91). The structural validity of EPCD was also generally supported, although the uncrystallized identity difficulty category was located empirically under the self-concept and identity cluster in the Israeli sample in accordance with the theoretical model but was categorized into the anxiety cluster in the American sample. In another study, Oztemel (2013) tested the structure of EPCD with a sample of Turkish high school students. The exploratory cluster analysis showed a similar structure of EPCD-Turkish version with the hypothesized theoretical model by Saka et al. (2008) with one difference: The difficulty category “anxiety about the outcomes” was included in the self-concept and identity instead of anxiety cluster. However, the results of confirmatory factor analysis (CFA) still provided sufficient support for the hypothesized structure of EPCD in another study with a Turkish sample (Oztemel, 2013). The third study was conducted in Asian context with Korean university students (Jin et al., 2014). The results supported the factor structure of EPCD–Short Form with the exception of two low loading items in the conflictual attachment and separation difficulty category. The findings also provided evidence for the concurrent validity of EPCD in the Korean sample, given its significant correlations with relevant career decision-making measures like career stress, trait anxiety, and career decision self-efficacy.

These reviewed three studies, taken together, generally supported the reliability, structural validity, and concurrent validity of EPCD in various cultures (United States, Turkey, and Korea) other than the original Israeli context in which it was developed. However, minor discrepancies were also found that showed certain overlaps between the anxiety cluster and the self-concept and identity cluster, which called for necessary revisions of items under these two clusters. Given these findings and especially the shared cultural similarity between Korea and China (Sun, 2009), it could be speculated that the general factor structure of EPCD might generalize to the Chinese population and it might have good reliability and concurrent and predictive validity, but it is less certain that all the items would be cross-culturally valid. Moreover, as argued previously, a comprehensive measure of career decision-making difficulty is instrumental, given the current socioeconomic context of China. Research in the Chinese context suggested that career decision-making difficulties related to personality (Li, Hou, & Feng, 2013) and emotions (Li, Hou, & Jia, 2015) were very typical in Chinese university students, who particularly displayed anxiety and intolerance about the ambiguity and uncertainty in decision-making processes (Li, Wu, Tao, & He, 2012) due to the unique socioeconomic changes (Hou & Zhang, 2007). Therefore, EPCD may serve as a desirable option for assessing Chinese students’ career decision-making difficulties because of its relatively comprehensive theoretical bases as well as the inclusion of personality and emotion-related components.

In sum, while the reviewed evidence (sufficient reliability and satisfactory structural, concurrent, and predictive validity) suggested EPCD to be a potentially promising measure of career decision-making difficulties, most research supporting the use of EPCD was conducted with samples from the Israeli culture and only three published studies applied it to other cultural contexts. Therefore, more evidence is clearly called for to support the cross-cultural validity of EPCD and serve as the empirical base for its broader use in other cultures. Combined with the practical need to adapt a comprehensive measure to help assess Chinese university students’ career decision-making difficulties, this study was conducted to test the reliability and validity of EPCD in the Chinese social context. Specifically, we tested (1) the factor structure of EPCD–Chinese version; (2) its relationships with the established measure of CDS, trait anxiety, and vocational identity; and (3) associations with demographic variables (gender and academic year). We hypothesized that (1) EPCD–Chinese version would be compatible with the model consisting of three major clusters and 11 difficulty categories as originally hypothesized in Saka et al. (2008), (2) students scoring higher on EPCD would show higher level of trait anxiety and lower level of vocational identity, and (3) decision-making difficulties may show some variations across gender and academic year.

Participants

Five hundred and sixty students from two universities in northern China participated in this study. Data from 20 participants were excluded in the preliminary analysis (see details in the Analysis section), resulting in 540 valid responses for the data analysis (253 males and 287 females). Participants’ age ranged from 16 to 25 years (M = 20.82, SD = 1.28). With regard to academic grade, 3.5% (n = 19) were freshmen, 32.2% (n = 174) were sophomores, 38.3% (n = 207) were juniors, and 25.9% (n = 140) were seniors. About 50% (n = 273) of participants majored in science and technology, 49.2% (n = 266) in liberal arts, and one student did not declare major. To collect test–retest data, 80 students were asked to fill out the whole set of questionnaires again after 2 weeks. Seventy-seven questionnaires were completed and valid.

Instruments

The EPCD scale

The EPCD is a self-report questionnaire with 53 items and includes 1 warm-up item, 2 validity items, and 50 items representing the 11 specific categories (Saka et al., 2008). Participants were asked to respond to each item on a 9-point Likert-type scale from 1 (does not describe me at all) to 9 (describes me well). Higher rating represents higher level of difficulties. The Cronbach’s α internal consistency reliabilities of the 11 scales, the three major clusters, and the total EPCD, as well as the 2-week test–retest reliabilities are presented in Table 1.

OPEN IN VIEWER

Table 1.

Means, Standard Deviations, and Internal Consistency Coefficients for EPCD–China Form.

EPCD Scales	No. of Itemsa	M	SD	Cronbach’s α	Test–Retest Correlation (n = 77)
Pessimistic views about	12	4.92	1.34	.78	.81
The process	3	5.88	1.97	.64	.74
The world of work	5	4.82	1.69	.69	.74
One’s control	4	4.05	1.74	.74	.70
Anxiety about	21	5.17	1.42	.91	.93
The process	4	5.54	1.76	.79	.83
Uncertainty	6	4.80	1.82	.82	.84
The choice	7	5.41	1.76	.85	.86
The outcomes	4	4.92	2.07	.83	.84
Self-concept and identity	14	4.17	1.45	.85	.83
General anxiety	3	4.83	2.12	.77	.61
Self-esteem	4	5.22	1.95	.81	.80
Uncrystallized identity	4	4.26	1.89	.75	.69
Conflictual attachment and separation	3	3.39	1.88	.71	.70
Total EPCD score	47	4.88	1.20	.93	.94

Note. N = 540. EPCD = Emotional and Personality-Related Career Difficulties.

^aAfter deleting 3 items.

Procedure

The first author translated the 53 original EPCD items from English into Chinese. Then, three master students in a counseling psychology program reviewed the translation to ensure language clarity in Chinese. After obtaining group consensus, a Chinese doctoral student who was studying counseling psychology in United States translated the Chinese items back into English. After that, the fourth author who participated in the construction of the original version of EPCD checked the correspondence between original English items and the back-translated items. The translation of 5 items was questioned because of phrasing. The main problems of these 5 items back-translation were related to the positive or negative phrasing a sentence. For example, the fourth author questioned the correspondence between the back-translated item “Most occupations are not really interesting” and the original item “Few careers are really interesting.” In fact, these two sentences were almost identical in Chinese. Based on such comments from the fourth author, the translation team discussed and modified the translation, until all research team members reached consensus. This procedure resulted in the EPCD–Chinese version, which was then administrated to university students.

All participants completed a battery of instruments, which consisted of the EPCD–Chinese version, VI scale (from MVS; Holland et al., 1980), CDS (Osipow et al., 1976), TAS (Spielberger et al., 1983), and a demographic questionnaire. Students were informed of the purpose of this study and voluntarily completed the questionnaires in classroom settings. Souvenirs were given to them afterward to express appreciation for their participation. Two weeks later, 80 students from the total sample were asked to complete the EPCD–Chinese version again. Students for retest were selected because they took the same course and would be convenient to follow up for research purposes. Seventy-seven of them actually filled out all instruments for the second time.

To examine the structural validity of EPCD with the Chinese sample, CFA with maximum likelihood estimation was conducted using Mplus 5.1 (Muthén & Muthén, 1998–2012). According to the theoretical EPCD structure (Saka et al., 2008), we hypothesized that the 50 items would load on 11 scales representing the 11 categories of difficulties and the 11 scales would load on the three major clusters and then one general factor.

The Internal Consistency and Test–Retest Reliability

Means, standard deviations, and Cronbach’s α values of the 11 scales, three major clusters, and EPCD total score are presented in Table 1. The reliability coefficients of all 11 subscales were from moderate to high (ranging from .64 to .85), the reliability of the three clusters were .78, .91, and .85, and the reliability of the total EPCD score was .93. These coefficients, though slightly lower than those derived from Israeli and American samples, are psychometrically acceptable. Two weeks test–retest reliabilities ranged from .61 to .86 for the 11 scales; .81, .93, and .83 for three major clusters, respectively, and .94 for the total EPCD score. These findings supported the reliability of the Chinese version of the EPCD, including that of its 11 scales and the three main clusters.

Structural Validity

The intercorrelations among the 11 scale scores of EPCD, the three cluster scores, and the total EPCD score are presented in Table 2. As can be seen in Table 2, all correlation coefficients between the 11 scale scores are statistically significant (ps < .01), ranging from .19 to .60. We conducted CFA to test the structural validity of EPCD. Specifically, we tested the third-order factor model (50-11-3-1), as originally proposed in Saka et al. (2008): 50 items loaded onto the 11 categories as the first-order factors, 11 categories loaded onto the 3 clusters as the second-order factors, and the 3 clusters loaded onto one general EPCD as third-order factor. The overall fit of the model was satisfactory: χ² = 3584.75, df = 1,161, root mean square error of approximation (RMSEA) = 0.066 (confidence interval [CI] = [0.064, 0.069]), comparative fit index (CFI) = 0.93, and non-normed fit index (NNFI) = 0.93. The factor loading of 11 categories on three major clusters all exceeded .50 and the loading of these three clusters on the 11 second-order categories were from .82 to .99. For all the three categories within the cluster of pessimistic views, the loading coefficients of 12 items on their corresponding categories ranged from .44 to .73. For the four categories within the cluster of anxiety, the loading coefficients of 21 items on the categories ranged from .51 to .91. Finally, for the three categories (self-esteem, general anxiety, and uncrystallized identity) within the cluster of self-concept and identity, the loading coefficients of 11 items ranged from .53 to .83. However, for the last category (conflictual attachment and separation) within this cluster, only three loading coefficients of items were above .40 (.59, .66, and .80, respectively), whereas the other three were too low (.06, .18, and .30, respectively). Therefore, we modified the model by deleting the three lower loadings items, and 3 items left in the conflictual attachment and separation category. The overall model fit for this 47-item questionnaire was also satisfactory: χ² = 3175.66, df = 1,020, RMSEA = 0.066, CI = [0.064, 0.068], CFI = 0.94, NNFI = 0.94, and all loadings of 47 items on the 11 categories were higher than .44. Therefore, we used 47-item model for the following analysis. These results supported our first hypothesis.

OPEN IN VIEWER

Table 2.

Intercorrelations Among the 11 Scales, the Three Clusters, and Total in EPCD-China.

Scales	1	2	3	4	5	6	7	8	9	10	11	12	13	14
1. pv_pr	—
2. pv_ww	.32	—
3. pv_cont	.26	.45	—
4. Ax_pr	.36	.33	.33	—
5. Ax_un	.30	.45	.52	.45	—
6. Ax_ch	.34	.38	.32	.46	.60	—
7. Ax_out	.26	.27	.19	.32	.37	.50	—
8. Gen_ax	.22	.29	.31	.35	.36	.40	.30	—
9. SE	.38	.37	.42	.31	.43	.41	.33	.52	—
10. Uncrys_id	.36	.40	.43	.20	.54	.41	.29	.30	.51	—
11. Conf_att	.16	.25	.30	.20	.27	.25	.26	.27	.30	.36	—
12. Pv Cluster	.74	.77	.75	.45	.56	.46	.32	.36	.52	.52	.31	—
13. Ax Cluster	.40	.46	.44	.71	.78	.83	.74	.46	.48	.47	.32	.58	—
14. Si Cluster	.34	.43	.48	.34	.55	.49	.38	.71	.60	.80	.70	.55	.57	—
15. Total	.53	.64	.60	.63	.79	.79	.61	.61	.61	.66	.47	.79	.92	.78

Note. N = 540. All the intercorrelations are significant with p < .01. EPCD = Emotional and Personality-Related Career Difficulties; Pv_pr = pessimistic views about the process; Pv_ww = pessimistic views about the world of work; Pv_cont = pessimistic views about one’s control; Ax_pr = anxiety about the process; Ax_un = anxiety about uncertainty; Ax_ch = anxiety about the choice; Ax_out = anxiety about the outcomes; Gen_ax = general anxiety; SE = self-esteem; Uncrys_id = uncrystallized identity; Conf_att = conflictual attachment and separation; Pv = pessimistic views; Ax = anxiety; Si = self-concept and identity; Total = total score of EPCD–Chinese Version.

To further test the structural validity of the original EPCD model, we conducted another set of CFA to rule out a possible alternative structure. After an exploratory factor analysis, Brown et al. (2012) suggested that the 11 difficulty categories of EPCD could be restructured under their four-factor model: pessimistic views about the process, anxiety about the process, anxiety due to uncertainty, anxiety about choosing, anxiety about outcome, and uncrystallized identity all loaded on Factor 1 (neuroticism/negative affectivity); general anxiety and self-esteem loaded on Factor 2 (choice/commitment anxiety); pessimistic views about the process, pessimistic views about world of work, pessimistic views about personal control loaded on Factor 3 (lack of readiness); and conflictual attachments loaded on Factor 4 (interpersonal conflicts). To test if this model provided a better fit, we performed a CFA according to this alternative structure. The model fit indices were χ² = 237.61, df = 38, RMSEA = 0.099, CI = [0.087, 0.111], CFI = 0.896, and Tucker–Lewis index (TLI) = 0.849. Comparing these fit indices with those of the original 11-3 model (χ² = 223.52, df = 41, RMSEA = 0.091, CI = 0.079, 0.103], CFI = 0.905, and TLI = 0.872), it could be seen that the original EPCD structure had better fit (manifested by the smaller χ² and RMSEA and higher CFI and TLI) as well as greater parsimony. Therefore, the data supported the original EPCD structure over the four-factor alternative model as suggested in Brown et al. (2012) in the Chinese population.

Construct Validity

Table 3 presents the correlations between each of the three major clusters of EPCD, the EPCD total score, CDS total score, the TAS total score, and VI total score. As hypothesized, TAS positively correlated with the three difficulty clusters of EPCD. Furthermore, the correlation of TAS was higher with the self-concept and identity cluster of the EPCD, which includes the general anxiety scale (.53) than with the two other clusters (.37 and .37), Z = 4.62, p < .001.

OPEN IN VIEWER

Table 3.

Correlations Between the Three Clusters and Total EPCD, TAS, CDS, and VI.

	M	SD	1	2	3	4	5	6
1.EPCD_Pv	4.92	1.34
2.EPCD_Ax	5.17	1.42	.58
3.EPCD_Si	4.17	1.45	.55	.57
4.EPCD Total	4.88	1.20	.79	.92	.78
5.TAS	2.20	0.35	.37	.37	.53	.48
6.CDS	2.10	0.37	.32	.43	.40	.46	.24
7.VI	9.40	3.70	−.48	−.56	−.59	−.64	−.39	−.44

Note. N = 540. All the correlation coefficients are statistically significant p < .001. EPCD = Emotional and Personality-Related Career Difficulties; EPCD_Pv = pessimistic views; EPCD_Ax = anxiety; EPCD_Si = self-concept and identity; EPCD Total = the total score of EPCD–China Form; TAS = Trait Anxiety Scale; CDS = Career Decision Scale; VI = vocational identity.

Result also showed that CDS positively but only moderately correlated with the three difficulty clusters (r = .32, .43, and .40, respectively), and this result lent some support to EPCD’s discriminant validity since it means EPCD detect career decision making differently from CDS. However, the correlation of CDS with the pessimistic views cluster, that is more cognitively focused, was lower than with the other two major clusters (Z = 3.05, p < .001 and Z = 2.13, p < .05, for anxiety and self-concept and identity, respectively).

As can be seen in Table 3 and as expected, the VI had negative correlations with the three clusters of the EPCD. Furthermore, the associations between the three clusters of the EPCD were stronger with the VI than with the CDS (Z = 3.94, p < .001, Z = 3.44, p < .001, and Z = 5.05, p < .001, respectively). Finally, similarly as with respect to the CDS, the associations of the VI were stronger with the anxiety and the self-concept and identity clusters than with the pessimistic views cluster (Z = 2.47, p < .01 and Z = 3.34, p < .05, respectively). These results supported the convergent and divergent validity of EPCD. In sum, the pattern of associations between three EPCD clusters and the criterion measures (TAS, CDS, and VI) indicated good construct validity of the Chinese version of the EPCD.

Demographic Differences in EPCD Scores

Gender difference was only found in the self-concept and identity cluster of the EPCD: Male students (M = 4.35, SD = 1.50) reported higher difficulties than female students (M = 4.01, SD = 1.38) in this cluster, t(538) = 2.74, p < .01, Cohen’s d = 0.24. When we further examined gender difference for the four difficulty categories in this cluster, we found gender difference in two difficulty categories: Male students (M = 4.47, SD = 1.89) had higher score than females (M = 4.08, SD = 1.87) in the categories of uncrystallized identity, t(538) = 2.41, p < .05, Cohen’s d = 0.21, and male students (M = 3.63, SD = 1.97) also reported higher level of difficulties than female students (M = 3.17, SD = 1.77) in the cluster of conflictual attachment separation, t(538) = 2.86, p < .01, Cohen’s d = 0.25.

In addition, students from countryside, in comparison to those from cities, reported slightly higher levels of difficulties in anxiety related to uncertainty category, M = 4.91 vs. M = 4.58, t(538) = −2.07, p < .05, Cohen’s d = 0.18, and uncrystallized identity category, M = 4.44 vs. M = 3.93, t(538) = 3.01, p < .01, Cohen’s d = 0.28. Finally, no statistically significant differences were found across the academic year for both total score and three cluster scores.