Introduction

Alternatives to traditional forms of employment are becoming commonplace (Cappelli & Keller, 2013). For example, approximately 7.2 million Americans are classified as a dual jobholder (U.S. Bureau of Labor Statistics, 2016): an individual who works for and receives income from two different organizations or who works for and receives income from one organization and is self-employed in another job. Despite the prevalence with which dual jobholding occurs, research on the topic has been described as “quite barren” (Zickar, Gibby, & Jenny, 2004, p. 234) as most studies in the organizational behavior literature examine individual work experiences in only one job.

Better insight into the experiences of dual jobholders is important for a number of reasons. First, more than 50% of men engage in dual jobholding at some point in their lives (Paxson & Sicherman, 1996) and men and women currently participate in dual jobholding at equal rates (U.S. Bureau of Labor Statistics, 2016). Second, researchers, practitioners, and popular press often imply dual jobholders differ from single jobholders in detrimental ways. For example, Rodell (2013) suggests that holding a second job can distract a person because dual jobholders juggle different and sometimes competing demands. The Society for Human Resource Management noted that many employers believe dual jobholding can have detrimental effects on individual productivity and organizational loyalty (Hirschman, 2000), and popular press articles have suggested that holding two jobs may leave individuals tired and devoid of energy (Davidson, 2016; Mattis, 2008). However, others have noted that positive benefits may accompany dual jobholding such as increased income, the acquisition of new skills, or personal fulfillment unavailable in one job (Hochschild, 1997; Kimmel & Conway, 2001). Because many of the notions regarding dual jobholding are speculative rather than scientifically tested, a theoretical and empirical examination of dual jobholders would prove beneficial to the organizational behavior literature and organizations that commonly employ dual jobholders.

In the present study, we seek to shed light on the understudied phenomenon of dual jobholding. The notions that dual jobholding may be a distraction for employees (Rodell, 2013) and has detrimental effects on employee productivity and loyalty (Hirschman, 2000) imply dual jobholders prioritize one job at the expense of the other job or that the jobs interfere with each other in some way. Thus, we compare the individual work experiences of a dual jobholder across both of his/her jobs. If dual jobholders are distracted and allow one job to interfere with the other job, dual jobholders should exhibit poorer work behaviors in one of the jobs. We also seek to understand if dual jobholders and single jobholders experience work and life differently due to the added demands of a second job. One way to identify if dual jobholders differ from single jobholders in the workplace is to compare the two populations on important work behaviors and attitudes. By examining potential differences between an individual’s experiences in the primary job and second job, and by comparing the experiences of dual jobholders to single jobholders, we advance understanding of dual jobholding and provide an important contribution to the understanding of alternative work arrangements.

Theoretical Rationale and Hypotheses

Conceptualization of a Dual Jobholder

Although some studies have employed datasets that consist of dual jobholders (e.g., Inness, Barling, & Turner, 2005; Schaubroeck, Judge, & Taylor, 1998), we only identified one study in the human resources and organizational behavior literatures that specifically examined the topic of dual jobholding (i.e., Zickar et al., 2004). To date, the economics literature has devoted the most, albeit slim, attention to the topic of dual jobholding. Most economic scholars who examine dual jobholding in the USA obtain data from the Current Population Survey, which has been conducted by the U.S. Bureau of Labor Statistics every month for the past 100 years. Academic research has yet to put forth a formal definition of dual jobholding. Thus, we present a conceptual and specific definition of the employment arrangement in which an individual is formally considered a dual jobholder, which we adapted from the Current Population Survey. We categorize an individual as a dual jobholder if he or she answers “yes” to the question: “Last month, did you have more than one job (or business), including part-time, evening, or weekend work?” (Hipple, 2010). Further, the individual must meet one of two criteria: holds (a) a paid job with two or more employers or (b) combines a paid job from one employer with self-employment in another job (Hipple, 2010).

We conceptualize a dual jobholder as a manifest, demographic variable. A dual jobholder must concurrently receive some income from two distinct employers. If one received no income for performing a particular job, one would be classified as a volunteer; which is beyond the scope of the present study. The timeframe to be considered a dual jobholder is 1 month because the majority of important, aggregate employment data tracked by the U.S. Bureau of Labor Statistics is presented in monthly intervals (e.g., unemployment rates, jobs added, jobs lost).

We define the “primary job” to be the job at which the greatest number of hours per week is worked. Accordingly, the “second job” is the job at which the second greatest number of hours per week is worked (Hipple, 2010). In the present study, our focus is concerned only with dual jobholders rather than individuals who may have three or more jobs. Therefore, if an individual works more than two jobs, we focus on the two jobs at which the greatest number of hours per week is worked. This approach is similar to traditional organizational behavior research which does not inquire about employees’ experiences in more than one job.

Dual Jobholding Scenarios

An example of a dual jobholder is a teacher who works during the day but also works as a retail associate at a clothing store a few nights per week or on weekends. Other examples are an individual who works as an accountant during the week and as a bartender on weekends or a university professor who engages in consulting work in his or her area of expertise. It is important to note that we do not limit dual jobholding to occur during a pre-specified timeframe. For example, a dual jobholder may work two different jobs only during the daytime, two different jobs only at night, one job during the day and one job at night, or one job during the week and one job during the weekend. What is important, for purposes of the present study, is that an individual works two jobs concurrently to be considered a dual jobholder.

It is prudent to illustrate questionable scenarios in which an individual would not be considered a dual jobholder. First, an individual is not considered a dual jobholder if he or she earns income from one employer and is a parent. Because an individual does not receive income for directly performing the duties of a parent, he or she is not a dual jobholder and falls outside the scope of our study. Also, individuals who occupy two positions within the same organization are not considered to be dual jobholders. For example, a high school science teacher may also serve as the high school track and field coach. Even in the event that this individual is paid for work as a science teacher and work as a track and field coach, this would fall outside our operationalization of a dual jobholder because the individual still receives income from only one employer (i.e., the school district) rather than from two different employers. Finally, it is important to remember the month time span component of the dual jobholder conceptualization. An individual may routinely obtain a part-time second job for only 2 months out of the year during the holidays (e.g., November and December). This individual would be considered a dual jobholder only during the months that he or she works for and receives income from two different employers.

It is important to acknowledge that dual jobholders are likely to work more hours than single jobholders. Indeed, the samples in the present study confirm this as well as aggregate data from the BLS (Hipple, 2010). However, it is also important to consider that one’s status as a dual jobholder entails more than simply working longer hours. For example, dual jobholders must learn multiple and different job tasks and must switch job tasks on a daily/weekly basis. Research in cognitive psychology suggests that multi-tasking or switching between two tasks presents a cognitive demand and potential stressor that can hinder one’s ability to perform well, especially when compared to single-task completion (Leroy, 2009; Monsell, 2003). Thus, the task switching nature of dual jobholding might make dual jobholders more prone to performance hindrance than simply one who works longer total hours.

Ashforth, Kreiner, and Fugate (2000) note that individuals recognize organizations believe employees should concentrate only on work roles and responsibilities during work time. Thus, dual jobholders will likely feel pressure to concentrate only on work roles and responsibilities in the job at hand rather than their other job. However, research on work-family conflict and domain switching (e.g., Ashforth et al., 2000) suggests the more one attempts to psychologically separate domains, the impact of the change that occurs when one crosses boundaries between two domains increases, making the switch to participation in a new domain more difficult. It is more difficult for individuals who try to keep “domain A” and “domain B” separate to leave participation in domain A and begin participation in domain B.

Dual jobholders must also balance schedules between two jobs (in addition to the family and leisure aspects of life). In addition, the exertion of energy in one job may deplete energy available for the other job (Edwards & Rothbard, 2000). All of these elements help distinguish dual jobholders, conceptually, from the concept of simply working longer hours and from single jobholders.

Prior Research on Dual Jobholders

The scant research that exists on dual jobholders examines dual jobholding from a labor economics theory perspective (Dickey, Watson, & Zangelidis, 2011; Hipple, 2010; Wu, Baimbridge, & Zhu, 2009) to suggest individuals undertake a second job to improve their income/leisure combination (Dickey et al., 2011; Johannessen & Hagen, 2014; Livanos & Zangelidis, 2012; Socha & Bech, 2011). That is, individuals obtain a second job as a result of an unsatisfactory level of income at the primary job or to obtain something non-pecuniary at the second job (e.g., leisure, gain new experience). Although an individual is thought to hold two jobs to obtain the preferred income/leisure combination, some research contends that holding two jobs may be detrimental. For instance, Socha and Bech (2011) investigated how physicians employed in the public sector and in the private sector might negatively impact wait times and quality of care. Their review of the literature demonstrated there was not enough evidence to support this assumption but acknowledged that there were no carefully controlled studies that compared dual-practice physicians with those working only in the public sector (i.e., the single-practice physicians). Additionally, Walsh, Dahling, Schaarschmidt, and Brach (2016) found that surface acting at the primary job and second job interact to predict levels of emotional exhaustion.

In terms of experiencing differences in work outcomes between single jobholders and dual jobholders, one empirical study (Zickar et al., 2004) showed that dual jobholders were more satisfied and more committed to their primary job than their second job. Thus, there is some evidence to suggest that dual jobholders prioritize or devote more attention to one job over the other. Further, qualitative evidence suggests teachers who hold a second job perceive that the second job negatively impacts performance in their teaching job (Parham & Gordon, 2011). Although they did not investigate dual jobholding, other researchers have speculated that dual job holding is a source of stress (e.g., Lu & Kao, 2013). The steady occurrence of dual jobholding (U.S. Bureau of Labor Statistics, 2016) and the lack of research, theory, and empirical clarity regarding dual jobholders necessitate scientific inquiry (Sliter & Boyd, 2014; Zickar et al., 2004).

Partial Inclusion Theory

In line with anecdotal assertions from researchers and popular press, we draw from partial inclusion theory (Katz & Kahn, 1966) to make predictions that holding a second job is detrimental. Specifically, we believe dual jobholding is detrimental to organizations in the form of dual jobholders exhibiting lower levels of work engagement at the second job. We focus on the construct of work engagement because the construct connotes high levels of personal involvement at work (Rich, LePine, & Crawford, 2010) and is a theoretical construct in accordance with the tenets of partial inclusion theory. The theory of partial inclusion focuses on involvement at work and suggests that individuals hold membership and roles in numerous social groups (Katz & Kahn, 1966). The more involved one is in a role, the more one identifies with that role (Greenhaus & Beutell, 1985; Hall, 1976). Because dual jobholders work less hours at the second job (approximately 20 to 40% of total working hours), they are likely to be less involved and identify less with the second job compared to the primary job (Paxson & Sicherman, 1996). Dual jobholders should be less involved at the second job as they have less time to practice and engage in job tasks, should be less socially integrated with coworkers (Peters, Jackofsky, & Salter, 1981), and should have fewer experiences with and less knowledge of the organization compared to the primary job (Martin & Hafer, 1995; Wetzel, Solosky, & Gallagher, 1990). The more one is involved with a role, the more motivated one is to fulfill the duties of that role (Greenhaus & Beutell, 1985; Hall, 1976). As individuals become involved with activities in one role and devote more time and energy to that role, their preoccupation with that role is likely to decrease their ability to satisfy requirements of other roles (Greenhaus & Beutell, 1985). As a result of lower involvement with the second job, dual jobholders may be inclined to exhibit less energy, effort, and engagement toward the second job than compared to the primary job. This notion is consistent with the finding that part-time employees had lower job involvement than full-time employees (Eberhardt & Shani, 1984; Martin & Hafer, 1995; Wetzel et al., 1990) and were less satisfied than full-time employees (Miller & Terborg, 1979). Because being engaged at work involves a personal connection and involvement with one’s work (Christian, Garza, & Slaughter, 2011), we hypothesize the following:

  • Hypothesis 1. Dual jobholders report lower work engagement at the second job than the primary job.

Partial inclusion theory may also help explain potential workplace differences between dual jobholders and single jobholders (Katz & Kahn, 1966). Single jobholders, compared to dual jobholders, spend a greater portion of their working time (i.e., 100% of their work time) at a single organization than dual jobholders (approximately 60% to 80% of total working hours is spent at the primary job; Paxson & Sicherman, 1996). Thus, at the same job, single jobholders should be more involved with their job compared to a dual jobholder. The more one identifies and is involved with a role, the more one strives to satisfy the duties of that role (Greenhaus & Beutell, 1985; Hall, 1976). According to partial inclusion theory, single jobholders should be more involved with their job and exert more effort to satisfy the duties of their job. Although hypothesis 1 predicted that the efforts of dual jobholders will be focused more on the primary job than the second job, the second job still requires attention, effort, and resources. Thus, compared to single jobholders without the distraction of a second job, dual jobholders should be less engaged than single jobholders in their primary job. Therefore, single jobholders are also likely to devote more of their efforts to help other employees and focus on fulfilling their roles as an employee. Therefore, we predict that dual jobholders will engage in fewer organizational citizenship behaviors (OCB) and receive lower job performance ratings than single jobholders. We offer the following hypotheses:

  • Hypothesis 2. Within the same job, dual jobholders report lower levels of work engagement than single jobholders.

  • Hypothesis 3. Within the same job, dual jobholders exhibit less organizational citizenship behaviors than single jobholders.

  • Hypothesis 4. Within the same job, dual jobholders receive lower job performance ratings than single jobholders.

A pressing concern for dual jobholders is likely to be the balance between their work lives and their family lives. Indeed, dual jobholders work an average of 46.8 h per week whereas the average worker in the USA works 38.6 h per week (Hipple, 2010). By working longer hours than single jobholders, dual jobholders may be putting themselves at risk as longer working hours have been shown to be associated with higher work interference with family (Brett & Stroh, 2003; Major, Klein, & Ehrhart, 2002). Although longer working hours do not necessarily hinder family relationships for everyone, dual jobholders and single jobholders may experience work and life differently due to the added demands of a second job. Antecedents to work-family conflict include involvement in work and time demands (Carlson & Kacmar, 2000). Because dual jobholders experience more work hours (U.S. Bureau of Labor Statistics, 2016; Hipple, 2010), increased responsibilities, continual balance of schedules, and must switch among domains constantly, it is likely their involvement with and time devoted to work, as a whole, is higher than that of single jobholders. As more time and energy is devoted toward work, as a whole, less time and energy should remain to devote to one’s family life. The susceptibility for work to conflict with one’s family life should be higher for dual jobholders than for single jobholders. Thus, we offer the following hypothesis:

  • Hypothesis 5: Dual jobholders report higher work-family conflict than single jobholders.

We compare differences between dual jobholders and single jobholders with regard to work engagement, organizational citizenship behaviors (OCBs), job performance, and work-family conflict. We selected these particular variables to examine because they are important and common workplace constructs, they are manifestations of the suggested problems associated with dual jobholding, and they represent the tenets of partial inclusion theory. The notion that the second job interferes with individual productivity (Hirschman, 2000) and that holding two jobs may leave individuals tired and devoid of energy (Mattis, 2008) maps onto work engagement, OCB, and job performance. The notion that longer working hours may hinder family and social relationships (Brett & Stroh, 2003) is tested by examining work-family conflict. To our knowledge, these variables have not been examined in regard to dual jobholders (Zickar et al., 2004 examined some workplace constructs in regard to work-family conflict among dual jobholders). Thus, we believe the inclusion of these variables makes an important contribution to the literature in that we extend knowledge of workplace experiences related to dual jobholders. We test hypothesis 1 in study 1. This sample is comprised of dual jobholders from an array of organizations and industries. We test hypotheses 2–5 in study 2. Two samples are used in study 2: a teacher sample and a bar employee sample consisting of dual jobholders and single jobholders.

Method

Study 1

Procedures

Working adults employed in multiple organizations and industries were recruited by undergraduate business students at a public, Midwestern university to participate in the study. Students were asked to recruit one-to-two working adults known to work two jobs in exchange for extra class credit. Students forwarded an internet survey link to the working adults to complete the study. This procedure for data collection has been used in several studies (Grant & Mayer, 2009; Greenbaum, Mawritz, & Eissa, 2012; Morgeson & Humphrey, 2006) and allows for examination of individuals in a variety of organizations and industries to enhance generalizability. Participants completed two surveys: a “time 1” survey and a “time 2” survey. During the time 1 survey, the participant provided demographic information, an email address, and information related to the primary job. Approximately 1 week later, we emailed all participants the time 2 survey, which contained information related to the second job. We administered two surveys spaced 1 week apart to help mitigate concerns of same source bias (Podsakoff, MacKenzie, Lee, & Podsakoff, 2003). Before providing demographic information on the sample, it is important to first clarify how we categorized participants as a dual jobholder.

Classifying a Dual Jobholder

To be classified as a dual jobholder, the participant first had to answer “yes” to the question, “Last month, did you have more than one job (or business), including part-time, evening, or weekend work?” If the participant answered “no” to this question, the participant was classified as a single jobholder and was not included in the analysis. If the participant answered “yes” to the preceding question, the participant had to indicate that his or her employment arrangement met one of two scenarios: (a) receive a wage from two or more employers or (b) receive a wage from one employer and be self-employed in another job. If the participant selected a third scenario, (c) receive a wage from one employer, the participant was not classified as a dual jobholder but rather a single jobholder and was not used for analysis. By specifying that a wage must be received from an employer (or from self-employment), we eliminate volunteering work. And, by specifying the time component (i.e., “last month”), we ensure the individual currently, at the present time, indeed works two jobs.

We also collected additional data to verify the accuracy of our operational definition of a dual jobholder. Specifically, we asked participants classified as a dual jobholder to list the titles for their two jobs and the average number of hours worked per week at each job. We presented the definition of a primary job and a second job to participants, each of which was defined by the number of hours worked per week. In an open-ended question, the participant provided the job title of the primary job and the job title of the second job. Obtaining the job title was a second validity check to ensure that the individual was a dual jobholder. For instance, if a participant labeled “parent” as a second job the individual was eliminated from consideration as having parent as one’s second job does not align with the definition of a dual jobholder. To verify that each participant provided the correct primary job title and second job title, we asked the participant to then provide the number of hours worked per week at each job. Thus, we cross-checked the job titles with the number of hours to verify that participants understood which job was the primary job and which job was the second job. For the constructs of interest, the computer displayed the participant’s open-ended response to the job title question above each scale to which the participant should be referring. For example, if a participant typed “teacher” in the open-ended response to the primary job title, the instruction stem for each primary job scale stated: “The following questions should be answered with respect to your PRIMARY job (teacher).” We used this approach to ensure that the participant clearly understood which job was the referent for each scale.

Participants

A total of 236 individuals completed the time 1 survey. After omitting participants who indicated they received a wage from only one employer, omitting participants who indicated they worked more hours at the second job, and attrition from the time 1 survey to the time 2 survey, we had 128 usable responses (81 female, 46 male, 1 not provided) for a response rate of 54%. Primary job titles (e.g., claims adjuster, network analyst, speech language pathologist) and second job titles (e.g., flight instructor, pastor, physical therapy assistant) varied.

Measure

Work Engagement

An 18-item measure developed by Rich et al. (2010) was used to measure work engagement. Work engagement at the primary job was collected during the time 1 survey and work engagement at the second job was collected during the time 2 survey. Participants indicated how often they feel a certain way at work on a 7-point scale (1 = never; 7 = always or every day). An example item from this scale is “I exert a lot of energy on my job.” The measure was scored by obtaining the mean rating of the items. Cronbach’s alpha (α) for work engagement scores at the primary job was 0.97 and Cronbach’s alpha (α) for work engagement scores at the second job was 0.97.

Control Variables

In accordance with Becker’s (2005) recommendations on inclusion of control variables, we were careful to control for variables that may be a priori theoretically related to work engagement. We controlled for the self-reported number of hours worked per week at each job as hours worked at each job may affect levels of work engagement exhibited. We also controlled for the annual income received from each job as pay received may affect one’s proclivity to be more or less engaged at work. Income was treated a nominal variable ranging from “$500 or less” to “$10,000” in $1000 increments. After “$10,000,” the range extended to “$500,000 or more” in $10,000 increments.

Results

In addition to administering two surveys spaced 1 week apart to help mitigate concerns of same source bias (Podsakoff et al., 2003), we also conducted a Harman’s one factor test to examine concerns of common method issues. The un-rotated principal component factor analysis on the primary job work engagement and second job work engagement items showed the presence of four eigenvalues greater than one: 15.11, 10.04, 1.55, 1.27. The four factors produced a cumulative variance of 77.68% of the total variance. The largest factor did not account for a majority of the variance (41.96%). Given four eigenvalues greater than one, no factor accounting for a majority of the variance, the low correlation between the primary job work engagement and second job work engagement composites (r = 0.21, p < .05), and the separation of at least 1 week between data collection as recommended by Podsakoff et al. (2003), we believe common method issues and same source bias concerns are minimal.

The means, standard deviations, and intercorrelations are presented in Table 1. Hypothesis 1 predicted dual jobholders would report lower work engagement at the second job than the primary job. We tested hypothesis 1 using a repeated measures analysis of covariance (ANCOVA). The ANCOVA tested for a statistically significant difference between primary job work engagement and second job work engagement. The control variables were primary job hours, second job hours, primary job income, and second job income.

Table 1 Means, standard deviations, and intercorrelations of variables in study 1
Full size table

Results from the repeated measures ANCOVA (see Table 2) suggest that dual jobholders do not report lower levels of work engagement at the second job (M = 6.04, SD = 0.82) compared to the primary job (M = 5.97, SD = 1.00), F(1, 123) = 2.67, p = ns, η2 = 0.02. A pairwise comparison of the mean difference revealed the difference between primary job work engagement and second job work engagement was not statistically significant (p = .50, lower 95% confidence interval = −.13, upper 95% confidence interval = .27). Thus, hypothesis 1 is not supported in that dual jobholders did not differ in terms of their reported work engagement at the second job and primary job.

Table 2 Repeated measures analysis of covariance results
Full size table

In addition to our repeated measures ANCOVA, we attempted to uncover any hidden patterns in the relationship between engagement in the primary job and engagement in the second job. First, we computed a difference score of primary job work engagement and second job work engagement and correlated it with variables that may be related to differences in engagement between the two jobs—how long they have been a dual jobholder; hours per week at the primary job and the second job; annual income at the primary job and the second job; the extent to which the jobs are in the same job family; gender and age. We chose these variables in accordance with Becker’s (2005) recommendations on inclusion of control variables in that we were careful to control for variables that may be a priori theoretically related to work engagement. The only variable significantly correlated with the difference score was age (r = .18; p < .05). Thus, older participants were more likely to experience a larger difference between engagement at the primary job and second job. All other correlations were below .10. Therefore, age can explain some of the difference between the rank orders of the primary job and second job engagement scores. We examined the partial correlation of primary job work engagement and second job work engagement partialling out age (r = 0.21; p < .05), which is identical to the bivariate correlation of primary job work engagement and second job work engagement (r = 0.21; p < .05). Thus, the impact of age on differences between engagement scores at jobs is minimal.

Discussion of Study 1

Study 1 provided a first glimpse into dual jobholding by demonstrating dual jobholders do not differ on levels of work engagement between the primary job and the second job. The high means on a 7-point scale suggest dual jobholders are not necessarily forced to prioritize one job at the expense of the other job. However, in study 2, we compare dual jobholders to single jobholders to better understand if holding a second job is detrimental to organizations and to the individual employee.

Method

Study 2

To test hypotheses 2–5, we collected two samples consisting of dual jobholders and single jobholders. In a “school sample” (N = 306), we examined hypotheses 2 and 5 by comparing 86 dual jobholders and 220 single jobholders working in three school districts in three different states across the Midwest. In the school sample, all 86 dual jobholders indicated their primary job was at the school district. This sample allows for a comparison of dual jobholders and single jobholders at a job in which all individuals are working in their primary job. In a “bar sample” (N = 77), we examined hypotheses 3–5 by comparing 31 dual jobholders and 46 single jobholders working in 20 bars in three different states in the Midwest. In the bar sample, all but three of the dual jobholders indicated work performed in the bar was their second job rather than their primary job. Thus, the bar sample allows for a comparison of dual jobholders and single jobholders at a job in which nearly all of the dual jobholders are at their second job and the single jobholders are at their primary job. The results of the study do not change regardless of whether the three dual jobholders who indicated the bar was their primary job are included or excluded from the analysis. Holding constant the job was a methodological control we thought was important to make meaningful comparisons across populations of single jobholders and dual jobholders. Furthermore, we investigated our hypotheses at both the primary job (teachers) and the second job (bartenders) in separate samples, which allowed us to investigate our hypotheses from the perspective of both the primary job and the second job.

The school districts would not allow us to collect performance data or allow us to collect coworker ratings of workplace constructs so we collected self-report data in the teacher sample (e.g., work engagement and work-family conflict) and other-rated performance data in the bar sample (e.g., OCB and job performance), along with self-ratings of work-family conflict. Therefore, we answered research questions which contained constructs traditionally measured via self-report in the teacher sample and answered research questions which contained constructs traditionally measured via other ratings in the bar sample. In so doing, we provide a more holistic examination of the comparison between dual jobholders and single jobholders by using both self-report and other-rated data and collecting samples from two different occupations.

Classifying a Dual Jobholder and a Single Jobholder

We classified dual jobholders and single jobholders using the same manner described in study 1. We again collected data related to job title and number of hours worked per week to verify the accuracy of our operational definition of a dual jobholder.

School Sample Participants and Procedures

The United States Bureau of Labor Statistics found that teaching is an occupation that contains high rates of dual jobholding (Hipple, 2010). Thus, we sent emails to all employees of three school districts in three different states in the Midwest (a total of 2845 emails) with an introduction to the survey and internet survey link. We had permission to survey employees of each school district; however, school administrators did not alert school employees we would be sending emails inviting participation in our study. Each school district contained elementary schools, middle schools, and high schools. In exchange for their voluntary participation, participants were offered the opportunity to win a $101 prize in a drawing. Three hundred ninety-seven employees at least started to take the survey. After applying our criteria for classifying dual jobholders and single jobholders, using listwise deletion (i.e., each scale measure had to be completed to form a composite score), and accounting for individuals who began the survey but only answered one-to-two questions and withdrew, we had a final usable sample of 306 (86 dual jobholders and 220 single jobholders) employees for a response rate of 11%. The rate of dual jobholders to single jobholders in our sample (28%) is in line with previous studies, which found dual jobholding rates for teachers at 33% (Wisniewski & Hilty, 1987) and 34% (Betts, 2004). Primary job titles (e.g., teacher, administrative, counselor, principal) and second job titles (e.g., cashier, mortgage broker, retail sales) varied.

Bar Sample Participants and Procedures

Other occupations that afford individuals the ability to hold two jobs are those found in bars (e.g., bartender, server, bouncer), likely because of the late hours of operation that would not interfere with a day job. Thus, we sought to obtain data from employees of bars in three different states. The first author and two research assistants entered the bars in the early evening; approximately 8:00 p.m. This time was chosen as it allowed the researcher to approach bar employees while on the job but not during peak business hours. The researcher first asked to speak with the manager of the bar to obtain permission to administer paper surveys to any employees currently in the bar. None of the bar managers we approached denied us permission to survey employees. All of the employees we asked to participate freely volunteered. Thus, our response rate in the bar sample was 100%. We obtained responses from 77 individuals (31 dual jobholders and 46 single jobholders) working in 20 bars. Employees of the bars were offered the chance to win a $101 cash prize in exchange for their participation.

We asked each employee to complete a paper survey individually. Further, we asked the employee to recruit a coworker from the bar to complete a coworker survey. The coworker survey measured the target employee’s OCB and job performance in his or her job at the bar. This process of asking an employee to select a coworker to complete a survey to obtain other-rated data has been used by a number of researchers (e.g., Greenbaum et al., 2012; Mawritz, Mayer, Hoobler, Wayne, & Marinova, 2012) and helps mitigate concerns of same source bias (Podsakoff et al., 2003). The coworker completed the coworker survey at the same time in which the employee completed the employee survey. All of the employee participants and coworker participants completed the survey during, shortly before, or shortly after their work shift at the bar. Both the employee and the coworker individually returned the survey to the researcher after completion. Because the majority of these data were collected while participants were working on the job and during hours in which the bar was open for business, our intention was to keep the survey as short as possible.

Measure Administered Only to the School Sample

Work engagement was administered to both dual jobholders and single jobholders of the school sample. The work engagement scale was measured with respect to the individual’s job in the school district. This was important as we wanted to compare the single jobholder’s level of work engagement in the teaching job to the dual jobholder’s level of work engagement in the teaching job so that the comparison would take place at the job in which the organizational setting was held constant. In other words, we did not want to confound a single jobholder’s level of work engagement as a teacher to a dual jobholder’s level of work engagement as a mortgage broker (in their second job), for example.

Work Engagement

We used a 9-item measure developed by Schaufeli, Bakker, and Salanova (2006) to measure work engagement. Participants indicated how often they feel a certain way at work on a 7-point scale (1 = never; 7 = always or every day). Dual jobholders and single jobholders responded to this scale with respect to their job in the school district. An example item from this scale is “At my work, I feel bursting with energy.” The measure was scored by obtaining the mean rating of the items. Cronbach’s alpha (α) for work engagement scores was .91.

Measures Administered Only to the Bar Sample

Organizational citizenship behavior and job performance scales were administered to coworkers of both dual jobholders and single jobholders of the bar sample. Each scale was measured with respect to the job at the bar so as to ensure the comparison would take place at the job in which the organizational setting was held constant.

Organizational Citizenship Behavior

We used a 16-item measure developed by Lee and Allen (2002). Eight items consisted of OCB directed toward colleagues (OCB-I) and eight items consisted of OCB directed toward the organization (OCB-O). The coworkers of dual jobholders and single jobholders in the bar sample completed the OCB scale by indicating the frequency with which the employee displays the behavior portrayed by the items on a 7-point scale (1 = never; 7 = always or every day). Example items from this scale are “Willingly gives his/her time to help others who have work-related problems” and “Demonstrates concern about the image of the organization.” The measure was scored by obtaining the mean rating of the items. Cronbach’s alpha (α) for the OCB scores was .95. The results of the study do not differ whether OCB-I and OCB-O are tested separately or together. Therefore, we combined OCB-I and OCB-O into one, “OCB” variable (e.g., LePine, Erez, & Johnson, 2002).

Job Performance

Job performance was measured using a 7-item scale developed by Williams and Anderson (1991). The coworkers of dual jobholders and single jobholders in the bar sample indicated how much they agreed with statements on a 7-point scale (1 = strongly disagree; 7 = strongly agree), with respect to the employee. An example item from this scale is “This individual adequately completes assigned duties.” The measure was scored by obtaining the mean rating of the items. Cronbach’s alpha (α) for the job performance scores was .97.

Measures Administered to Both the School Sample and Bar Sample

Work-Family Conflict

Work-family conflict was measured using a 5-item measure developed by Netemeyer, Boles, and McMurrian (1996). Participants in both the school sample and the bar sample self-reported their levels of work-family conflict. Participants were asked to rate the extent to which they agreed with the statements on a 7-point scale (1 = strongly disagree; 7 = strongly agree). An example item from this scale is “The demands of (both) my job(s) interferes with my home family life.” The measure was scored by obtaining the mean rating of the items. Cronbach’s alpha (α) for work-family conflict scores was .94 in the school sample and .90 in the bar sample.

Control Variables Used in the School Sample and Bar Sample

We were careful to control for variables that may be a priori theoretically related to work engagement, OCB, job performance, and work-family conflict (Becker, 2005). We controlled for the amount of time the dual jobholder has been a dual jobholder and the amount of time the single jobholder has been a single jobholder (i.e., “dual jobholder/single jobholder tenure” variable). Although we may be statistically controlling for relevant variance in the dependent variable, we considered this to be a strict test for answering our research question. We asked for the participants’ organizational tenure, in years, at both the primary job and the second job. Whichever value was smallest was used as the amount of time the individual has been simultaneously employed in both jobs. Although it is possible the individual has been a dual jobholder for a longer period of time, this measure allowed us to capture how long the individual was a dual jobholder in the two specific jobs for which he or she is currently working. Thus, the dual jobholder tenure variable captures how long each individual has had to balance responsibilities and adjust to their specific jobs and holds constant the time frame for operationally defining dual jobholder tenure. To create this variable for single jobholders, we used their reported organizational tenure at their current job. We also controlled for the self-reported number of hours worked per week at the job in which the comparison took place.

In addition, we adhered to methods used by Zickar et al. (2004) by controlling for negative affect. Negative affect is an individual characteristic present in both the primary job and the second job and should have an influence on reported work attitudes in both jobs.

Negative Affect

We used the 5-item version of the negative affect schedule developed by `Mackinnon et al. (1999). Participants were asked to rate how often they felt a certain way, in general, over the past few weeks (1 = never; 7 = always or every day). Example items include “upset” and “scared.” The measure was scored by obtaining the mean rating of the items. Cronbach’s alpha (α) for negative affect scores in the school sample was .81 and Cronbach’s alpha (α) for negative affect scores in the bar sample was .81.

We controlled for the annual income received at the job at which the comparison took place in the same manner used in study 1.

Results

Measurement Equivalence

To make meaningful group comparisons, we first established the measurement equivalence of the self-reported dependent variables across dual jobholders and single jobholders. The most widely used method for establishing measurement equivalence is metric invariance whereby we tested the equivalence of factor loadings. We compared a baseline model where the factor loadings, intercepts, and uniquenesses were freely estimated in each group (i.e., configural model) to a model where the factor loadings were constrained to be equal across the two groups (i.e., metric model). In the school sample, a comparison of the chi-square values for work engagement and work-family conflict (Δχ2 = 10.29, ns; Δdf = 12) showed no significant differences. In the bar sample, a comparison of the chi-square values for work-family conflict (Δχ2 = 0.38, ns; Δdf = 4) showed no significant difference. We did not submit OCB and job performance from the bar sample to a metric invariance test as these measures were reported by a single group—coworkers.

The means, standard deviations, and intercorrelations for the variables of both the school sample and the bar sample are presented in Tables 3 and 4, respectively. Analysis of covariance (ANCOVA) was used to test hypotheses 2–5. The independent variable was a categorical variable (i.e., single jobholder or dual jobholder) and the control variables were dual jobholder/single jobholder tenure, hours worked at the comparison job, negative affect, and income from the comparison job.

Table 3 Means, standard deviations, and intercorrelations of study 2: school sample
Full size table
Table 4 Means, standard deviations, and intercorrelations of study 2: bar sample
Full size table

Hypothesis 2 predicted dual jobholders report lower levels of work engagement than single jobholders. Results from the school sample suggest that dual jobholders (M = 5.39, SD = 0.88) do not report lower levels of work engagement than single jobholders (M = 5.38, SD = 0.80), d = 0.01, F(1, 300) = 0.74, ns. Thus, hypothesis 2 is not supported. Hypothesis 3 stated dual jobholders engage in lower levels of OCB than single jobholders. Results from the bar sample suggest dual jobholders (M = 5.99, SD = 0.74) do not engage in lower levels of OCB than single jobholders (M = 5.96, SD = 1.00), d = 0.03, F(1, 71) = 0.68, ns. Thus, hypothesis 3 is not supported. Hypothesis 4 predicted dual jobholders receive lower job performance ratings than single jobholders. Results from the bar sample showed that dual jobholders (M = 6.18, SD = 1.04) did not receive lower job performance ratings than single jobholders (M = 6.33, SD = 0.86), d = − 0.16, F(1, 71) = 0.78, ns. Thus, we do not find support for hypothesis 4.

Hypothesis 5 predicted dual jobholders would report higher levels of work-family conflict than single jobholders. We tested this hypothesis in both the school sample and the bar sample. In the school sample, results revealed that dual jobholders (M = 3.83, SD = 1.46) reported higher levels of work-family conflict than single jobholders (M = 3.45, SD = 1.62), d = 0.25, F(1, 300) = 4.78, p < .05. In the bar sample, dual jobholders (M = 4.23, SD = 1.47) also reported higher levels of work-family conflict than single jobholders (M = 2.98, SD = 1.32), d = 0.89, F(1, 71) = 35.49, p < .05. Thus, hypothesis 5 was supported in both samples and the effect sizes were moderate-to-large.

Post hoc Analysis

An important variable affecting the degree to which dual jobholders are engaged in their work may be the motivation for holding a second job. Although the primary focus of study 2 was to compare dual jobholders to single jobholders, we felt a worthwhile follow-up of study 1 would be to include a measure of the motive for undertaking the second job. Indeed, the research question for study 1 attempted to understand if one job interferes with the other job in the form of work engagement. Thus, we felt it prudent to collect information on motives for holding a second job in the teacher sample as motives may shed further light on the experiences of dual jobholders.

The economics literature has suggested two primary motives to engage in dual jobholding: to earn more money or to gain something non-pecuniary at the second job (Conway & Kimmel, 1998; Kimmel & Conway, 2001; Paxson & Sicherman, 1996). The extent to which an individual holds a second job simply to increase income or the extent to which an individual holds a second job to fulfill a passion, for enjoyment, or to gain experience, may have an important impact on individual work experiences. The Bureau of Labor Statistics collects data on motives for holding a second job using a 7-item scale (see Appendix). Items 1, 2, 5, and 7 capture financial motives (i.e., finances scale) for holding a second job and items 3, 4, and 6 capture non-pecuniary reasons for holding a second job (i.e., non-finances scale). This approach for considering these items as “finances” and “non-finances” has been used in the economics literature (e.g., Kimmel & Powell, 2001). Although the items were not originally developed for academic research, we did find it interesting to report some initial post hoc results related to motives of dual jobholders. We included this 7-item scale in the teacher sample to gain some glimpse into how motives affect dual jobholders’ experiences at work. We were unable to collect the motives data in the bar sample as our intention was to keep the survey as short as possible given that the participants were working on the job and during hours in which the bar was open for business when they completed the survey. Our intention in collecting the motives data was to shed additional light on results from study 1 by considering the motives for holding the second job. Thus, we used the dual jobholders in the teacher sample (N = 86) to examine the correlations of the finances and non-finances scales with primary job work engagement and second job work engagement (we asked the dual jobholders in the teacher sample to provide work engagement information for both of their jobs). Results show the finances scale was not related to primary job engagement (r = .04, p = ns) or with second job engagement (r = − .13, p = ns). The non-finances scale was positively correlated with primary job engagement (r = .20, p < .10) and positively correlated with second job engagement (r = .43, p < .05). Although correlational, these post hoc analyses suggest non-financial reasons for taking a job are significantly and positively associated with work engagement at both the primary job and the second job.

Discussion of Study 2

Study 2 provided insight into whether holding a second job is detrimental by comparing dual jobholders to single jobholders. The pattern of results was quite interesting in that work engagement, OCB, and job performance did not seem to suffer as a result of working two jobs, but work-family conflict did increase as a result of working two jobs. Two samples, one in which dual jobholders and single jobholders were compared at the primary job and one in which dual jobholders and single jobholders were compared at the second job, provided some evidence that dual jobholders exhibit few differences compared to single jobholders on several important work-related variables. Contrary to conventional wisdom and the popular press, dual jobholders do not seem to suffer, compared to single jobholders with regard to work-related experiences. Although dual jobholders do not appear to be hurting the organizations in which they work, in the form of lower work engagement, lower OCB, and lower job performance, they may instead be hurting their lives outside of work. In both samples, dual jobholders reported experiencing significantly higher levels of work-family conflict than single jobholders. Thus, the present study presents a novel and relevant empirical finding regarding the experiences of dual jobholders.

Discussion

The present study provides an early and important step to studying individuals who are employed in two jobs. In study 1, we failed to show that dual jobholders prioritize one job over the other job, in terms of work engagement, as levels of work engagement at the primary job and second job did not significantly differ. To provide a more holistic examination of dual jobholders, study 2 showed dual jobholders and single jobholders exhibit few meaningful differences in the workplace. However, two samples replicated the finding that dual jobholders report higher levels of work-family conflict than single jobholders.

We highlight that findings of few significant differences between a dual jobholder at the primary job and second job, and between single jobholders and dual jobholders, are important and interesting (e.g., Landis, James, Lance, Pierce, & Rogelberg, 2014). Conventional wisdom says that holding two jobs is problematic for several reasons but there is no empirical data to support this position. Given an organization’s proclivity to enact policies preventing dual jobholding (Hirschman, 2000), it is important to understand if these policies are justified from an evidenced-based perspective—it is important that empirical data drive such organizational policies rather than conventional wisdom. We therefore view both studies as making an important and novel empirical contribution to the organizational behavior and human resource management literatures.

The pattern of results in the present study is quite interesting, considering the general negative perceptions of dual jobholding. Nonetheless, we are confident in the story conveyed by our results. First, the pattern of results held across multiple samples and industries (i.e., school and bars) and multiple sources of data (i.e., self and coworkers). Second, the effect sizes of the null hypotheses (i.e., 0.00 ≤ d ≤ 0.16) were consistent and small according to Cohen’s (1988) standards. Landis et al. (2014) and many others have warned against bias of statistically significant findings because finding no differences can be quite meaningful. One valuable example they provide is when a pattern of relationships between meaningful variables are not significantly different between two groups (e.g., men and women). For example, the Roth, Purvis, and Bobko (2012) finding that gender differences in supervisors’ job performance ratings are small (i.e., between men and women) is a noteworthy and desired finding. We should note that we cannot unequivocally commit to there being no significant differences between people who work one job and those who do not (e.g., Edwards, 2008; Kluger & Tikochinsky, 2001). We do, however, draw attention to a lack of evidence for differences among a variety of workplace variables and workplace settings.

Research Implications

An important implication of the present study is that we observed only small-magnitude differences between dual jobholders and single jobholders for the work-related constructs—work engagement, organizational citizenship behaviors, and job performance. Our findings are in contrast to the stereotype that working two jobs interferes with performance at one or both of the jobs. Thus, we contributed to a growing body of work on alternative employment arrangements. Indeed, recent calls within the management literature emphasize the need to examine alternative, or non-traditional, forms of employment as “alternatives to full-time regular employment are now so prominent that all research based on the workplace needs to consider them” (Cappelli & Keller, 2013, p. 593). Thus, the present study may relate to previous work on contingent, contract, and even temporary employees.

The present study offers a novel context in which to study employees in organizations. Although the economics literature has devoted some research to dual jobholding (e.g., Kimmel & Conway, 2001; Paxson & Sicherman, 1996), the present study was the first to put forth a formal classification of when an individual is classified as a dual jobholder and examine individual, psychological constructs related to holding two jobs. Thus, we have helped spur the formal study of dual jobholding. The present study may prompt researchers to consider the possibility that employees may have more than one job, more than one boss, and more than one set of colleagues that all may have a significant impact on relevant work experiences.

Drawing from partial inclusion theory, we predicted that dual jobholders would be more involved with and exhibit higher work engagement toward their primary job than their second job. Based on this same rationale, we predicted that compared to single jobholders, dual jobholders would report less work engagement, exhibit fewer OCB, receive lower job performance ratings, and report higher levels of work-family conflict. The pattern of results supported partial inclusion theory for work-family conflict but not for work engagement, OCB, and job performance. Thus, employees did not report being more strongly involved with one job over the other. Instead, we did find that holding two jobs negatively affected work-life balance.

Practical Implications

Approximately 7.2 million people, both men and women, hold more than one job in the USA and this number has been steady over the past two decades (Bureau of Labor Statistics, 2016). Thus, results of the study are likely to be pertinent to a wide range of individuals. Although the possible addition of a new job may be overwhelming to an individual, the results from the present study may prove encouraging to individuals worried about their ability to handle and balance job tasks at both jobs.

The present study provides the first scientific examination of the popular press notion (e.g., Hirschman, 2000) that holding two jobs is detrimental to individuals and organizations. Indeed, the present study shows that organizations may not need to enact policies to prevent individuals from undertaking a second job. However, given the negative, personal effects of holding two jobs (i.e., higher work-family conflict), organizations may be inclined to enact policies that help dual jobholders strike a healthy balance between work life and home life. Certain organizations (e.g., education, restaurant) may employ a high rate of dual jobholders. These organizations in particular may want to develop such policies and encourage managers to engage in an open dialog regarding the benefits and consequences of holding two jobs.

Limitations and Future Directions

Although the present study contains a number of strengths, there are also limitations. First, our samples in study 2 were constrained to two specific types of organizations (i.e., school district and bars). We chose to collect data from these two samples as each organizational setting was likely to consist of a significant number of dual jobholders. Because of this, the sample was not drawn using a random sampling methodology. Thus, the generalizability of the present study may not extend to other organizations and dual jobholding arrangements. However, we believe this limitation in study 2 is somewhat combatted by surveying dual jobholders from an array of organizations and industries in study 1. Nevertheless, future research could benefit from studying other occupations that employ high rates of dual jobholders such as firefighters, social workers, or dental hygienists (Hipple, 2010) to better understand the experiences this working population may encounter. In addition, the response rate for the teacher sample (11%) is lower than that obtained in other studies administering surveys by mail (i.e., 19%; Baruch, 1999). We believe the low response rate is because we sent the study via an email invitation and many of those potential participants may not have seen the invitation email (e.g., it went to spam). We computed our response rate against the total number of emails submitted (2845) which may not be an accurate denominator of the actual number of potential participants that made the conscious effort not to participate. Other evidence for our conclusion is that the response rate for the bar sample was 100% because the survey instruments were administered by the researchers face-to-face. Despite the low response rate in the teaching sample, the proportion of dual jobholders to single jobholders in the teaching sample (28.7%) was consistent with previous studies which found dual jobholding rates for teachers at 33% (Wisniewski & Hilty, 1987) and 34% (Betts, 2004). Therefore, we have reason to believe that a non-response bias in our sample did not affect the representativeness of the dual jobholding rate found in our sample.

Although meta-analytic evidence demonstrates that student-recruited samples do not differ from non-student-recruited samples (Wheeler, Shanine, Leon, & Whitman, 2014), some readers may have concerns with study 1 in that working adults were recruited by undergraduate students rather than the researchers themselves. We obtained no data to suggest that our student-recruited sample lacked representativeness of the working adult population of dual jobholders and indeed our theoretical rationale for study 1 was also supported in study 2 with researcher-recruited samples. Nevertheless, future research may benefit from sample recruitment similar to the ones we used in study 2 to verify the generalizability and stability of our effects.

In both studies, we were not able to pinpoint the specific time at which the primary job and second job were worked. We did not know if a dual jobholder worked the second job late at night or on weekends, nor did we know whether the primary job was a full-time, “traditional” 8 a.m. to 5 p.m. job. Thus, we were unable to detect the amount of time between the end of one’s primary job shift and the beginning of one’s second job shift (and vice versa). We could not capture how long the dual jobholder had to rest and recover after one job before starting work in the other job. The time gap between jobs may be important in determining how dual jobholders may differ from single jobholders on important workplace attitudes and behaviors. Future research may benefit from collecting data regarding the extent to which a dual jobholder has to rest between jobs or from an experience sampling methodology that allows researchers to investigate dual jobholders over the course of time to observe instances in which dual jobholders may exhibit differences from their single jobholding counterparts. Additionally, it would be informative to measure psychological processes (e.g., stress, emotions) associated with task switching and switching jobs. This may further elucidate potential differences between dual and single jobholders. Relatedly, a reviewer highlighted another interesting avenue for future research in that it would be useful to examine the exempt status of jobs with respect to the Fair Labor Standards Act. Exempt jobs often entail non-traditional work hours. Thus, the decision to enter into a second job when the primary job is exempt, and potentially requires non-traditional work hours, may alter the extent to which one job affects the other job. Indeed, it may be more manageable for a dual jobholder to balance work schedules and juggle job demands when at least one job occurs in a more routine and certain timeframe.

Conclusion

The present study responded to the call to continue research on alternative work arrangements and served to spur research on dual jobholders. In study 1, we showed that dual jobholders seem to be able to exhibit the same (and high) levels of work engagement at both the primary job and the second job. Across two samples in study 2, we found few differences between single jobholders and dual jobholders on a host of important work attitudes and behaviors. We believe both studies provide a novel and important empirical contribution to the literature. Although the present study represents an initial step toward the examination of dual jobholders, the unique circumstances they face in the workplace have yet to be fully understood. Due to the regularly occurring nature with which dual jobholding occurs, the present research is especially important for theory and practice.