Predicting Career Stability and Mobility

Abstract

This study tested occupational embeddedness and boundaryless career attitudes as a complementary explanation for career stability and mobility, in occupations previously known for lifetime employment. Current and former Australian teachers and police officers (n = 315) completed an online survey about their careers. Consistent with the hypothesis, logistic regression analyses confirmed that embeddedness-related variables such as financial responsibility and age predicted having made an active decision to stay in the chosen occupation, and boundaryless career attitudes predicted having left that occupation to change careers. Contrary to the hypothesis, years in the career predicted a history of mobility and years of career-specific education did not add predictive utility to the model. Overall, the findings partially supported the complementary explanation for career stability and mobility. By including current and former occupational members, and identifying predictors of career behavior, this study contributed to deeper understanding of the changing nature of previously lifelong careers.

Keywords

career stability career mobility occupational embeddedness boundaryless careers career change career decision making

Career stability and mobility decisions at the individual level have an impact on retention and attrition at the organizational level. Employee turnover from an organization includes employee migration to other jobs within the organization, involuntary attrition (e.g., age retirement and death), and voluntary attrition (e.g., occupational change and early retirement; Ingersoll, 2003; Lynch & Tuckey, 2004). High attrition rates are undesirable for organizations because they result in the loss of experienced mentors (Ewing & Manuel, 2005) and the high-level knowledge associated with experienced practitioners, impeding service delivery (Lynch & Tuckey, 2008). Furthermore, high attrition rates flow on to increased economic costs in recruiting, training, and employing new staff and hinder long-term workplace planning and continuity (Brill & McCartney, 2008; Lynch & Tuckey, 2008). Voluntary attrition is of particular interest to organizations because it constitutes potentially preventable attrition (Lynch & Tuckey, 2004). This article is concerned with voluntary attrition preceding career change, specifically referring to instances where employees leave one occupation and enter another.

Explanations for Voluntary Employee Attrition

A meta-analysis of research about leaving careers revealed that strong predictors of voluntary employee attrition were decreased job satisfaction, decreased organizational commitment, and withdrawal behaviors (Griffeth, Hom, & Gaertner, 2000). Withdrawal behaviors consisted of job seeking, comparison of alternatives, withdrawal cognitions, and quit intentions prior to a decision to leave (Griffeth et al., 2000). Decreased job satisfaction was an important variable associated with both intention to leave (Mitchell, Holtom, Lee, & Erez, 2001) and leaving (Griffeth et al., 2000; but see Carless & Arnup, 2011).

Characteristics of the work environment which predicted turnover with small to moderate effect sizes were perceptions of high stress, low autonomy, poor work group cohesion, poor leadership, and poor promotional opportunities (Griffeth et al., 2000). Also predictive of turnover were lack of job security (Carless & Arnup, 2011) and work exhaustion (Blau, 2007). Career change was associated with younger age and high scores on the personality constructs of openness to experience and extraversion (Carless & Arnup, 2011). Decisions about careers were made in the context of family and friends who would be affected by those decisions (Motulsky, 2010). These studies provided valuable insight into voluntary employee attrition; however, they did not explicitly consider the changing nature of careers, or attitudes about them, as a contributor to voluntary attrition.

Boundaryless Careers and Occupational Embeddedness

Recent changes to workplace agreements have led to changes in notions of careers (Orrange, 2003; Smithson & Lewis, 2000), including employer and employee perceptions of implicit reciprocal obligations or the psychological contract (Rousseau, 1989). Traditional relational contracts implicitly depending upon trust, loyalty, and job security have been increasingly replaced by transactional contracts wherein employees seek, for example, training and development to secure future “employability security” (Orrange, 2003). New psychological contracts, in turn, have led to new conceptualizations of careers, including that of boundaryless careers. Boundaryless careers represent a shift away from traditional careers that held lifetime employment as the norm (Hodkinson & Sparkes, 1997) and rewarded seniority with incremental increases in pay and benefits (Stone, 2002).

The notion of boundaryless careers has been defined in various ways (Inkson, Gunz, Ganesh, & Roper, 2012). Definitions have included boundarylessness measured by job change frequency and perceptions of the desirability of increased mobility (Feldman & Ng, 2007). But Briscoe, Hall, and DeMuth (2006) emphasized that job change frequency is a poor measure of boundaryless attitudes and preferences. Instead, boundaryless careers are characterized by working across or beyond organizational boundaries in terms of both physical and psychological movement (Arthur & Rousseau, 1996; Sullivan & Arthur, 2006). Although physical movement refers to employment mobility, psychological movement refers to the degree to which someone willingly and actively develops and maintains working relationships across organizational boundaries (Briscoe et al., 2006). In empirical tests over three studies, Briscoe et al. demonstrated convergent validity between the Boundaryless Career Attitudes scale and a number of measures. These measures included proactive personality, openness to experience, and mastery goal orientation, together suggesting an association between boundaryless career attitudes and agency (see Briscoe et al., 2006). Similarly, willingness to change occupations was associated with uncertainty tolerance and mobility self-efficacy (Otto, Dette-Hagenmeyer, & Dalbert, 2010).

Despite ongoing academic discussion of boundaryless careers as an existing phenomenon, not everyone changes jobs, workplaces, or career paths. Shniper (2005) argued that the claim that graduates today will have numerous occupational careers is unsupported by objective data. Instead, many people remain in a career for various reasons.

Time spent in a profession is associated with human capital, the accumulation of job-specific skills, and with occupational embeddedness, “the totality of forces that keep people in their occupations” (Ng & Feldman, 2007, p. 338). In cases of career change, the totality of forces, or the fit, links, and sacrifice are framed in terms of the occupation, as opposed to the job or the organization (Adams, Webster, & Buyarski, 2010). Fit refers to the extent to which the occupation complements the rest of one’s life; links refer to the ties to other people and activities within the occupation; and sacrifice refers to the ease with which links could be broken to make an occupational change (Feldman & Ng, 2007). Although there are similarities between occupational embeddedness and various forms of occupational commitment, unlike occupational commitment, occupational embeddedness does not contain a sense of moral obligation or lack of alternatives as factors influencing staying in the occupation (Adams et al., 2010).

Past research has supported the notion of occupational embeddedness as an explanation for deciding to stay in an occupation. A qualitative study by Cooper and Davey (2011) explored narrative accounts of nine female teachers in their 40s and 50s with 15–30 years’ teaching experience, who had considered leaving teaching but had not done so. Participants’ accumulated skills, relationships, and the positive aspects of teaching, including salary and lifestyle, meant that changing careers was a less plausible option. Occupational embeddedness was also shown to decrease turnover intentions in a study of law enforcement personnel (Johnson, Sachau, & Englert, 2010).

Ng and Feldman (2007) proposed that occupational embeddedness be considered in conjunction with boundaryless careers, arguing that career mobility and stability decisions would likely be affected by life- and career stage. Specifically, in mid-life and mid-career, financial and family responsibilities may make career mobility more challenging than in early or late career (Ng & Feldman, 2007). Supporting this notion, past research demonstrated that both lower age and fewer years in an occupation were associated with career change (Carless & Arnup, 2011). However, marital status and the number of children were found to be unrelated to career change (Carless & Arnup, 2011).

Research on other potentially embedding variables has been more equivocal. For example, some investigations suggested that employees with more education were more likely to leave (Carless & Arnup, 2011); others found no relationship between education and turnover (Jones, Jones, & Prenzler, 2005). According to Feldman and Ng (2007), occupation-specific training would constitute a human capital investment, which may contribute toward occupational embeddedness, whereas investment in generalizable skill development may increase career mobility. Overall, past research indicates that age or stage of life, financial considerations, and accumulated career-specific education and skills may act as embedding factors that influence employees’ decisions to stay in their careers.

Present Study

The aim of this study was to contribute to an understanding of employee attrition and retention in traditionally long-term occupations, in light of the changing nature of careers. We aimed to test Ng and Feldman’s (2007) proposition that the concepts of occupational embeddedness and boundaryless career attitudes could be considered in tandem, by using them to enhance a predictive model of career stability and mobility. We decided not to restrict the study to one occupation and considered both teaching and policing careers. Past research identified numerous occupation-specific stressors that contribute to decisions about resignations from policing (Adams & Buck, 2010; Brough & Frame, 2004; Haarr, 2005; Lynch, 2007) and teaching careers (Anthony & Ord, 2008; Borman & Dowling, 2008; Brill & McCartney, 2008; Buchanan, 2009). In contrast, we aimed to consider a complementary explanation of both voluntary attrition and retention.

Although policing and teaching differ in a number of respects, both traditionally offered lifetime employment and increasingly face difficulties in retaining staff. The issue of attrition of police officers and teachers has been documented in Australia, New Zealand, the United States, the United Kingdom, and Europe (Borman & Dowling, 2008; Brill & McCartney, 2008; Brough & Frame, 2004; Ewing & Manuel, 2005; Koper, Maguire, Moore, & Huffer, 2001; Lynch & Tuckey, 2008; White, Cooper, Saunders, & Raganella, 2010). In addition, changes in the nature of careers can be seen in increased numbers of mature-aged entrants into policing (Jennett, Islam, Bull, & Woolston, 2008) and teaching (Richardson & Watt, 2006; Watt & Richardson, 2008).

It has been noted that past research has predominantly explored the career attitudes of employees, rather than their career behavior (Briscoe et al., 2006; Ng, Eby, Sorenson, & Feldman, 2005). To address this in part, we aimed to generate a predictive model of career stability and mobility based on the attitudes and behavior of current and former occupational members. We decided to focus on participants who had made active decisions about staying in—or leaving—their policing or teaching careers. This meant withholding the group of participants currently engaged in a career decision-making phase from analyses. Once the predictive model had been developed, we would be able to use it to predict the hypothetical outcome of career stability or mobility for participants who were currently undecided about remaining in—or resigning from—their careers.

Hypotheses

Based on the reviewed research, differences were not expected in mobility and stability for police officers compared with teachers. We developed two hypotheses. First, that demographic variables associated with occupational embeddedness such as higher age, more years in career, presence of substantial financial responsibility, and higher levels of career-specific education would predict career stability. Second, boundaryless career attitudes were chosen as a measure to capture participants’ attitudes about physical and psychological movement beyond organizational boundaries. We hypothesized that higher boundaryless career attitudes would predict career mobility and enhance the predictive capacity of the model beyond that accounted for by embedding variables.

Method

Participant Recruitment and Characteristics

Participants were currently employed within Australia as teachers or police officers or had been previously employed within Australia as teachers or police officers and had made (or were making) transitions into other careers. To recruit participants, we sent letters introducing the research via e-mail to the journal editors of Police and Education Unions or Associations of all Australian states and territories; to the presidents of a number of police social clubs; and we posted invitations on online university student discussion forums. A number of Police and Education unions and associations in Australian states and territories placed information about the study and a link to the online survey in their newsletters or journals. With support of police jurisdictions, police officers were recruited via e-mail invitation. At the conclusion of the survey, participants were invited to forward the survey link to other potential participants (snowball sampling).

Three hundred and eighty-nine people commenced the survey. Twenty-three people did not meet eligibility criteria (e.g., they had left careers to retire, rather than to change careers), and a further 51 respondents did not complete survey protocols, leaving a total of 315 eligible and complete response sets. The sample of 315 participants comprised 133 teachers (93 current, 40 former) and 182 police officers (153 current, 29 former). Participants had spent between 1 and 40 years in their policing and teaching careers (M = 14.91 years; SD = 10.30). There were 146 (46.3%) female respondents and 169 (53.7%) male respondents. Of total teachers, 68.4% were female and 31.6% were male (reflecting gender proportions of teachers nationally; Australian Bureau of Statistics, 2010); 24.1% were primary trained, 66.9% were secondary trained, and 9.0% were other (e.g., both primary and secondary). Of total police officers, 69.8% were male and 30.2% were female (nationally women account for just over 24% of police officers; Prenzler, Fleming, & King, 2010); 35.2% were general duties officers and 64.8% were in specialized areas of policing.

Participants ranged in age from 22 to 70 years (M = 42.21 years; SD = 10.11), 89.2% were born in Australia and of these 3.5% identified as people of Aboriginal or Torres Strait Islander origin. English was spoken at home by 98.7% of respondents (3.1% spoke both English and other language/languages; 1.3% spoke other language/languages only). Those cohabiting with a spouse or partner accounted for 75.9% of the sample, and 50.5% had dependent children.

Materials

The online Career Decision Making Survey comprised six parts: (1) eligibility; (2) career history and career decision making; (3) either (a) for current teachers and police officers or (b) for former police officers and teachers; (4) feelings about your current career; (5) career attitudes; and (6) demographic data. The survey was designed to be responsive to participants’ individual experience: Not all questions were seen by all participants. A pilot version of the online survey was tested on a small sample (n = 6) of participants. Based on respondents’ feedback, minor changes were made to the wording of three questions, and additional multiple-choice answer options were added to two questions; one technical flaw was corrected.

Respondents specified career (teaching = 0, policing = 1) and membership (current = 0 or former = 1). Career decision status was assessed by asking, “Have you ever considered a career change out of your (policing or teaching) career?” Response categories were collapsed such that 1 = career-stayer (current teacher or police officer, never considered a career change, or considered one but decided to stay); 2 = career-decider (current teacher or police officer, currently considering or working toward a career change but still working as police officer or teacher); and 3 = career-changer (former teacher or police officer, left policing or teaching and working toward entering or working in another field). Participants were grouped according to their career decision status.

Participants reflected on their current career field, current workplace, and current job. As a measure of job satisfaction, participants completed the Workplace Wellbeing Index (WWBI; Page, 2005), selected as it was normed on an Australian sample. The WWBI was derived from the Personal Wellbeing Index (PWI; International Wellbeing Group, 2006) and is scored using the manual of the PWI. The 15-item scale (α = .93) consists of questions about both extrinsic and intrinsic satisfiers (Page, 2005). Sample items include “How satisfied are you with the following in your career: the amount of responsibility you have? the use of your abilities and knowledge? and your promotional opportunities?” For each item, participants select a number from 0 (completely dissatisfied) to 10 (completely satisfied). Scores range from 0 to 150 but may be converted to percentages to facilitate comparisons. Higher percentages reflect greater job satisfaction. In this study, Cronbach’s α for the WWBI was .92 (n = 308).

Participants were asked to reflect on their career attitudes. Boundaryless career attitudes were measured using the Boundaryless Career Attitudes Scale (Briscoe et al., 2006). The 13-item scale includes items tapping the boundaryless mind-set and organizational mobility preference. Examples of items include “I would enjoy working on projects with people across many organizations,” “I am energized in new experiences and situations,” and “In my ideal career I would work for only one organization.” For each item on the scale, participants select from 1 (to a limited extent) to 5 (to a great extent). Scores on the scale can range from 13 to 65, with higher scores indicating greater boundaryless career attitudes. In this study, Cronbach’s α was .86 (N = 315).

Demographic information relevant to occupational embeddedness was collected. This included years in career and years of career-specific education. We asked participants the number of years in the career to take into account the increasing number of mature-aged entrants into policing and teaching careers who would have shorter tenure than those who had commenced training directly after high school. Career-specific education, rather than education in total was selected on the basis that career-specific education was more potentially relevant to embeddedness and was also more likely to have been achieved prior to exit from the career. We asked participants to indicate financial responsibility (e.g., home loan repayments; no = 0, yes = 1) to take into account long-term ongoing financial commitment from participants’ perspectives regardless of amount. Demographic data from the time of career exit were used for former teachers and police officers for years in career, financial responsibility, and age.

Intended career duration was assessed. Participants were asked, “While completing your training, how long did you think that you would stay in your career?” Response categories were collapsed such that 1 = short- to mid-term intentions (from <1 year to 11 years); 2 = long-term intentions (from 12 years until retirement); and 3 = unsure. Finally, to assess whether participants had made career changes into policing or teaching, we asked whether they had worked full time in another career for 3 years or more, including in unpaid work such as child rearing, elder care, or volunteer work prior to entering policing or teaching careers (yes = 0, no = 1).

Research Design and Sample Size

We used binary logistic regression to predict two career decision outcomes: decision to stay or leave. To do this, we used participants’ career decision status (as stayers, leavers, or undecided) as the grouping variable and deselected the undecided group. Predictors for the model were based on Ng and Feldman’s (2007) suggestion of considering boundaryless career attitudes and embedding variables in tandem. For logistic regression analyses, it has been suggested that for each predictor variable added, approximately 10 participants are needed to avoid bias in the coefficients obtained (Peduzzi, Concato, Kemper, Holford, & Feinstein, 1996). In this study, the smaller subsample of former career members (n = 69) could support no more than seven predictor variables.

Results

Preliminary Results

A one-way, between-groups analysis of variance (ANOVA) was used to compare job satisfaction scores of career-stayers (n = 155), career-deciders (n = 91), and career-changers (n = 69). The purpose of this was to check that career-deciders formed a distinct group of current employees, before using career-decision status as the grouping variable for developing a predictive model of career stability and mobility. Seven members of the career-changer group did not complete the job satisfaction measure, as they were in full-time study or were searching for employment at the time of the survey. The assumptions of normality and homogeneity of variance were supported for each group. The ANOVA was significant, confirming that mean level of satisfaction differed by career decision status, F(2, 305) = 44.75, p < .001, η² = .23, f = .54. This effect can be considered large (Cohen, 1988). Post hoc analyses with Tukey’s honest significant difference revealed that the career-deciders (M = 57.97, SD = 15.72; means expressed as percentages) had significantly lower job satisfaction than did the career-stayers (M = 74.26, SD = 13.38) and career-changers (M = 75.85, SD = 14.10). Effect sizes for these comparisons were 1.00 and 0.86, respectively. However, there was no significant difference between the job satisfaction scores of the career-stayer and career-changer groups (d = 0.06).

Thus, career-deciders were considered a distinct group of current employees and withheld from the next stage of analysis, which aimed to develop a predictor model of career stability and mobility based on the responses of career-stayers and career-changers. The benefit of withholding responses of the career-decider group was that the resulting model could later be used to predict the career-deciders’ future membership as career-stayers or career-changers.

Predictive Model

Means, standard deviations (SDs), and intercorrelations of predictor variables by career-decision status (stayer and leaver) are reported in Table 1.

Table 1.
Summary of Intercorrelations and Descriptive Statistics for All Predictor Variables as a Function of Career Decision Status (Stayers and Leavers).

Variable^a 1 2 3 4 5 6 Career-Changer, Mean (SD)

1. Career — .45 .11 .23 −.14 −.11 —

2. Years in career −.34 — .16 .84 .11 −.11 16.7 (11.6)

3. Financial responsibility .22 .03 — .17 .14 −.15 —

4. Age −.40 .78 −.05 — .15 −.08 39.6 (11.0)

5. Career-specific education −.49 .25 .01 .21** — −.12 4.1 (2.4)

6. Boundaryless career attitudes^b −.08 .13 .13 .08 .13 — 48.8 (8.4)

Career-stayer, Mean (SD) 15.1 (10.5) 41.8 (9.8) 3.2 (2.0) 42.3 (8.8) —

Note. Intercorrelations are presented below the diagonal for career-stayers (n = 155) and above the diagonal and for career-changers (n = 69); correlations are two-tailed. Means and standard deviations for career-stayers are presented in the bottom row and for career-changers are in the far right-hand column. ^aCareer (0 = teacher, 1 = police officer), 38% of career-stayers and 58% of career-changers in the sample were teachers; financial responsibility (0 = no, 1 = yes), 79.4% of career-stayers and 59.4% of career-changers reported financial responsibility. ^bRange of possible Boundaryless Career Attitude scores was 13–65. For other (nonpredictor) variables, 61.9% of career-stayers and 50.7% of career-changers reported that they had intended to have long-term careers in teaching or policing; 53.5% of career-stayers and 29.0% of career-changers had worked in another role (including in unpaid roles) full time for 3 years or more, prior to entering teaching or policing careers.

p < .05; p < .01; p < .001.

A hierarchical logistic regression analysis was used to develop a predictive model of career mobility and stability. The outcome variable was binary, predicting membership as a current staying employee or as a former employee who changed career. Of career-decision status, the selection variable, two subgroups (career-stayer and career-changer) were selected and the other (career-decider) was deselected. Hierarchical regression was used to identify not only the unique variance explained by different predictor variables but also changes in the predictive utility of the final model. Before interpreting the results of the analysis, a number of assumptions were tested and checks performed (Hosmer, Taber, & Lemeshow, 1991). In terms of multicollinearity, Myers (1990) suggested that variance inflation factor (VIF) values above 10 were cause for concern. The highest VIF value in the present study was 2.69, indicating no such issue for interpretation.

Demographic and occupational embeddedness variables entered at Step 1 were career (teacher or police officer), years in career, age (in years), financial responsibility (e.g., home loan repayments: no, yes), and career-specific education (years). Occupational embeddedness variables together accounted for 24% of the variance in career mobility and stability (Nagelkerke R ² = .24; model χ² = 42.15, df = 5, p < .001).

Boundaryless career attitudes were added to the regression equation at Step 2 and accounted for an additional 10% of variance (Nagelkerke R ² = .34, ΔR ² = .10, Model χ² = 61.296, df = 6, p < .001). In addition, the nonsignificant result of the Hosmer and Lemeshow’s test (χ² = 8.06, df = 8, p = .43) indicated that the null hypothesis (that a model without the predictor variables would be no different to a model with the predictors) could be rejected (La Valley, 2008).

B coefficients (B) and odds ratios or exponentiations of B coefficients (Exp B), with 95% confidence intervals are reported in Table 2 for each predictor in the final model. As can be seen in Table 2, significant predictors of career mobility and stability were years in career (Exp B = 1.13, p < .001), financial responsibility (Exp B = 2.69, p = .009), age (Exp B = .87, p < .001), and boundaryless career attitudes (Exp B = 1.09, p < .001). For each additional year in the career, the odds of a career change increased, the absence of financial responsibility such as mortgage repayments increased the odds of a career change, and for each year younger in age the odds of career change increased. In addition, higher scores on boundaryless career attitudes increased the odds of a career change. Whether a person was a teacher or a police officer and years of career-specific education did not add predictive utility to the model.

Table 2.
Hierarchical Logistic Regression Analyses Predicting Career-Change From Embeddedness-Related Demographic Variables and Boundaryless Career Attitudes.

Step Predictor^a B (SE) Odds Ratio/Exp(B) 95% CI

1 Career .60 (.37) 1.82 [0.88, 3.77]

Years in career .12** (.03) 1.13 [1.05, 1.20]

Financial responsibility .99* (.38) 2.69 [1.29, 5.61]

Age −.14 (.04) .87 [0.81, 0.94]

Years career-specific education .12 (.08) 1.13 [0.96, 1.32]

2 Boundaryless career attitudes .09 (.02) 1.09 [1.05, 1.14]

Constant −2.12 (1.34) .12

Note. n = 224 (n = 155 career-stayer and n = 69 career-changer; unselected cases were n = 91 career-decider); R ² = .34 (Nagelkerke’s). Model χ² = 61.30, df = 6, p = < .001. CI = confidence interval; SE = standard error.

^aCareer was coded 0 = police officer, 1 = teacher; financial responsibility was coded 0 = yes, 1 = no.

*p < .01; **p < .001.

A model without predictors correctly identified all career-stayers as current employees but failed to identify any career-changers as former employees, overall providing 69.2% correct classifications. The model with predictors improved upon the overall classification rate, correctly predicting 75.9% of cases, which can be considered acceptable discrimination (Hosmer & Lemeshow, 2000). The model was better able to predict career-stayer (88.4%) than career-changer (47.8%) group membership. When career-deciders were tested with the model, 79.1% were predicted as career-stayers and the remaining 20.1% more closely matched the profiles of career-changers.

Additional Analyses

Additional analyses were performed to further understand boundaryless career attitudes as a predictor of career mobility. For our sample as a whole, we tested whether mean boundaryless career attitude scores differed depending upon whether participants had or had not had full-time employment for 3 years or more prior to entering a policing or teaching career. Assumptions of normality and equal variance were met. The mean boundaryless career attitudes score (M = 46.80, SD = 8.92) for participants who had not had prior employment was significantly higher than was the mean score (M = 44.32, SD = 9.21) for the prior employment group, with a small effect, F(1, 313) = 5.88, p = .016, η² = .02, f = .14.

We tested whether boundaryless career attitude scores differed depending upon intended career duration. The assumption of normality was met, but Levene’s statistic was significant indicating that the assumption of equal variances was not met (p = .032). Therefore, Welch’s F test was used. The mean boundaryless career attitude scores differed depending upon whether participants had reported short-mid-term or long-term intended career duration or were unsure about intended career duration, Welch’s F(2, 155.26) = 9.38, p < .001, estimated w² = .05. This is a small–medium effect. Post hoc analyses using the Games-Howell’s test revealed that the mean boundaryless career attitude score for participants with short-mid-term (M = 49.03, SD = 6.88) career intentions was significantly higher than were the mean score for the long-term (M = 44.20, SD = 9.13) group. The effect size for the comparison was medium (d = 0.60). The mean scores of the unsure (M = 46.01, SD = 9.97) group did not differ significantly from either the short-mid-term or the long-term groups (with effect sizes of 0.35 and 0.15, respectively).

Discussion

The aim of this study was to test occupational embeddedness and boundaryless career attitudes as a complementary explanation for career stability and mobility. To do so, we invited current and former members of two occupations, previously known for lifetime employment to reflect upon their careers and career decisions. Reflecting past research indicating that low job satisfaction was associated with leaving (Griffeth et al., 2000) and intention to leave (Mitchell et al., 2001), job satisfaction was significantly lower for participants who were considering career-change than for those who had made a decision to stay or to leave. This finding supported our proposition that police officers and teachers in a career decision-making phase formed a distinct group of current employees and should be considered separately to career-stayers who had actively decided to stay. On this basis, we developed a predictor model of career stability and mobility using the responses of participants who had actively decided either to stay in or to leave their careers.

First, the hypothesis that demographic variables associated with occupational embeddedness would predict career stability was partially supported. Higher ages and the presence of financial responsibility predicted career stability. In contrast, contrary to the hypothesis, a higher number of years in the career predicted mobility. Years of career-specific education predicted neither career stability nor mobility. Second, the hypothesis that boundaryless career attitudes would add predictive utility to a model of career stability and mobility was supported; boundaryless career attitudes predicted mobility.

Occupational Embeddedness

As expected, career mobility or stability was not predicted by membership in a policing as opposed to a teaching career. The factors influencing decisions about staying in or leaving the careers applied to members of both of the careers previously renowned for lifelong membership. Financial responsibility, years in career, and age predicted career mobility and stability. Participants who had financial responsibility, such as home loan repayments, were less likely to leave their careers, and thus financial responsibility acted as an embedding factor. Participants who had been in the careers for longer were more likely to leave, and people who were younger were also more likely to leave; this suggests early- and late-career mobility. According to Ng and Feldman’s (2007) definition of occupational embeddedness, we might have expected that accumulated skills and relationships, acquired with more years in the career, would act as embedding factors. The finding that career mobility was associated with lower ages and with more years in the career is consistent with past research on policing and teaching careers (e.g., Borman & Dowling, 2008; Lynch & Tuckey, 2004) and may reflect the challenge for career mobility faced in mid-life (acknowledged by Ng & Feldman, 2007). In policing and teaching, the effectiveness of the number of years in career as an embedding factor may be affected by the prevalence of early retirement, encouraged by the structures of superannuation benefits schemes (contributory retirement pension funds) to which many teachers and police officers belong and of medical retirement (early retirement on medical grounds) in policing in particular (Audit Office of NSW, 2008a, 2008b).

Career-specific education did not predict mobility or stability. Although it was anticipated that more years undertaking education and training for a career might contribute to occupational embeddedness (Feldman & Ng, 2007), this does not appear to have been the case. Arguably education, even when it is specific to one career, is beneficial in other careers and therefore did not contribute to embedding people within one career. It is possible that the distinction between career-specific education and education in general may not be meaningful in practical terms. The results of one study indicated, for example, that police managers viewed the generic or transferable skills of problem solving, decision making, and communication, rather than the policing-specific skills developed in their studies, the most relevant to their work (Trofymowych, 2008).

Boundaryless Career Attitudes

Boundaryless career attitudes added predictive utility to the model of career stability and mobility. This predictor of mobility appeared to be of high importance. Given the relationship of boundaryless career attitudes to the constructs of openness to experience, proactive personality, and mastery goal orientation, (Briscoe et al., 2006), and that openness to experience was associated with career change (Carless & Arnup, 2011), it is perhaps unsurprising that boundaryless career attitudes predicted career mobility. Boundaryless career attitudes reflected a flexible and adaptive approach of individuals to the labor market in the broader social context. Police officers and teachers with higher scores on the boundaryless career attitudes scale may be more likely to move into new careers than those with lower scores on the boundaryless attitudes scale in response to career and personal circumstances.

The limitation inherent in a cross-sectional survey requires that this result be interpreted with caution, as it is not entirely clear when boundaryless career attitudes develop. To address this issue in part, we considered whether having had previous careers was associated with boundaryless career attitudes. For our sample, having had careers prior to entering teaching or policing careers was significantly associated with lower scores for boundaryless career attitudes. This suggested that people who had already had other careers were not more open to another career change to move out of policing or teaching careers than those who had not had prior careers. This was consistent with Briscoe et al.’s (2006) finding that mobility behavior was a poor measure of boundaryless career attitudes. We also considered participants’ initial intended career durations. As may have been anticipated, boundaryless career attitudes were significantly higher for people with short-mid-term intended membership in the career.

Although some police officers and teachers may have hoped for a life of diverse career opportunities from the outset, decision-making processes are complex. Many factors are undoubtedly taken into consideration when making decisions about careers. As Ng and Feldman (2007) suggested, priorities may change at different life stages. Past research has shown that plans are modified during the course of the career (Rinke, 2009) and indicated the role of chance or happenstance in career decisions (Hancock, 2009; Krumboltz, 2009). Furthermore, individual differences in personality and decision-making style would affect the decision-making process (Drew, Carless, & Thompson, 2008; Thunholm, 2004).

Job Satisfaction

The finding that job satisfaction did not differ significantly for those who decided to stay and those who decided to leave may be encouraging for teachers and police officers who are in a decision-making phase. As Ng and Feldman (2007) suggested, many people who consider career changes ultimately reject the idea and decide to stay in a career. Indeed, this study revealed that of the participants considering career change, approximately four fifths (79.9%) more closely matched the profiles of those who had decided to stay in their chosen careers.

Although the results of cross-sectional studies must be interpreted with caution, the suggestion for people in a decision-making phase is that they cannot make a wrong decision about their career: Ultimately, there was no difference in level of job satisfaction for those who had decided to stay and those who had decided to leave. Past research found that traditional careers were associated with higher job satisfaction than were more boundaryless careers (Dries, Van Acker, & Verbruggen, 2012). But it was also found that low job satisfaction improved after a career change was made (Carless & Arnup, 2011). It may be that considering career change is a period of crisis within the career, which for some is resolved in such a way that job satisfaction is regained in the career; while for others it is regained in a new career setting. Feldman (2002) argued that occupational change may be associated with self-fulfilling prophecies, as high expectations of new roles are actualized, and cognitive dissonance, as career-changers are likely to report high job satisfaction in new roles to justify the sacrifices made in leaving previous roles. However, it could equally be argued that such cognitive dissonance may also apply to decisions to stay in an occupation, having foregone other potential opportunities.

Limitations and Implications for Future Research

This study responded to the need (Ng et al., 2005) for research based on career behavior rather than on behavioral intention of both current and former occupational members. However, we acknowledge that our sample of participants was not randomly selected. For the smaller group of former teachers and police officers, in particular, respondents may have been more likely to maintain interest in their past careers, reading association journals or staying in touch with former colleagues, and may therefore differ from the total population of former Australian teachers and police officers in other ways not accounted for in the research.

The current study was limited by a relatively small group of career-change (former career member) participants. Although we were able to use the subsample of undecided current career members to predict decision outcomes, the sample size constraint prevented a random subsample of the two main groups (stayers and changers) from being kept aside to use for validation of the results (Babyak, 2004). Ideally, a larger sample would have been obtained, even in order to use the number of predictor variables used here. Therefore, there is a need for replication across future samples of police officers and teachers, and with samples drawn from other occupational populations, to validate the results. A larger sample size would also permit inclusion of a greater number of predictor variables.

In this study, we used correlates of embeddedness—derived from theory and past research—as proxy measures of occupational embeddedness. Future research could include a measure of the construct of occupational embeddedness, such as that proposed by Adams, Webster, and Buyarski (2010). This would enhance commensurability across studies that incorporate the construct of occupational embeddedness. For the variable of financial responsibility, different levels of financial commitment and satisfaction with the financial benefits of an occupation could be considered. Future research could include a group of preservice teachers and police recruits and incorporate the boundaryless career attitude scale to establish its utility as a predictor of career mobility from the outset of the career. Use of the scale during occupational training could further clarify whether boundaryless career attitudes are stable over time or are modified in response to the experience of career membership and subsequent career change. There is a need to identify and include additional predictors of mobility. Some such potential predictors may include the Protean Career Attitudes Scale (Briscoe et al., 2006) to measure preferences for self-managed as opposed to organization-managed careers. Follow-up interviews with participants from this study indicated that for some who left careers, there was a sense of betrayal or employer breach to the psychological contract (Howes & Goodman-Delahunty, 2014). Future research could include a measure of employee perception of employer breach of psychological contract.

Because each of the three groups (stayers, deciders, and changers) was comprised of two subgroups, investigation of within-group differences in each group may be a fruitful area for future research. Further, given that levels of job satisfaction did not differ between career-stayer and career-changer groups, longitudinal research of occupational members in a career decision-making phase is warranted. Unfortunately, we did not make the provision of asking participants who were currently in a decision-making phase whether they would be willing to participate in a later follow-up study. Although it has been previously found that career plans are modified over the career course (Rinke, 2009), future research could compare and contrast the decision-making processes and considerations of those who ultimately decide to leave and those who decide to stay in their careers.

Conclusion

This study contributed to the understanding of career stability and mobility among members of two previously lifelong careers by examining the predictive utility of two complementary concepts. We showed that boundaryless career attitudes predicted career mobility and certain demographic variables embedded people in their careers. Specifically, financial responsibility and age embedded people in their careers, contributing to career stability. In contrast, increased number of years in the career predicted mobility; career-specific education predicted neither stability nor mobility. Our approach to prediction of group membership, surveying both current and former members of two careers and using career decision status as a selection variable, allowed valuable insight. We observed that approximately four fifths of police officers and teachers who were considering career change more closely matched the profiles of police officers and teachers who had decided to stay than those who changed careers. These findings contribute to the understanding of employee attrition and retention in previously lifelong careers in light of changes to careers.

Variable^a	1	2	3	4	5	6	Career-Changer, Mean (SD)
1. Career	—	.45**	.11	.23	−.14	−.11	—
2. Years in career	−.34**	—	.16	.84**	.11	−.11	16.7 (11.6)
3. Financial responsibility	.22**	.03	—	.17	.14	−.15	—
4. Age	−.40**	.78**	−.05	—	.15	−.08	39.6 (11.0)
5. Career-specific education	−.49**	.25**	.01	.21**	—	−.12	4.1 (2.4)
6. Boundaryless career attitudes^b	−.08	.13	.13	.08	.13	—	48.8 (8.4)
Career-stayer, Mean (SD)		15.1 (10.5)		41.8 (9.8)	3.2 (2.0)	42.3 (8.8)	—

Step	Predictor^a	B	(SE)	Odds Ratio/Exp(B)	95% CI
1	Career	.60	(.37)	1.82	[0.88, 3.77]
Years in career	.12**	(.03)	1.13	[1.05, 1.20]
Financial responsibility	.99*	(.38)	2.69	[1.29, 5.61]
Age	−.14**	(.04)	.87	[0.81, 0.94]
Years career-specific education	.12	(.08)	1.13	[0.96, 1.32]
2	Boundaryless career attitudes	.09**	(.02)	1.09	[1.05, 1.14]
Constant	−2.12	(1.34)	.12

Footnotes

Acknowledgement

The authors would like to thank Dr. Mike Garry for a helpful discussion of an earlier draft of the paper.

Authors’ Note

The current affiliation of the author Loene M. Howes is Psychology, School of Medicine; and Tasmanian Institute of Law Enforcement Studies (TILES), School of Social Sciences, University of Tasmania, Private Bag 22, Hobart, TAS 7001, Australia.

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.

References

Adams

G. A.

Buck

(2010). Social stressors and strain among police officers: It’s not just the bad guys. Criminal Justice and Behavior, 37, 1030–1040. doi:10.1177/0093854810374282

Adams

G. A.

Webster

J. R.

Buyarski

D. M.

(2010). Development of an occupational embeddedness measure. Career Development International, 15, 420–436. doi:10.1108/13620431011075312

Anthony

Ord

(2008). Change-of-career secondary teachers: Motivations, expectations, and intentions. Asia-Pacific Journal of Teacher Education, 36, 359–376.

Arthur

M. B.

Rousseau

D. M.

(Eds.). (1996). The boundaryless career: A new employment principle for a new organizational era. New York, NY: Oxford University Press.

Audit Office of NSW. (2008a). Ageing workforce – teachers: Department of education and training (Performance Audit). Sydney, Australia: Author.

Audit Office of NSW. (2008b). Managing injured police: NSW police force (Performance Audit). Sydney, Australia: Author.

Australian Bureau of Statistics. (2010). Schools Australia 2009 (4221.0; Reissue). Retrieved from http://www.abs.gov.au.

Babyak

M. A.

(2004). What you see may not be what you get: A brief, nontechnical introduction to overfitting in regression-type models. Psychosomatic Medicine, 66, 411–421.

Blau

(2007). Does a corresponding set of variables for explaining voluntary organizational turnover transfer to explaining voluntary occupational turnover? Journal of Vocational Behavior, 70, 135–148. doi:10.1016/j.jvb.2006.07.007

10.

Borman

G. D.

Dowling

N. M.

(2008). Teacher attrition and retention: A meta-analytic and narrative review of the research. Review of Educational Research, 78, 367–409. doi:10.3102/0034654308321455

11.

Brill

McCartney

(2008). Stopping the revolving door: Increasing teacher retention. Politics and Policy, 36, 750–774.

12.

Briscoe

J. P.

Hall

D. T.

DeMuth

R. L. F.

(2006). Protean and boundaryless careers: An empirical exploration. Journal of Vocational Behavior, 69, 30–47. doi:10.1016/j.jvb.2005.09.003

13.

Brough

Frame

(2004). Predicting police job satisfaction and turnover intentions: The role of social support and police organisational variables. New Zealand Journal of Psychology, 33, 8–16.

14.

Buchanan

(2009). Where are they now? Ex-teachers tell their life-work stories. Issues in Educational Research, 19, 1–13.

15.

Carless

S. A.

Arnup

J. L.

(2011). A longitudinal study of the determinants and outcomes of career change. Journal of Vocational Behavior, 78, 80–91. doi:10.1016/j.jvb2010.09.002

16.

Cohen

(1988). Statistical power analysis for the behavioral sciences (2nd ed.). Hillsdale, NJ: Lawrence Erlbaum.

17.

Cooper

Davey

K. M.

(2011). Teaching for life? Midlife narratives from female classroom teachers who considered leaving the profession. British Journal of Guidance & Counselling, 39, 83–102. doi:10.1080/03069885.2010.531386

18.

Drew

Carless

S. A.

Thompson

B. M.

(2008). Predicting turnover of police officers using the sixteen personality factor questionnaire. Journal of Criminal Justice, 36, 326–331. doi:10.1016/j.jcrimjus.2008.06.003

19.

Dries

Van Acker

Verbruggen

(2012). How ‘boundaryless’ are the careers of high potentials, key experts, and average performers? Journal of Vocational Behavior, 81, 271–279. doi:10.1016/j.jvb.2011.10.006

20.

Ewing

Manuel

(2005). Retaining quality early career teachers in the profession: New teacher narratives. Change: Transformations in Education, 8, 1–16.

21.

Feldman

D. C.

(2002). Second careers and multiple careers. In Cooper

C. L.

Burke

(Eds.). The new world of work: Challenges and opportunities (pp. 75–94). Oxford, England: Blackwell.

22.

Feldman

D. C.

T. W. H.

(2007). Careers: Mobility, embeddedness, and success. Journal of Management, 33, 350–377. doi:10.1177/0149206307300815

23.

Griffeth

R. W.

Hom

P. W.

Gaertner

(2000). A meta-analysis of antecedents and correlates of employee turnover: Update, moderator tests, and research implications for the next millennium. Journal of Management, 26, 463–488.

24.

Haarr

R. N.

(2005). Factors affecting the decision of police recruits to “drop out” of police work. Police Quarterly, 8, 431–453. doi:10.1177/1098611103261821

25.

Hancock

(2009). The effects of chance, turning points and routine on male career development. Journal of Education and Work, 22, 121–135.

26.

Hodkinson

Sparkes

A. C.

(1997). Careership: A sociological theory of career decision making. British Journal of Sociological Education, 18, 29–44.

27.

Hosmer

D. W.

Lemeshow

(2000). Applied logistic regression (2nd ed.). New York, NY: John Wiley.

28.

Hosmer

D. W.

Taber

Lemeshow

(1991). The importance of assessing the fit of logistic regression models: A case study. American Journal of Public Health, 81, 1630–1635.

29.

Howes

L. M.

Goodman-Delahunty

(2014). Life course research design: Exploring career change experiences of former school teachers and police officers. Journal of Career Development, 41(1), 62–84. doi: 10.1177/0894845312474370

30.

Ingersoll

(2003). Is there really a teacher shortage? [Research Report]. Seattle: University of Washington.

31.

Inkson

Gunz

Ganesh

Roper

(2012). Boundaryless careers: Bringing back boundaries. Organization Studies, 33, 323–340. doi:10.1177/0170840611435600

32.

International Wellbeing Group (2006). Personal wellbeing index (4th ed.). Melbourne, Australia: Australian Centre on Quality of Life, Deakin University. Retrieved from http://www.deakin.edu.au/research/acqol/instruments/wellbeingindex.htm.

33.

Jennett

Islam

Bull

Woolston

(2008, September). Occupational identity of police recruits. WACE/ACEN Asia Pacific Conference 2008 E-Proceedings, Sydney, Australia, 254–261.

34.

Johnson

A. E.

Sachau

D. A.

Englert

(2010). Organizational and occupational embeddedness of federal law enforcement personnel. Journal of Police and Criminal Psychology, 25, 75–89. doi:10.1007/s11896-009-9063-x

35.

Jones

Prenzler

(2005). Tertiary education, commitment, and turnover in police work. Police Practice and Research, 6, 49–63.

36.

Koper

C. S.

Maguire

E. R.

Moore

G. E.

Huffer

D. E.

(2001). Hiring and retention issues in police agencies: Readings on the determinants of police strength, hiring and retention of officers, and the federal COPS program. Washington, DC: Urban Institute.

37.

Krumboltz

J. D.

(2009). The happenstance learning theory. Journal of Career Assessment, 17, 135–154. doi:10.1177/1063072708328861

38.

La Valley

M. P.

(2008). Logistic regression. [Special section: Statistical primer for cardiovascular research] Circulation: Journal of the American Heart Association, 117, 2395–2399. doi:10.1161/CIRCULATIONAHA.106.682658

39.

Lynch

(2007). Burnout and engagement in probationary police officers: A scoping paper (ACPR Scope No. 1). Payneham, South Australia: Australasian Centre for Policing Research.

40.

Lynch

Tuckey

(2004). Understanding voluntary turnover: An examination of resignations in Australasian police organisations (Report Series No. 143.1). Adelaide, SA: Australasian Centre for Policing Research (ACPR).

41.

Lynch

J. E.

Tuckey

(2008). The police turnover problem: Fact or fiction? Policing: An International Journal of Police Strategies and Management, 31, 6–18. doi:10.1108/13639510810852549

42.

Mitchell

T. R.

Holtom

B. C.

Lee

T. W.

Erez

(2001). Why people stay: Using job embeddedness to predict voluntary turnover. Academy of Management Journal, 44, 1102–1121.

43.

Motulsky

S. L.

(2010). Relational processes in career transition: Extending theory, research, and practice. The Counseling Psychologist, 38, 1078–1114.

44.

Myers

(1990). Classical and modern regression with applications (2nd ed.). Boston, MA: Duxbury.

45.

T. W. H.

Eby

L. T.

Sorenson

K. L.

Feldman

D. C.

(2005). Predictors of objective and subjective career success: A meta-analysis. Personnel Psychology, 58, 367–408.

46.

T. W. H.

Feldman

D. C.

(2007). Organizational embeddedness across career stages. Journal of Vocational Behavior, 70, 336–351. doi:10.1016/j.jvb.2006.10.002

47.

Orrange

R. M.

(2003). The emerging mutable self: Gender dynamics and creative adaptations in defining work, family, and the future. Social Forces, 82, 1–34.

48.

Otto

Dette-Hagenmeyer

D. E.

Dalbert

(2010). Occupational mobility in members of the labor force: Explaining the willingness to change occupations. Journal of Career Development, 36, 262–288. doi:10.1177/0894845309345842

49.

Page

(2005). Subjective wellbeing in the workplace (Unpublished Honours thesis). Deakin University, Melbourne, Australia. Retrieved from Deakin University website http://www.deakin.edu.au/research/acqol/instruments/index.htm

50.

Peduzzi

Concato

Kemper

Holford

T. R.

Feinstein

A. R.

(1996). A simulation study of the number of events per variable in logistic regression analysis. Journal of Clinical Epidemiology, 49, 1373–1379.

51.

Prenzler

Fleming

King

A. L.

(2010). Gender equity in Australian and New Zealand policing: A five year review. International Journal of Police Science and Management, 12, 584–595. doi: 10.1350/ijps.2010.12.4.211

52.

Richardson

P. W.

Watt

H. M. G.

(2006). Who chooses teaching and why? Profiling characteristics and motivation across three Australian universities. Asia-Pacific Journal of Teacher Education Research, 34, 27–56. doi:10.1080/13598660500480290

53.

Rinke

C. R.

(2009). Finding their way on: Career decision-making processes of urban science teachers. Science Education, 93, 1096–1121. doi:10.1002/sce.20339

54.

Rousseau

D. M.

(1989). Psychological and implied contracts in organizations, Employee Rights and Responsibilities Journal, 2, 121–139.

55.

Shniper

(2005, December). Research summary: Occupational mobility, January 2004. Monthly Labor Review , 30–35.

56.

Smithson

Lewis

(2000). Is job insecurity changing the psychological contract? Personnel Review, 29, 680–702.

57.

Sullivan

S. E.

Arthur

M. B.

(2006). The evolution of the boundaryless career concept: Examining physical and psychological mobility. Journal of Vocational Behavior, 69, 19–29. doi:10.1016/j.jvb.2005.09.001

58.

Stone

K. V. W.

(2002). Knowledge at work: Disputes over the ownership of human capital in the changing workplace. Connecticut Law Review, 34, 721–763.

59.

Thunholm

(2004). Decision-making style: Habit, style, or both? Personality & Individual Differences, 36, 931–944. doi:10.1016/S0191-8869(03)00162-4

60.

Trofymowych

(2008, September). Police managers’ views about their university education and professional status. WACE/ACEN Asia Pacific Conference 2008 E-Proceedings, Sydney, Australia, 543–550.

61.

Watt

H. M. G.

Richardson

P. W.

(2008). Motivations, perceptions, and aspirations concerning teaching as a career for different types of beginning teachers. Journal of Learning and Instruction, 18, 408–428. doi:10.1016/j.learninstruc.2008.06.002

62.

White

M. D.

Cooper

J. A.

Saunders

Raganella

A. J.

(2010). Motivations for becoming a police officer: Re-assessing officer attitudes and job satisfaction after six years on the street. Journal of Criminal Justice, 38, 520–530.