A model of career success: A longitudinal study of emergency physicians

q
Sarah Pachulicz, Neal Schmitt
*
, Goran Kuljanin
Department of Psychology, Michigan State University, 262 Psychology, E. Lansing, Michigan 48824-1116, USA
a r t i c l e i n f o
Article history:
Received 22 January 2008
Available online 12 June 2008
Keywords:
Career success
Emergency physicians
Salary change
Career satisfaction
Leisure and career success
Work centrality
Gender difference in careers
a b s t r a c t
Objective and subjective career success were hypothesized to mediate the relationships
between sociodemographic variables, human capital indices, individual difference vari-
ables, and organizational sponsorship as inputs and a retirement decision and intentions
to leave either the specialty of emergency medicine (EM) or medicine as output variables.
Objective career success operationalized as the number of leadership positions held did not
mediate the relationship, but income change and career satisfaction mediated the relation-
ship between the hours worked and years employed in emergency medicine. Work central-
ity was signicantly related to subjective career success more so for men than women and
perceptions of success or self-efcacy were positively related to subjective career success
for women, but not for men. The expected pattern of women indicating more difculties
with personal time and family time did not emerge; but women did indicate less perceived
support from the organization, fewer EM leadership positions, less perceived control over
their work situation and less organizational support than did men.
2008 Elsevier Inc. All rights reserved.
1. Introduction
Career success has been of interest to both individuals and organizations. For individuals, a sense of career success has
been related to life satisfaction as well as general mental health (Hall, 1976, 2002). For organizations, the prevailing view
is that individual career success is to some extent coincident with organizational success as well (Hall, 2002; Judge, Higgins,
Thoresen, & Barrick, 1999). Consequently, both the determinants and the outcomes associated with career success have been
of interest to researchers. Recently, Ng, Eby, Sorensen, and Feldman (2005) have meta-analyzed the data regarding predictors
of career success. They considered career success from two vantage points. Objective career success or mobility was indexed
by salary level and promotional advancement. Subjective career success was measured by self-reports of career satisfaction.
In this paper, we examine the same issues for individuals who are engaged in a particularly demanding career (i.e., emer-
gency medicine) in which conict can and does arise between family concerns, personal and leisure activities, and demands
related to careers.
1.1. Overview of Ng et al. meta-analysis
Ng et al. (2005) considered both subjective and objective indices of career success. Subjective career success is a persons
subjective judgment about her career attainments such as career satisfaction. On the other hand, objective career success is
measured by extrinsically observable factors, and includes salary progression and promotions. According to Ng et al. (2005),
0001-8791/$ - see front matter 2008 Elsevier Inc. All rights reserved.
doi:10.1016/j.jvb.2008.05.003
q
We acknowledge the nancial support of the American Board of Emergency Medicine and the considerable effort of Dr. Mary Ann Reinhart and her
colleagues in the development of the survey used in this project and for their many helpful comments on our work.
* Corresponding author. Fax: +1 517 432 2476.
E-mail address: schmitt@msu.edu (N. Schmitt).
Journal of Vocational Behavior 73 (2008) 242253
Contents lists available at ScienceDirect
Journal of Vocational Behavior
j our nal homepage: www. el sevi er . com/ l ocat e/ j vb
the components of career success can be inuenced by four categories of predictors: Human capital, organizational sponsor-
ship, sociodemographic status, and stable individual differences.
Human capital is comprised of an individuals educational, personal, and professional experiences. Across different
occupations, it may be comprised many different indicators including the number of hours worked, work centrality
(job involvement or the psychological investment in work or centrality of work for self-identity or self-image), job ten-
ure, organization tenure, work experience (total years in the workforce), willingness to transfer, international work expe-
rience, education level, career planning (I have a strategy for achieving my career goals, I have a plan for my career),
political knowledge and skills, and social capital (quantity and quality of accumulated contacts, quantity of people an
employee knows of in other functions or at higher levels, and the extent to which an employee engages in networking
activities).
Organizational sponsorship includes career sponsorship (the extent to which employees received sponsorship from indi-
viduals within the organization, including senior managers and mentors). This included the self-reported career-enhancing
functions of being assigned challenging tasks, obtaining exposure and visibility, receiving protection, sponsorship, and
coaching, supervisor support (extent to which supervisors provide emotional and work related support), training and skill
development opportunities (self-reported perceptions of the extent to which their company provided opportunities for
training and skill acquisition), and organizational resources (measured by organization size, number of employees in
organization).
Sociodemographic predictors include demographic and social background, such as gender, race, marital status, and age.
Stable individual differences factors are personality factors including the Big Five factors (i.e., neuroticism, conscientious-
ness, extroversion, agreeableness, and openness) as well as factors such as proactivity, locus of control, and cognitive ability.
Human capital and sociodemographic predictors were found to have stronger relationships with objective success, and orga-
nizational sponsorship and stable individual differences had stronger relationships with subjective career success. Gender
and time were found to be moderators of some relationships.
1.2. Medical studies of career success
In the medical literature, studies of career success have been almost totally focused on career satisfaction and often trea-
ted satisfaction not as an outcome variable, but rather as a predictor of another variable such as intention to leave medicine
or intention to reduce the hours of ones practice. However, several studies of medical personnel have addressed the corre-
lates of career success. Number of hours worked was consistently found to be related to satisfaction, such that the higher the
number of hours worked, the lower the satisfaction (Frank, McMurray, Linzer, & Elon, 1999; Landon, Reschovsky, & Blumen-
thal, 2003; Landon, Reschovsky, Pham, & Blumenthal, 2006). However, Williams et al. (2001) found that the dissatisfaction
with ones job and career were not related when individuals reported working under 60 h a week, but dissatisfaction was
signicantly greater when physicians reported working over 60 h a week. This raises the question of potential curvilinearity
between the hours worked and satisfaction. In addition, the effect of number of hours worked could be perceived differently
according to specialty; for example, Lepnurm, Danielson, Dobson, and Keegan (2006) found that the number of hours worked
per week was a negative predictor of satisfaction for psychiatrists, but a positive predictor of satisfaction for surgeons.
Glisson and Durick (1988) measured variables called task identity and task signicance, which were similar to Ng et al.
(2005) work centrality variable. Statistically signicant (p < .01) correlations between task identity, task signicance, and ca-
reer satisfaction were .44 and .47, respectively (both ps < .001).
Organization tenure was not found to correlate with career satisfaction (Glisson & Durick, 1988; Lepnurm, Dobson, Back-
man, & Keegan, 2006), and neither were work experience and educational level (Glisson & Durick, 1988). These relatively
objective indicators of human capital were not related to subjective career success.
Evidence in the medical literature for the contribution of organizational resources is mixed. Glisson and Durick (1988)
used a variable called workgroup budget that did not correlate with satisfaction. Lepnurm, Dobson, Backman, and Kee-
gan (2006b) determined that access to and quality of the health system in which the physicians worked did contribute
to career satisfaction. While these two variables appear to be organizational resources, their operationalization is quite
different from that used in Ng et al. (2005), where organizational resources was simply dened as size of the
organization.
A few medical researchers have documented the role of sociodemographic variables. Frank et al. (1999) reported that
marital status was not signicantly related to satisfaction; but this study only included women physicians. Age was found
to be positively related to satisfaction (Frank et al., 1999; Glisson & Durick, 1988), and gender was related to satisfaction such
that males generally were more satised than females (Glisson & Durick, 1988; Lepnurm et al., 2006b).
Locus of control was also consistently found to be related to satisfaction (Frank et al., 1999); for example, primary care
physicians who became owners of their practice reported increased satisfaction (Landon et al., 2003); female physicians who
perceived more control over their work situation were more satised (Robinson, 2004); and individuals who perceived they
had a high level of inuence over decisions felt more satised (Lepnurm et al., 2006b).
Some variables that are mentioned frequently in the medical literature were not included in the Ng et al. (2005) meta-
analysis. For example, Lepnurm et al. (2006b) found that self-reported health predicted 13% of the variance in career satisfac-
tion for psychiatrists, and 18.9% of the variance in career satisfaction for surgeons. Similarly, Williams et al. (2001) reported
that physical health was correlated .26 with satisfaction, and mental health correlated .56 with satisfaction (p < .001).
S. Pachulicz et al. / Journal of Vocational Behavior 73 (2008) 242253 243
Rather than using income as part of career success, Landon et al. (2003) treated it as a correlate (implicitly a determinant)
of career satisfaction. For medical specialists, income was not related to satisfaction, but for primary care physicians, satis-
faction was signicantly related to income. In addition, an increase in the level of patient acuity (perhaps an indirect measure
of stress) that physicians were expected to handle without referral was related to decreased satisfaction. In a similar vein,
Frank et al. (1999) reported that income was related to higher satisfaction, and that work stress was related to lower
satisfaction.
Lepnurm, Danielson, Dobson, and Keegan (2006a) used some of the same constructs as included in the Ng et al. (2005)
meta-analysis but applied a different analytical approach. They conducted a survey and computed loadings of their survey
items on four factors of career satisfaction that they labeled personal, professional, performance, and inherent. Control of
work schedule loaded on the personal factor; relationship with administrators, relationship with nurses, authority to get
your decisions carried out, and earnings loaded on the professional factor; access to resources to treat patients and ability
to keep up with advances in ones specialty loaded on a performance factor; and interactions with other physicians and ca-
reer advancement loaded on an inherent factor. All of these variables were considered in past research (Ng et al.), but Lep-
nurm et al.(2006a, 2006b) did not assess the relationship between these variables and overall career satisfaction or salary or
promotion outcomes.
Interestingly, personality variables such as the Big Five received no mention in the medical literature at all though they
have been the focus of many studies in the broader literature on career success. The only individual difference variable that
received some attention was perception of locus of control.
The career success literature, especially as it relates to success in medical professions, leaves several unanswered ques-
tions, some of which we try to address in this paper. First, we consider both objective and subjective measures of career suc-
cess and its correlates. Promotions in a medical career are likely different than in a traditional business organization. In
medical careers, a promotion likely means some change in ones job to include more administrative than medical tasks
which may actually decrease career satisfaction for those heavily invested in the practice of emergency medicine. On the
input side, objective human capital measures likely include specialty certications of some type since all will have medical
degrees. Second, we examine the correlates of career success over a span of 10 years. While almost required conceptually in
career success studies, longitudinal studies of career success (Bray, Campbell, & Grant, 1974) especially among medical pro-
fessions, are rare. Use of cross time measures also reduces the common method explanation of observed relationships when
measures are collected by self-report. Third, we examine the role of involvement in leisure activities in career success stud-
ies. While considered an effective source of relief or buffer of the stress and time required of people in demanding profes-
sions, very few studies have considered the role of leisure activities as correlates of career success. Excessive involvement
in leisure activities may actually be a detriment to objective career success indicators while opportunities to engage in lei-
sure activities may contribute to ones satisfaction in a career. Fourth, like Ng et al. (2005), we examine the degree to which
observed relationships are moderated by gender. In this sample of emergency medical physicians, we believe that gender
will be more important as a moderator of relationships with career success because of the demanding nature of the job
and the fact that responsibilities for family and children often fall disproportionately on women. Fifth, we go beyond the
Ng et al. meta-analysis in considering some outcomes of career success (or the lack thereof) such as retirement, leaving
emergency medicine, or intending to leave the profession. Finally, we consider some individual difference correlates of ca-
reer success that have been largely neglected in the medical literature.
1.3. Hypotheses
Consistent with the meta-analysis reported above, the medical literature on career success, and the situation faced by the
emergency medical physicians that are the subject of our investigation, we propose the following hypotheses.
Hypothesis 1. Four human capital indices (number of hours worked, number of years working in emergency medicine,
number of special certications received, and the level of self-reported work excitement) are related positively to objective
career success.
Measures of these variables were collected at the second of three data collection efforts separated by 5-year intervals
since we wanted to assess the degree to which the presence/absence of human capital would lead to potential career-ending
decisions as well as career success. Using an earlier measure of human capital seemed inappropriate as there would likely be
important changes in human capital in a 5-year period that should be reected in the relationship. In addition, all human
capital variables and the objective career success indicators with the exception of the excitement variable (see description
below) are relatively objective and should not be subject to the usual common method bias criticism of self-report variables.
Hypothesis 2. Sociodemographic variables (age, race, and marital status) are related positively to objective career success
variables.
These variables, all collected in the third survey, are unlikely to change with the possible exception of marital status. Since
marital status was collected at Time 3, any attributions concerning the direction of causality are unwarranted.
Hypothesis 3. Individual difference variables (planning, sociability, self-efcacy or success, health, leisure activity will be
related positively to subjective career success. Personal conict will be negatively related to subjective career satisfaction.
244 S. Pachulicz et al. / Journal of Vocational Behavior 73 (2008) 242253
Measures of planning, sociability, and self-efcacy were collected at all three time points. We used the rst set of avail-
able responses, because these variables were conceived of as stable individual difference variables that should precede and
contribute to the subjective appraisals of ones career. Measures of health, leisure activity, and level of personal conict were
collected simultaneous with the measure of subjective career success at the second data collection since it was felt that these
variables would change with time and that the most recent measure of these variables would be most relevant to percep-
tions of career success.
Hypothesis 4. Organizational sponsorship variables including organizational support items (compensation, benets, job
security, a sense of ownership, opportunity for specializations), further educational opportunities, and perceptions of control
over ones work environment are related positively to subjective career success.
The input variables described in this hypothesis were measured in the Time 2 survey on the thesis that the effects of such
factors would take place over time, but that they would be relatively proximal in their effects on subjective career success.
Hypothesis 5. Objective career success is related positively to career outcomes such as leaving medicine, intentions to leave
emergency medicine, and retirement.
Objective career success (number of academic and emergency medicine leadership positions held, salary change over
the 10-year period) and the outcome measures were all assessed at Time 3. We felt the objective career success vari-
ables should include all events including recent ones that occurred prior to or simultaneously with outcomes. Since
all these items were relatively objective, the likelihood of common response bias should be minimal. The three outcome
variables are not conceptually related and, in fact, empirical correlations between the three ranged from .08 to .21.
Hence each outcome was treated separately in our analyses rather than considering each as indicators of an outcome
construct.
Hypothesis 6. Subjective career success is related to career outcomes such as intentions to leave medicine, intentions to
leave emergency medicine, and retirement.
Subjective career success was measured at Time 2 and outcome variables were taken from the Time 3 survey.
Hypothesis 7. Relationships between variables are moderated by gender.
Both the Ng et al. (2005) meta-analysis and various medical studies of career success suggest that the hypothesized rela-
tionships above are moderated by gender. There is a general lack of specicity as to which of these relationships will be mod-
erated; hence, the examination of this hypothesis is largely exploratory. However, we would surmise that the relationship
between the marital status variable and objective career success may vary as a function of gender. Given the usual societal
roles associated with marital relationships, we would expect marriage to increase the responsibilities of women and de-
crease the career success they experience. For men, a spouse may serve a more supportive role and increase their career suc-
cess. It may also be the case that personal-work conicts will play a greater role in the career success of women than men.
The rst six hypothesized relationships are depicted in Fig. 1; Hypothesis 7 suggests the relationships depicted in this gure
will vary as a function of gender.
2. Method
2.1. Sample and procedures
Details of the sampling procedure were described in Reinhart, Munger, and Rund (1999). Representative emergency phy-
sicians (EPs) in four cohorts (as represented by the year they completed a residency from 1979 to 1993) were selected. These
EPs were then contacted and invited to become members of the study for their lifetimes. Future cohorts were identied and
invited to participate in the 1999 and 2004 survey with the stipulation that all must be graduates of professionally approved
residency programs.
Participants were solicited based on the recommendations of Dillman (1978). A letter of invitation with a response post-
card was followed by a postcard, then a second letter of invitation and response postcard and nally a certied letter. Indi-
viduals who declined to participate were asked to specify the reason for their refusal. New invitations were extended until
the targeted sample size of 1008 was met. A similar four-step procedure was used to get the surveys returned at all three
waves with the result that there were 95% rates of return across the three waves of surveys. To retain the desired number
of participants in each cohort, new recruits to the survey were solicited to replace those who refused to answer the 1999 and
2004 surveys.
The resulting sample consisted of a total of 1269 EPs that responded to one or more of the three surveys. Most had med-
ical (MD) degrees with about 7% having Doctor of Osteopathy (DO) degrees and approximately 1012% with other advanced
degrees such as law degrees, masters, and doctoral degrees. Over 80% were male. Participant ages at the rst survey were
primarily between 35 and 49 with substantially smaller numbers in the older age groups (i.e., 55 to 65+). As would be ex-
pected, the sample aged so that by the 2004 survey there was a much more equal representation of participants across age
with approximately 6% of the sample reporting that they were over 65. The sample was predominantly (89% or greater)
S. Pachulicz et al. / Journal of Vocational Behavior 73 (2008) 242253 245
white with 5% reporting that they were Asian and only 12% reporting that they were Black or Latin American. Over 80%
reported that they were married and nearly 80% reported that they had one or more children living with them. Most of
the respondents reported income between $100,000 and $300,000. As expected, the portion reporting they were retired in-
creased to 7.3% in the 2004 group.
2.2. Measures
With some minor exceptions, the same 38-page survey was administered at all three times. For the purpose of this report,
we used items and created scales that were relevant to the career success questions described above and at times in the data
collection process that allowed for causal attributions consistent with the model in Fig. 1. We recognize that our data are all
correlational, but timing of measurement does allow some weak causal attributions.
2.2.1. Subjective career success
Career satisfaction was measured with four items that were considered indicators of a career satisfaction construct. These
questions inquired as to the degree to which the EM specialty met their expectations, whether it measured up to the type of
career they wanted when they selected it, whether they would select the specialty again, and how satised they were overall
with their career in EM. Responses were made on four- or ve-point Likert-type scales that were appropriate in content to
the item stem. This measure like the indices of objective career success was collected at Time 2.
2.2.2. Objective career success
Consistent with the Ng et al. (2005) denition of career success, we employed three objective career success indicators.
The rst of these was the number of academic leadership roles the person reported holding (e.g., academic dean, academic
department chair, residency director). The second was the number of leadership roles the individual reported in leading
emergency medicine groups (e.g., medical director of emergency department, emergency medical service physician director,
or manager of emergency medicine group). The third was a salary progression variable. Income change was computed as the
difference between respondents self-reported medical income from a variety of sources at Times 1 and 3 and was used to
indicate their salary progression over a 10-year period. All three were objective variables self-reported by the participants.
The income change variable at each time point was measured on a 10-point scale ranging from 1 (less than $50K per year) to
10 (more than $500K per year). Income change was computed using this 10-point scale.
Human
Capital
No. of hours worked (2)
Work excitement (2)
No. of years in EM (2)
Certifications (2)
Age (1)
Race (1)
Marital Status (3)
Socio-
Demographic
Planful (1)
Social (1)
Self efficacy (success) (1)
Health (2)
Leisure (2)
Personal conflict (2)
Individual
Difference
Org Support (2)
Perc control (2)
Further education
Opportunities (2)
Org.
Sponsorship
Obj. career success
No. of academic leads
(2)
No. of EM leads (2)
Salary change (3-1)
Subjective career (2)
success
Outcomes (3)
(Retirement, LV.
EM, LV. Prof.)
H
1
H
2
H
3
H
4
H
6
H
5
Fig. 1. Hypothesized model of career success and outcomes. Numbers in parentheses represent the time at which data were collected.
246 S. Pachulicz et al. / Journal of Vocational Behavior 73 (2008) 242253
2.2.3. Human capital indices
The number of hours worked per week on various work activities was summed as a single measure. Three indicators
of Work centrality or excitement assessed the respondents level of agreement that emergency medicine was challeng-
ing, exciting, and overall work is fun. The number of years respondents reported that they spent working in emer-
gency medicine was a single item measure. Investment in educational efforts (referred to as Certications in Fig. 1) was
the number of graduate degrees, residency specialties, fellowship training, and certications participants reported that
they had received.
2.2.4. Organizational sponsorship
Respondents indicated which of a set of 11 conditions were available to them on their jobs. These included both extrinsic
items (e.g., fair compensation, fringe benets, job security) and items intrinsic to their work (e.g., exciting work, sense of
ownership) as well as items that reected the degree of perceived control respondents had over their work situation (e.g.,
autonomy in work, control over working conditions, dened working hours), and further education opportunities (e.g.,
opportunity to attend conferences, opportunity for subspecialization, and teaching and research opportunities). Three indi-
cators of each of these constructs were formed by summing randomly selected items to form parcels when more than three
items were available.
2.2.5. Sociodemographic variables
Gender, race, marital status, and age were collected. Gender was proposed as a moderator of the hypothesized relation-
ships. Because of the small number of non-Caucasian respondents, race was coded 1 for Caucasians, 0 for non-Caucasians.
Marital status was coded 1 for married, 0 for not married (single, widowed, etc.).
2.2.6. Individual differences
Respondents were asked to describe themselves on a group of 13 bipolar adjectives on a scale ranging from 1 to 6. These
self-descriptions comprised three scales. A success or general self-efcacy scale consisted of six items including active
inactive, incompetentcompetent, successfulunsuccessful, and strongweak. A social scale consisted of four items includ-
ing coldwarm, interested in self-interested in others, and openclosed. A nal set of three items we labeled planful in-
cluded conventionalunconventional, dreamerpractical, and impulsivedeliberate. As was true for other constructs in
the model, three indicators were computed for each construct. A single item requested their self-appraisal of their health
(i.e., how would you describe your current health?). Response options for this scale ranged from one (some serious health
concerns) to four (exceptionally healthy for my age). Two items assessed the degree to which enough time for family or per-
sonal life were serious problems. Response options for these two items ranged from not a problem (1) to serious problem
(5). As they were correlated .90, they were summed to form a measure of personal conict. Finally, a set of 30 leisure activ-
ities was rated on a seven-point scale with anchors ranging from Not very enjoyable to Extremely enjoyable. We formed
three leisure activity indicators by randomly assigning items to the three indicators and summing them. Those who re-
sponded not very enjoyable or do not participate for a particular activity received a score of zero for that activity, and
those who rated the activity enjoyable to extremely enjoyable received a score of one.
2.2.7. Career success outcomes
We also considered three potential outcomes of career success. These included having thought of leaving the specialty of
emergency medicine in the next year on a scale ranging from very unlikely (1) to very likely (5) and whether or not they
hoped to leave medical practice for another career in the next 5 years (yesno). Finally, respondents were asked if they had
retired from medicine (yesno). For the latter two variables, a yes response was coded 1, no was coded 0.
2.3. Data analysis
Because of missing data across the three waves of data collection, we used the SPSS Missing Values Analysis (Version
15.0) to impute missing values for those respondents who failed to respond to all the items. This analysis provided data
for 1269 participants. Using a likelihood function that assumes the missing data structure is conditional only on the variables
in the model and not unspecied variables external to the model (i.e., data are missing at random; Schafer, 1997), the expec-
tationmaximization algorithm (Dempster, Laird, & Rubin, 1977) in SPSS was used to generate missing values for cases on
which we did not have complete data. This data set was used as the basis of the structural equation modeling analyses done
with LISREL. It should be noted that the maximum likelihood estimation procedure we used assumes multivariate normality.
As is true for many actual data sets (Lei & Lomax, 2005), this assumption was not met by our data. Research on the impor-
tance of violations of this assumption has indicated that nonnormality has little effect on the parameter estimates them-
selves (e.g., Fan & Wang, 1998), but with relatively extreme levels of skewness and kurtosis, the standard errors of the
parameters may be underestimated (Finch, West, & MacKinnon, 1997; West, Finch, & Curran, 1995). A similar nding by
Lei and Lomax (2005, p. 16) led them to conclude that the usual interpretation of SEM parameter estimates can be accepted
even under the severe nonnormality conditions examined in their simulations. Violations of nonnormality inate the v
2
test, but seem to have little effect on t indices (Lei & Lomax, 2005; West et al., 1995). All of these problems also seem to
be greater when sample sizes are much smaller (<200) than is the case in our study.
S. Pachulicz et al. / Journal of Vocational Behavior 73 (2008) 242253 247
To test the proposed hypotheses, we used structural equation modeling with full information maximum likelihood esti-
mation (du Toit & du Toit, 2001) in which the model depicted in Fig. 1 was tested. Responses of both males and females
were included in this rst step. Four model t indices (Hu & Bentler, 1998) were used to evaluate the adequacy of these
models: Root mean square error of approximation (RMSEA), the standardized root mean residual (SRMR), the comparative
t index (CFI), and the nonnormed t index (NNFI). Assessment of the output of this initial analysis strongly suggested that
the Work excitement construct was related to subjective career success, rather than objective career success indices. This
seemed conceptually reasonable as well, though inconsistent with Ng et al. (2005), so this modication was made and in-
cluded in the remainder of the analyses reported in this paper. Second, we tried to test a two-group model with male and
female respondents to examine gender differences, but it did not converge. Hence, separate analyses of the male and fe-
male covariance matrices were conducted and t-tests of the difference between the two sets of parameters were con-
ducted. These t-tests represented tests of Hypothesis 7. Signicance of the parameters indicated in Fig. 1 constituted
tests of the rst six hypotheses.
3. Results
The means, standard deviations, and intercorrelations of all study variables are presented in Table 1. Support for some of
the hypotheses is evident in these correlations. Intentions to leave EM and intentions to leave medicine are signicantly cor-
related (r = 0.21, p < .01). Income change and leadership positions are correlated with retirement status as well as intent to
leave EM. Indicators of organizational support, perceptions of control over ones work situation, and further educational
opportunities are correlated relatively highly with the subjective career success measure and to a lesser extent with the
objective career success variables. Human capital variables (number of hours worked, number of years in EM, and number
of certications earned) with the exception of work excitement/centrality are related to objective career success variables as
hypothesized.
3.1. Inputmediator relationships (Hypotheses 14)
Overall, analysis of the model suggested by Fig. 1 t the data very well: v
2
(673) = 3273.06, p < .01, RMSEA = .05,
SRMR = 0.05, CFI = 0.96, NNFI = 0.95. All t indices indicate good t based on commonly accepted criteria (Hu & Bentler,
1998). The parameter estimates associated with this model are presented in Table 2. Input to mediator results are provided
in the left half of the table while mediator to outcome parameters for each of the three outcomes are presented in the right
half of the table.
Hypothesis 1. Four human capital indices (number of hours worked, number of years working in emergency medicine,
number of special certications received, and the level of self-reported work excitement) are related positively to objective
career success.
Number of hours worked was related to all three objective career success areas: academic leadership positions held (stan-
dardized parameter estimate = .18, p < .05), EM leadership positions held (standardized estimate = .24, p < .05), and income
change (standardized estimate = .13, p < .05).
The number of years working in emergency medicine was not related signicantly to academic leadership positions held
(standardized path estimate = .07, p > .05), but was related signicantly to EM leadership positions held (standardized
estimate = .24, p < .05), and negatively to income change such that individuals who had worked more years in EM experi-
enced less increase in their income over the 10-year period than newcomers (standardized estimate = .11, p < .05).
The number of special certications received (an index of educational effort/investment) was not related to any of the
three objective career success areas (standardized estimates: academic leadership positions held, .03, ns; EM leadership posi-
tions held, .01, ns; income change: standardized estimate = .02, ns).
As noted earlier, work excitement, as pictured in Fig. 1 (Ng et al., 2005) was misspecied. It was not related to any of the
objective career indices, but was related signicantly to subjective career success (.66).
Hypothesis 2. Sociodemographic variables (age, race, and marital status) are related signicantly to objective career success
variables.
Age was not related to EM leadership positions held (standardized estimate = .05, ns) nor academic leadership positions
held (standardized estimate = .06). Income change, however, was related to age (standardized estimate = .29, p < .05) such
that younger individuals reported higher salary increases.
Race was not related to academic leadership positions (standardized estimate = .05, ns). Race was related to EM lead-
ership positions and income change (standardized estimates = .06 and .08, respectively, p < .05) such that non-White
EPs were more likely to hold EM leadership positions and experienced greater salary changes over the 10-year period of
the surveys.
Marital status was related only to EM leadership positions (standardized estimate = .01, p < .05), but unrelated to aca-
demic leadership positions (standardized estimate = .04, ns) or income change (standardized estimate = .02, ns).
248 S. Pachulicz et al. / Journal of Vocational Behavior 73 (2008) 242253
Table 1
Correlations and standard deviations of study variables
Variables 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
1. Retired .29
2. Leave EM .04 1.04
3. Leave Med .01 .21 .30
4. CarSat1 .07 .11 .19 .85
5. CarSat2 .07 .10 .13 .66 .67
6. CarSat3 .01 .07 .11 .63 .72 .68
7. CarSat4 .03 .10 .15 .73 .79 .79 1.07
8. EM Lead .04 .04 .04 .12 .13 .19 .19 1.20
9. AC Lead .05 .02 .02 .17 .23 .28 .31 .10 .26
10. Inc. Chg. .04 .14 .07 .09 .02 .02 .04 .04 .06 2.88
11. Yrs EM .06 .19 .07 .00 .14 .18 .17 .14 .31 .36 1.28
12. Certs .02 .01 .01 .14 .10 .12 .13 .05 .09 .08 .17 .52
13. Hrs. Work .07 .07 .03 .32 .33 .35 .41 .21 .28 .11 .09 .19 28.5
14. Race .06 .03 .04 .07 .12 .13 .09 .07 .10 .06 .06 .02 .09 .31
15. Age .01 .00 .06 .00 .07 .04 .07 .08 .06 .02 .08 .07 .11 .02 10.8
16. MarStat .07 .24 .09 .05 .17 .23 .24 .14 .28 .36 .82 .01 .13 .05 .05 .39
17. Workex1 .04 .12 .11 .26 .31 .28 .33 .06 .10 .08 .00 .02 .12 .10 .05 .02 .75
18. Workex2 .04 .07 .14 .47 .48 .50 .59 .15 .13 .05 .03 .19 .31 .08 .07 .10 .54 .91
19. Workex3 .01 .03 .12 .49 .54 .57 .68 .18 .19 .02 .14 .19 .36 .03 .09 .22 .31 .59 .96
20. Suc1 .03 .05 .04 .15 .19 .20 .22 .02 .09 .01 .03 .04 .05 .01 .00 .04 .13 .14 .18
21. Suc2 .04 .05 .08 .13 .24 .21 .21 .01 .07 .01 .07 .07 .05 .05 .03 .08 .16 .18 .15
22. Suc3 .07 .03 .14 .17 .23 .20 .25 .04 .06 .01 .08 .09 .05 .01 .10 .07 .16 .21 .23
23. Soc1 .00 .05 .03 .11 .11 .10 .12 .02 .03 .05 .17 .08 .03 .07 .03 .13 .13 .16 .12
24. Soc2 .01 .04 .01 .10 .05 .05 .06 .01 .05 .02 .06 .07 .06 .09 .02 .02 .04 .05 .02
25. Soc3 .03 .05 .02 .03 .07 .07 .05 .00 .02 .04 .06 .02 .02 .01 .04 .06 .09 .08 .09
26. Plan1 .04 .04 .07 .06 .04 .04 .02 .08 .07 .05 .03 .03 .01 .13 .10 .03 .07 .03 .01
27. Plan2 .02 .03 .02 .04 .05 .02 .05 .02 .01 .02 .07 .03 .01 .03 .03 .02 .02 .02 .02
28. Plan3 .02 .00 .04 .04 .02 .00 .01 .03 .02 .03 .00 .02 .04 .03 .03 .03 .02 .05 .00
29. Leis1 .01 .06 .03 .28 .30 .35 .39 .19 .26 .00 .14 .24 .38 .13 .11 .24 .13 .31 .42
30. Leis2 .02 .07 .02 .25 .31 .33 .38 .14 .25 .04 .24 .20 .37 .12 .07 .34 .14 .30 .42
31. Leis3 .01 .08 .04 .29 .35 .37 .44 .14 .27 .06 .25 .24 .41 .13 .10 .36 .18 .34 .46
32. Health .04 .06 .04 .19 .19 .18 .20 .01 .09 .04 .08 .09 .10 .03 .01 .05 .07 .11 .14
33. Personal .06 .02 .07 .28 .30 .29 .36 .01 .03 .13 .13 .10 .11 .02 .01 .21 .08 .25 .34
34. OrgSup1 .07 .01 .04 .41 .44 .49 .55 .22 .40 .06 .13 .19 .49 .12 .09 .19 .19 .43 .53
35. OrgSup2 .08 .02 .03 .47 .51 .54 .62 .19 .32 .04 .18 .20 .50 .17 .11 .26 .21 .47 .57
36. OrgSup3 .09 .04 .07 .39 .43 .46 .51 .20 .32 .06 .14 .18 .45 .12 .13 .19 .16 .38 .49
37. Control1 .07 .02 .02 .42 .44 .48 .55 .18 .32 .04 .18 .23 .53 .14 .15 .26 .15 .41 .54
38. Control2 .06 .04 .06 .37 .41 .44 .49 .16 .31 .06 .12 .14 .32 .07 .10 .18 .15 .35 .48
39. Control3 .08 .02 .02 .37 .39 .44 .50 .17 .29 .03 .21 .17 .47 .12 .11 .28 .15 .38 .46
40. FurEd1 .04 .03 .02 .17 .22 .25 .27 .15 .11 .04 .08 .14 .22 .02 .01 .08 .09 .22 .28
41. FurEd2 .09 .02 .02 .36 .41 .45 .50 .22 .26 .07 .16 .23 .48 .11 .09 .23 .14 .38 .49
42. FurEd3 .04 .04 .03 .35 .39 .43 .47 .26 .21 .09 .09 .18 .48 .13 .13 .12 .18 .36 .46
20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39
20. Suc1 1.80
21. Suc2 .40 1.81
22. Suc3 .44 .49 1.36
23. Soc1 .33 .26 .40 1.89
24. Soc2 .08 .04 .15 .28 1.04
25. Soc3 .36 .32 .27 .49 .16 1.29
26. Plan1 .07 .02 .00 .06 .02 .14 1.35
27. Plan2 .01 .08 .09 .03 .12 .05 .25 1.32
28. Plan3 .01 .05 .11 .05 .05 .09 .21 .38 1.16
29. Leis1 .16 .06 .12 .15 .05 .09 .04 .06 .07 7.82
30. Leis2 .15 .08 .10 .08 .00 .06 .05 .03 .04 .80 7.93
31. Leis3 .12 .10 .11 .08 .04 .06 .02 .00 .01 .78 .83 8.71
32. Health .32 .21 .26 .14 .02 .07 .02 .02 .07 .13 .08 .08 .86
33. Personal .16 .12 .19 .07 .01 .02 .00 .03 .04 .24 .27 .28 .08 1.19
34. OrgSp1 .12 .11 .11 .05 .02 .04 .01 .02 .02 .54 .52 .54 .11 .23 1.13
35. OrgSp2 .10 .08 .08 .01 .01 .04 .03 .02 .04 .60 .58 .60 .10 .32 .79 .91
36. OrgSp3 .12 .10 .06 .03 .04 .03 .02 .05 .02 .50 .47 .48 .11 .23 .73 .74 .91
37. Control1 .09 .07 .05 .01 .00 .03 .01 .03 .05 .59 .59 .62 .08 .30 .73 .83 .70 .48
38. Control2 .13 .13 .11 .05 .03 .05 .02 .01 .01 .42 .37 .40 .06 .25 .59 .61 .56 .61 .49
39. Control3 .06 .02 .02 .00 .02 .03 .01 .06 .09 .55 .54 .55 .09 .29 .66 .76 .67 .77 .55 .49
40. FurEd1 .04 .05 .05 .03 .03 .06 .05 .06 .05 .26 .20 .22 .01 .07 .43 .30 .34 .29 .28 .27
41. FurEd2 .08 .07 .06 .01 .02 .05 .03 .06 .06 .54 .49 .53 .08 .21 .72 .72 .67 .71 .54 .66
42. FurEd3 .07 .06 .04 .03 .01 .05 .03 .00 .02 .45 .43 .43 .06 .12 .65 .62 .62 .58 .45 .50
(continued on next page)
S. Pachulicz et al. / Journal of Vocational Behavior 73 (2008) 242253 249
Hypothesis 3. Individual difference variables (planning, sociability, self-efcacy or success, health, leisure activity), will be
related positively to subjective career success, and Personal conict will be related negatively to subjective career
satisfaction.
Three of the six individual difference variables showed a signicant positive relationship with subjective career success
consistent with Hypothesis 3. Self-efcacy was positively related to subjective career success, such that individuals higher in
this trait also reported more subjective career success (standardized estimate = .11, p < .05). Personal conict or time prob-
lems was negatively related to subjective career success, such that individuals who reported having more conicts for per-
sonal and family time also reported lower subjective career success (standardized estimates = .07, p < .05). Health was
positively related to career satisfaction (standardized path estimate = .06, p < .05), which indicated that individuals who re-
ported better health were more satised with their careers. Those who were engaged in fewer leisure activities also reported
more career satisfaction (standardized path estimate = .08, p < .05). This relationship is opposite that hypothesized. A post
hoc explanation may be that some individuals perceive that their work interferes with their leisure activity. Self-reports of
planfulness were not related signicantly to career satisfaction and sociability was related signicantly, but negatively to
career satisfaction (standardized estimates of .03 and .09, p < .05). Both of the latter coefcients were inconsistent with
Hypothesis 3.
Hypothesis 4. Organizational sponsorship variables including organizational support items, perceptions of control over
ones work environment, and further educational opportunities are related positively to subjective career success.
This hypothesis was supported for the organizational support variable (standardized estimate = .38, p < .05), indicating
that the more support a person perceived, the higher the reported subjective career success. Educational opportunity was
related negatively to career success (standardized estimate = .02, NS) and perceived control was related negatively to ca-
reer success (standardized path estimate = .15, NS). Both of the latter negative and nonsignicant relationships were likely
a function of suppressor effects as the relationships between the indicators of these constructs and career satisfaction indi-
cators were all positive (see Table 1).
Table 1 (continued)
40 41 42
40. FurEd1 .38
41. FurEd2 .75 .70
42. FurEd3 .46 .68 .83
Note: Correlations above .06 are statistically signicant, p < .05. Values on the diagonal are standard deviations. CarSat, career satisfaction; Inc. Chg., income
change; Yrs EM, years employed as an emergency physician; Certs, certications; MarStat, marital status; Workex, work excitement; Suc, self-efcacy or
success; Soc, social; Plan, planful; Leis, leisure; Personal, personal life-work conicts; OrgSp, organizational support; and FurEd, further educational
opportunities.
Table 2
Parameter estimates for hypothesized structural model
Input Input to mediator relationships Mediator to outcome relationships
Subj carsat AcadL EMLead SalaryD Mediator Retired Leave Med Leave EM
Human capital Acadl .05 .01 .01
Hours Worked .18
a
.24
a
.13
a
EML .01 .03 .00
Yrs EM .07 .24
a
.11
a
SalaryD .44
a
.08
a
.16
a
Ed Effort (Certs) .03 .01 .00 CarSat .11
a
.17
a
.10
a
Socio-Demogr.
Age .06 .05 .29
a
Race .05 .06
a
.08
a
MaritStat .04 .01 .02
Ind diffs
Planful .03
Social .09
a
Self-eff .11
a
Health .06
a
Leisure .08
a
Pers con .07
a
Work excitement .66
a
Org spon
OrgSup .38
a
Perccont .15
Fur ed opp .02
Note: In the interest of space, we do not present the parameters associated with the measurement model, but these are available upon request from the
second author.
a
p < .05.
250 S. Pachulicz et al. / Journal of Vocational Behavior 73 (2008) 242253
3.2. Mediatoroutcome relationships
The career success variables were each hypothesized correlates of the three outcome variables and tests of those relation-
ships constituted a test of Hypothesis 5.
Hypothesis 5. Objective career success is related negatively to career outcomes such as intentions to leave medicine,
intentions to leave emergency medicine, and retirement.
The parameter estimate for income change was statistically signicant, such that the more respondents income in-
creased, the less likely the person was to have retired or report that they intended to leave medicine or the EM specialty
(standardized estimates = .44, .08, and .16, respectively, p < .05, respectively). None of the relationships between lead-
ership positions held and the three outcome variables were signicant statistically.
Hypothesis 6. Subjective career success is related signicantly to career outcomes such as intentions to leave medicine,
intentions to leave emergency medicine, and retirement.
Signicant (p < .05) standardized parameters were .11, .17, and .10 for retirement status, intent to leave medicine,
and intent to leave emergency medicine, respectively. In all cases, EPs were more likely to indicate that they would leave
their practice when their career satisfaction was low.
Hypothesis 7. Relationships between variables are moderated by gender.
A multigroup analysis of the model was attempted but did not converge. Separate models of covariance matrices repre-
sented by male and female responses were then evaluated. The model t the male responses well (v
2
(673) = 2892.01,
p < .01, RMSEA = .05, SRMR = .06, CFI = .96, NNFI = .95), but t of the female covariance matrix was not as good (v
2
(673) = 1970.23, p < .01, RMSEA = .08, SRMR = .08, CFI = .90, NNFI = .87). Because the nonconvergence of the multigroup mod-
el precluded the usual v
2
difference tests across gender groups, we computed t-tests of the difference between correspond-
ing male and female parameters. Parameter estimates for the two gender groups are presented in Table 3. Six of the 40
structural parameter estimates were statistically signicantly (p < .05) different. The work excitement variable was more
strongly related to mens career satisfaction than it was for women, whereas social and self-efcacy variables were more
strongly related to female (as opposed to male) career satisfaction levels. Mens ages were more strongly and negatively re-
lated to income changes than were womens ages. Race to income change relationships differed signicantly in direction for
men and women. For women, it was the case that income changes were greater for Caucasians than Blacks. For men, the
reverse was true in that income changes were greatest for minority respondents as opposed to Caucasian males. For women,
additional educational and professional certicates actually resulted in smaller salary changes, whereas the relationship for
men was nonsignicant.
Table 3
Parameter estimates for hypothesized model for men and women
Input Input to mediator relationships Mediator to outcome relationships
Subj carsat AcadL EMLead SalaryD Mediator Retired Leave Med Leave EM
M F M F M F M F M F M F M F
Human Capital Acadl .05 .00 .00 .01 .01 .04
Hours Worked .19
a
.21
a
.23
a
.13
a
.14
a
.06 EML .04 .12 .04 .01 .00 .01
Yrs EM .09 .19 .29
a
.18 .17
a
.11 SalaryD .45
a
.39
a
.10
a
.14
a
.19
a
.04
Ed Effort (Certs) .02 .09 .00 .14
a
.05 .19
a
CarSat .10
a
.04 .18
a
.09 .10
a
.09
Socio-Demogr
Age .06 .11 .00 .22 .29
a
.07
Race .06
a
.04 .04 .14
a
.09
a
.14
a
MaritStat .04 .01 .03 .10 .01 .10
Ind diffs
Planful .01 .02
Social .02 .70
a
Self-eff .02 .97
a
Health .06
a
.05
Leisure .07 .31
Pers con .05
a
.01
Work excitement .75
a
.46
a
Org spon
OrgSup .49
a
.93
Perccont .31 1.74
Fur ed opp .03 .41
Note: M and F refer to male and female. Bolded estimates are signicantly different, p < .05.
a
p < .05.
S. Pachulicz et al. / Journal of Vocational Behavior 73 (2008) 242253 251
4. Discussion
Perhaps the most important contribution of our study was the overall conrmation of the model of career success based
on the meta-analytic work of Ng et al. (2005). Each of the hypothesized links in their model was conrmed, though not for all
variables in each of their sets of career success determinants. Human capital in the form of hours worked and years as an EP
was related to leadership positions and salary changes held though the number of educational or professional certications
was not. The latter nding may be unique to the EM profession. The number and variance of these certications was not high
limiting the possibility of large correlations with other variables. Of the demographic variables, age and race were negatively
related to salary change though the latter relationship was moderated by gender as noted above. With the exception of a
relatively small negative relationship between race and EM leadership positions, there was little evidence of demographic
correlates of the number of leadership positions that respondents held.
Individual difference constructs were signicantly related to subjective career success with the exception of the plan-
ning variable though the social variable was negatively related to career satisfaction. Self-efcacy or perceptions of success
were the most signicant predictor of career satisfaction 5 years later, but tests of gender differences indicated this nding
was particularly strong for women and nonsignicant for men. The perception that job responsibilities led to conicts
with ones personal life was negatively related to career satisfaction as expected. Inconsistent with the Ng et al. (2005)
model, however, was the nding that engagement in leisure activities was negatively related to career satisfaction. This
nding may be consistent with the personal conictcareer satisfaction relationship. Perceptions that ones organization
was supportive in various ways was the only one of three organizational sponsorship variables that were related to career
satisfaction. This set of variables was highly intercorrelated and the ndings here may represent a suppressor effect rather
than differences in the constructs measured in this set of predictors. Finally, work excitement was not related to objective
career satisfaction as hypothesized, but it was strongly related to subjective satisfaction, more so for men than women.
In terms of the mediatoroutcome relationships examined, it is clear that leadership positions held played a minimal role
in reported intentions to leave medicine or EM or retire. On the other hand, changes in ones income were particularly
strongly related to the decision to retire, and less so for intentions to leave EM and medicine. Career satisfaction was also
related to all three outcomes in the expected negative direction. These mediatoroutcome relationships are particularly
impressive as the outcomes were collected 5 years after the mediator variables.
A second major contribution of the study was the fact that the relationships suggested by Ng et al. (2005) were evaluated
and at least partly conrmed using longitudinal data. Ng et al. based their model on meta-analyses of the correlates of career
success, but most of the primary studies in this area are cross-sectional. Our data shows that, in general, these relationships
hold over a relatively long period of time. Future research should explore over what period of time these relationships exert
an impact. Surveys in this study were separated by 5 years, but no theoretical rationale exists to support this period of time
or any other. It is also the case, however, that the data were correlational. Hence the temptation to derive strong causal attri-
butions should be avoided. Second, sampling of respondents from the existing EP population was done carefully, but we do
not have evidence of the generalizability of our results to members of other professional groups is limited.
Because of space considerations, we did not report the means of variables for male and female subgroups. That analysis
did reveal the expected pattern of women indicating more difculties with personal time and family time and that women
did indicate less perceived support from the organization, fewer EM leadership positions, and less perceived control over
their work situation than did men. These are all problems that apparently continue to confront women in modern organi-
zations; awareness of this situation must also be accompanied by appropriate interventions if the contributions of women
are to be fully realized.
The results also provide a basis for some practical suggestions as well. Change in salary levels appears to be an important
signal that an individual either will leave or intends to leave this occupation. This may be a way in which organizations signal
to people that they are no longer as supportive of their role or that these individuals are performing at a level that no longer
merits signicant raises. It also means that if organizations want to keep valued experienced talent, it should take steps to
ensure desired compensation levels are maintained. Second, a sense that the organization is supportive of employees efforts
appears to be a central factor in their career satisfaction, and indirectly, in their long term retention. Perhaps self-evident, on
the individual level, those who report being excited and challenged by their work are likely to report career satisfaction and
remain with the occupation. Finally, for employees themselves, it appears that investments in ones job in the form of hours
and years worked do pay dividends in terms of the objective indicators of career success.
4.1. Limitations
Perhaps the most signicant limitation of the study was the reliance on archival data and the fact that this meant we were
forced to use existing measures to operationalize some of the intended constructs. For example, planful is certainly quite
similar in concept to conscientiousness in the broader literature on personality, but the items used in this study may not
appear on any existing measure of conscientiousness. The social items are similar in concept to extraversion and the self-
efcacy or success items reect a similar construct in the personality literature. Similar statements can be made about
the work excitement items and the organizational support and perceptions of control measures. However, in spite of the fact
that we used ad hoc items and measures, the results were consistent with the general model.
252 S. Pachulicz et al. / Journal of Vocational Behavior 73 (2008) 242253
A second possible limitation relates to the timing of the collection of the data. As mentioned above, there is no theory to
guide the timing of measurements in longitudinal research and the 5-year interval used in this study was guided primarily
by the resources of the organization that sponsored the research. Organizational research, and possibly much psychological
research, would benet greatly from some attention to estimating the time required for various processes to exert an impact
on behavioral or attitudinal outcomes.
A nal limitation may be the relatively small sample size of the female group in our study. While proportionately repre-
sentative of the population of EPs, the sample size of 222 meant that some of the parameters estimated were likely unreli-
able and would be very different in a new sample. The model t the female sample less well than the male sample also. We
examined modication indices and residuals and tried various model modications, but none increased t substantially nor
were they convincing conceptually so only results for the one model are presented. As in any modeling effort, though, there
are certainly other models that t the data equally well.
4.2. Conclusions
Overall, the results provide support for the hypotheses formulated in the introduction to the paper. This study does con-
rm previous work on career success and satisfaction among physicians that indicates that hours worked was positively re-
lated to career satisfaction, but also to a number of other outcomes. However, one previous study (Williams et al., 2001)
suggested that satisfaction may decrease when the number of hours worked went over 60 h a week. Mean number of hours
in our study was small partly because some individuals were no longer working as EMs (for those working, the mean number
of hours working in EM was 30 a week). Consistent with past research, organizational sponsorship variables played an
important role in subjective career success and a role in intentions to leave the profession. Work excitement/centrality
was important to subjective career success rather than objective career success. Results of past studies on income have indi-
cated that income plays a minimal role in career satisfaction, but the results of the present study indicate that income change
plays a signicant role in intentions to leave EM, intentions to leave medicine, and retire.
References
Bray, D. W., Campbell, R. J., & Grant, D. L. (1974). Formative years in business. New York: Wiley.
Dempster, A., Laird, N., & Rubin, D. (1977). Maximum likelihood from incomplete data via the EM algorithm. Journal of the Royal Statistical Society, 39, 138.
Dillman, D. (1978). Mail and telephone surveys. New York: Wiley.
Du Toit, M., & Du Toit, S. (2001). Interactive LISREL: Users Guide. Lincolnwood, IL: Scientic Software, International.
Fan, X., & Wang, L. (1998). Effects of potential confounding factors on t indices and parameter estimates for true and misspecied models. Structural
Equation Modeling, 5, 701735.
Finch, J. F., West, S. G., & MacKinnon, D. P. (1997). Effects of sample size and nonnormality on the estimation of mediated effects in latent variable models.
Structural Equation Modeling, 4, 87105.
Frank, E., McMurray, J. E., Linzer, M., & Elon, L. (1999). Career satisfaction of U.S. women physicians: Results from the Women Physicians Health Study.
Archives of Internal Medicine, 159, 14171426.
Glisson, C., & Durick, M. (1988). Predictors of job satisfaction and organizational commitment in human service organizations. Administrative Science
Quarterly, 33(1), 6181.
Hall, D. T. (1976). Careers in organizations. Pacic Palisades. CA: Goodyear.
Hall, D. T. (2002). Careers in and out of organizations. Thousand Oaks. CA: Sage.
Hu, L., & Bentler, P. M. (1998). Cutoff criteria for t indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural
Equation Modeling, 6, 155.
Judge, T. A., Higgins, C. A., Thoresen, C. J., & Barrick, M. R. (1999). The Big Five personality traits, general mental ability, and career success across the life
span. Personnel Psychology, 52, 621651.
Landon, B. E., Reschovsky, J., & Blumenthal, D. (2003). Changes in career satisfaction among primary care and specialist physicians, 19972001. Journal of the
American Medical Association, 289(4), 442450.
Landon, B. E., Reschovsky, J. D., Pham, H. H., & Blumenthal, D. (2006). Leaving medicine: The consequences of physician dissatisfaction. Medical Care, 44(3),
234242.
Lei, M., & Lomax, R. G. (2005). The effect of varying degrees of nonnormality in structural equation modeling. Structural Equation Modeling, 12, 127.
Lepnurm, R., Danielson, D., Dobson, R., & Keegan, D. (2006a). Cornerstones of career satisfaction in medicine. Canadian Journal of Psychiatry, 51(8), 512522.
Lepnurm, R., Dobson, R., Backman, A., & Keegan, D. (2006b). Factors explaining career satisfaction among psychiatrists and surgeons in Canada. Canadian
Journal of Psychiatry, 51(4), 243257.
Ng, T. W. H., Eby, L. T., Sorensen, K. L., & Feldman, D. C. (2005). Predictors of objective and subjective career success: A meta-analysis. Personnel Psychology,
58, 367408.
Reinhart, M. A., Munger, B. S., & Rund, D. A. (1999). American board of emergency medicine longitudinal study of emergency physicians. Annals of Emergency
Medicine, 33, 2232.
Robinson, G. E. (2004). Career satisfaction in female physicians. Journal of the American Medical Association, 295, 635.
Schafer, J. L. (1997). Analysis of incomplete multivariate data. Monographs on statistics and applied probability 72, Chapman and Hall/CRC.
West, S. G., Finch, J. F., & Curran, P. J. (1995). Structural equation models with nonnormal models. In R. H. Hoyle (Ed.), Structural equation modeling. Thousand
Oaks.
Williams, E. S., Konrad, T. R., Scheckler, W. E., Pathman, M. L., McMurray, M. G., & Schwartz, M. (2001). Understanding physicians intentions to withdraw
from practice: The role of job satisfaction, job stress, mental and physical health. Health Care Management Review, 26(1), 719.
S. Pachulicz et al. / Journal of Vocational Behavior 73 (2008) 242253 253