You are on page 1of 15


Confirmatory Factor Analysis of the Career Factors Inventory on a Community College Sample
Merril A. Simon California State University, Northridge Esau Tovar Santa Monica College
A confirmatory factor analysis was conducted using AMOS 4.0 to validate the 21item Career Factors Inventory on a community college student sample. The multidimensional inventory assesses types and levels of career indecision antecedents. The sample consisted of 512 ethnically diverse freshmen students; 46% were men and 54% were women. Participants ranged in age from 16 to 50 (M = 19.3, SD = 3.43). Goodness-of-fit statistics revealed that a minor revision of the four-factor structure proposed by Chartrand and Robbins fit the data, whereas the original model did not. The revision entailed dropping one item while retaining the four factors. Item loadings ranged from .44 to .89; factor intercorrelations ranged from .20 to .80. The internal consistency for the subscales ranged from .77 to .86 and .87 for the total inventory. Keywords: factor analysis, structural equation modeling, career indecision, college students, community colleges

Career indecision has been broadly conceptualized and studied in the college setting over the past 40 years (e.g., Ashby, Wall, & Osipow, 1966; Gordon, 1995; Lewallen, 1993; Luzzo, 2000; Osipow, 1999; Simon, 1990; Vondracek, Hostetler, Schulenberg, & Shimizu, 1990). Most of the studies have involved 4-year college or university studentshence validations of the existing assessment instruments
Merril A. Simon, Department of Educational Psychology and Counseling, California State University, Northridge. Esau Tovar, Counseling Department, Santa Monica College. This research was partially supported by a Faculty Instructional Improvement grant from the California Community Colleges Chancellors Office. The authors wish to thank Dr. John Gonzalez, Dean Judith Penchansky, Dean Brenda Benson, and Melissa Edson for their support in carrying out this research and Dr. Charles Healy and Dr. Adele Gottfried for reviewing an earlier manuscript. Correspondence concerning this article should be addressed to Merril A. Simon, Department of Educational Psychology and Counseling, California State University, Northridge, Michael D. Eisner College of Education, 18111 Nordhoff Street, Northridge, CA 91330-8265; e-mail: merril. simon@csun. edu. JOURNAL OF CAREER ASSESSMENT, Vol. 12 No. 3, August 2004 DOI: 10.1177/1069072703261538 2004 Sage Publications



are primarily based on that population. To meaningfully study the relationships of decidedness about career or major in other college populations, measures of decidedness need to be validated for those populations. Specifically, there is a need to study decidedness of community college students as they represent a significant majority of college freshmen. In 2000, 37% of all undergraduate students were enrolled in public 2-year colleges (U.S. Department of Education Institute of Education Sciences, 2002), and in California, the state with the largest college population in the United States, 83% of fall 2001 college freshmen students studied at a community college (California Postsecondary Education Commission [CPEC], 2003). The question to be considered is whether these community college students career development needs are the same as students at 4-year collegeswhere the majority of studies have been conducted. Community college students have been found to complete bachelors degrees at much lower rates than their peers at 4-year schools as well as to reap the associated benefits (Pascarella & Terenzini, 1991, p. 641) of such educational attainment at a lower rate. It is likely that community college students needs may be different because their characteristics are different. Community college students are typically employed at higher rates (A. Cohen & Brawer, 2002), are often older (CPEC, 2002; U.S. Department of Education Institute of Education Sciences, 2002), represent more students of color (CPEC, 2002; Chenoweth, 1998; Nora & Rendon, 1990), have greater family responsibilities (A. Cohen & Brawer, 2002), commuteoften by public transportation in urban colleges (Gonzalez, 2000), spend more time in classroom with college peers than socialize outside the classroom (Hagedorn, Maxwell, & Hampton, 2001; Tinto, 1993), attend college on a less than full-time and/or more varied daytime/evening basis (A. Cohen & Brawer, 2002; Grimes, 1997; U.S. Department of Education Institute of Education Sciences, 2002), and are generally poorer than 4-year college students (Tinto, Russo, & Kadel, 1994; U.S. Department of Education Institute of Education Sciences, 2002). More of these students also have personal issues such as child care needs, greater job demands, financial issues (Sandler, 2000), or family problems that may also affect their career development process while in college (Hagedorn et al., 2001; Kerka, 1995). In addition, they are retained and persist at lower rates than 4-year college/university students (Kahn, Nauta, Gailbreath, Tipps, & Chartrand, 2002; Pascarella & Terenzini, 1991; Tinto, 1993). Because community college students are different, assessment instruments designed to assess students readiness and commitment to the career selection process must be adequately validated with this population. The Career Factors Inventory (CFInv; Chartrand & Robbins, 1997; Chartrand, Robbins, Morrill, & Boggs, 1990) is a promising measure based on its multidimensional orientation to assess both level and type of career indecision (Chartrand & Robbins, 1990, 1997; Hartman & Fuqua, 1983), its validation studies (Chartrand & Robbins, 1997; Lewis & Savickas, 1995), and utility with college populations in general (Chartrand, Martin, Robbins, & McAuliffe, 1994; Chartrand & Nutter, 1996).



Chartrand and Robbinss (1990) early research compared the use of the CFInv and the Career Decision scale (Osipow, Carney, Winer, Yanico, & Koschier, 1976) and discussed how a multidimensional assessment tool can be used to help counselors differentiate between simple indecision, indecisiveness, and maladjustment in conjunction with interview data (p. 176). Kelly and Lee (2002) found that the CFInv give[s] the broadest coverage to the domain of career decision problems . . . [and] also the most efficient measure (Gati, Osipow, Krausz, & Saka, 2000, p. 324) as compared to the Career Decision scale and the Career Decision Difficulties Questionnaire upon in-depth factor analysis. They indicated the factor structure of the CFInv is superior to that of the Career Decision scale based on its stable factor structure (Kelly & Lee, 2002). Chartrand and Nutter (1996) described the theoretical constructs, purposes, and psychometric properties of the CFInv. In addition, they provided illustrative examples of its utility in various career counseling situations that included precounseling or consultation, problem definition, intervention selection and design, educative framework, client/student feedback, and evaluation and verification. A number of studies have used the CFInv with college students. C. Cohen, Chartrand, and Jowdy (1995) found that four cluster groups of career-undecided college students differed along their level of successful Eriksonian ego identity resolution, thus supporting the need for supportive, ongoing career development interventions. The CFInv and other career assessment instruments were used by Gaffner and Hazler (2002) with a traditional-aged university sample (N = 111), finding that lack of career readiness was the best single predictor of indecisiveness than any combination of variables in the sample. As a second generation (Osipow, 1999) assessment instrument tapping into the very complex construct of career indecision, the CFInv represents a significant improvement over the instruments discussed earlier. However, although normative data and reported psychometric properties of the CFInv have generally been supported across a few studies, it is not without shortcomings. First, the CFInv factor structure was derived from a sample consisting of primarily Caucasian students (74%) of whom only 19% were freshmen (Chartrand et al., 1990) and later validated with a similar sample (83% Caucasian, 28% freshmen). Thus, these two samples do not completely represent the college student population in the United Statesespecially community collegesor its racial and ethnic composition. Second, at least two studies reported in the literature have found cross-loading of two items on two different factors (Stead & Watson, 1993) and a failure of one item to load at all (Kelly & Lee, 2002). Lastly, several university studies have also reported significantly higher scores on the CFInvs Need for Career Information subscale and to a greater extent in the Need for SelfKnowledge subscale (Chartrand & Nutter, 1996; Kahn et al., 2002; Kelly & Lee, 2002) compared to the normative groups discussed by Chartrand and Robbins (1997). Whereas higher scores may be entirely reflective of the individual samples studied, it is also plausible that the normative groups do not accurately



reflect college students at large, including community college students for whom there are no established norms. Given the shortcomings of the CFInv discussed earlier, the lack of community college samples in normative data, and calls for further research in career indecision with diverse samples (Chartrand & Nutter, 1996; Chartrand et al., 1990; Kelly & Lee, 2002), this study sought to test the hypothesized factor structure of the CFInv as presented by Chartrand and Robbins (1997) on a community college sample using confirmatory factor analysis and to provide additional descriptive and normative data for this understudied population.

METHOD Participants
The sample consisted of 512 randomly selected, first-time college students attending a large, urban, community college on the West Coast. Participants were drawn from a larger pool of freshmen students invited to participate in a college orientation 1 month before school began. Of these, 54% were women and 46% were men. The majority of participants were Latino (39%), followed by Caucasian/White (22%), Asian American (17%), African American (13%), and other (9%). Participants ranged in age from 16 to 50, with an average age of 19.3 years (SD = 3.43). In addition, 58% reported coming from a household with a combined annual income of less than $30,000, 12% between $30,000 and $40,000, and 30% earning more than $40,000. At the time of their orientation, 12% of participants reported an interest in obtaining at least an associate degree, 67% a baccalaureate degree, and 21% a masters or higher degree; 78% indicated needing assistance deciding on a major.

Career Factors Inventory. The CFInv (Chartrand & Robbins, 1997) is a selfscorable and interpretable instrument consisting of 21 items designed to assess an individuals readiness to engage in the career decision-making process. The CFInv consists of 10 items rated on a 5-point Likert scale anchored by 1 (strongly disagree) and 5 (strongly agree), with low scores indicating less career indecision. The 11 remaining items are answered using a semantic differential scale. For example, Item 7 reads, When I think about actually deciding for sure what I want my career to be, I feel. Individuals are then given the answer choice of 1 (confident) to 5 (frightened). The 21 items load on the following four factors: Need for Career Information (6 items), Need for Self-Knowledge (4 items), Career Choice Anxiety (6 items), and Generalized Indecisiveness (5 items), upon which four subscales of the same names are based. Psychometric informa-



tion for the CFInv was first reported by Chartrand et al. (1990). Subsequently, various studies for college students (Lewis & Savickas, 1995), high school students (Bizot, 1996), female offenders (Chartrand, 1997), and adult (Chartrand & Nutter, 1997) samples have been conducted and their reliability and validity information reported. Internal consistency for the four subscales ranges from .73 to .91 and .92 for the total inventory (Chartrand & Robbins, 1997). Adequate discriminant, concurrent, and convergent validity have also been reported (Chartrand & Nutter, 1997; Chartrand & Robbins, 1990; Lewis & Savickas, 1995). Student demographic questionnaire. The student demographic questionnaire consisted of 35 items written by the authors designed to assess various background variables, including educational background, academic pursuits, parental information, finances, and work plans.

The CFInv and a student demographic questionnaire were completed during a daylong college orientation program in a classroom setting. Participants were informed of the voluntary nature of data collection and were assured complete confidentiality. Informed consent forms were collected from all participants. Students were thoroughly debriefed on the nature of the study and were offered the opportunity to meet with an academic counselor to discuss the results of the inventory.

Statistical Analyses
AMOS 4.0 (Arbuckle & Wothke, 1999), a program for structural equation modeling techniques, was used to evaluate how well the specified models adequately described the data. Statistical analyses conducted included confirmatory factor analyses and reliability analyses to assess the internal consistency of the inventory and its four subscales. Given our wish to extend the literature on the CFInv and validate the instrument on a community college sample, we decided a priori to use model misspecification information to revise and improve the model if warranted. All statistical analyses were set with an alpha level of .05.

Model Estimation
Maximum likelihood (ML) estimation was used to perform the confirmatory factor analyses in this study. In assessing model fit, the independence model was compared to the hypothesized and respecified models. Given the significant lack of consensus present in the literature for preferred indices of fit (e.g., Bentler,



Table 1 Summary of Goodness-of-Fit Indices for Alternate Models for the Career Factors Inventory
Goodness-of-Fit Measures Models 2 df 183 189 2/df < 3.0

NFI < .90


CFI < .95


TLI < .95


RMSEA < .06d .07 .14

Optimal value Hypothesized fourfactor model 588.87** One-factor model 2156.39** Respecified Model 2: four-factor model, Item 21 eliminated 563.98** Respecified Model 3: four-factor model, Item 21 eliminated; error 11 and 12 correlation 384.17**e

3.2 11.41

.86 .50

.90 .52

.89 .47













Note. NFI = normed fit index; CFI = comparison fit index; TLI = Tucker-Lewis index; RMSEA = root mean square error of approximation. a. Kline (1998). b. Bentler and Bonett (1980). c. Bentler (1990). d. Hu and Bentler (1999). e. 2 (1, N = 490) = 179.81, p < .001. ** p < .001.

1990; Bentler & Bonett, 1980; Hu & Bentler, 1995, 1999; Kline, 1998; MacCallum, Browne, & Sugawara, 1996; Ullman, 1996), it was decided that multiple goodness-of-fit indices, residual error terms, modification indices, and their accompanying expected parameter change (Arbuckle & Wothke, 1999) would be used to assess model fit. The indices used included the 2 to df ratio, the normed fit index (NFI), the comparative fit index (CFI), the Tucker-Lewis index (TLI), and the root mean square error of approximation (RMSEA). Optimal values desired for these indices are reported in Table 1.

Model Specification
The hypothesized model for this validation study of the CFInv on a community college student sample was based on that described in the Career Factors Inventory Application and Technical Guide (Chartrand & Robbins, 1997). The inventory is hypothesized as a four-factor model consisting of four to six indica-



tor variables, each loading in the following pattern: Need for Career Information (NCI: Items 1, 13, 14, 16, 20, and 21), Need for Self-Knowledge (NSK: Items 2, 3, 15, and 17), Career Choice Anxiety (CCA: Items 7, 8, 9, 10, 11, and 12), and Generalized Indecisiveness (GI: Items 4, 5, 6, 18, and 19). The hypothesized model allowed for the possibility of covariance among factors. The first congeneric variable for each factor was fixed to a 1.0 loading and to a zero loading on all other factors. The error terms associated with the observed variables were assumed to be independent of all other error terms.

RESULTS Assessment of Normality and Outliers

Descriptive statistics for each item were derived and evaluated for univariate and multivariate normality. There was evidence suggesting that both univariate and multivariate nonnormality were present. Skewness (SK) values ranged from 1.513 to 0.435, with a mean SK of 0.419. Kurtosis (KU) values ranged from 0.921 to 1.807, with a mean KU of 0.172. Mardias coefficient (measure of multivariate kurtosis) was 89.19 (normalized estimate = 32.47). Multivariate outliers were evaluated using the square Mahalanobis distance for all cases. This measure is interpreted like a 2 statistic with degrees of freedom equal to the number of variables and a p value equal to .001 (Tabachnick & Fidell, 1996). This resulted in 22 cases with values near to or in excess of 46.78 being deleted from the analyses. The resulting normalized estimate for Mardias coefficient was 18.77, still indicating multivariate nonnormality. For this reason, bootstrapping was employed to assess model fit. Bootstrapping (Efron & Tibshirani, 1993) is a resampling procedure whereby a sample is treated as the population and from which subsamples equal in size are drawn randomly with replacement for X number of times (generally > 1,000 bootstrap samples are recommended) (Efron, 1988) to determine parameter estimates under nonnormal conditions. Although traditional ML estimation is subjected to meeting multivariate normality, bootstrapping techniques do not require meeting this assumption (Arbuckle, 1999; Efron & Tibshirani, 1993; Zhu, 1997). Bootstrapping is particularly helpful given that ML underestimates standard errors when population distributions are skewed (Ichikawa & Konishi, 1995). Model fit was further assessed by using the Bollen-Stine corrected p value associated with the 2 instead of the usual ML p value. The Bollen-Stine approach modifies the 2 goodness-of-fit statistic by transforming the sample data in such a manner as to fit the hypothesized model perfectly (Byrne, 2001). Bootstrapping was used throughout this study with a final sample size of 490 cases and 2,000 bootstrap subsamples.



Evaluation of Model Fit

Results for the confirmatory factor analyses conducted are presented next. We first describe the results of the hypothesized four-factor model, followed by a onefactor model, and two respecified four-factor models. Independence model and the hypothesized four-factor model. Table 1 presents a summary of bootstrap goodness-of-fit statistics for the independence and hypothesized models. Results indicated the independence model, which tests the hypothesis that all variables are uncorrelated, was a poor fit for the data and should be rejected, 2(210, N = 490) = 4476.50, p < .001. Given that the 2 statistic is heavily influenced by sample size (Byrne, 2001), other goodness-of-fit indices were examined to evaluate the hypothesized model. We next tested the hypothesized four-factor model, also of poor fit, 2(183, N = 490) = 628.13, p < .001, 2 / df = 3.43, NFI = .86, CFI = .90, TLI = .88, RMSEA = .07. A close examination of the standardized residuals and the modification indices (MI) further supported the models significant misfit. The standardized residual covariance matrix revealed that nine residual values were in excess of 2.58. Values above this are considered large and indicative of model misfit (Jreskog & Srbom, 1988). The largest residual of 8.80 was between indicators 11 and 12. The other eight ranged from 2.62 to 2.99 and involved indicators 21 with 4, 5, 11, and 20; 10 with 13; 16 with 15 and 17; and 6 with 14. In reference to the MI, the output revealed a large covariance between Error 11 and Error 12 (MI = 154.28, EPC = 0.54) and a cross-loading of Item 21 on the generalized indecisiveness and career choice anxiety factors. Taking into account the large standardized residuals, the large error covariance, and the cross-loading described earlier, respecification of the model was pursued. One-factor model. Given that the four-factor CFInv represented significant misfit, a one-factor model evaluation was conducted. However, no support for it was found, 2(189, N = 490) = 2312.97, p < .001, 2 / df = 12.24, NFI = .48, CFI = .50, TLI = .45, RMSEA = .15. Thus, respecification of the model was pursued. Respecified Model 2. Post hoc model adjustments were conducted in an effort to develop a better fitting model. Fit statistics for the hypothesized model discussed earlier and the respecified model are presented in Table 1 for comparison. Following the intent of Chartrand et al. (1990) to create pure measures of each factor (p. 495) by allowing items to load on only one factor, Item 21 (Before choosing or entering a particular career area, I need to seek advice from others regarding my choice) was deleted from the respecified model analysis. The indices of fit indicated a marginal improvement for Model 2 over the hypothesized model. Only two standardized residual covariance values in excess of 2.58 were found (indicators 11 with 12 and 16 with 17). In addition, a large error covariance between Items 11 and 12 remained present (MI = 154.32, EPC = 0.54). This suggested that allowing the two errors to correlate would significant-



Table 2 Descriptive Statistics, Internal Reliability, and Scale Intercorrelations for Career Factors Inventory
Scale 1. Need for Career Information 2. Need for Self-Knowledge 3. Career Choice Anxiety 4. Generalized Indecisiveness Total Inventory M 20.38 15.82 15.76 SD 3.67 3.65 5.38 .77 .81 .86 .80 .87 2 .80**/.68a/.65b . 3 .20*/.54a/.45b .16*/.46a/.44b . 4 .24*/.41a/.53b .22*/.42a/.30b .69**/.50a/.53b .

14.11 3.88 66.07 11.59

Note. Present study in bold (N = 490). a. Chartrand (1990), N = 331. b. Lewis and Savickas (1995), N = 214. *p < .01. **p < .001.

ly improve model fit. MacCallum (1995) cautioned that substantive and theoretical consideration must guide error correlations. Measurement error covariances may derive from characteristics of either the items in question or the respondents (Aish & Jreskog, 1990). For example, error correlations may be indicative of item content redundancy. In examining Items 11 (When I think about actually deciding for sure what I want my career to be, my hands feel choices: dry/wet) and 12 (When I think about actually deciding for sure what I want my career to be, my breathing feels choices: loose/tight), it was clear they were attempting to measure anxiety related to career selection as manifested physically. Given the redundancy seen in Items 11 and 12, Model 2 was respecified as discussed in the following. Respecified Model 3. As seen in Table 1, Model 3 represents a substantial improvement in model fit over Model 2, 2(1, N = 490) = 179.81, p < .001. The modification indices indicated respecification might still be possible by allowing Errors 2 and 3 to correlate. However, in the interest of parsimony and given an EPC of .26, Model 3 is proposed as the best-fit model. This final model, including significant standardized coefficients, is presented in Figure 1. The four-factor loadings and variance/covariance parameter estimates indicate they are reasonable and statistically significant compared to the parameter estimates described for the original CFInv (Chartrand et al., 1990). Standardized item loadings range from .57 to .68 for the NCI factor, .66 to .75 for the NSK factor, .44 to .89 for the CCA factor, and .55 to .76 for the GI factor. Table 2 presents mean scores, standard deviations, and intercorrelations for each scale. In general, scores for the subscales in our community college sample were higher than those reported by




.43 .32 .45 .47 .36 .54 .57 .56 .43

q1 q13 q14 q16 q20 q2 q3 q15 q17

.65 .57** .67** .60** .68**

e13 e14 e16 e20


e2 e3 e15 e17

.74 .75** .75** .66**




e7 e8 e9 e10 e11

.58 .78 .67 .67 .19 .24

q7 q8 q9 q10 q11 q12



.89** .82** .82** .49** .44** .69**



e4 e5 e6 e18 e19

.57 .54 .46 .31 .37

q4 q5 q6 q18 q19

.76 .73** .68** .55** .61**


Figure 1. Bootstrap standardized coefficients for the revised model of the Career Factor Inventory for a community college student sample.

Not freely estimated. *p < .01. **p < .001.

Chartrand and Robbins (1997) as norms for the CFInv, namely, C. R. Cohen (1994) and Perrone (1994). Unpaired samples t tests revealed significant differ-



ences in the NCI subscale between our sample and the C. R. Cohen university sample, t(865) = 7.35, p < .0001. Significantly different scores were also found between our sample and the C. R. Cohen and Perrone samples in the NSK subscale, t(865) = 18.48, p < .0001 and t(893) = 16.04, p < .0001, respectively. No other statistically significant scores were found on the other subscales. Correlations among the four factors range from a low .16 between the NSK and CCA factor to a high .80 between the NCI and the NSK factor. Consistent with previous findings, some scale intercorrelations remain moderately high (Chartrand et al, 1990; Lewis & Savickas, 1995).

Although this study did not specifically seek to address all aspects of validity, our findings do lend further support to the use of the CFInv with a college population, specifically, to community college students. A two-way multivariate analysis of variance (MANOVA) was conducted to assess the effects of gender and ethnicity on decidedness as measured by the four subscales of the CFInv. Boxs M test, which had a value of 101.78, revealed that equal variances could not be assumed, F(70, 102082) = 1.40, p < .05; for this reason, Pillais Trace was selected to evaluate overall significance. Results indicated a significant difference only for ethnicity, Pillais Trace = .050, F(12, 1314) = 1.87, p < .05, partial 2 = .017. A follow-up analysis of variance (ANOVA) indicated a statistically significant ethnic group difference in the CCA subscale, F(3, 439) = 3.39, p < .05, partial 2 = .023. A post hoc analysis using Tukeys honestly significant differences test indicated that African American students had lower CCA scores (M = 13.74) than Asian (M = 16.56), Latino (M = 15.58), and White (M = 15.93) students, who did not differ from each other.

Internal consistency for the CFInv subscales and total inventory were calculated using Cronbachs coefficient alpha. These were .77 for the NCI subscale, .81 for the NSK subscale, .86 for the CCA subscale, .80 for the GI subscale, and .87 for the total inventory. These reliability estimates are consistent with those found in two other college samples (Chartrand et al., 1990; Lewis & Savickas, 1995).

The four-factor model of the CFInv consisting of 21 indicator variables presented in previous research (Chartrand & Robbins, 1997) was not precisely repli-



cated in our large, diverse, community college student sample. Respecified Model 3, which eliminated one item from the analyses due to cross-loading on two factors and allowed two error terms to correlate, is proposed as a better fitting model for community college students. With the exception of these changes, the factor structure of the inventory remains unchanged. As noted in Table 2, two pairs of subscales on the CFInv were found to be highly correlated in contrast to previous research (Chartrand et al., 1990; Lewis & Savickas, 1995). In reference to the NCI and NSK subscales, their high correlation may indicate that the community college students in this sample did not distinguish need for self-knowledge (NSK) from the need for career information (NCI). This may mean that they feel their occupational choices are constrained and tied more to who they presently are because they generally are not as academically prepared as university students, will have lower credentials when they seek jobs, and will have to take account of more personal circumstances in taking a job. Or it may mean that they do not yet understand the career development process as they were assessed before the start of their freshman year. In any regard, the linkage of the scales suggests a need for research of whether community college students need additional assistance in distinguishing information needs. As with the NCI and NSK subscales, a moderate correlation between the CCA and GI subscales was also found. This may mean that the imminence of working and therefore choosing the right work preparation is higher for community college students and so weighs more on their general anxiety than other student groups. In contrast, both male and female African American students CCA subscale scores were significantly lower than students of other ethnic groups. This may be attributed to their overall lower anxiety about higher education and to the belief that the specific major of the student is less critical than the completion of the degree itself. Certainly more research is needed of such possibilities to more fully understand the decidedness of community college students. Findings from this study suggest that the CFInv is an instrument that can help in such research. Essential replication of the factor structure of the CFInv suggests that counselors can use the inventory in helping to understand the level and type of career decidedness of community college students so long as they recognize the need to evaluate whether the factors are linked as shown in these participants. Furthermore, from a counseling perspective, it is incumbent that community college counselors carefully interpret the CCA subscale score as our research demonstrated that two items (11 and 12) were indicative of content redundancy. Specifically, the resulting score may be higher than expected and thus provide an inaccurate picture of the students true state of anxiety related to career selection. An elevated subscale score may also occur with the incorporation of Item 21 when using the published CFInv. Our finding that Item 21 did not load on the NCI factor may also artificially inflate that subscale and indicate more need for career information than is truly reflective of their level of indecision on that subscale.



Although we believe our findings further the research on career indecision, particularly as measured by the multidimensional CFInv, caution should be exercised in interpreting our findings. Specifically, this study used only one community college sample to test and validate the factor structure of the inventory. Until additional studies with community college students are conducted and our proposed model is replicated, our findings should be considered tentative. Future career indecision research may now be conducted using the CFInv with a community college population with greater confidence having established its validity with this population. Research with college students (Hagedorn et al., 2001; Osipow, 1999; Pascarella & Terenzini, 1991) has found connections between level of career decidedness and students academic experience as discussed earlier. For example, the relationship between career indecision with retention, persistence, and academic success of community college students as well as research that would explore the meaning of being undecided while attending community college may now be more meaningfully conducted with this important and understudied population. Given Mitchell, Levin, and Krumboltzs (1999) recent work on planned happenstance with university students, it would be of interest to also consider the potential value of indecision with community college students as well. Future research could also extend the work of Healy and Reilly (1989) in their extensive study of career development needs of community college students. With a greater understanding of the construct of career indecision in community college students, more meaningful interventions may be developed and used to positively support these students development, decision making, and potentially greater success as college students.

Aish, A. M., & Jreskog, K. G. (1990). A panel model for political efficacy and responsiveness: An application of LISREL 7 with weighted least squares. Quality and Quantity, 19, 716-723. Arbuckle, J. L. (1999). Amos, Version 4. Chicago: SmallWaters. Arbuckle, J. L., & Wothke, W. (1999). Amos 4.0 users guide. Chicago: SPSS. Ashby, J., Wall, H., & Osipow, S. (1966). Vocational uncertainty and indecision in college freshmen. Personnel & Guidance Journal, 44, 1037-1041. Bentler, P. M. (1990). Comparative fit indexes in structural models. Psychological Bulletin, 107, 238-246. Bentler, P. M., & Bonett, D. G. (1980). Significance tests and goodness of fit in analysis of covariance structures. Psychological Bulletin, 88, 588-606. Bizot, E. (1996, August). Career Factors Inventory reliability and correlates in ninth graders. Poster session presented at the annual meeting of the American Psychological Association, Toronto, Canada. Byrne, B. M. (2001). Structural equation modeling with AMOS: Basic concepts, applications, and programming. Hillsdale, NJ: Lawrence Erlbaum. California Postsecondary Education Commission. (2002). First time freshmen. Retrieved December 7, 2002, from



Chartrand, J. M. (1997). [Career Factors Inventory results life/career-planning intervention for female offenders]. Unpublished raw data. Chartrand, J. M., Martin, W., Robbins, S., & McAuliffe, G (1994). Testing a level versus an interactional view of career indecision. Journal of Career Assessment, 2, 55-69. Chartrand, J. M., & Nutter, K. J. (1996). The Career Factors Inventory: Theory and applications. Journal of Career Assessment, 4, 205-218. Chartrand, J. M., & Nutter, K. J. (1997). [Career Factors Inventory results from career decision courses]. Unpublished raw data. Chartrand, J. M., & Robbins, S. (1990). Using multidimensional career decision instruments to assess career decidedness and implementation. Career Development Quarterly, 39, 166-177. Chartrand, J. M., & Robbins, S. (1997). Career Factors Inventory application and technical guide. Palo Alto, CA: Consulting Psychologists Press. Chartrand, J. M., Robbins, S. B., Morrill, W. H., & Boggs, K. (1990). Development and validation of the Career Factors Inventory. Journal of Counseling Psychology, 37, 491-501. Chenoweth, K. (1998). The surging degree wave. Black Issues in Higher Education, 15(10), 20-22. Cohen, A., & Brawer, F. (2002). The American community college (4th ed.). San Francisco: JosseyBass. Cohen, C., Chartrand, J., & Jowdy, D. (1995). Relationships between career indecision subtypes and ego identity development. Journal of Counseling Psychology, 42, 440-447. Cohen, C. R. (1994). Relationships between career indecision subtypes and ego development. Unpublished masters thesis, Virginia Commonwealth University, Richmond. Efron, B. (1988). Bootstrap confidence intervals: Good or bad? Psychological Bulletin, 104, 293296. Efron, B., & Tibshirani, R. J. (1993). An introduction to the bootstrap. New York: Chapman & Hall. Gaffner, D., & Hazler, R. (2002). Factors related to indecisiveness and career indecision in undecided college students. Journal of College Student Development, 43, 317-326. Gati, I., Osipow, S. H., Krausz, M., & Saka, N. (2000). Validity of the Career Decision-Making Difficulties Questionnaire: Counselee versus career counselor perceptions. Journal of Vocational Behavior, 56, 99-113. Gonzalez, J. M. (2000). Community colleges: Open or revolving door? Intervention strategies to reduce the number of probationary students and to increase persistence (Doctoral Dissertation, University of California, Los Angeles, 2000). Dissertation Abstracts International, 61, 65. Gordon, V. N. (1995). The undecided college student: An academic and career advising challenge (2nd ed.). Springfield, IL: Charles C Thomas. Grimes, S. K. (1997). Underprepared community college students: Characteristics, persistence, and academic success. Community College Journal of Research and Practice, 21, 47-56. Hagedorn, L. S., Maxwell, W., & Hampton, P. (2001). Correlates of retention for AfricanAmericans in community colleges. Journal of College Student Retention, 3, 243-263. Hartman, B. W., & Fuqua, D. R. (1983). Career indecision from a multidimensional perspective: A reply to Grites. The School Counselor, 30, 340-349. Healy, C. C., & Reilly, K. C. (1989). Career needs of community college students: Implications for services and theory. Journal of College Student Development, 30, 541-545. Hu, L. T., & Bentler, P. M. (1995). Evaluating model fit. In R. H. Hoyle (Ed.), Structural equation modeling: Concepts, issues, and applications (pp. 76-99). Thousand Oaks, CA: Sage. Hu, L. T., & Bentler, P. M. (1999). Cutoff criteria to fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling: A Multidisciplinary Journal, 6, 1-55. Ichikawa, M., & Konishi, S. (1995). Applications of the bootstrap methods in factor analysis. Psychometrika, 60, 77-93. Jreskog, K. G., & Srbom, D. (1988). LISREL 7: A guide to the program and applications. Chicago: SPSS.



Kahn, J. H., Nauta, M. M., Gailbreath, R. D., Tipps, J., & Chartrand, J. M. (2002). The utility of career and personality assessment in predicting academic progress. Journal of Career Assessment, 10, 3-23. Kelly, K., & Lee, W. (2002). Mapping the domain of career decision problems. Journal of Vocational Behavior, 61, 302-326. Kerka, S. (1995). Adult learner retention revisited. Columbus, OH: ERIC Clearinghouse on Adult, Career and Vocational Education. (ERIC Reproduction Service No. ED 289880) Kline, R. B. (1998). Principles and practice of structural equation modeling. New York: Guilford. Lewallen, W. (1993). The impact of being undecided on college-student persistence. Journal of College Student Development, 34, 103-112. Lewis, D., & Savickas, M. (1995). Validity of the Career Factors Inventory. Journal of Career Assessment, 3, 44-56. Luzzo, D. A. (2000). Career counseling of college students: An empirical guide to strategies that work. Washington, DC: American Psychological Association. MacCallum, R. C. (1995). Model specification: Procedure, strategies, and related issues. In R. H. Hoyle (Ed.), Structural equation modeling (pp. 16-36). Thousand Oaks, CA: Sage. MacCallum, R. C., Browne, M. W., & Sugawara, H. M. (1996). Power analysis and determination of sample size for covariance structure modeling. Psychological Methods, 1, 130-149. Mitchell, K., Levin, A., & Krumboltz, J. (1999). Planned happenstance: Constructing unexpected career opportunities. Journal of Counseling and Development, 77, 115-124. Nora, A., & Rendon, L. (1990). Determinants of predisposition to transfer among community college students: A structural model. Research in Higher Education, 31, 235-255. Osipow, S. (1999). Assessing career indecision. Journal of Vocational Behavior, 55, 147-154. Osipow, S., Carney, C., Winer, J., Yanico, B., & Koschier, M. (1976). The Career Decision scale (3rd Rev. ed.). Columbus, OH: Marathon Consulting & Press. Pascarella, E., & Terenzini, P. (1991). How college affects students. San Francisco: Jossey-Bass. Perrone, K. M. (1994). Role model influence on the career aspirations of young adult college students. Unpublished masters thesis, Virginia Commonwealth University, Richmond. Sandler, M. E. (2000). Career decision-making self-efficacy, perceived stress, and an integrated model of student persistence: A structural model of finances, attitudes, behavior, and career development. Research in Higher Education, 41, 537-580. Simon, M. A. (1990). Career indecision in graduate business students: Correlates and moderators. Unpublished doctoral dissertation, University of Southern California. Stead, G. B., & Watson, M. B. (1993). How similar are the factor structures of the Career Decision scale, the Career Decision Profile, and the Career Factors Inventory? Educational and Psychological Measurement, 53(1), 281-290. Tabachnick, B. G., & Fidell, L. S. (1996). Using multivariate statistics (3rd ed.) New York: HarperCollins. Tinto, V. (1993). Leaving college: Rethinking the causes and cures of student attrition (2nd ed.). Chicago: University of Chicago Press. Tinto, V., Russo, P., & Kadel, S. (1994). Constructing educational communities: Increasing retention in challenging circumstances. Community College Journal, 64, 26-29. Ullman, J. B. (1996). Structural equation modeling. In B. G. Tabachnick & L. S. Fidell (Eds.), Using multivariate statistics (3rd ed., pp. 748-753). New York: HarperCollins. U.S. Department of Education Institute of Education Sciences. (2002, May). Projections of educational statistics to 2012 (Tables A1.6 and A1.7). Retrieved December 7, 2002, from http://nces. Vondracek, F. W., Hostetler, M., Schulenberg, J. E., & Shimizu, K. (1990). Dimensions of career indecision. Journal of Counseling Psychology, 37, 98-106. Zhu, W. (1997). Making bootstrap statistical inferences: A tutorial. Research Quarterly for Exercise and Sport, 68, 44-55.