Confirmatory Factor Analytical Study of the Revised Developmental Work Personality Scale

Abstract

This study investigated psychometric properties of the Revised Developmental Work Personality Scale (RDWPS). Results yielded a 14-item three-factor model that aligns with the original DWPS and fits the data very well. RDWPS scores were useful in predicting the resolution of Erikson’s fourth stage of development, indicating construct validity.

Keywords

developmental work personality confirmatory factor analysis vocational behaviors psychosocial development role model

Organizational psychology, career, and vocational research consistently demonstrate that job success is related to strong work completion and strong interpersonal skills (Herr & Cramer, 1996; Ployhart, Schneider, & Schmitt, 2006). In today’s labor market, this translates to employees’ ability to demonstrate qualities such as adaptability, reliability, teamwork, and the ability to take on new tasks or new technology when needed (Blustein, 2006; Friedman, 2005; Ployhart et al., 2006). This can be challenging for many employees but may be particularly difficult for persons who struggle to demonstrate these qualities positively. Multiple explanations exist for why some people struggle to express these important traits positively and consistently, or alternately, why some people are at a higher risk for being perceived as expressing these traits negatively. One emerging and promising construct, the developmental work personality, may explain how, when, where, and with whom work behavioral traits are developed and subsequently expressed.

Overview of Developmental Work Personality Theory and Model

The Developmental Work Personality (DWP) model synthesizes Bandura’s (1986, 1989) social learning theory, Erikson’s (1968) developmental stage theory, and Neff’s (1986) ideas on workplace personality to explain how the home and school environments, and the role models embedded in these environments, influence identity during the grade school years, which continues into adult work-role behavior. The DWP model outlines how role models in childhood influence the developing identity related to (a) completion of school and home tasks and (b) social and interpersonal skills of relating to peers and authority figures. These two primary domains of childhood development at this stage are believed to strongly influence critical adult work behaviors of task completion and social skills (O’Sullivan & Strauser, 2010; O’Sullivan, Strauser, & Wong, in press). The positive expression of these two behavioral domains was found to be desirable to a range of employers and human resource managers and necessary for job tenure and length of employment (Fan, Thompson, & Wang, 1999; Ployhart et al., 2006).

According to Erikson, people develop in the life course during eight distinct stages that present crises or conflicts (Erikson, 1968). How a person resolves a given crisis will contribute to the progression to the next stage and overall personality development. If a person is able to resolve the major conflict during a developmental stage, the person is thought to carry that resolution throughout the lifespan (Erikson, 1963, 1968). Erikson used the term crisis or conflict, because in his view, each stage presents the individual with life circumstances that have not yet been navigated, which creates a sense of vulnerability. Each new challenge is experienced by the individual as a crisis until it is resolved, either mostly positively or mostly negatively. Erikson was careful to point out that positive resolution should not be interpreted as “achievement,” implying complete and final resolution (Erikson, 1959). Rather, Erikson would say that a child acquires a ratio of positive to negative resolution, where, if the balance is tipped toward the positive, then the later developmental crises are more likely to be navigated with the advantage of strengths, and these strengths are integrated into the overall development and personality formation (Erikson, 1959). Additionally, Erikson believed that resolution of a mostly negative conflict can be revisited at ages typically beyond the initial developmental stage when the conflict first occurred, and a rebalancing toward a mostly positive resolution can be realized given appropriate intervention. According to Erikson (1959, 1963, 1968), middle childhood provides opportunities for children to view themselves as being either mostly Industrious or mostly Inferior, depending on the child’s resolution of self-mastery of tasks and a view of the self as being either competent or not. Across cultures, children in this stage (ages 6–12) receive formal instruction in one way or another. Erikson succinctly summarizes that children during this stage center their identity crisis resolution on the theme of “I am what I learn” (Erikson, 1959).

Bandura (1969, 1977a) explained that children learn to identify in one way or another by watching role models behave in either competent or incompetent ways and by incorporating the feedback from role models into their repertoire of behaviors via rehearsal. According to Bandura’s observational learning theory (Bandura, 1977a), children must be able to attend to and retain observed events for later rehearsal. Children are most likely to retain and rehearse events performed by role models that are perceived as salient and attractive to the child, such as parents and teachers. When children rehearse events, their motivation to continue the behavior depends largely on the role model’s feedback to the child, which can include verbal praise, correction, or explanation of consequences to continue or discontinue behavior. For children in the middle childhood stage of development, rehearsal takes the form of play and work at home and at school. At school, rehearsal occurs in the classroom and on the schoolyard. Most children at this stage attend school, which requires punctuality and regular attendance as well as homework completion, classroom behavior management, and observation of teachers; playground interactions and involvement in extracurricular activities provide peer interaction and development of social skills. The home environment provides opportunities to complete chores and observe parents’ and older siblings’ behavior. A rich social environmental structure embedded with positive role models and opportunities for observation, rehearsal, and feedback will likely contribute to the development of an industrious identity. Bandura suggested that as children progress through this stage, their self-efficacy related to learning new behaviors will increase if sufficient support for learning occurs (Bandura, 1977a). Higher self-efficacy likely contributes to the industrious identity at this stage. Inconsistent or inadequate role model feedback regarding the learned behaviors reduces self-efficacy (Bandura, 1977a). Lower self-efficacy related to task completion contributes to a view of the self as being less competent.

The DWP model was conceptualized from an identity-formation perspective with an understanding that barriers to positive development include impoverished environments, lack of positive role models, and reduced opportunities to rehearse social and work tasks. For example, the experience of disability during this stage can disrupt the identity development. Adult survivors of childhood illness and disability often spent much of their childhood experiencing symptoms and/or recovering, which kept them away from school and many typical childhood activities that foster strong social skills and work ethic (Spelten, Sprangers, & Verbeek, 2002; Taskila & Lindbohm, 2007). For instance, because of functional limitations, absence from school, or overprotective parents and teachers, these children are likely to miss out on opportunities that teach them to get along with others on the playground, be accountable for homework and other tasks completed on time, and follow classroom and home rules. According to several theorists and work-behavior researchers, these learned behaviors in middle childhood contribute to a strong work personality in adulthood (Erikson, 1968; Neff, 1986; Strauser, Waldrop, & Ketz, 1999; Vaillant & Vaillant, 1981). A brief instrument was constructed from the model to assess childhood experiences believed to relate to adult work behavior.

Overview of Recent Research Results Using DWP Scale

The original version of Developmental Work Personality Scale (DWPS) was developed by Strauser and Keim (2002). The DWPS is a 26-item self-report measure addressing social behaviors, role models, and tasks that individuals encounter during middle childhood and which are consistent with Erikson’s developmental stage of Industry vs. Inferiority (Erikson, 1959, 1963). The DWPS asks participants to reflect on their childhood experiences related to task completion, social interactions, and role models. A recent three-phase study presented evidence for robust psychometric properties of data from scale scores for use in research to better understand personality development related to work behaviors for persons with disabilities (O’Sullivan & Strauser, 2010). Results of this study found strong construct and convergent validity evidence of the model and scale scores in a sample of adults with disabilities. Three distinct factors emerged that aligned with the model, and positive and significant correlations with other measures of work behavior and personality were found. The internal consistency of each subscale score was adequate to good in this sample of adults with disabilities: Work Tasks (α = .89), Role Model (α = .75), and Social Skills (α = .90; O’Sullivan & Strauser, 2010). A student sample was used to test reliability. The Role Model subscale score revealed inadequate internal consistency (α = .40) and low test–retest reliability (r = .29) in the student sample. Although the correlation of the Role Model scale at Time 1 was positive and significantly related to Time 2, the reliability coefficients of this factor clearly need enhancement for use in samples of people without disability.

The problems with the Role Model scale in the student sample may have been caused by the low number of items (three), as well as the item content. On further analysis of the full scale and the item alignment with the DWP model, we hypothesized that the items relating to role model in the DWPS were inadequately capturing the construct and behavior of role models, as theorized by Bandura (1977, 1986). To improve the psychometric properties of the scale scores for use in broader populations, we revised the scale and added items that more accurately reflect the different types of role models that children in this stage encounter, the feedback and interaction component of role model–child relationship, and the feelings that children with positive role models develop as a result of the relationship.

Role Models

Role models have been conceptualized and operationalized in a variety of ways, including as either real or hypothetical persons worthy of imitation, persons who inspire others to be like them, and simply salient older or more advanced respected mentors. According to social cognitive and career-development researchers (Bandura, 1977a; Nauta & Kokaly, 2001; Hackett & Betz, 1981), people who observe role models performing behaviors beyond their current developmental stage improve their efficacy for performing the observed behavior later. This is true for students who observe role models performing tasks in career tracks that are beyond the students’ current developmental abilities who later achieve similarly in their careers to their role models or mentors (Hackett, Esposito, & O’Halloran, 1989). Others have postulated that role models do more than demonstrate and teach but actually inspire others to behave in ways that may not have been attractive before (Nauta & Kokaly, 2001). Essentially, positive role models must be relevant to the child, provide consistent feedback regarding learning of new behaviors, and inspire a person to persist at task improvement despite difficulty or barriers (Nauta & Kokaly, 2001).

Potential Application of RDWPS to Population Without Disabilities

Although the development of the original scale was primarily tested and validated on a group of people with disabilities, we believe this scale could be applied in a general population of young adults as the theoretical underpinnings of DWP model was synthesized based on Bandura’s (1969, 1977a) social learning theory and Erikson’s (1968) developmental stage theory, both of which explain normative experiences in diverse populations and explain how contextual environments during childhood influence the work-role behavior in adults. Therefore, one purpose of this study was to confirm if the modified scale could be expanded to the general population of young adults to explain their work personality development, taking the contextual demands and developmental challenges of this stage into account.

A modification of the DWPS was constructed for the present study with the intention of confirming the factor structure in a sample of persons without disability and improving items related to role models that better reflect the theoretical underpinnings of Bandura’s social cognitive and observational learning theories. Therefore, the goals of the present study were threefold: (a) explore and confirm the factor structure of the Revised DWPS (RDWPS) in two samples of students without disabilities and provide psychometric properties of the data from the revised scale, (b) investigate the properties of the revised items and their relationship with other measures of role model influence, and (c) investigate construct validity of the RDWPS by analyzing its ability to predict the degree to which the probability of successful/unsuccessful resolution of Erikson’s fourth stage (as assessed by Measures of Psychosocial Development [MPD] scores).

Method

Participants and Procedures

Two samples of college students were recruited for this study. Eligible individuals were at least 18 years of age, enrolled in classes at a large Midwestern university, were male or female, voluntarily agreed to participate, and indicated that they did not have a disability. Sample 1 was used for the exploratory factor analysis and first confirmatory factor analysis and consisted of 295 college students. These participants ranged in age from 18 to 28 years (Mean = 20.93, SD = 1.20). The majority of students were female (70.5%). Racial composition of the sample was as follows: Caucasian (73.9%), African American (6.5%), Asian (8.9%), Hispanic (7.9%), and biracial (2.7%). The majority of students are seniors (50.5%), followed by juniors (28.5%), sophomores 13.9%), and freshman (7.1%). Based on a classification model provided by National Science Foundation (National Science Foundation, 2010), 16.6% of students studied in Science Technology Engineering Mathematics (STEM) programs and 83.4% of student studied in non-STEM programs. The second sample (n = 255) of college students were demographically comparable to the first sample and ranged in age from 18 to 28 years (Mean = 20.43, SD = 1.23). Sample 2 was used for a second confirmatory factor analysis to confirm the factor and structural model obtained in the first confirmatory factor analysis. The majority of students in this sample are female (64.3%). Racial composition of Sample 2 is as follows: Caucasian (69.4%), African American (11.0%), Asian (11.0%), Hispanic (5.5%), and biracial (2.7%). The majority of students in Sample 2 were seniors (43.5%), followed by juniors (26.7%), sophomores (22.0%), and freshman (7.8%). For Sample 2, 14.1% of students studied in STEM programs and 85.9% of student studied in non-STEM programs. Independent t test and chi-square analyses were conducted to determine if there were any significant differences in demographic variables between two groups. No significant differences were found among the two groups on gender, χ²(1) = 2.398, p = .122; ethnicity, χ²(5) = 6.391, p = .270; year of education, χ²(3) = 6.708, p = .082; and major of study χ²(1) = .650, p = .420. A significant difference was found on age, t(548) = 4.797, p < .001, and the age difference between two groups was around half a year (mean difference = .497).

Protocol packets including consent form, demographic form, and research instruments were disturbed to students during class with informed consent. Students in both samples were enrolled in the same large undergraduate class that is taught twice a year in the department where one of the investigators is a faculty member. The data from the two samples were collected approximately 1 year apart. No student was enrolled in this class more than once; therefore, no student participated in this study more than once. Students who completed the protocol packets were given time to complete them in class and were allowed to use the completed packets to make up one missed in-class assignment. Students were informed of this in writing and were told verbally and in writing that they could skip any items on any part of the packet and still earn the credit for class. This study had the institutional review board’s approval.

Measures

Revised Developmental Work Personality Scale

The original version of Developmental Work Personality Scale was developed (Strauser & Keim, 2002) as a 26-item self-report measure addressing behaviors, role models, and tasks that individuals encounter during middle childhood and which are consistent with Erikson’s developmental stage of industry versus inferiority (Erikson, 1959, 1963). Individuals reflect on their experiences in childhood and rate DWPS item responses on a scale ranging from 0 (not at all like me) to 5 (very much like me), with reverse scoring for some items. Example items include the following: “In school I completed my work on time” and “I got into fights frequently at school.” A factor analysis of this scale on a group of individuals with disabilities found three distinct constructs of developmental work personality aligned with subscales of the DWPS, each with adequate to good internal consistency ranging from .75 to .90 in a sample of adults with disabilities (O’Sullivan & Strauser, 2010). Results from this study also found that two subscale scores had adequate test–retest reliability in a sample of students (Work Tasks r = .78; Social Skills r = .66), but the role model factor test–retest reliability coefficient was inadequate (r = .29) and inadequate internal consistency (α = .40) in the sample of college students. Convergent validity was found among the five-factor model of personality and the DWPS. Specifically, the Work Tasks subscale of the DWPS was found to positively and significantly correlate with the NEO-FFI subscale of Conscientiousness and Social Skills and was positively and significantly correlated with the NEO-FFI subscale Agreeableness (Costa & McCrae, 1992). The DWPS total significantly and positively correlated with subscales of the Work Personality Profile (Bolton, 1992).

A modification of the role model factor was constructed for the present study. One of the authors (DO) was responsible for creating new items to be included in the revised scale. The new item content was based on the core principles that form the theoretical background of the DWP and literature focusing on role model influence on students’ academic and vocational outcomes (Hill & Tyson, 2009; Nauta & Kokaly, 2001; Rhodes, Grossman, & Resch, 2000). All authors discussed and came to a consensus of the final wording of the new items. A total of 9 items were added to the original 26 items, resulting in a 35-item self-report measure. Examples of new items include the following: “When I needed help with my homework, one of my parents was available to help,” “Growing up, I was responsible for chores at home,” “Growing up, I had someone in my life who inspired me.”

Measures of Psychosocial Development

The MPD is a self-report inventory assessing personality development according to Erikson’s theory of human development (Hawley, 1988). The inventory contains 112 statements that are rated on a 5-point scale ranging from Very much like me to Not at all like me. The MPD has 27 scales, including positive, negative, and resolution scales for each of the eight psychosocial stages, as well as 3 scales that measure the total positive, negative, and resolution scores. Positive and negative scales assess positive and negative attitudes, in which Erikson proposed as the basic constituents of personality. Positive scales include trust, autonomy, initiative, industry, identity, intimacy, generativity, and ego integrity, whereas negative scales include mistrust, shame and doubt, guilt, inferiority, identity confusion, isolation, stagnation, and despair. The resolution scales assess the degree and direction of a resolved crisis, which represent the status of conflict resolution for each of Erikson’s psychosocial stages (Hawley, 1988). A low resolution score suggests inadequate resolution of the stage. Persons who successfully resolved a given stage will have high positive scale scores and low negative scale scores. Reliability of the MPD was found to be adequate to strong, with test–retest coefficients ranging from .75 to .87 and internal consistency coefficients ranging from .75 to .91 (Hawley, 1988). Convergent validity for the MPD was also supported by high intercorrelations of positive scales (.75 to .85) and negative scales (.67 to .89; Hawley, 1988). For the purpose of this study, only the items related to the fourth stage of development (Industry vs. Inferiority) were used, for a total of 14 items. Healthy resolution of this stage reflects an active orientation toward learning, competence, and production as opposed to despair of one’s own skills and abilities. The test–retest and internal consistency coefficients of “Industry versus Inferiority” subscale are .77 and .83, respectively.

Influence of Others on Academic and Career Decisions Scale (IOACDS)

This 15-item instrument assesses individuals’ perceptions of the amount and types of career role model influence and support from others (Nauta & Kokaly, 2001). All items are rated on a 5-point scale ranging from 1 (Strongly disagree) to 5 (Strongly agree) to assess the amount of Support/Guidance (eight items) and Inspiration/Modeling (seven items) individuals perceived from influential others when making academic or career decisions. Sample items include the following: “There is someone who helps me consider my academic and career options” and “There is someone I can count on to be there if I need support when I make academic and career choices.” Internal consistency reliability for the Support/Guidance and Inspiration/Modeling subscales scores were reported to be .85 and .87, respectively (Nauta, Saucier, & Woodard, 2001). Validity was supported by significant correlations with measures of general social support, career indecision, career certainty, and occupational information, as well as nonsignificant correlation with a measure of social desirability (Nauta & Kokaly, 2001).

Overview of Statistical Analyses

Item evaluation

In the first stage, research protocols were distributed to Sample 1, and the 35 items of RDWPS were evaluated with regard to variance and frequency distribution. The items that were not sensitive for use on a normal adult population were deleted. Items showing ceiling effect (extreme means and close to zero variances) were excluded for subsequent factor analysis. The frequency distributions were examined for skewness and kurtosis to assess the normality assumption of the remaining RDWPS items (Weston, Gore, Chan, & Catalano, 2008). For skewness index, absolute values more than 2.0 are considered extreme (Curran, West, & Finch, 1996). For kurtosis index, values more than 7.0 suggest a problem (Curran et al., 1996) and more than 20.0 are considered extreme (Kline, 2005).

Exploratory factor analysis (EFA)

In the second stage, a sample of 92 participants was randomly selected using the randomization function on SPSS version 17.0. An EFA, using principal component analysis, was then conducted on this subset of participants to determine the factor structure of the RDWPS. Kaiser–Meyer–Olkin’s measure of sample adequacy (MSA; Kaiser, 1974) and Barlett’s test of sphericity (Bartlett, 1954) were used to test for a satisfactory factor analysis to proceed. Guttman’s criterion (eigenvalues greater than 1; Guttman, 1954) and Cattell’s scree test (Cattell, 1966) were used to assess the number of factors for extraction (Gorsuch, 1983; Green & Salkind, 2005). After the number of factors was initially determined, the residual correlation matrix was then computed to examine retaining factors did not further improve the fit to the data (Tabachnick & Fidell, 2001). Based on theoretical conceptualization of the DWPS, our previous finding (O’Sullivan & Strauser, 2010), and recommendation in psychometric literature (Costello & Osborne, 2005), we expected some correlations among factors in the RDWPS; thus, oblique rotation using direct oblimin was applied for best interpretation of factor structure. Items with primary factor pattern coefficients ≥.40 (including values that round to .40) and secondary factor pattern coefficients ≤.30 and those that did not load on more than one factor were retained. Items not meeting these criteria were removed one at a time. Analyses were repeated until a solution in which all items retained in the analysis met all criteria (Pai et al., 2007). Item–total correlation and Cronbach’s coefficient alpha were estimated to assess internal consistency, with an α ≥.70 as recommended minimal value (Nunnally & Bernstein, 1994). The factors’ correlations, means, and standard deviations for the subscales in the final solution were also reported.

Confirmatory factor analysis (CFA)

In the third stage, by using Analysis of Moment Structures(AMOS), version 18.0 (Guttman, 1954), a CFA, using the generalized least squares method, was conducted on the remaining 203 participants of Sample 1 to satisfy two purposes. The first purpose is to compare the exploratory model from EFA with several nested models to determine which one is the best-fit model. The second purpose is to determine whether the factor structure required modification using a separate sample of participants. Chi-square (χ²) was computed for comparing the hypothesized model with other models through their fit to the data (Byrne, 2001). Given that significant χ² for all the competing models is not uncommon in the case of large samples, additional measures of fit that are less sensitive to sample size were used (Bentler, 1990; Chan, Lee, Lee, Kubota, & Allen, 2007; Fan et al., 1999; Hu & Bentler, 1999; Steiger, 1990; Weston et al., 2008) and are as follows: relative chi-square (χ²/df; cutoff value for good fit: <2–5), goodness-of-fit index (GFI; ≥.90), comparative fit index (CFI; ≥.90 minimally acceptable, ≥.95 good, and at or close to 1 excellent; Bollen, 1990; Browne & Cudeck, 1993), and root mean square error of approximation (RMSEA, approximating .06 close fit, 0 exact fit of the model; Browne & Cudeck, 1993; Hu & Bentler, 1999). One additional fit index, Akaike information criterion (AIC), was used to determine which of two or more competing models fit best (Akaike, 1974). AIC is considered a predictive fit index and indicates how well models would be expected to fit sample data drawn from the population. It is generally used to choose between models. Smaller AIC values reflect the better fitting model.

Relationship between work personality and role model influence

In the fourth stage, correlation analyses were conducted to evaluate the relationship between the RDWPS and IOACDS, a measure of role model influence.

Relationship between work personality and Erikson’s fourth stage of psychosocial development

In the fifth stage, correlation analyses were conducted to evaluate the relationship between the RDWPS and MPD. Logistic regression was used to assess the discriminative validity of RDWPS in classifying participants with high and low levels of the Erikson’s fourth stage of psychosocial development. The goodness-of-fit of the model was assessed by using the Hosmer–Lemeshow test (Hosmer & Lemeshow, 1989).

Confirmation of model structure

In the sixth stage, a second CFA was conducted on the second sample to confirm the factor and structural model.

Results

Item Evaluation

The scoring of all negative-oriented items was reversed before the item evaluation. Analysis of the individual items indicated that a total of 8 items with extreme means and variances close to zero were discarded for subsequent analyses. Items 2, 9, 10, 22, 23, 24, 25, and 31, had means ranging from 4.75 and 4.95, and variances between .07 and .58, which were believed not salient for use with young adult normal population and were deleted for subsequent data analysis. Analysis of all 27 remaining items indicated that the distribution is skewed negatively to the left. Among these 27 items, 7 items including 7, 8, 13, 15, 17, 19, 20, 26, 27, 33, and 36 had absolute values slightly greater than 2.0 for skewness (range from 2.056 to 2.400) and another 5 items including 5, 6, 21, 29, and 34 had values ranging from 2.545 to 3.135. Six items including 5, 6, 20, 21, 29, and 36 had values more than 7.0 for kurtosis (range from 7.049 to 11.765). Hence, the inverse transformation of all 27 items was taken to enhance normality of the items (Tabachnick & Fidell, 2001). As suggested by Tabachnick and Fidell (2001), the inverse transformation when compared with other transformations (e.g., square root and logarithmic) was the most powerful method to improve the normality. Both skewness and kurtosis values of all items were within the recommended ranges after linear transformation (skewness, 0.006 to 1.980; kurtosis, 0.025 to 2.069).

Exploratory Factor Analysis

The transformed scores of all 27 items were included. In the initial EFA operation, the observed indices MSA = .753 and Bartlett’s χ²= 1227.05 (df = 351, p < .001) indicated a strong relationship among variables and that the factor analysis was satisfactory to proceed. We first used the Guttman’s criterion (i.e., eigenvalue greater than 1) to determine the number of factors to be retained. An 8-factor solution was indicated with several trivial factors toward the end: eigenvalues and percentages of variance (in parentheses) for components 1 to 8 in this initial EFA operation were 7.420 (27.480%), 2.631 (9.744%), 2.180 (8.075%), 1.638 (6.068%), 1.340 (4.961%), 1.214 (4.495%), 1.145 (4.241%), and 1.106 (4.095%), respectively. We then used Catell’s scree test as an alternative to determine the number of factors to be retained. This time, a four-factor solution was indicated. The resulting four-factor solution was judged by virtue of parsimony. The residual correlation matrix further indicated that correlations in the residual matrix were small (with absolute values <.05), suggesting that a good factor structure was obtained and that retaining factors did not improve the fit to the data. In the subsequent operations, each item was removed one at a time on the basis of predetermined criteria. Items were removed from the measure if they did not have primary factor pattern coefficients that were ≥.40 (including values that rounded to .40) and secondary factor pattern coefficients ≤.30 or if the item loaded on more than one factor. Based on these criteria, 13 items, including 3, 4, 7, 8, 11, 12, 14, 15, 16, 26, 27, 29, and 34, were deleted one at a time. In the final EFA operation, 14 items were retained and this 14-item measure contained a four-factor solution, which accounted for 67.279% of the total variance: 5 items measuring Factor 1 (Work Tasks in School Environment) accounted for 30.279% of the total variance, 3 items measuring Factor 2 (Role Model) accounted for 14.640%, 4 items measuring Factor 3 (Social Skills) accounted for 13.511%, and 2 items measuring Factor 4 (Work Tasks in Home Environment) accounted for additional 8.848%. Of these four factors, only three had at least 3 items and are thus considered stable as recommended by methodologists (Costello & Osborne, 2005; Fabrigar, Wegener, MacCallum, & Strahan, 1999). To enhance the stability and conceptual meaning, the first and fourth were merged into a single factor (Work Tasks) and yielded a final 14-item measure with a three-factor solution (Table 1). The internal consistency for each factor was estimated using Cronbach’s coefficient and all of them were over the recommended value: for Factor 1 (Work Tasks), α = .713; for Factor 2 (Role Model), α = .734; and for Factor 3 (Social Skills), α = .809. For comparison purpose, all pattern and structure coefficients, item–total correlations, and Cronbach’s coefficients of the final 14-item measure with a three-factor solution obtained by nontransformed scores are also presented in Table 1.

Table 1.

Factor Pattern and Structure coefficients from Exploratory Factor Analysis of Revised Developmental Work Personality Scale, Item–Total Correlation, and Cronbach’s Alpha of the Subscales (n = 92)

	Factor 1: Work Tasks		Factor 2: Role Model		Factor 3: Social Skills		Item–Total Correlation
Items	T	Non-T	T	Non-T	T	Non-T	T	Non-T
19. It was important for me to complete all my school work.	.909 (.876)	.930 (.902)					.583	.790
21. I completed all my assignments in school.	.827 (.795)	.780 (.749)					.542	.542
17. In school I completed my work on time.	.773 (.788)	.819 (.832)					.591	.704
20. It made me feel good when I completed all my school work.	.730 (.758)	.716 (.730)					.491	.625
18.* In school I tried my best even if I didn’t like what I was doing.	.608 (.703)	.668 (.774)	(−.485)			(.541)	.535	.714
30*. Growing up, I was responsible for chores at home.	.888 (.879)			.660 (.638)			.204	.399
35*. If I did not do my homework or chores, I got into trouble.	.812 (.824)			.714 (.682)			.208	.464
33*. There was someone in my life whom I admired.			.936 (.895)	.739 (.750)			.644	.528
32*. Growing up, I had someone who inspired me.			.906 (.910)	.820 (.846)			.691	.635
28*. When I needed help with my homework, one of my parents was available to help.			.545 (.597)	.356 (.381)			.373	.184
5. I was in trouble a lot with my teachers.					−.901 (−.913)	.932 (.943)	.767	.837
6. When I was in school I got in trouble a lot.					−.867 (-.882)	.922 (.926)	.725	.804
36*. I got in arguments a lot with classmates when I was in school.					−.841 (−.811)	.856 (.833)	.670	.706
13. When I was in school, I had problems getting along with classmates.					−.504 (−.543)	.307 (.330)	.378	.201
Cronbach’s α	.713	.841	.734	.665	.809	.796

Note: Asterisks represent new added/modified items. Structure coefficients are presented in parentheses. Ts represent values obtained by transformed scores. Non-Ts represent values obtained by nontransformed scores. Pattern and structure coefficients on factors <.400 were suppressed and therefore not presented in the table. Items 30 and 35 were initially loaded on Factor 4, but were added to Factor 1 because of theoretical alignment and good fit in CFA. Item 18 was loaded on Factors 1 and 2 based on structure coefficient. Items 5, 6, 36, and 13 are reverse scored. Negative factor pattern coefficients were caused by questions that are negatively oriented to the factor.

Pearson correlations were then computed to examine the intercorrelations among all three factors. By Bonferroni-type adjustment, the alpha level was set at less than or equal to .017 for the three pairs of correlations to control for the possibility of a Type I error. All three factors are moderately correlated with each other (r = .262 to .352, ps ≤ .017), providing evidence for the semiautonomous nature of the DWP (Table 2).

Table 2.

Intercorrelations of Revised Developmental Work Personality Scale, Means, and Standard Deviations of Subscales (n = 92)

Scales	Work Tasks	Role Model	Social Skills	Mean	SD
Work Task	—	.314**	.352***	4.38	.56
Role Model		—	.262*	4.33	.85
Social Skills			—	4.63	.69

Note: Raw scores were used to estimate means and standard deviations.

p ≤ .017. **p ≤ .01. ***p ≤ .001.

Confirmatory Factor Analysis

We hypothesized that these three factors would be intercorrelated and as such identified a three-factor oblique model from the EFA. For comparative purposes, we analyzed, besides this model, a one-factor global model and a three-factor orthogonal model. Again the transformed item scores were used for estimates. The χ² was significant for all three competing models (Table 3), although this does not necessarily reflect a poor model fit since chi-square analysis is sensitive to sample size (Bentler, 1990; Browne & Cudeck, 1993; Fan et al., 1999; Steiger, 1990). To better accommodate our adequate sample size, we continued with other measures of fit indices: relative chi-square (χ²/df; cutoff value for good fit: <2–5), goodness-of-fit index (GFI; ≥.90), CFI (≥.90 minimally acceptable, ≥.95 good, and at or close to 1 excellent; Bollen, 1990; Browne & Cudeck, 1993), and RMSEA (approximating .06 close fit, 0 exact fit of the model; Browne & Cudeck, 1993; Hu & Bentler, 1999). One additional fit index, AIC, was used to determine which of two or more competing models fit best, in which smaller AIC values reflect the better fitting model (Akaike, 1974). All the models performed well for the fit index, relative chi-square (χ²/df), with values that ranged from 1.757 (three-factor oblique model) to 3.298 (one-factor global model), well below the recommended cutoff value of 5. The hypothesized three-factor oblique model had RMSEA value approximating .06, indicating that this model did not have significant error. Only the hypothesized three-factor oblique model had GFI value above .9, although the other two models were appreciably high, being above .8. However, the CFI values of all models were below the recommended .9 threshold. Chi-square difference test between three-factor orthogonal and oblique models was performed (chi-square change = 0.334, df = 3, p = .954), indicating no statistical difference on the model fit of data between these two models. By comparing the AIC values, the best performing model is the three-factor oblique model, which has the smallest AIC value (192.036).

Table 3.

Fit Indices From Confirmatory Factor Analyses of RDWPS (n = 203)

Model	χ²	df	p	χ²/df	GFI	RMSEA	CFI	AIC
A. One-factor	253.94	77	<.001	3.298	.820	.107	.304	309.944
B. Three-factor orthogonal	160.99	77	<.001	2.091	.886	.073	.670	216.990
C. Three-factor oblique	130.0	74	<.001	1.757	.908	.061	.780	192.036

We also reviewed the factor pattern coefficients of CFA across three possible models to justify the selection of best fit model (Table 4). For the single-factor model, all factor coefficient estimates for a single scale in the prediction of RDWPS items were significantly different from zero (p ≤ .05), except three items, 30, 35, and 28. For the three-factor orthogonal model, all factor coefficient estimates for subscales in the prediction of RDWPS items were significantly different from zero (p ≤ .05), except two items, 30 and 35. For the three-factor oblique model, all factor coefficient estimates for subscales in the prediction of RDWPS items were significantly different from zero (p ≤ .05), except one item, 35. The review of coefficient estimates of these three models suggested that the three-factor oblique model is likely to represent the best fitting model. In combination, considering all measures of fit, theoretical constructs in previous studies and literature, and factor pattern coefficient estimates, the hypothesized three-factor oblique model is likely to represent the best fitting model.

Table 4.

Standardized Factor Pattern Coefficients From Confirmatory Factor Analyses of Revised Developmental Work Personality Scale (n = 203)

		Three Factor Orthogonal			Three Factor Oblique
Items	Single Factor	Work Tasks	Role Model	Social Skills	Work Tasks	Role Model	Social Skills
19. It was important for me to complete all my school work.	.637^	.760^			.812^ (.812)	(.241)	(.417)
21. I completed all my assignments in school.	.564*	.690*			.757* (.757)	(.225)	(.389)
17. In school I completed my work on time.	.589*	.577*			.737* (.737)	(.219)	(.378)
20. It made me feel good when I completed all my school work.	.448*	.476*			.604* (.604)	(.179)	(.310)
18. In school I tried my best even if I didn’t like what I was doing.	.431*	.393*			.553* (.553)	(.164)	(.284)
30. Growing up, I was responsible for chores at home.	.151	.125			.235* (.235)	(.070)	(.120)
35. If I did not do my homework or chores, I got into trouble.	.091	.021			.141 (.141)	(.042)	(.072)
33. There was someone in my life whom I admired.	.503*		1.063^		(.288)	.970^ (.970)	(.138)
32. Growing up, I had someone who inspired me.	.427*		.715*		(.241)	.812* (.812)	(.116)
28. When I needed help with my homework, one of my parents was available to help.	.083		.280*		(.095)	.322* (.322)	(.046)
5. I was in trouble a lot with my teachers.	.940*			.882^	(.487)	(.135)	.948^ (.948)
6. When I was in school I got in trouble a lot.	.944*			.968*	(.483)	(.134)	.941* (.941)
36. I got in arguments a lot with classmates when I was in school.	.593*			.497*	(.318)	(.088)	.619* (.619)
13. When I was in school, I had problems getting along with classmates.	.492*			.297*	(.277)	(.077)	.540* (.540)

Note: Standardized factor pattern coefficients are presented for all three models. All estimations were calculated based on factor pattern coefficients.

Path from a latent construct to a selected indicator for that construct fixed to 1.0. Structure coefficients of three-factor oblique model are also presented in parentheses.

p ≤ .05.

As a logical next step, we performed a post hoc model modification to develop a better-fitting model based on the characteristics of the data. The determination of whether to add a path to a model is based on a combination of theoretical, logical, and empirical indications (Quintana & Maxwell, 1999). The examination of modification indices, as suggested from AMOS, guided path additions to improve the goodness-of-fit of the model (Kline, 2005). The content analysis based on the theoretical foundation was then conducted to justify the addition of each path. First, a path of covariance was added between error terms for Items 13 and 36, resulting in an improved fitting model (Table 5). The examination of these two items revealed that both items contained content regarding communication problems with classmates. Next, another path of covariance was added between error terms for Items 30 and 35 as both items’ content related to the engagement of household chores. This again resulted in an improved model. Finally, a path of covariance was added between Items 19 and 20 since both items contained content related to school work accomplishment. This final model (Model 4, Figure 1) indicated an excellent fit to our data, where χ² = 83.134 (df = 71, p > .154), χ²/df = 1.171, GFI = .941, RMSEA = .029, CFI = .952, and AIC = 151.134.

Table 5.

Fit Indices From Each Model Based on Modification Indices and Tested for the Confirmatory Factor Analyses of RDWPS (n = 203)

Model	χ²	df	p	χ²/df	GFI	RMSEA	CFI	AIC
1. Initial model (three-factor oblique)	130.036	74	<.001	1.757	.908	.061	.780	192.036
2. Add path between error terms for Items 13 and 36	105.073	73	.008	1.439	.926	.047	.874	169.073
3. Add path between error terms for Items 30 and 35	94.609	72	.038	1.314	.933	.039	.911	160.609
4. Add path between error terms for Items 19 and 20	83.134	71	.154	1.171	.941	.029	.952	151.134
5. Model 4 using new sample	88.504	71	.078	1.247	.950	.031	.946	156.504

Note: RDWPS = Revised Developmental Work Personality Scale; GFI = goodness-of-fit index; RMSEA = root mean square area of approximation; CFI = comparative fit index; AIC = Akaike information criteria. Estimation of model fit to data (Model 5) was based on the nontransformed items of new sample (n = 255).

Figure 1.

Final confirmatory factor analysis model of the Revised Developmental Work Personality Scale (RDWPS; Model 4; n = 203)

Since the final model (Model 4) emerged from an examination of modification indices, this model may be less generalizable to other populations. Additionally, the model was estimated using the transformed data, which further limited the practical utilization of the instrument. To address these two issues, we used another sample of 255 college students who enrolled in the same class in the following academic year and conducted the second CFA on the 14-item measure in which the nontransformed item scores were used for estimates. Demographic characteristics of the new sample were comparable to the primary sample. See participants section for demographic data regarding this comparable sample. Results from the second CFA indicated that the new model (Model 5; Tables 5 and 6) performed well for the fit index, where χ² = 88.504 (df = 71, p = .078), χ²/df = 1.247, GFI = .950, RMSEA = .031, CFI = .946, and AIC = 156.504. The relative chi-square (χ²/df) was below the recommended cutoff value (<2). The RMSEA value was less than .06, indicating that this model did not have significant error. GFI and CFI values were rounded to .95, indicating that this model is likely the best fitting model. In addition, all factor coefficient estimates for subscales in the prediction of RDWPS items were significantly different from zero (p ≤ .05), further indicating that this model demonstrated an excellent fit to the nontransformed data in Sample 2.

Table 6.

Standardized Factor Pattern Coefficients of Final CFA Model With Modification Indices Using New Sample (n = 255)

	Three-Factor Oblique
Items	Work Tasks	Role Model	Social Skills
19. It was important for me to complete all my school work.	.832^ (.832)	(.360)	(.431)
21. I completed all my assignments in school.	.679* (.679)	(.294)	(.351)
17. In school I completed my work on time.	.860* (.860)	(.372)	(.445)
20. It made me feel good when I completed all my school work.	.565* (.565)	(.244)	(.292)
18. In school I tried my best even if I didn’t like what I was doing.	.778* (.778)	(.337)	(.403)
30. Growing up, I was responsible for chores at home.	.206* (.206)	(.089)	(.107)
35. If I did not do my homework or chores, I got into trouble.	.200* (.200)	(.086)	(.103)
33. There was someone in my life whom I admired.	(.370)	.855^ (.855)	(.222)
32. Growing up, I had someone who inspired me.	(.430)	.995* (.995)	(.259)
28. When I needed help with my homework, one of my parents was available to help.	(.182)	.421* (.421)	(.109)
5. I was in trouble a lot with my teachers.	(.469)	(.235)	.906^ (.906)
6. When I was in school I got in trouble a lot.	(.487)	(.245)	.942* (.942)
36. I got in arguments a lot with classmates when I was in school.	(.157)	(.079)	.304* (.304)
13. When I was in school, I had problems getting along with classmates.	(.202)	(.101)	.390* (.390)

Note: Standardized factor pattern coefficients are presented. All estimations were calculated based on factor pattern coefficients.

Path from a latent construct to a selected indicator for that construct fixed to 1.0. Structure coefficients are also presented in parentheses. Estimation of model fit to data (Model 5) was based on the nontransformed items of new sample (n = 255).

p ≤ .05.

Relationship of RDWPS Factors to Measures of Role Model Influence

Pearson correlations were computed to examine the relationships between the RDWPS factors and the two IOACDS subscales for the entire sample one. To control for the possibility of a Type I error, we adjusted the overall alpha level for the six pairs of correlations. The alpha level was set at less than or equal to .008. The Role Model factor was significantly correlated with both the Support/Guidance (r = .408, p < .001) and Inspiration/Modeling (r = .437, p < .001) IOACDS subscales. The Work Tasks and Social Skills factors were significantly correlated with the Support/Guidance IOACS subscale only (r = .191, p = .001, and r = .207, p < .001, respectively).

Relationship of RDWPS Factors to Measures of Psychosocial Development

Pearson correlations were computed to examine the relationships between the RDWPS factors and measures of MPD for the entire sample. The alpha level was set at less than or equal to .006 for the nine pairs of correlations to control for the possibility of a Type I error. All nine pairs of correlations were significant (all ps < .001). Negative correlations were found between all RDWPS factors and the inferiority score (for Work Tasks, r = −.304; for Role Model, r = −.294; and for Social Skills, r = −.332). Positive correlations were found between all RDWPS factors and the industry score (for Work Tasks, r = .459; for Role Model, r = .363; and for Social Skills, r = .264). Resolution scores positively correlated with RDWPS factors and indicate that persons with high levels of developmental work personality had successful resolution of the fourth stage of development (for Work Tasks, r = .427; for Role Model, r = .369; and for Social Skills, r = .336).

Logistic regression was then used to test how well the RDWPS factors would predict Erikson’s fourth stage of psychosocial development (i.e., industry versus inferiority) in our sample. All three RDWPS factors served as predictor variables, and the participants’ group membership in either high or low resolution category served as the dependent variable. Group membership was determined according to the high (T60) or low (T40) resolution using the T-score interpretation guidelines provided in the manual (Hawley, 1988). People scoring in the extreme ranges were not included in our analysis. The overall model, containing one constant and three RDWPS factors, was significant (χ² = 44.75, df = 3, p < .001; Table 7). The accuracy of the prediction model was 90.1%, yielding a sensitivity of .731 and a specificity .969. The Hosmer–Lemeshow statistic did not reject the goodness-of-fit (χ² = 8.86, df = 8, p = .354), indicating that our observed distribution of categorical variables fits the theoretical distribution predicted in our logistic regression model.

Table 7.

Logistic Regression Analysis on Erikson’s Fourth Stage of Psychosocial Development (n = 91)

	Variables in the Equation
Variables	B	SE	df	p	Exp(B)
Work Tasks	−.663	.290	1	.022	.515
Role Model	−1.438	.446	1	.001	.237
Social Skills	−.995	.506	1	.049	.370
Constant	9.470	2.421	1	<.001
Classification:
	Predicted
Observed	High (T60) group		Low (T40) group

High (T60) group	63		2
Low (T40) group	7		19

Note: Dependent variable = group membership using resolution T-score.

Discussion

Results of this multiphase study provide evidence for construct validity of the DWP model, convergent and divergent validity of the RDWPS factors to other relevant measures, and internal reliability of the revised factor subscales. We believe the results of all the analyses taken together provide evidence for use of the DWP model with populations of people with and without disability, for the purpose of enhanced understanding of the developmental nature of work personality. Counselors and clients wishing to better understand vocational identity may find application of the developmental work personality three-factor model to be a useful construct for exploring how, when, where, and with whom this aspect of identity forms. An understanding of the DWP construct may be particularly useful in contexts where an individual experiences chronic or consistent employment difficulties, despite a strong person–environment fit.

Initial item response analysis revealed that 8 of the 35 items should be discarded. An EFA revealed a three-factor structure with 14 of the remaining 27 items loading unambiguously to one of three factors (Work Tasks, Social Skills, Role Model). The factor structure confirmed in the current sample of students is consistent with the three-factor structure found in the EFA done on a sample of adults with disabilities (O’Sullivan & Strauser, 2010). A follow-up CFA on a second sample confirmed the factor structure that has been reported in prior research and is consistent with the confirmed factor structure in Sample 1 of the current study. These factors align with the conceptualized model based on Erikson’s developmental theory and Bandura’s observational learning and social cognitive theories. After comparing all possible models, the three-factor oblique model was confirmed, and modification to the model provided an excellent fit to the data. The three factors are moderately correlated with each other, providing evidence for the semi-autonomous nature of the DWP (Neff, 1986). This semiautonomous nature allows for the possibility of distinguishing components of the work personality for possible intervention goals. For vocational intervention purposes, this means that one component, such as social skill expression, may need strengthening, while the work task component can be leveraged to improve work outcomes.

The results of the EFA and two CFAs provide evidence for the stability and generalizability of the revised scale for use in research contexts. Specifically, the confirmation of the three-factor oblique solution with modifications in a second sample of college students provides reasonable evidence that this model is applicable to a broad population and is not the result of sample specific data. The end result yielded a three-factor scale that contains seven of the original items and seven new and/or revised items, with three of the new items contained in the role model factor. The work task factor in the confirmed model contains seven items pertaining to completion of homework and chores. Four of these seven items remained from the original version of the DWPS, two items are new (Items 30 and 35), and one is revised (Item 8). The role model factor contains three items pertaining to positive role model influence; all items are new, with no original items remaining from the DWPS. The social skills factor contains four items describing negative interpersonal interactions at school, with one revised item (Item 36), and others remaining from the original DWPS. All factor scores reflect adequate to good internal reliability (α ranging from .713 to .809).

The new role model factor correlated highly, positively, and significantly with two subscales of another measure of role model influence, indicating convergent validity. Specially, this revised factor significantly correlated with the Support/Guidance and the Inspiration/Modeling subscales of the IOACDS, indicating that the revised role model factor is more accurately capturing the role model–child interaction. Additionally, the other factors of work task and social skills had only modest correlations with the Support/Guidance subscale, and no significant correlations were found among the Inspiration/Modeling subscale and the other DWP factors. The significant but modest correlations between the works tasks and social skills factors and Support/Guidance subscale is not surprising. These relationships are likely due to the existence of support and guidance in a child’s life as one mechanism that fosters strong social and work skills. The nonsignificant correlation among the other DWP factors and Inspiration/Modeling subscale can be interpreted as evidence for discriminant validity, as we would only expect the revised role model factor to significantly correlate with both IOACDS subscales, and in particular, the Inspiration/Modeling subscale since this subscale is reflective of the emotional dynamic of the child-role model relationship.

Finally and importantly, the RDWPS factors significantly predicted individuals’ responses to successful or unsuccessful resolution of the fourth stage of development, providing further support for construct validity of the DWP. Students in our sample who reported high levels of developmental work personality also reported positive resolution of Erikson’s fourth stage of development, Industry versus Inferiority. This provides preliminary evidence that the items contained in the RDWPS are reflective of important developmental experiences that contribute to the formation of the individual’s identity pertaining to industriousness.

The RDWPS that yielded the best fit comprised fewer items than the original version, representing a more efficient measure. The shorter version of the RDWPS could mitigate the burden associated with assessing the work personality and other career constructs in the future. The revised scale builds on the original scale with the addition of improved items related to the Role Model factor that more accurately reflects the nature and quality of role model interactions with children during the fourth stage of development. The revised role model subscale is positively and significantly correlated with the role model influence on academic and career decision as measured by IOACDS. This indicates that students who reported having a positive role model during the fourth stage of development also reported having someone in their present lives with whom they consider a positive influence. This relationship provides preliminary support for the notion that positive role model influence in childhood relates to positive role model interaction in later developmental stages. Given the importance of positive mentorships on career success (Hacket & Betz, 1981; Hacket et al., 1989; Nauta & Kokaly, 2001) this relationship is worth exploring further to better understand how early role model relationships influence adult mentoring relationships.

The developmental work personality appears to be a viable model for understanding childhood developmental experiences believed to contribute to adult work personality expression. The overall results of this study lend further support to the developmental nature of the work personality. The construct of DWP contributes to our understanding of how, where, when, and with whom people develop their self-concept related to industry and competence. The results of this study provide clarity regarding important developmental experiences that may influence overall vocational identity. Much of the vocational and career development research has emphasized late adolescence and early adulthood while underemphasizing childhood. Our study provides evidence for the need to refocus some of these efforts to an earlier time not previously believed critical for vocational behavior.

Assessment of this stage of development is expected to provide a better understanding of the origins of work personality identity and expression in order to design targeted interventions for adults at risk of reduced employment outcomes. Our results indicate that scores on the RDWPS can predict successful or unsuccessful resolution of the fourth stage of development as measured by the MPD, providing evidence of construct validity. Furthermore, the majority of our sample’s scores on the MPD reflect high resolution, indicating successful navigation of the fourth stage of development. This is not surprising given that our sample consisted of college students. Persons who successfully resolved this stage of development would be more likely to work toward higher education (an industrious endeavor) compared with people who negatively resolved the fourth stage of development. Results from our sample provide initial evidence that individuals who positively resolved the fourth stage of development scored high on the RDWPS and are demonstrating a positive vocational behavior (i.e., attending college) expected for industrious people.

Use of the DWP Model and Scale for Counseling Interventions

While the results of this study provide evidence for construct validity, convergent validity, and improved reliability, we do not recommend use of the scale in clinical settings at this time. Research with clinical samples and established norms are needed before use of the instrument is recommended for that purpose. However, we do believe the DWP model, and use of select items, can be useful when designing counseling interventions to strengthen vocational behavior related to a mostly negative resolution of the fourth stage of development. At this time, we are only recommending a conceptual use of the DWP model in clinical settings, with scale items serving as a guide to counselors seeking ways to connect relevant childhood experiences with adult vocational experiences. We believe this construct is very relevant to several counseling disciplines, including rehabilitation counselors, career counselors, and school counselors working with individuals who experienced disability, impoverished home and school environments, or impaired role models during the fourth stage of development, all of which relate to negative adult work outcomes. The original scale was validated using a sample of individuals with disabilities; that iteration of the scale reflected the same three factors, and more than half of the items from the DWPS were retained or revised in the RDWPS. For this reason, we believe the revised scale maintains its relevance to this population.

Disability is not the only barrier to successful resolution of the fourth stage. Use of the DWP construct may be important for understanding environmental barriers including poverty, role model scarcity, and school and neighborhood factors as contributors to vocational outcomes for nondisabled populations. The counseling process allows for the exploration of past events and connecting them to present life experiences for the purposes of identifying critical experiences and influential people, such as teachers and parents, who contribute to our developing identity. Connecting the past with the present is a common factor in counseling that facilitates insight and opportunity to express emotion (Wampold, 2001). Career counselors, rehabilitation counselors, school counselors, and organizational psychologists frequently work with individuals wishing to improve academic and vocational outcomes. The DWP may be useful in these contexts as counselors explore with their clients the people, places, and events that contribute to vocational identity.

Because the scale items pertain to childhood experiences, we believe application of the model and instrument will be less intrusive than assessing a troubled academic or work history. In particular, the stigma of unemployment and underemployment may prevent many people from openly discussing their experiences in the workplace, which can contribute to failed counseling interventions. Exploration of environmental factors during childhood as significant contributors to negative work personality expression in adulthood is likely to ease the experience of self-stigma for many people seeking vocational and career counseling.

Limitations

Interpretation of results must be considered with caution. One purpose of our study was to assess the validity of our construct for use with a nondisabled population. Although our results provide evidence for use with nondisabled college students, we acknowledge that a college student sample is not representative of the general population. Therefore, we cannot determine from the current analyses whether the factor structure would differ in a group of nondisabled noncollege student population. Second, it should be noted that the use of modification indices to guide CFA analysis increases the probability of the results being influenced by chance. Furthermore, self-report data are problematic, and we must consider that the results are reflecting our sample’s interpretation of their experiences during the fourth stage of development rather than accurate recollections of reality. Having said this, from a psychodynamic perspective, it could be argued that an individual’s perspective of the past is more important for understanding that person’s view of the self more so than an accurate account of experiences. Several of the items in the confirmed model reflect low loading coefficients. Although the structural model is consistent, we cannot recommend use of all the items in each factor at this time. In a previous study, the test–retest reliability of the role model factor scale was found to be inadequate in a sample of college students. We were unable to confirm if the test–retest reliability was improved for the revised role model subscale in the current study. Future studies should aim to report the test–retest reliability coefficient of this revised factor scale. Finally, use of transformed data prohibits ease of score interpretation, and thus, we cannot recommend use of the scale in clinical settings.

Future Research

While the results of this study provide initial evidence for the construct validity of the DWP for use with broad samples, we do not recommend use of the instrument for intervention purposes at this time. While this study provided useful information regarding item retention and factor structure, follow-up studies are needed on diverse samples to make decisions about final item retention. Additional psychometric and normative data are needed before this instrument should be used clinically. Similarly, the validity and clinical utility of the measure should be further examined by determining if the three-factor model of the RDWPS is related to other critical career- or vocational-related outcomes such as career maturity, career indecision, vocational aspirations, and choices. Since we do not have any recommendations for score interpretation or threshold scores reflective of high, medium, or low DWP, we can only recommend use of the model, and perhaps selected scale items as discussion points in a counseling context, to better understand a framework for the developing work personality. Future studies should aim to investigate ways to implement the model and scale for use with persons at risk for unemployment or underemployment. We anticipate use of the scale in the future to guide interventions aimed at strengthening the developmental work personality. Finally, longitudinal studies aiming to verify the experiences of children during the fourth stage of development will improve our understanding of disruptions to development, including the experience of disability, role model interactions, and school and home environments.

Footnotes

Authors’ Note

Alex W. K. Wong and Deirdre O’Sullivan are co–lead authors.

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.

Bios

Alex W. K. Wong obtained a PhD from The Hong Kong Polytechnic University, and is a current doctoral student in the Department of Kinesiology and Community Health at the University of Illinois at Urbana-Champaign. His research is related to occupational development for people with disabilities.

Deirdre O’Sullivan is an Assistant Professor in the Department of Educational Psychology, Counseling, and Special Education at the Pennsylvania State University. Her research interests focus on work personality and stigma for persons living with disabilities.

David R. Strauser is an Associate Professor in the Department of Kinesiology and Community Health at the University of Illinois at Urbana-Champaign. His primary research focus is the occupational development of young adult CNS survivors and their families.

References

Akaike

(1974). A new look at the statistical model identification. Automatic Control, IEEE Transactions on Automatic Control, 19, 716–723. doi:10.1109/TAC.1974.110070510.1109/TAC.1974.1100705

Bandura

(1969). Social Learning Theory of identification processes. In Goslin

D. A.

(Ed.), Handbook of Socialization Theory and Research. Chicago, IL: Rand McNally & Co.

Bandura

(1977a). Self-efficacy: Toward a unifying theory of behavioral change. Psychological Review, 84, 191–215.

Bandura

(1977b). Social learning theory. Englewood Cliffs, NJ: Prentice-Hall.

Bandura

(1986). Social foundations of thought and action: A social cognitive theory. Englewood Cliffs, NJ: Prentice-Hall.

Bandura

(1989). Social cognitive theory. Annals of Child Development, 6, 1–60.

Bartlett

M. S.

(1954). A note on the multiplying factors for various chi square approximations. Journal of the Royal Statistical Society, Series B, 16, 296–298.

Bentler

P. M.

(1990). Comparative fit indexes in structural models. Psychological Bulletin, 107, 238–246.

Blustein

D. L.

(2006). The psychology of working: A new perspective for career development, counseling, and public policy. Mahwah, NJ: Erlbaum.

10.

Bollen

K. A.

(1990). Overall fit in covariance structure models: Two types of sample size effects. Psychological Bulletin, 107, 256–259. doi:10.1037/0033-2909.107.2.25610.1037/0033-2909.107.2.256

11.

Bolton

(1992). Manual for the Work Personality Profile: Self Report version. Fayetteville: Arkansas Training and Research Center for Rehabilitation Studies.

12.

Browne

M. W.

Cudeck

(1993). Alternative ways of assessing model fit. In Bollen

K. A.

Long

J. S.

(Eds.), Testing structural equation models (pp. 136–162). Newbury Park, CA: Sage.

13.

Byrne

B. M.

(2001). Structural equation modeling with AMOS: Basic concepts, applications, and programming. Mahwah, NJ: Lawrence Erlbaum.

14.

Cattell

R. B.

(1966). The scree test for number of factors. Multivariate Behavioral Research, 1, 140–161.

15.

Chan

Lee

G. K.

Lee

E. J.

Kubota

Allen

C. A.

(2007). Structural equation modeling in rehabilitation counseling research. Rehabilitation Counseling Bulletin, 51, 44–57.

16.

Costa

P. T.

Jr. McCrae

R. R.

(1992). The Revised NEO Personality Inventory (NEO PI-R) and NEO Five-Factor Inventory (NEO-FFI) professional manual. Odessa, FL: Psychological Assessment Resources.

17.

Costello

A. B.

Osborne

J. W.

(2005). Best practices in exploratory factor analysis: Four recommendations for getting the most from your analysis. Practical Assessment, Research & Evaluation, 10. Retrieved from http://pareonline.net/pdf/v10n7.pdf

18.

Curran

P. J.

West

S. G.

Finch

J. F.

(1996). The robustness of test statistics to nonnormality and specification error in confirmatory factor analysis. Psychological Methods, 1(1), 16–29.

19.

Erikson

E. H.

(1959). Identity and the life cycle: Selected papers. Oxford, England: International Universities Press.

20.

Erikson

E. H.

(1963). Childhood and society. New York, NY: W. W. Norton.

21.

Erikson

E. H.

(1968). Identity: Youth and crisis. New York, NY: W. W. Norton.

22.

Fabrigar

L. R.

Wegener

D. T.

MacCallum

R. C.

Strahan

E. J.

(1999). Evaluating the use of exploratory factor analysis in psychological research. Psychological Methods, 4, 272–299.

23.

Fan

Thompson

Wang

(1999). Effects of sample size, estimation methods, and model specification on structural equation modeling fit indexes. Structural Equation Modeling: A Multidisciplinary Journal, 6, 56–83. doi:10.1080/1070551990954011910.1080/10705519909540119

24.

Friedman

T. L.

(2005). The world is flat: A brief history of the 21st century. Waterville, ME: Thorndike Press.

25.

Gorsuch

R. L.

(1983). Factor analysis. Hillsdale, NJ: Lawrence Erlbaum.

26.

Green

Salkind

(2005). Using SPSS for Windows and Macintosh: Understanding and analysing data. Upper Saddle River, NJ Prentice Hall.

27.

Guttman

(1954). Some necessary conditions for common factor analysis. Psychometrika, 19, 149–161.

28.

Hackett

Betz

N. E.

(1981). A self-efficacy approach to the career development of women. Journal of Vocational Behavior, 18, 326–339. doi:10.1016/0001-8791(81)90019-110.1016/0001-8791(81)90019-1

29.

Hackett

Esposito

O’Halloran

M. S.

(1989). The relationship of role model influences to the career salience and educational and career plans of college women. Journal of Vocational Behavior, 35, 164–180. doi:10.1016/0001-8791(89)90038-910.1016/0001-8791(89)90038-9

30.

Hawley

G. A.

(1988). Measures of psychosocial development. Odessa, FL: Psychological Assessment Resources.

31.

Herr

E. L.

Cramer

S. H.

(1996). Career guidance and counseling through the lifespan: Systematic approaches. New York, NY: HarperCollins College.

32.

Hill

N. E.

Tyson

D. F.

(2009). Parental involvement in middle school: A meta-analytic assessment of the strategies that promote achievement. Developmental Psychology, 45, 740–763. doi:10.1037/a001536210.1037/a0015362

33.

Hosmer

D. W.

Lemeshow

(1989). Applied logistic regression. New York, NY: John Wiley.

34.

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, 1–55. doi:10.1080/1070551990954011810.1080/10705519909540118

35.

Kaiser

H. F.

(1974). An index of factorial simplicity. Psychometrika, 39, 31–36.

36.

Kline

R. B.

(2005). Principles and practice of structural equation modeling (2nd ed.). New York, NY: Guildford Press.

37.

National Science Foundation. (2010). NSF STEM classification of instructional programs crosswalk. Retrieved from http://www.anamp.org/AMP%20Operations%20Manual%20/CIP_STEM_Crosswalk.pdf

38.

Nauta

M. M.

Kokaly

M. L.

(2001). Assessing role model influences on students’ academic and vocational decisions. Journal of Career Assessment, 9, 81–99. doi:10.1177/10690727010090010610.1177/106907270100900106

39.

Nauta

M. M.

Saucier

A. M.

Woodard

L. E.

(2001). Interpersonal influences on students’ academic and career decisions: The impact of sexual orientation. Career Development Quarterly, 49, 352–362.

40.

Neff

W. S.

(1986). Work and human behavior (3rd ed.). New York, NY: Aldine de Gruyter.

41.

Nunnally

J. C.

Bernstein

I. H.

(1994). Psychometric theory. New York, NY: McGraw-Hill.

42.

O’Sullivan

Strauser

D. R.

(2010). Validation of the developmental work personality model and scale. Rehabilitation Counseling Bulletin, 54, 46–56. doi:10.1177/003435521037804510.1177/0034355210378045

43.

O’Sullivan

Strauser

D. R.

Wong

A. W. K.

(in press). Examining differences in developmental work personality across disability category: Implications for individuals with psychiatric disabilities. Work: A Journal of Prevention, Assessment, and Rehabilitation.

44.

Pai

A. L. H.

Mullins

L. L.

Drotar

Burant

Wagner

Chaney

J. M.

(2007). Exploratory and confirmatory factor analysis of the Child Uncertainty in Illness Scale among children with chronic illness. Journal of Pediatric Psychology, 32, 288–296.

45.

Ployhart

R. E.

Schneider

Schmitt

(2006). Staffing organizations: Contemporary practice and theory (3rd ed.). Mahwah, NJ: Lawrence Erlbaum.

46.

Quintana

S. M.

Maxwell

S. E.

(1999). Implications of recent developments in structural equation modeling for counseling psychology. Counseling Psychologist, 27, 485–527.

47.

Rhodes

J. E.

Grossman

J. B.

Resch

N. L.

(2000). Agents of change: Pathways through which mentoring relationships influence adolescents’ academic adjustment. Child Development, 71, 1662–1671. doi:10.1111/1467-8624.0025610.1111/1467-8624.00256

48.

Spelten

E. R.

Sprangers

M. A. G.

Verbeek

J. H. A. M.

(2002). Factors reported to influence the return to work of cancer survivors: A literature review. Psycho-Oncology, 11, 124–131. doi:10.1002/pon.58510.1002/pon.585

49.

Steiger

J. H.

(1990). Structural model evaluation and modification: An interval estimation approach. Multivariate Behavioral Research, 25, 173–180.

50.

Strauser

D. R.

Keim

(2002). Developmental work personality scale: An initial analysis. Rehabilitation Counseling Bulletin, 45, 105–113. doi:10.1177/00343552020450020510.1177/003435520204500205

51.

Strauser

D. R.

Waldrop

D. G.

Ketz

(1999). Reconceptualizing the Work Personality. Rehabilitation Counseling Bulletin, 42, 290–301.

52.

Tabachnick

B. G.

Fidell

L. S.

(2001). Using multivariate statistics (4th ed.). Boston, MA: Allyn & Bacon.

53.

Taskila

Lindbohm

M. L.

(2007). Factors affecting cancer survivors’ employment and work ability. Acta Oncologica, 46, 446–451. doi:10.1080/0284186070135504810.1080/02841860701355048

54.

Vaillant

G. E.

Vaillant

C. O.

(1981). Natural history of male psychological health, X: Work as a predictor of positive mental health. American Journal of Psychiatry, 138, 1433–1440.

55.

Wampold

B. E.

(2001). The great psychotherapy debate: Models, methods, and findings. Mahwah, NJ: Lawrence Erlbaum.

56.

Weston

Gore

P. A.

Chan

Catalano

(2008). An introduction to using structural equation models in rehabilitation psychology. Rehabilitation Psychology, 53, 340–356.