Psychological Contracts of Multiple Jobholders: A Multilevel Analysis

Organizational research has paid little attention to multiple jobholders. This study investigated the psychological contracts of multiple jobholders in relation to each employment contract and the nonstandard work arrangement of multiple jobholding. The sample consisted of 141 multiple jobholders with two or more jobs and either employed or self-employed status. The results of the multilevel analyses indicated that, with one exception, psychological contract obligations vary more between employment relationships than between individuals. Multiple jobholders’ psychological contracts are only partly related to the terms of their employment contracts and their nonstandard work arrangement. This study connects mainly economic knowledge on multiple jobholders to the organizational variable psychological contracts. Investigating the psychological contracts within and between multiple jobholders helps researchers better understand psychological contracts and how they are related to the job. Typical patterns of multiple jobholders’ employment situation and psychological contracts can thus be identified.


Introduction
Multiple jobholding "is an important labor market phenomenon in most Western industrialized countries" (Dickey, Watson, & Zangelidis, 2011, p. 3767).It affects labor market estimates about part-time work because multiple jobholders work in two or more jobs (Bamberry & Campbell, 2012).In some European countries, the rate of multiple jobholders has reached nearly 10% (e.g., the United Kingdom, the Netherlands, Denmark, Switzerland; Andrews, Caldera Sánchez, & Johansson, 2011;Panos, Pouliakas, & Zangelidis, 2014;Swiss Federal Statistical Office, 2010), and it is 5% to 6% in the United States and Australia (Bamberry & Campbell, 2012; Bureau of Labor Statistics & U.S. Department of Labor, 2015;Hipple, 2010).Although economic studies have investigated the rate of multiple jobholders in the working population (Hipple, 2010) and their motivation to work in more than one job (Dickey et al., 2011), organizational research has paid little attention to multiple jobholders (Sliter & Boyd, 2014).Knowledge is thus limited about how multiple jobholders differ from workers with one job and how the jobs combine into one portfolio.
Multiple jobholders work in more than one employment relationship and develop psychological contracts in each of their employment relationships.Psychological contract theory states that these contracts develop through reciprocal inducements and contributions in the employment relationship and do not depend on employees' prior expectations (Rousseau, 1989(Rousseau, , 2001)).Although formed in the employment relationship, psychological contracts are perceived with subjectivity.These contracts are thus related to the employment relationship and to the individual, but current research has not attempted to assess the relative importance of both aspects.The relationship between psychological contracts and the employment situation has predominantly been regarded in terms of temporary or permanent contracts and full-time and part-time jobs (e.g., Gakovic & Tetrick, 2003;Isaksson, De Cuyper, Oettel, & De Witte, 2010;McDermott, Heffernan, & Beynon, 2013), without considering more complex, nonstandard forms of employment.Existing studies have compared psychological contracts between groups of employees (e.g., Conway & Briner, 2002;Isaksson et al., 2010), but particular types of employees hold particular employment contracts; for example, more women work part time, and temporary employees are more likely to be less educated (Organisation for Economic Co-operation and Development [OECD], 2014[OECD], , 2015)).Considering predictors from the employment situation of multiple jobholders and two of their jobs helps identify patterns in psychological SAGE Open contracts that are typical for this form of nonstandard employment.
In this study, the employment relationships of multiple jobholders were investigated, focusing on their psychological contracts in their two most important jobs.The sample of multiple jobholders provided the opportunity to contrast psychological contracts between and within employees, and thus the differences in psychological contracts between and within study participants could be disentangled.Multiple jobholders' psychological contracts were analyzed in relation to their employment situation in each job and to their nonstandard work arrangement of multiple jobs.Highly diverse work arrangements and particular demographic patterns have been found among multiple jobholders (Panos et al., 2014); thus, psychological contracts might differ between multiple jobholders (Figure 1).
This study contributes to the existing research in three aspects.First, it reviews the role of psychological contracts in the employment relationship by investigating them within and between multiple jobholders, thus improving the understanding of psychological contracts and their association with jobs and individuals.Second, it provides empirical evidence on the employment relationships of multiple jobholders and on how these individuals combine their jobs into one portfolio.It connects existing, mainly economic, knowledge on multiple jobholders to an organizational variable, namely, psychological contracts.It thus adds to the yet scarce research on multiple jobholders and their nonstandard work arrangements.Third, it compares existing research on the relationship between employment contracts and psychological contracts among employees with one job to typical employment relationships among employees holding multiple jobs.Finally, this study provides the first evidence on psychological contracts of multiple jobholders.

Multiple Jobholders and Their Psychological Contracts
Research on multiple jobholding has typically focused on the multiple jobholder or the work situation without distinguishing between the jobs (e.g., Brown & Gold, 2007;Guest, Oakley, Clinton, & Budjanovcanin, 2006;Jamal, 1986;Jamal, Baba, & Rivière, 1998;Sliter & Boyd, 2014;Totterdell, Wood, & Wall, 2006), but few studies have addressed differences between the jobs of multiple jobholders and related attitudes (Sliter & Boyd, 2014).Zickar, Gibby, and Jenny (2004) reported little consistency in job attitudes across two jobs, particularly when jobs were different.Multiple jobholders were more stressed and demonstrated more continuance commitment in their primary job but experienced similar levels of affective commitment and job satisfaction in both jobs.The authors concluded that considering the reasons for multiple jobholding when assessing job attitudes could improve the understanding of differences between jobs.Schaubroeck, Judge, and Taylor (1998) found that a full-time and an additional part-time job differed in contextual factors, commitment, interpersonal stressors, and quantitative workload.They argued that the two jobs were assessed differently based on the work situation, while the disposition of the multiple jobholders was less influential.Both studies indicated that multiple jobholders perceived their jobs in relation to their employment situation in each job, while their attitude toward multiple jobholding was of less importance.
It can hence be expected that the employment relationships of multiple jobholders differ from one another because they are perceived based on the work situation in each job.Psychological contracts reflect these differences in employment relationships because they are specific to each job.The psychological contract consists of mutual promises, obligations, and expectations between the employee and the employer and is not part of the legal employment contract (Rousseau, 1989(Rousseau, , 2001)).Psychological contract theory emphasizes the assumption that the psychological contract is associated with the employment relationship and that its perception is based on promises made by both parties in the employment relationship.The psychological contract is not supposed to reflect employees' expectations prior to employment.Although this contract originates in the employment relationship, subjectivity plays a role in how the contract is perceived by the employee.Subjectivity thus introduces an individual component that reaches beyond each employment relationship.Psychological contract perceptions thus vary between multiple jobholders to the degree that their subjective view is relevant.
The psychological contract of multiple jobholders should hence differ between jobs and relate to the particular employment relationship.Consequently, the psychological contracts of multiple jobholders are expected to differ according to the employer, although they are perceived by one multiple jobholder.Psychological contracts should further differ between multiple jobholders, thus representing their individual, subjective component.It is important to know to what degree psychological contracts differ between employment relationships and between employees, that is, multiple jobholders.Such information allows researchers to determine what portion of variance can be explained through predictors related to each employment relationship or to the individual.This issue is explored with the following research question:

Research Question 1:
To what degree do psychological contract obligations vary within or between multiple jobholders?

Employment Contracts
The nonstandard work arrangements of multiple jobholders are highly diverse, but they are still distinct from the employment situation of permanent or temporary and full-time or part-time employees with one job.Some multiple jobholders work full time with an additional part-time job; others combine two part-time jobs, perform self-employed work for several customers, or are employed in a sequence of temporary jobs (Hipple, 2010;Swiss Federal Statistical Office, 2010).Although multiple jobholders in the United States are more likely to combine a full-time job with a part-time job, multiple jobholders in Europe more often work in two parttime jobs.Multiple jobholders can occupy managerial and nonmanagerial positions and jobs in different occupations, and also occupations they have not been trained or educated for (Panos et al., 2014;Sliter & Boyd, 2014;Wu, Baimbridge, & Zhu, 2009).Relating employment contracts and the nonstandard work arrangement of multiple jobholders to their psychological contracts contributes to identifying whether their psychological contracts reflect typical patterns of their employment.
Previous research has shown that for employees with one job, psychological contracts depend on employment contracts and positions employees are hired for.Psychological contracts were found to vary between employees with permanent and temporary contracts, between employees with full-time and part-time employment, between employees in managerial and nonmanagerial positions and according to the qualifications for the job (Conway & Briner, 2002;Coyle-Shapiro & Kessler, 2000;Luksyte, Maynard, & Spitzmueller, 2011;Van Dyne & Ang, 1998).These studies compared groups of employees with different employment contracts to find differences between their psychological contracts.The types of employment contracts can, however, vary between groups of employees; for example, more women are found in part-time jobs, and temporary employees are more likely to be less educated (OECD, 2014(OECD, , 2015)).Because multiple jobholders have one psychological contract in each employment relationship, differences between multiple jobholders and within the employment portfolio of multiple jobholders can be investigated simultaneously.
Existing studies have used various terms for psychological contracts, such as fulfillment or breach, obligations and expectations, or transactional and relational contracts (Conway & Briner, 2002;Millward & Brewerton, 1999;Tekleab & Taylor, 2003;Van Dyne & Ang, 1998).Fulfillment and breach refer to the quality of the employment relationship, denoting whether obligations or promises are fulfilled (Conway & Briner, 2002;McDermott et al., 2013).Investigating the level of obligations or expectations toward the employer or the employee aims to determine how broad or narrow psychological contract obligations are (Coyle-Shapiro & Kessler, 2002;Van Dyne & Ang, 1998).When obligations are split up into dimensions of psychological contract contents, often transactional and relational contracts are distinguished that cover economic or socioemotional aspects of psychological contract contents (Chambel & Alcover, 2011;Millward & Brewerton, 1999).Some researchers have used more than two dimensions of psychological contract contents to provide more differentiation particularly in relational aspects of psychological contracts (Blomme, van Rheede, & Tromp, 2010;Dabos & Rousseau, 2013;De Vos, Buyens, & Schalk, 2003;Freese, Schalk, & Croon, 2011;Lub, Bijvank, Bal, Blomme, & Schalk, 2012;Raeder, Wittekind, Inauen, & Grote, 2009).Such differentiation is conducive to research that places its main emphasis on psychological contracts.In line with this body of research, this study investigates dimensions of employer obligations and multiple jobholder obligations.This literature review thus aims to interpret empirical evidence with a focus on obligations but includes studies with other terms if they support the development of the hypotheses.
Empirical research comparing permanent employees and temporary workers has noted that permanent employees report higher obligations (Coyle-Shapiro & Kessler, 2002;McDermott et al., 2013;Van Dyne & Ang, 1998) and more relational psychological contracts with a socioemotional, long-term exchange relationship (Chambel & Alcover, 2011;De Cuyper & De Witte, 2006;Isaksson et al., 2010;Millward & Brewerton, 1999).Only one investigation failed to find significant differences in obligations between permanent and temporary employees (McDonald & Makin, 2000).Permanent employees reported an overall higher fulfillment of obligations (McDermott et al., 2013;Scheel, Rigotti, & Mohr, 2013) but a lower fulfillment of relational obligations (De Cuyper & De Witte, 2006;Isaksson et al., 2010).Temporary employees were more likely to perceive transactional obligations-with a short-term, economic focus-as fulfilled.Overall, a permanent employment contract was associated with higher obligations, a more relational contract and higher fulfillment of obligations.
Comparing psychological contracts of full-time and parttime employees yielded fewer significant differences than were found for permanent and temporary employees (Conway & Briner, 2002;Gakovic & Tetrick, 2003).In contrast to part-time employees, full-time employees were more likely to report higher obligations and higher fulfillment (McDermott et al., 2013).Compared to full-time employees, part-time employees perceived fewer employee obligations, but a similar number of organizational obligations (Gakovic & Tetrick, 2003).Part-time employees in a bank were more likely to perceive their psychological contracts as fulfilled by the employer, but no such effect was found for employees in a supermarket (Conway & Briner, 2002).The evidence regarding full-time and part-time employees is hence less clear, but it suggests lower obligations for part-time employees.Part-time employees are very heterogeneous, with some working few hours every week and others 80% or 90% of a full-time equivalent.Working hours in part-time employment and the related psychological contract thus appear as a continuum between a combination of few hours and lower obligations and a combination of more hours and higher obligations.Given that multiple jobholders most likely combine two or more part-time jobs, employment contracts with more working hours presumably rather resemble a full-time job.
Research has identified typical psychological contract perceptions of managers and employees.Compared to employees, managers reported higher relational obligations of the employer and a higher fulfillment of relational and transactional obligations (Coyle-Shapiro & Kessler, 2000).Managers perceived a lower fulfillment of employee obligations and a higher fulfillment of employer obligations than employees (Aydin, Yilmaz, Memduhoğlu, Oğuz, & Güngör, 2008;Lester, Turnley, Bloodgood, & Bolino, 2002).Employees and their supervisors agreed on employer obligations and their breach (Tekleab & Taylor, 2003).This research demonstrates a negative bias of managers toward employees and their psychological contracts and a tendency for mangers to perceive higher obligations and higher fulfillment.
Multiple jobholders hold jobs in the same occupation or in different occupations or industries (Hipple, 2010;Panos et al., 2014).If they work in different occupations, they have probably not been trained or educated for all their occupations.For example, one job might correspond to their occupational training, the second job might require other knowledge and skills, and the third job may require lower qualifications.Only one study investigated overqualification and psychological contracts, finding a negative association with a balanced or relational psychological contract and a nonsignificant association with a transactional psychological contract (Luksyte et al., 2011).This finding was interpreted in a way that the employment relationship lacks long-term orientation.Due to the scarce research investigating psychological contracts and under-or overqualification, it is tentatively expected that multiple jobholders would perceive higher obligations if the job corresponded with their qualifications than if the job required lower or higher qualifications.
In line with previous research, it is expected that the psychological contracts of multiple jobholders are related to each employment contract.A higher level of obligation was hypothesized to be associated with a permanent employment contract, more work hours, a supervisor position and correspondence between job and qualifications.
Hypothesis 1: The employment contract is associated with the psychological contract obligations such that (a) a permanent contract, (b) working hours, (c) a manager position, and (d) correspondence between job and qualifications are positively related to psychological contract obligations.

Nonstandard Work Arrangements
Economic studies have primarily investigated types of jobs and portfolios and the motivation and demographic characteristics of multiple jobholders (Dickey et al., 2011;Panos et al., 2014;Wu et al., 2009), but they have not considered behavioral or attitudinal outcomes.Multiple jobholding has not been linked to the psychological contract.Research in labor economics has presented multiple jobholders as a heterogeneous group of workers with various combinations of jobs (Hipple, 2010;Swiss Federal Statistical Office, 2010).The work arrangements vary regarding the type of employment contract (permanent/temporary, working hours) in each job but also regarding the number of jobs of one multiple jobholder.It could be expected that multiple jobholders perceive lower obligations in each job if they work in several smaller jobs.If a multiple jobholder is employed in several jobs, fewer opportunities for developing promises and obligations exist in jobs with few working hours.When working in several jobs, the other jobs can compensate for obligations that are not fulfilled in one job.Correspondingly, multiple jobholders probably perceive on average higher obligations if they work in two jobs of equal importance or one larger and one smaller job as compared with several smaller jobs.
The reasons of multiple jobholders for their nonstandard work arrangement have been widely researched in economic studies that usually distinguished between financial and nonfinancial reasons (Averett, 2001;Dickey et al., 2011;Kimmel & Powell, 1999;Martel, 2000;Wu et al., 2009).Multiple jobholders gave primarily financial reasons for their nonstandard work arrangement, also known as constraints related to their main job, such as "meet regular household expenses," "pay off debts," "save for the future," or "insure against job insecurity" (Averett, 2001;Dickey et al., 2011;Kimmel & Powell, 1999;Martel, 2000).Nonfinancial considerations, also known as heterogeneous job motives, such as "gain experience," "enjoy the work of the second job," or "family responsibility," were cited less often as motives for multiple jobholding (Brown & Gold, 2007).Researchers have recommended considering motives when investigating multiple jobholding and behavioral or attitudinal outcomes (Zickar et al., 2004).
Economic research concentrated on financial reasons to determine whether employees need a second income to earn their living but did not aim to further differentiate nonfinancial motives (e.g., Dickey et al., 2011).Self-determination theory has in a similar manner identified a continuum between controlled and autonomous motivation, with controlled motivation being related to extrinsic rewards and autonomous motivation to an individual's volition and interest (Gagne & Deci, 2005;Ryan & Deci, 2002).In contrast to psychological theories of motivation, reasons for multiple jobholding only focus on why an individual opts for this form of employment but not how such reasons contribute to work outcomes.According to self-determination theory, intrinsic motivation is linked to the satisfaction of three basic needs: autonomy, competence, and relatedness.The satisfaction of needs in an employment relationship entails that resources are available and guarded by obligations of the psychological contract (Dabos & Rousseau, 2013).It is thus expected that multiple jobholders with more intrinsic motivation or more nonfinancial reasons for multiple jobholding perceive a higher level of obligations.Accordingly, a lower level of obligations presumably correlates with financial reasons.
Hypothesis 2: The nonstandard work arrangement of multiple jobholders relates to psychological contract obligations such that (a) the number of jobs and (b) more financial reasons for multiple jobholding are negatively related to psychological contract obligations.

Sample
The sample of this study consisted of 141 multiple jobholders in Switzerland.Because multiple jobholders account for less than 10% of the working population in Switzerland and work in various types of organizations and work contexts, several routes were used to access the sample: (a) 25 organizations that might employ multiple jobholders were contacted, including temporary work agencies, radio stations, vocational schools, colleges, cleaning companies, hotels, and mountaineering offices.(b) Multiple jobholders who were portrayed in the media or on the Internet were approached.(c) Furthermore, multiple jobholders who agreed to participate were asked to spread information about this research in their network.The questionnaire was finally sent to 347 multiple jobholders and was filled out by 148 respondents (response rate = 43%).
In this sample, 64 respondents were employed in two or more jobs, 77 respondents were simultaneously employed and self-employed, and seven respondents were selfemployed, with two businesses.Data from all multiple jobholders with salaried positions (N = 141) were used because their job portfolio represents a realistic employment situation of multiple jobholders.Psychological contracts for selfemployed positions were not included because in these positions, multiple jobholders lack an employer who can make promises on which a psychological contract is built.Hence, 64 respondents provided information on two psychological contracts and 77 on one psychological contract.The multiple jobholders reported in total 205 employment relationships and psychological contracts.
The multiple jobholders were working in a broad range of occupations (International Labour Organization, 2010), the most frequent being teaching professionals, business professionals, clerks, artists, social science professionals, and journalists.Multiple jobholders were either working in one occupation or combined two or more occupations.Examples of combinations of different occupations were software developer and farmer, teacher and goldsmith, artist and engineer, employee at a cinema and photographer, puppeteer and psychologist, and innkeeper and physiotherapist.The respondents were aged 20 years to 62 years (M = 41.92,SD = 9.85) and had worked as multiple jobholders for 1 month up to 33 years (M = 7.12, SD = 6.20).Of the 141 multiple jobholders, 69 were female, 70 were male, and two provided no information regarding their gender.The respondents had an above-average education; 96 held a university or college degree.

Measures
The psychological contract was measured with five validated scales capturing employer and multiple jobholder obligations (Raeder, Wittekind, et al., 2009).These scales have been used in other studies (e.g., Gerber, Grote, Geiser, & Raeder, 2012) and have a similar structure and contents as other measures of psychological contracts that distinguish dimensions of contents (Dabos & Rousseau, 2013;De Vos et al., 2003;Freese et al., 2011;Lub et al., 2012).Respondents were instructed to focus on obligations that were promised in the context of the employment relationship and were then presented with a list of obligations.Respondents indicated on a 5-point Likert-type scale ranging from 1 (strongly disagree) to 5 (strongly agree) whether these obligations were promised by the employer or by the multiple jobholder.The multiple jobholders filled out identical scales for up to two salaried jobs but not for self-employed work.The questionnaire would have become overly long and complex if data on more than two employment relationships were provided in detail.
Employer obligations were covered with the scales security and retention, support for career and skill development, and participation.Security and retention was measured with four items.A sample item was "Opportunities for identification."One item was deleted because it negatively affected the internal consistency.Cronbach's alpha for the scale consisting of three items was .69.Support for career and skill development was evaluated with seven items (sample item, "Opportunities for promotion").Cronbach's alpha was .78.Participation was captured with three items, for example, "Involvement in decision making."Cronbach's alpha for this scale was .83.
Multiple jobholder obligations were captured with the scales loyalty and performance and responsibility for career and skill development.Loyalty and performance was measured with five items (sample item, "Achievement orientation").Cronbach's alpha was .71.Responsibility for career and skill development was measured with two items, for example, "Responsibility for skill development."Cronbach's alpha for this scale was .70.
Nonstandard work arrangements of multiple jobholders are complex and diverse.This study hence aimed to identify the most important aspects with standardized questions that could be answered by all respondents.Additional open questions served to understand the complexity of the work arrangements and to verify the quantitative data.With one exception, reasons for multiple jobholding, only the quantitative data were used for this analysis.
The employment contract was measured with four objective variables: permanent/temporary contract, working hours, manager position, and correspondence between job and qualifications.The multiple jobholders indicated whether their employment was permanent or temporary on a dichotomous scale (0 = temporary, 1 = permanent).Information on working hours agreed upon in their employment contract was given as a percentage of a full-time employment between 40 hr and 44 hr because this is the standard form of defining part-time employment in Switzerland and some other European countries.For example, 80% indicates that someone is hired for 34 hr a week, with 40 hr full-time employment.Respondents provided information on managerial responsibility (0 = no, 1 = yes).Whether the job corresponded to qualifications and skills was measured on a dichotomous scale (0 = no, 1 = yes).This item was used to measure the correspondence of education and occupation in an objective way and to avoid comparing each multiple jobholder's occupations and education.In some cases, it would have been difficult to decide about the match of occupation and education based on job titles and education and training activities.
To explain the idea of corresponding occupations and education, the following examples are provided: (a) An example of correspondence between qualifications and occupation is a multiple jobholder trained as a sports teacher and working as a sports teacher, (b) an example of a higher qualification than necessary in a job is a person who studied agriculture at a college and is working as a farmer, (c) an example of a lack of qualification for a job is a goldsmith working as a graphic designer without having been trained in graphic design.
The nonstandard work arrangement was captured with two variables.First, the multiple jobholders indicated whether their work arrangement consisted of more than two employment relationships on a dichotomous scale (0 = no, 1 = yes).This question aimed to distinguish between multiple jobholders with the minimum number of jobs required for such a work arrangement and multiple jobholders with more complex work arrangements.It is difficult to capture a complex work arrangement in a questionnaire because job portfolios are diverse and can consist of a sequence of temporary jobs or a number of parallel jobs with few hours worked or a mix of both.The information that participants provided in an open question showed that some work arrangements would be too complex to measure in a standardized way.Basic information about the different jobs was further captured with the items on the employment contract (temporary/permanent, working hours).
Second, multiple jobholders were asked to provide reasons for multiple jobholding in an open question, and these were coded as financial and nonfinancial, in line with earlier research (Dickey et al., 2011;Kimmel & Powell, 1999;Martel, 2000).Financial reasons, also known as reasons related to constraints in one job, comprised motives in the categories "provide security for income," "additional income," "part-time employment in one job," and "career advancement."Nonfinancial reasons, also known as heterogeneous jobs motives, were "enjoy work arrangement," "diversity," "good combination," "personality development," "social contact," "health," "habit," and "chance" (Raeder, Mutz, Widmer, Wittekind, & Grote, 2009).The qualitative results were quantified by counting the number of financial and nonfinancial reasons provided by each respondent.Because some multiple jobholders reported many reasons and some very few, the proportion of financial and nonfinancial reasons was transformed into a scale of 1 to 5, with 1 indicating exclusively financial reasons, 3 indicating a balance between financial and nonfinancial reasons, and 5 indicating exclusively nonfinancial reasons.
Educational level, age, and gender were included as control variables because they were found to be related to different patterns of psychological contracts and multiple jobholding.Psychological contracts varied according to employees' education, age, and gender (Bal & Kooij, 2011;Bellou, 2009;Sok, Blomme, & Tromp, 2013).Women and more highly educated people were more likely to work in more than one job, whereas younger and older workers were less likely to work in more than one job (Hipple, 2010;Swiss Federal Statistical Office, 2010).Men and women combined different types of jobs; for example, men were more often self-employed in their secondary job.The reasons for multiple jobholding were shown to vary according to the gender and age of multiple jobholders (Dickey et al., 2011;Wu et al., 2009).Younger workers were more likely to hold multiple jobs due to financial reasons (Dickey et al., 2011).Two levels of education were distinguished (0 = only junior secondary school or junior secondary school plus an apprenticeship or senior secondary school with university entrance certificate, 1 = college, bachelor's degree, higher vocational education, master's degree, or PhD) because differences in multiple jobholding between these two levels were found to be most pronounced (Dickey et al., 2011;Hipple, 2010;Swiss Federal Statistical Office, 2010).

Analysis
The data were analyzed with multilevel modeling because the multiple jobholders reported one or two psychological contracts, resulting in unbalanced data that could be handled with multilevel analysis.Multilevel analysis was also suitable because the jobs could not be ranked into a primary and a secondary job (Kenny, Kashy, & Cook, 2006).The data of all multiple jobholders were used in this analysis because (a) the data represent the real employment situation of multiple jobholders and (b) a larger sample size at higher levels is beneficial for multilevel models (Maas & Hox, 2005).To avoid overly complex models that would be based on sparse data, the analysis did not contain random slopes.The high number of singletons (i.e., multiple jobholders with only one psychological contract) could be criticized, but simulation studies found low levels of bias for models with up to 70% singletons and 50 to 500 clustering units (Bell, Ferron, & Kromrey, 2008;McNeish, 2014).These simulation studies showed that the proportion of singletons was less influential than the sample size at the highest level and supported the decision to include data of all multiple jobholders.With a sample size of 141 multiple jobholders, this study was well above the critical threshold for the number of clustering units.
The research question and the hypotheses were tested in SPSS 24 in two sets of multivariate multilevel models predicting employer obligations (security and retention, support for career and skill development, and participation) and multiple jobholder obligations (loyalty and performance and responsibility for career and skill development).Through the use of multivariate models, all employer obligations and all multiple jobholder obligations could be tested in one model each, without producing different results than univariate models would provide.The multivariate multilevel models contained three levels (Figure 1): the multivariate measurement of obligations of the employer or multiple jobholder (Level 1), the employment contract (Level 2), and the multiple jobholder (Level 3).For each multivariate model, an empty model without predictors (Model 0) was calculated to investigate the partitioning of variance.In Model 1 the control variables were added, in Model 2 the predictors at Level 2 measuring the employment contract and in Model 3 the predictors at Level 3 measuring the work arrangement.The model fit was assessed with Akaike's information criterion (AIC) on a smaller-is-better basis.

Results
Descriptive statistics and correlations are presented in Table 1 for Levels 1 and 2 and in Table 2 for Level 3. Disaggregation of data that would occur if the Level 3 variables (work arrangement and demographics) were correlated with the variables at the lower levels (employment and psychological contract) could thus be avoided.
Research Question 1 explored the distribution of variance within and between multiple jobholders.When comparing the amount of unexplained variance between and within multiple jobholders in the empty models (Model 0), clear differences between the dependent variables were found.The intraclass correlation (ICC) was between .32 and .58for the different dependent variables, indicating that 32% to 58% of the unexplained variance resided between multiple jobholders.For the responsibility for career and skill development, the ICC was clearly above .50,indicating a higher proportion of variance between multiple jobholders than within multiple jobholders.For all other obligations, the variance within multiple jobholders was larger than the variance between multiple jobholders.The comparably low variance within multiple jobholders for responsibility for career and skill development showed that the multiple jobholders perceived a similar amount of responsibility while referring to different jobs.This finding depicts an individual factor in their psychological contracts.For all other obligations, the proportion of unexplained variance between multiple jobholders ranged between .32 and .44,showing that these dimensions of the psychological contract are more clearly targeted toward a specific job.Nonetheless, psychological contracts contain more variance to be explained between employment relationships than between multiple jobholders.
Hypothesis 1 aimed to investigate the association of the employment contract with the psychological contract.The results show that a permanent contract (H1a) and more working hours (H1b) were related to higher employer obligations in participation (Table 3).Working hours (H1b) predicted slightly lower multiple jobholder obligations in loyalty and performance (Table 4).A manager position (H1c) was consistently associated with higher employer obligations and higher multiple jobholder obligations, whereas correspondence between job and qualifications (H1d) was associated with a higher multiple jobholder responsibility for career and skill development.Hypothesis 1c found full support and Hypotheses 1a, 1b, and 1d were partly supported.The variance explained within multiple jobholders was lower for their obligations (6%-9%) than for employer obligations, varying between 9% (support for career and skill development) and 35% (participation).The models explained up to 25% of the variance between multiple jobholders.
Hypothesis 2 predicted that the nonstandard work arrangement of the multiple jobholders would be related to the psychological contract.The number of jobs (H2a) was not associated with obligations, whereas the reasons for multiple jobholding (H2b) were negatively related to the employer's obligation to support career and skill development; employees with more financial reasons expected more employer support (Tables 3 and 4).Hypothesis 2a failed to find empirical support, whereas Hypothesis 2b was partly supported.Compared to Model 2, Model 3 explained up to 7% of additional variance between multiple jobholders.
In the two sets of models, Model 2 showed the best fit because it had the lowest Although Hypotheses 1 and 2b found partial support, the support was weaker for Hypothesis 2b due to the low model fit.
The large number of singletons in these models could be criticized because all multiple jobholders with at least one salaried job were included.Alternative models with a sample of 64 multiple jobholders with two salaried jobs were thus calculated and compared with the models presented above.The alternative models showed similar ICCs (± .02).The same variables were found to be significant when predicting multiple jobholder obligations, while more significant predictors were found when predicting employer obligations (correspondence of job and qualifications, more than 2 jobs).The alternative models explained more variance, particularly between multiple jobholders, although differences between the models using the full and the reduced sample size were overall small and do not lead to different conclusions.

Discussion
This study investigated psychological contracts in complex nonstandard work arrangements with simultaneous employment relationships, an approach that has not been taken   previously.In this sample of multiple jobholders, effects related to the employment contract and effects related to the multiple jobholder could be distinguished.Except for the multiple jobholder responsibility for career and skill development, larger variance was found within the psychological contracts of the multiple jobholders than between them.This indicates that psychological contracts are tied to each employment relationship, while the subjectivity in perception plays a minor role.The employment contract, measured through permanent/temporary contract, working hours, managerial responsibility and correspondence between job and qualifications, was related to some or all obligations.Of the nonstandard work arrangement of multiple jobholding, only the reasons for working in multiple jobs were related to one obligation.
This study contributes to research on multiple jobholders and psychological contracts in several ways.First, this study indicated that multiple jobholders differentiated between psychological contract obligations in different jobs, tying them more clearly to the employment contract than to their work arrangement (Coyle-Shapiro & Kessler, 2000;Gakovic & Tetrick, 2003;Van Dyne & Ang, 1998).The subjective component of psychological contracts remained visible through the respectable proportion of variance between multiple jobholders, which varied between different obligations.In particular, the responsibility for career and skill development contained a large individual component, indicating that this obligation is more closely connected to the individual than to the employment relationship.As the study examined the psychological contract within and between multiple jobholders, the findings demonstrated that the psychological contract is related to the employment contract, although it is interpreted with subjectivity (Rousseau, 1989(Rousseau, , 2001)).Second, this research expands existing, mainly economic, knowledge on multiple jobholders to an organizational context (Sliter & Boyd, 2014).The study demonstrated how variables that have often been researched in labor economic studies, such as reasons for multiple jobholding and nonstandard types of work arrangements (Dickey et al., 2011), relate to the employment relationships.Only financial reasons for multiple jobholding were associated with higher support for career and skill development, indicating that multiple jobholders with financial reasons care more about receiving support for their career.Despite this result, multiple jobholders' reasons for their work arrangement (Zickar et al., 2004) were overall of minor importance for their psychological contracts.The complexity of the work arrangement, measured through the numbers of jobs, failed to affect the psychological contract beyond the employment contract.Whether multiple jobholders work in many small jobs or in two larger ones does not change their perception of obligations in their employment relationships.
Third, this research contributes to a better representation of employment contracts for multiple jobholders' psychological contracts.Although previous research on temporary and permanent employment contracts has shown a clear pattern in favor of permanent contracts (Coyle-Shapiro & Kessler, 2002;Isaksson et al., 2010), permanent contracts of multiple jobholders accounted only for higher employer obligations in participation and not for other employer obligations.The situation of multiple jobholders, however, differs from that of employees with only one job because the secondary job can compensate for what is missing in the primary job and vice versa.Permanent and temporary jobs are not perceived as being significantly different due to the secondary job, and multiple jobholders might have similar expectations toward jobs with a temporary or permanent contract.The secondary job could, for example, provide them with opportunities for development and job security that are not offered in the primary job.In this study, some multiple jobholders cited reasons for multiple jobholding, such as the salaried job providing them with security of income for building up their own business in their secondary job.The effects of temporary and permanent contracts would thus be less evident than in a sample of employees with one job.
Earlier research found some effects on part-time and fulltime jobs (Conway & Briner, 2002;Gakovic & Tetrick, 2003), whereas the working hours of multiple jobholders hardly showed an influence, perhaps because none of the multiple jobholders was working full time in one of the jobs.Major differences could, for example, be expected between psychological contracts in a full-time job and in a 20% parttime job (8 hr per week).In this sample, the jobs of multiple jobholders differed less than a full-time and a part-time job, and multiple jobholders thus found other ways to prioritize their jobs.
Consistent with earlier research (Coyle-Shapiro & Kessler, 2000;Lester et al., 2002), a supervisor position was the most stable predictor of higher obligations in the employment relationship.Managers fill a different role in organizations with higher responsibility and higher mutual expectations, leading to a different quality of the employment relationship.This typical effect of a supervisor role also appeared when jobs were part time or temporary.
Over-and underqualification have rarely been researched (Luksyte et al., 2011).In this study, correspondence between job and qualifications played a role only for the multiple jobholder's responsibility for career and skill development and was thus related to career and training issues.Having a career in the occupation was possible only with the necessary skills and knowledge that are often acquired through education and training.This result was in line with earlier research (Luksyte et al., 2011) that showed that a match between job and qualification was positively related to a relational and balanced contract.Both studies conveyed the message that the appropriate qualification was related to a long-term employment relationship.

Limitations
There were limitations in this study that should be noted.The sample size at the group level (i.e., multiple jobholders) was sufficiently large, whereas each Level 2 unit contained only one or two elements (i.e., employment contracts), posing limits on adding additional random effects (Clarke, 2008;Kenny et al., 2006).Although a larger sample would have been beneficial for the statistical model, many people with a nonstandard work arrangement were reached who were willing to participate in the study.All multiple jobholders with at least one salaried job were retained in the sample because this sample better represented the employment portfolio of multiple jobholders than if partly self-employed respondents were excluded.It could be argued that those with only one salaried job should be excluded to avoid singletons, that is, groups with only one measurement.Singletons are, however, less problematic for the analysis than expected (Bell et al., 2008;McNeish, 2014), and a realistic representation of the work arrangement compensates for the disadvantages.
It proved to be difficult to perfectly represent the employment relationships of all types of multiple jobholders in a quantitative survey.It could be possible to investigate the psychological contracts of three salaried positions, instead of examining only two, if multiple jobholders work in three permanent part-time jobs.It might be difficult to examine psychological contracts for more than three smaller part-time jobs, for a sequence of temporary jobs or for a mix of all these types of jobs.If the same set of questions has to be filled out three or four times for three or four jobs, respondents most likely perceive this as redundant.Specifically, it is demanding to fill out identical scales for small jobs with few hours per week (e.g., language teacher) or short temporary assignments of a few weeks (e.g., actor).Due to fewer options for interaction, employers and employees probably share fewer promises and create fewer obligations.Thus, only the two main jobs were considered, also for multiple jobholders with more than two jobs, although the psychological contracts of additional jobs were not included.

Future Research
Although the models explained the variance between and within multiple jobholders well, a considerable proportion of unexplained variance remained, particularly regarding the obligations of multiple jobholders.The unexplained variance (Tables 3 and 4) shows at which level predictors should be sought.Variance within multiple jobholders could be explained through predictors related to the employment relationship, for example, organizational predictors such as organizational culture, perceived organizational support or Human Resource Management practices.Variance between multiple jobholders could be explained through predictors related to the individual (e.g., individual goals) and his or her career.
This study did not test whether the psychological contracts of multiple jobholders are distinct from the psychological contracts of workers with one job.It would be necessary to compare multiple jobholders and employees with one job and their psychological contracts to determine whether multiple jobholders are a distinct group of the working population.One basic research question to be answered could refer to the factorial structure of psychological contract measures.When differences between these two groups of the working population are large, items might not load on the same factors, as was the case for the scale security and retention.Further differences could then be explored in a next step.
Further research should also consider the development of multiple jobholding and psychological contracts over time.A focus on events of change could be used to examine how psychological contracts develop.Such change events could occur when multiple jobholders return to work in only one job or change one of their jobs.

Practical Implications
Multiple jobholders are a special type of employee who deserve the attention of employers.Multiple jobholders differentiate well between their jobs, especially regarding temporary/permanent contracts and manager positions.They are highly flexible regarding the use of their competencies, and they provide broad experience from their secondary jobs and often also a broad range of competencies due to education and training.Organizations can thus benefit from their flexibility and their broad range of competencies, and they can also influence psychological contracts through the choice of terms in their employment contracts.
Multiple jobholders have the flexibility to work in different jobs and different types of work tasks.They are resilient due to the other job(s) that gives them security, additional income, or options for developing their career.However, multiple jobholders can also feel stressed due to their other job (Zickar et al., 2004), and their flexible use of work time is probably limited.These are some of the few restrictions that employers have to be aware of when hiring multiple jobholders.

Table 1 .
Descriptive Statistics and Correlations Among Study Variables: Levels 1 and 2, Psychological Contract and Employment Contract.

Table 2 .
Level 3, Multiple Jobholder Characteristics and Work Arrangement.

Table 3 .
Multilevel Estimates for Multivariate Model Predicting Employer Obligations.

Table 4 .
Multilevel Estimates for Models Predicting Multiple Jobholder Obligations.