Journal of Vocational Behavior

Size: px
Start display at page:

Download "Journal of Vocational Behavior"

Transcription

1 Journal of Vocational Behavior 74 (2009) Contents lists available at ScienceDirect Journal of Vocational Behavior journal homepage: Psychological contracts and their implications for commitment: A feature-based approach q Kate J. McInnis *, John P. Meyer, Susan Feldman Department of Psychology, The University of Western Ontario, London, Ont., Canada N6A 5C2 article info abstract Article history: Received 12 December 2008 Available online 25 December 2008 Keywords: Psychological contract features Transactional and relational contracts Contract fulfillment Balanced contracts I-Deals Organizational commitment Affective commitment Normative commitment Two studies were conducted to examine the link between employee perceptions of the psychological contract and their affective and normative commitments to the organization. The authors adapt a new approach to the study of psychological contracts by developing a generalizable measure of contract features (e.g., scope; time frame). In Study 1 (N = 301), the authors predicted and found that employees perceptions of the contract s features contributed beyond perceptions of contract type (i.e., transactional; relational) and fulfillment to the prediction of affective and normative commitment. In Study 2 (N = 147), the features measure was refined and results from the first study were largely replicated. In both studies, affective and normative commitment were greater when employees viewed the contract as broad, trust-based, equal, negotiated, tangible, and long-term, and weaker when they saw it as unequal, imposed, and short-term. We also found evidence for patterns of features that correspond to transactional and relational contracts, as well as to recently proposed balanced contracts and I-Deals. Implications for theory and the management of psychological contracts are discussed. Ó 2008 Elsevier Inc. All rights reserved. 1. Introduction Change is arguably one of the few constants in many organizations today, and this change has important implications for the nature of the employer employee relationship. Two aspects of this relationship that have been viewed as particularly vulnerable to change are the psychological contract (Rousseau, 1998; Schalk & Roe, 2007) and employee commitment (Baruch, 1998; Mowday, 1998; Rousseau & Wade-Benzoni, 1995). Employee perceptions of the psychological contract and commitment to the organization are theoretically linked, with the former being viewed as a potentially important determinant of the latter (Meyer & Herscovitch, 2001). Although there is some empirical support for this relationship (e.g., Bunderson, 2001; Lester, Turnley, Bloodgood, & Bolino, 2002; Sels, Janssens, & Van den Brande, 2004), research has not kept pace with recent developments in the broader contracts and commitment literatures. Our objective in this research was to provide new insight into the relations between employees perceptions of the psychological contract and their commitment to the organization. Most research to date has focused on perceptions of contract fulfillment (or breach) and its relations with affective commitment (e.g., Bunderson, 2001). Other studies have examined relations between contract type (e.g., transactional and relational; social and economic) and commitment (e.g., King, 2003; Shore, Tetrick, Lynch, & Barksdale, 2006). More recently, however, an attempt has been made to determine how com- q This research was supported by a grant from the Social Science and Humanities Research Council of Canada to the second author. Data from Study 1 were presented at the 23rd annual conference of the Society for Industrial and Organizational Psychology, San Francisco, * Corresponding author. address: kmcinni3@uwo.ca (K.J. McInnis) /$ - see front matter Ó 2008 Elsevier Inc. All rights reserved. doi: /j.jvb

2 166 K.J. McInnis et al. / Journal of Vocational Behavior 74 (2009) mitment relates to perceptions of the features (i.e., defining characteristics) of the contract (Battisti, Fraccaroli, Fasol, & Depolo, 2007; Sels et al., 2004). Although relatively little attention has been given to the measurement and investigation of contract features, Rousseau and Tijoriwala (1998) argued that there are a number of potential advantages to a feature-oriented approach and that it warrants further investigation. We conducted two studies to examine how the features of the psychological contract, viewed from the employees perspective, relate to their commitment to the organization. In addition to being one of the few studies to examine the link between contract features and commitment, the present research makes a number of unique contributions. First, it introduces a modified measure of contract features that has several advantages over previous measures (e.g., Sels et al., 2004). Second, it goes beyond the typical emphasis on affective commitment (AC) to explore relations with the normative commitment (NC) mindset identified in Meyer and Allen s (1991, 1997) three-component model. Although NC, with its emphasis on obligation, should be particularly relevant to contract formation and/or breach (Meyer & Herscovitch, 2001; Wayne et al., 2008) it has received little attention to date. By comparing relations observed for AC and NC, we shed additional light on the issue of whether the two components are distinguishable. Finally, our findings have broader implications for the conceptualization and measurement of psychological contracts, particularly with regard to widely recognized transactional and relational contracts (Rousseau, 1990), as well as the more recently introduced balanced contract (Dabos & Rousseau, 2004; Ho, Rousseau, & Levesque, 2006; Rousseau, 2000) Conceptualization and measurement of psychological contracts Although the meaning of the term psychological contract has evolved somewhat since it was first introduced (Levinson, Price, Munden, Mandl, & Solley, 1962), and has been the subject of some debate (e.g., Guest, 1998; Rousseau, 1998), it is commonly considered to reflect an individual s belief in the terms and conditions of a reciprocal exchange agreement between the focal person and another group (Rousseau, 1989, p. 123). In the case of an employer employee relationship, this belief is predicated on the perception that a promise has been made (e.g., of employment or career opportunities) and a consideration offered in exchange for it (e.g., accepting a position, foregoing other job offers) (Rousseau & Tijoriwala, 1998, p. 679). This contract is psychological in the sense that it reflects an individual s perception of mutual obligation. Although agreement between the parties involved can have important implications for the enactment of the contract (Dabos & Rousseau, 2004), psychological contract research commonly focuses on the perceptions of one party, typically the employee (Coyle-Shapiro & Kessler, 2000; Lester et al., 2002; Robinson, 1996). This is also the approach taken in the present research because our interest is in determining how employees perceptions of their employers obligations relate to their commitment to the organization. Psychological contracts have been measured in different ways, dictated in part by the research questions being addressed. Rousseau and Tijoriwala (1998) identified three forms of measurement: content-oriented, evaluation-oriented, and featureoriented. Content-oriented measures focus on the terms of the contract (i.e., what has been promised) and typically measure specific obligations (e.g., job security and training; Shore & Barksdale, 1998). Others are designed to assess contract types reflecting distinct sets of terms (e.g., Rousseau, 2000). For example, transactional contracts tend to include economic or monetizable terms (e.g., pay and benefits), whereas relational contracts include socio-emotional terms (e.g., support and development). Although widely used (e.g., Herriot, Manning, & Kidd, 1997; King, 2003; McDonald & Makin, 2000), one concern about content measures is their lack of generalizability (DelCampo, 2007). The terms of a psychological contract can differ within and across organizations and therefore it is often necessary to tailor measures to the context (e.g., Montes & Irving, 2008). This makes comparison of findings across studies more difficult (Rousseau & Tijoriwala, 1998). Evaluation-oriented measures assess perceptions of the fulfillment or breach of a contract (e.g., Robinson & Morrison, 2000). In most cases, the focus has been on employees perception of fulfillment or breach on the part of the organization and its implications for employee attitudes and behavior. In their recent meta-analysis, Zhao, Wayne, Glibkowski, and Bravo (2007) distinguished between two commonly used forms of the breach measure: composite and global. Composite measures involve aggregation of perceived breach across specific terms of a contract and are therefore subject to the same generalizability concerns as the content measures described above. The global measures involve a subjective assessment of fulfillment versus breach in general, without reference to specific content. The feature-oriented approach to measurement is the least well-developed, but shows considerable promise (Battisti et al., 2007; DelCampo, 2007; Rousseau & Tijoriwala, 1998). Rather than focusing on the specific terms of a contract, feature measures address the characteristic properties that might be used to differentiate contracts regardless of their specific content. We believe that the feature-oriented approach has an important advantage over the content-oriented approach because it can be used to guide the development of general (i.e., content-free) measures that can be used across situations. It can also help in clarifying the distinction between different types of contract (e.g., transactional vs. relational). Interestingly, although the items in these type measures typically focus on content (e.g., Rousseau, 1990, 2000), theoretical distinctions among the types are usually made in terms of features. For example, transactional contracts have been described as short-term, tangible, specific, and static, whereas relational contract have been described as open-ended, intangible, loosely defined, and flexible (Conway & Briner, 2005; Rousseau, 1990). Similarly, the more recent notion of a balanced contract (e.g., Ho et al., 2006; Rousseau, 2000) has also been distinguished from the others in terms of its features (i.e., open-ended and specific). Finally, although the feature-oriented approach cannot substitute for an evaluation-oriented approach (i.e., fulfillment vs. breach), we believe that it might complement such an approach in explaining the nature of employees relationship with their organization, including their commitment.

3 K.J. McInnis et al. / Journal of Vocational Behavior 74 (2009) Psychological contracts and commitment Perhaps the most common approach to linking psychological contracts and commitment has been to examine the impact of contract fulfillment or breach. Two recent meta-analyses (Bal, De Lange, Jansen, & Van Der Velde, 2008; Zhao et al., 2007) reported moderate negative correlations between contract breach and AC (q =.39, q =.38, respectively). A somewhat smaller set of studies has examined the relations between perceptions of contract type (e.g., transactional and relational) and commitment. King (2003) proposed and confirmed that employees who view the contract as relational have relatively strong AC, whereas those who view the contract as transactional have strong continuance commitment (CC). King also found that NC and CC correlated positively with relational contract scores, albeit less strongly in the case of CC, and that AC correlated negatively with transactional contract scores. Other studies have also reported significant positive correlations between relational contracts and AC (Hughes & Palmer, 2007; Shore et al., 2006), but findings concerning relations between AC and transactional contracts have been mixed (e.g., Coyle-Shapiro & Kessler, 2000; Hughes & Palmer, 2007; Sloboda, 1999), as have findings involving CC and both transactional and relational contracts (e.g., Hughes & Palmer, 2007; Sloboda, 1999). The latter findings may be due, in part, to differences in the ways in which CC has been operationalized. To date, there have been only a few attempts to measure and investigate the implications of contract features for commitment (e.g., Battisti et al., 2007; Janssens, Sels, & Van den Brande, 2003; Sels et al., 2004). Although they make an important contribution, these studies are limited in several respects, including the nature of the contract measures and the exclusive focus on AC. Our research introduces a refined measure of contract features and examines relations with both AC and NC. Before elaborating on these objectives, we first provide a more general review of research examining the links between psychological contract features and commitment Psychological contract features and commitment Sels et al. (2004) measured employee perceptions of six bipolar feature dimensions of the psychological contract. Although they measured perceptions of both the employers and employees obligations, we focus here on the former. The first four dimensions were adapted from Rousseau and McLean Parks (1993), based on the works of Macneil (1985): tangibility, scope, stability, and time frame. To these, they added two additional dimensions based on the industrial relations literature (e.g., Visser, 1996): symmetry and level. They developed multi-item self-report measures of each dimension. The tangibility dimension (tangible vs. intangible) reflects the degree to which employees perceive the contract as clearly specified and observable from a third-party perspective. Scope (narrow vs. broad) refers to the extent to which the relationship between the employer and employee is restricted to the employment relationship or includes other aspects of one s life (e.g., implications for family). The stability dimension (flexible vs. stable) reflects the extent to which the terms of a contract can evolve and adapt in response to changing conditions versus being static and fixed at the time of formation. Time frame (short-term vs. long-term) refers to the perceived duration of the employment relationship. The symmetry dimension (equal vs. unequal) involves the degree to which the employer and employee are considered as equal partners in the relationship. For example, an equal symmetry contract would consider both parties needs, whereas an unequal symmetry contract would primarily focus on the needs of one party, typically the employer. Finally, contract level (individual vs. collective) was introduced to reflect the degree to which the employee perceives the contract as being individually or collectively established. That is, individual contracts tend to be relatively unique for each employee, whereas collective psychological contracts tend to be similar for all employees (Sels et al., 2004). Sels et al. (2004) found that AC was positively related to the time frame and level dimensions (see Janssens et al., 2003, for another analysis of the same data). That is, employees who were affectively committed perceived the employer s obligation as long-term and collective. Battisti et al. (2007) also found support for positive relations between AC and long-term feature ratings. To our knowledge, NC has yet to be examined with regard to the contract feature dimensions Overview of the present research Two studies were conducted to examine the relations between employee perceptions of the psychological contract and their AC and NC to the organization. Although our primary objective was to examine how contract features relate to employees commitment, a preliminary step was to address three potential limitations in the feature measures used by Sels et al. (2004): the assumed bipolarity of the features, reference to specific terms in some items, and restriction in the number of features. We address each of these in turn and then discuss the impact of the measurement revisions for our primary research question. In the Sels et al. (2004) measure, each of the six dimensions is considered to be bipolar and items focus on one pole. This assumes that the characteristics defining the opposite poles are mutually exclusive (i.e., the absence of one characteristic implies the presence of the other). However, it is not clear that this assumption holds in all cases. For example, although agreeing that a contract is long-term implies that it is not short-term, agreeing that a contract includes tangible terms does not rule out the possibility that it also includes intangible terms. Therefore, for purposes of the present study we developed measures of the characteristics reflecting each pole of the proposed dimensions, thereby allowing empirical evaluation of relations between the characteristics as well as their individual relations with commitment.

4 168 K.J. McInnis et al. / Journal of Vocational Behavior 74 (2009) Although the measures developed by Sels et al. (2004) were intended to be content-free, an examination of the items suggests that, in some cases, content and feature might be confounded. For example, one item makes specific reference to performance appraisal criteria, another to career development opportunities, and two others to benefits. Therefore, we developed measures that focus on feature characteristics without any reference to specific content. We measured all six of the features included in the Sels et al. (2004) measure, but relabeled one. Specifically, we use the term flexibility (flexible and static) rather than stability (flexible and stable) to avoid confusion with stability as an indication of the length of the contract (see Lester, Kickul, & Bergmann, 2007). In addition, we added three feature dimensions that we considered relevant in predicting employees commitment. The dimensions we added were explicitness, negotiation, and formality. The explicitness (explicit and implicit) dimension reflects the degree to which the terms of the contract were clearly stated (e.g., during recruitment, selection, or socialization processes) or had to be inferred from policies and practices of the organization or its agents. Explicitness was mentioned by Rousseau and Tijoriwala (1998) and Conway and Briner (2005) as an element that may differentiate psychological contracts from one another, but to our knowledge has not been studied previously. Negotiation (negotiated and imposed) refers to whether the terms of the contract were negotiated with employees or were imposed unilaterally by the organization. Again, this contract characteristic was mentioned in passing by Conway and Briner (2005) but not studied empirically. Sels and colleagues (2004) also discuss the role of collective bargaining in the employer employee relationship, but believed that this element of the contract would be captured within the symmetry and level feature dimensions. Formality (formal and trust-based) refers to the extent to which the terms of the contract are formally agreed upon as opposed to established on the basis of mutual respect and trust. Like negotiation, formality was added to recognize what we consider to be meaningful variations in the employee s involvement in determining the contract terms. As we explain below, such involvement can have implications for the level of attachment or obligation an employee experiences. Again, to our knowledge, the formality feature has not been studied in previous research. The specific feature items we used are presented in the Appendix. We conducted two studies to examine the relations between psychological contract features and organizational commitment. In Study 1, we developed measures of psychological contract features to address the issues described above. These measures were further refined in Study 2. In both studies, we also included measures of AC and NC to the organization, contract type transactional, relational, and balanced (the last in Study 2 only) and perceptions of employers contract fulfillment. The commitment measures were used to test hypotheses concerning the nature of the relations between contract features and employees AC and NC. These hypotheses are described in the introduction to Study 1. The contract type and fulfillment measures were included so we could determine whether our feature measures would account for unique variance in commitment. We also wanted to determine whether transactional, relational, and balanced contract measures related as expected to measures of the features they are purported to reflect. 2. Study 1 Our primary objective in Study 1 was to devise a measure of contract features and to examine relations between the contract features and employees AC and NC to the organization. Given the paucity of previous research on features, we relied primarily on theory and related research findings to generate testable hypotheses. Our review of theory (e.g., Meyer & Allen, 1991; Meyer & Herscovitch, 2001) and recent meta-analytic findings (e.g., Colquitt, Conlon, Wesson, Porter, & Yee Ng, 2001; Meyer, Stanley, Herscovitch, & Topolnytsky, 2002; Rhoades & Eisenberger, 2002) revealed two particularly strong correlates of AC that are relevant to its relations with psychological contract features: organizational support and organizational justice. Employees have a stronger desire to remain with an organization (AC) when they perceive that it is concerned about their needs and values employees contributions (Eisenberger, Huntington, Hutchison, & Sowa, 1986; Rhoades & Eisenberger, 2002) and when it implements policies that ensure that employees are treated fairly (Colquitt et al., 2001; Meyer et al., 2002). The impact of such treatments is particularly strong when it is attributed to the good will of the organization as opposed to outside influences (Koys, 1991; Meyer & Allen, 1997). There are several contract features that we believed might lead employees to view their employers as being voluntarily supportive and/or fair, and therefore should correlate positively with AC: long-term, tangible, broad, equal, trust-based, collective, and negotiated. That is, employers are more likely to be considered supportive when they are willing to (a) commit to a long-term relationship, (b) specify a clear set of terms by which they are willing to abide, (c) include concern for the employee s general well-being (and that of his/her family) among their commitments, (d) treat employees as equals in the relationship, and (e) demonstrate a willingness to trust and be trusted. Employers are more likely to be considered fair when (a) they treat employees as equals, (b) treat employees consistently, and (c) negotiate the terms of the contract with employees. There are other features that we believe will be seen as indicating low support and therefore be negatively related to AC: short-term, unequal, and imposed. Note that we do not expect that features reflecting opposite poles of a dimension will all necessarily correlate in opposite directions with commitment. For example, we did not expect that because tangible terms and trust-based contracts would be positively associated with AC, that the inclusion of intangible terms or formal contracts would be negatively related. It was for this reason that we measured these characteristics independently. In sum, we tested the following hypothesis with regard to AC.

5 K.J. McInnis et al. / Journal of Vocational Behavior 74 (2009) Hypothesis 1: Employees AC to the organization will be positively related to perceptions of the employers obligation as long-term, tangible, broad, equal, trust-based, collective, and negotiated, and negatively related to perceptions of the contract as short-term, unequal, and imposed. Given the emphasis of NC on felt obligation, it is surprising that it has received little attention in psychological contract research. Consequently, there have been calls for more research to assess its relevance to the psychological contract (Meyer & Herscovitch, 2001; Wayne et al., in press). Again, in the absence of prior research, we relied on theory and related research to generate preliminary hypotheses. It has often been noted that NC relates similarly, albeit less strongly, to the antecedents of AC, including support and justice (Meyer et al., 2002). Therefore, those conditions that contribute to a sense of desire to remain may also contribute to a sense of obligation to do so. However, NC can develop independently of AC when conditions contribute to feelings of obligation in the absence of desire (Meyer & Herscovitch, 2001). That is, an individual who receives a favor from another may experience a strong sense of obligation to reciprocate without necessarily wanting to do so. Therefore, all things considered, NC might relate positively with many of the same contract features as AC, but there could be some differences. In particular, although AC is expected to be stronger when the same obligations are made to everyone (i.e., collective level), NC might be stronger when the terms of the contract are individually negotiated (i.e., individual level), because these negotiations can increase the salience of the need to reciprocate. NC may also be more responsive to the explicitness of the terms and formality of the contract, again because it might increase the salience of the need to reciprocate. Therefore, we tested the following hypothesis. Hypothesis 2: Employees NC to the organization will be positively related to perceptions of the employers obligation as long-term, tangible, broad, equal, trust-based, negotiated, formal, individual, and explicit, and negatively related to perceptions of the contract as short-term, unequal, and imposed. A secondary purpose of this study was to determine whether a features measure of the psychological contract would account for variance in AC and NC beyond that explained by contract type (i.e., transactional and relational) and fulfillment measures. Therefore, we also measured contract type and fulfillment and entered these as control variables in multiple regression analyses to test the following hypothesis. Hypothesis 3: Perceptions of the features of a psychological contract will explain variance in AC and NC beyond that explained by measures of contract type (transactional and relational) and employer contract fulfillment. 3. Method 3.1. Sample and procedure Participants were recruited through StudyResponse. We selected this particular sample because participants represent a variety of employers and occupations. StudyResponse randomly selected 1957 individuals from their pool of participants to receive a recruitment that included a website link to the online survey for the present study. As an incentive, individuals who participated were given the chance to enter into a draw for gift-certificates to an online retailer. Two methods were used to identify possible non-purposeful responders in the present study. First, the length of time to complete the survey was examined. Individuals who took less than three minutes, the length of time it generally took to read the survey, were eliminated from analyses. The second method involved eliminating individuals who did not complete at least half of the items on both the commitment scale and psychological contract features scale. Of those who received the request to participate and met the above two requirements, 301 completed the survey, for a response rate of 15.4%. This is consistent with the estimated average response rate of 10 30% provided by StudyResponse and might be a conservative estimate because we do not know how many in the targeted sample actually received the request and met the requirement of current employment Demographics of participants The mean age of participants was 37.2 (SD = 10.7) with 53% of participants being men. The majority indicated that they worked full-time (67.6%). The participants had a variety of educational backgrounds: less than high school (1.0%), high school (24.8%), associates degree (i.e., a college degree that is less than four years) (11.2%), some college (23.1%), four-year college degree (22.0%), some graduate school (4.5%), and a post-graduate degree (13.2%). A variety of workplace sizes was represented: employees (62.7%), employees (11.4%), and over 250 employees (25.9%). The average tenure in the organization was 6.8 years and the average tenure in the current profession/occupation was 9.6 years. A comparison of demographic characteristics of responders with those for non-responders (provided by StudyResponse) revealed little difference in terms of gender, employment status (i.e., full-time or part-time), or occupational type. However, responders were significantly older, t(1936) = 3.50, p <.01, worked in their occupation longer, t(344) = 2.47, p <.05, and had more total work experience, t(1909) = 2.75, p <.01, compared to non-responders. Demographic comparisons between the

6 170 K.J. McInnis et al. / Journal of Vocational Behavior 74 (2009) target population (i.e., the U.S. working population using United States Department of Labor, 2007 statistics, and United States Census Bureau, 2005 statistics) and study participants were also examined. Overall, age, gender, employment status, and occupation type were all roughly equally proportioned. However, the responders were slightly more educated with approximately 74% having at least some college, compared to only 59.5% of the working population Measures Organizational commitment. Organizational commitment was measured using Meyer, Allen, & Smith s (1993) 6-item AC (a =.82) and NC (a =.87) measures. Responses were made on a 7-point scale (1 = strongly disagree; 7 = strongly agree). Psychological contract types. A shortened (four items per scale) and slightly reworded version of the Psychological Contract Inventory (PCI) developed by Rousseau (2000) was selected to measure the participants perceptions of their employers transactional (a =.66) and relational (a =.83) contracts. Responses were made on a 5-point scale (1 = not at all; 5 = to a great extent). Psychological contract fulfillment. Participants indicated the extent to which they believed their employer had fulfilled the terms of the psychological contract by two items (a =.91) adapted from Rousseau (2000). Responses were made on a 5-point scale (1 = not at all; 5 = to a great extent). Psychological contract features. The psychological contract feature measure was developed for the current study and included nine contract feature dimensions, each with two feature items: explicitness (explicit and implicit), flexibility (flexible and static), formality (formal and trust-based), level (individual and collective), negotiation (negotiated and imposed), scope (narrow and broad), symmetry (equal and unequal), tangibility (tangible and intangible), and time frame (short-term and long-term). Participants responded to a list of 18 feature items on a 5-point scale (1 = strongly disagree; 5 = strongly agree). Participants were asked to consider the commitments that they believed their employer had made to them. The feature items each began with the following statement: The commitments (explicit or implicit) made by my employer... Due to a wording error in the survey, the formal feature of the formality dimension was removed from the analyses. 4. Results and discussion 4.1. Examination of the features measure To examine the properties of the new features measure, we first conducted analyses to determine whether the feature pairings (e.g., short-term and long-term) reflect bipolar dimensions. We found significant negative correlations for five dimensions: symmetry (equal and unequal; r =.24, p <.01), explicitness (explicit and implicit; r =.21, p <.01), level (individual and collective; r =.19, p <.01), time frame (short-term and long-term; r =.18, p <.01), and negotiation (negotiated and imposed; r =.14, p <.05). Of the remaining feature pairings, those reflecting scope (narrow and broad; r =.34, p <.01) and flexibility (flexible and static; r =.32, p <.01) had significant positive relations, and there was no significant correlation between the tangibility feature pairings (i.e., tangible and intangible). Even features that seem clearly opposite (e.g., shortterm and long-term) were only moderately negatively correlated. This might be attributable to some participants being uncertain about the feature and therefore giving neutral to low ratings for both. In light of these correlations, we conducted subsequent analyses using the individual feature ratings rather than combining ratings to yield dimension scores. Next, we conducted a principal components analysis with varimax rotation to determine whether the 17 feature ratings could be reduced to a smaller set of factors. We expected there to be some dependencies among the features, but had no a priori predictions with regard to the factor structure. To determine how many factors to retain, a parallel analysis was first performed (Hayton, Allen, & Scarpello, 2004). In a parallel analysis, eigenvalues from the actual data set are compared to the eigenvalue average of 100 random data sets with the same number of variables and participants. Only the first three factors were larger than the corresponding random eigenvalues, suggesting that three factors are present in the data. These three factors accounted for 52% of the total variance. Factor 1 was defined by 10 of the feature scales: tangible, static, explicit, collective, narrow, equal, negotiated, broad, longterm, and flexible (see Table 1). Interestingly, some of these features (e.g., tangible, static, narrow) have been used to describe transactional contracts, whereas others (e.g., broad, long-term, flexible) have been used to describe relational contracts (Conway & Briner, 2005; Rousseau, 1990). Indeed, when we correlated scores on this factor with the transactional and relational contract scale scores, it correlated significantly positive with both (r =.34 and.47, p <.01, respectively). Consequently, this factor might best be described as reflecting the balanced contract that Rousseau (2000; Ho et al., 2006) suggested is becoming increasingly prevalent in modern organizations. Factor 2 was defined by the unequal, short-term, trust-based (negative), and imposed features. This might be considered a form of transactional contract, but one that is seen as dictated primarily by the interests of the organization. Interestingly, however, it did not correlate significantly with transactional scale scores (r =.09, ns), but did correlate significantly negative with relational scores (r =.46, p <.01). We labeled this factor organization-centered. Factor 3 was defined by three features individual, intangible, and implicit and may reflect a form of the individually-negotiated I-Deal proposed by Rousseau (2005). Therefore, we refer to this factor as individualized. Scores on this factor did not correlate significantly with either the transactional or relational contract scores (both r =.10, ns).

7 K.J. McInnis et al. / Journal of Vocational Behavior 74 (2009) Table 1 Factor loadings for psychological contract features in Study 1. Psychological contract feature Factors I II III Tangibility tangible Flexibility static Explicitness explicit Level collective Scope narrow Symmetry equal Negotiated negotiated Scope broad Time frame long-term Flexibility flexible Symmetry unequal Time frame short-term Formality trust-based Negotiated imposed Level individual Tangibility intangible Explicitness implicit Note: Loadings for features used to interpret the factors are in bold. In sum, the results of our preliminary assessment of the new feature measure suggest that the features are best considered individually rather than being combined to reflect feature dimension scores. It also appears that there are relations among the features that perhaps reflect different types of contract. Interestingly, two of the largest factors appear to reflect what Rousseau and her colleagues refer to as balanced contracts and I-Deals both of which are relatively new contract forms developed in response to changes in the modern workplace Test of hypotheses We tested our study hypotheses by examining the correlations between the individual feature ratings and both AC and NC. The means, standard deviations, and correlations are reported in Table 2. As predicted, we found a positive correlation between AC and the following contract feature ratings: trust-based, equal, broad, tangible, long-term, collective, and negotiated. We also predicted and found a negative correlation between AC ratings and the short-term, imposed, and unequal feature ratings. Although not predicted, AC correlated significantly positive with the narrow and explicit feature ratings and significantly negative with the implicit feature ratings. Interestingly, AC correlated positively with ratings of both the Table 2 Correlations between psychological contract features and organizational commitment Studies 1 and 2. Psychological contract feature Organizational Commitment Prediction Affective Prediction Normative Study 1 Study 2 Study 1 Study 2 Explicitness explicit NP.23 ** **.09 Explicitness implicit NP.18 **.30 ** NP Flexibility flexible NP.18 **.07 NP.23 **.10 Flexibility static NP.22 **.36 ** NP.34 **.41 ** Formality formal NP Formality trust-based +.51 **.38 ** +.53 **.42 ** Level individual NP * +.16 **.07 Level collective +.18 **.24 ** NP.24 **.20 * Negotiation negotiated +.14 *.21 * +.18 **.31 ** Negotiation imposed.27 **.28 **.19 **.20 * Scope narrow NP.23 **.28 ** NP.34 **.19 * Scope broad +.40 **.22 ** +.42 **.23 ** Symmetry equal +.47 **.47 ** +.49 **.59 ** Symmetry unequal.24 **.22 **.15 **.18 * Tangibility tangible +.29 ** **.27 ** Tangibility intangible NP * NP * Time frame short-term.31 **.36 **.27 **.29 ** Time frame long-term +.29 **.25 ** +.32 **.22 ** * p <.05 (2-tailed). ** p <.01 (2-tailed).

8 172 K.J. McInnis et al. / Journal of Vocational Behavior 74 (2009) flexible and static features two features presumably on opposite poles of the same dimension. Examination of the items led us to suspect that participants may have interpreted the static item as reflecting the length of the contract rather than stability of the terms over time. We addressed this issue in Study 2. Overall, our predictions with regard to AC were supported. Turning to NC, we found significant positive correlations with the following contract feature ratings: trust-based, equal, broad, tangible, long-term, explicit, negotiated, and individual. NC also correlated significantly negative, as predicted, with short-term, imposed, and unequal feature ratings. Although not predicted, yet similar to AC, NC also correlated significantly positive with the narrow, collective, flexible, and static feature ratings. Given that AC and NC are typically highly related, as they were in this study (r =.78, p <.01), it is interesting to compare the two in terms of their relations with the contract feature ratings. The correlations for AC and NC were similar for 15 of the 17 feature ratings. The features that correlated differently between the two commitment components were the individual and implicit ratings. As expected, a significant positive relation was found between the individual feature ratings and NC, but not AC. That is, when an employer provides an individually negotiated contract, the employee, in turn, feels obligated to the organization, but may not necessarily feel affectively attached to the organization. This finding may have important implications for I-Deals because it suggests that individualized contracts do not influence the employee s level of affective attachment to the organization. We initially predicted that the explicitness of the contract would relate more strongly to NC than to AC because it might instill a sense of the need to reciprocate. However, AC and NC were equally related to the explicit feature ratings (r =.23, p <.01). With respect to the implicit feature ratings, however, AC correlated significantly negative (r =.18, p <.01) and there was no significant relation with NC. To test Hypothesis 3, we conducted regression analyses to determine whether the feature measures contribute uniquely to the prediction of AC and NC when contract type (transactional and relational) and perceptions of employer contract fulfillment were controlled. We used the feature component scores obtained previously for this analysis to reduce the number of predictors and increase power. The results are reported in Table 3. As expected, the regression analyses revealed a significant increase in R 2 when the contract feature scores were added to the equation for AC (DR 2 =.09, p <.01) and NC (DR 2 =.10, p <.01). The feature component scores that contributed uniquely to AC were organization-centered (b =.46, p <.01) and balanced (b =.24, p <.01). For NC, all three feature component scores contributed significantly to prediction: organization-centered (b =.41, p <.01), balanced (b =.35, p <.01), and individualized (b =.24, p <.01). In sum, our hypotheses were generally supported. AC and NC were found to relate as expected to many of the contract features. There were also a few significant correlations that were not predicted, and these will require replication before attempting interpretation. The features scale might also need some refinement, particularly in light of the fact that commitment correlated similarly with the ratings of both the flexible and static features. As noted, participants may have interpreted the static item in terms of length rather than stability of the contract. In addition, refinement of the narrow and broad feature items may be warranted because these features correlated significantly positive with each other. We suspect that, as worded, respondents may have been indicating that their contracts included both narrow and broad terms. With respect to the regression analysis results, we found that perceptions of the features of a contract contribute over and above perceptions of fulfillment and the type (i.e., transactional and relational) to the prediction of AC and NC. Finally, the pattern of relations was similar for AC and NC, with a few exceptions, including one that was predicted in advance. 5. Study 2 We had two primary objectives for Study 2. The first was to replicate and extend the findings of Study 1. The second goal was to refine the features measure. In addition to correcting the wording error in the formal feature item, we modified two feature items static and narrow to better reflect the intended constructs. Specifically, in Study 1 we noted that the static item might have been interpreted as reflecting the length rather than the unchanging nature of the contract term. Therefore, we revised the wording of this item for Study 2. Similarly, in Study 1 we found that ratings of narrow correlated positively Table 3 Regression analyses for predicting affective and normative commitment in Study 1. Control variables Employer psychological contract fulfillment.32 **.20 ** Transactional psychological contract Relational psychological contract * Psychological contract factors Balanced.24 **.35 ** Organization-centered.46 **.41 ** Individualized ** DR 2.09 **.10 ** Note: Standardized beta coefficients, * p <.05, ** p <.01. AC b NC b

9 K.J. McInnis et al. / Journal of Vocational Behavior 74 (2009) rather than negatively with ratings of broad. Given that our intent was to determine whether the contracts placed an emphasis on narrow (job-focused) versus broad (person-focused) terms, we also modified the wording of the narrow feature item for Study 2 (see Appendix). Finally, we also revised several of the items to make them shorter and easier to read. The revised items are included in the Appendix Sample and procedure As in Study 1, participants were recruited through StudyResponse. Of 2000 individuals targeted by StudyResponse, 147 accepted the invitation to participate, and met our inclusion criteria (see Study 1). Again, this may be a conservative estimate because we cannot be sure how many individuals actually received the initial request to participate and were currently employed Demographics of participants The mean age of participants was 38.6 years (SD = 10.4) with 50.7% being men. The majority of participants indicated that they worked full-time (75.5%). A variety of educational backgrounds were represented: less than high school (9.3%), high school (10.7%), associates degree (22.9%), some college (5.0%), four-year college degree (24.3%), some graduate school (7.1%), and a post-graduate degree (20.7%). There was also a variety of workplace sizes among participants that was very similar to Study 1: employees (65.7%), employees (11.7%), and over 250 employees (22.6%). Participants had an average tenure of 5.8 years in their organization and 7.5 years in their current profession/occupation. As in Study 1, demographic comparisons between responders and non-responders, and between responders and the U.S. working population, revealed little difference in gender, employment status, or occupational type, similar to Study 1. However, the mean age of responders was significantly older than non-responders (38.6 and 34.6, respectively), t(1997) = 2.45, p <.05. Responders were also more educated than non-responders, with relatively more responders having at least a college degree, v 2 (7) = 21.91, p <.01. Similarly, responders were more educated than the U.S. working population (i.e., 75% vs. 39.6% had a least a college degree or diploma), and there were fewer individuals in the age group among responders (25.2% vs. 38.4%) Measures The organizational commitment (AC: a =.77, NC: a =.85), psychological contract type (transactional: a =.63, relational: a =.85), and psychological contract fulfillment (a =.91) measures were the same as in Study 1. For the psychological contract type measure, we also included a slightly reworded 6-item version of Rousseau s (2000) PCI balanced contract type (a =.91). As mentioned, we modified the wording of the narrow and static feature items and made the aforementioned changes to clarify the wording of the feature measures (see Appendix). 6. Results and discussion 6.1. Examination of the revised feature measure Again, we began by conducting correlation and principal components analyses to assess the bipolarity of the feature dimension pairings and dependencies among the feature ratings, respectively. We found significant negative correlations for three dimension pairings: level (individual and collective; r =.49, p <.01), explicitness (explicit and implicit; r =.29, p <.01), and symmetry (equal and unequal; r =.25, p <.01). Despite our efforts in modifying items, we did not find evidence for bipolarity for the flexibility (flexible and static; r =.01, ns) or scope dimension pairings (narrow and broad; r =.16, p <.05). The correlations between the features reflecting the remaining dimensions were near zero. These findings are generally consistent with those of Study 1 and therefore raise questions about the assumptions underlying the Sels et al. (2004) measure that features reflect bipolar dimensions, and that assessing the presence of one assumes the absence of another. Similar to Study 1, a parallel analysis concluded that three factors are present, accounting for 45% of the total variance in the feature ratings. Factor 1 was a unipolar factor defined by the broad, long-term, trust-based, equal, intangible, negotiated, and flexible ratings (see Table 4). This factor appears to reflect the relational contract as described by Rousseau (2000). Indeed, scores on this factor correlate significantly positive with relational (r =.45, p <.01) and balanced (r =.45, p <.01) contract scores. Although it also correlated positively with transactional scores, the correlation was considerably weaker (r =.24, p <.01). Factor 2 was also a unipolar factor defined by the unequal, short-term, individual, narrow, implicit, formal, and imposed ratings. This factor appears to be organization-centered but different across employees. Scores on this factor correlate significantly positive with transactional contract scores (r =.19, p <.05), significantly negative with relational contract scores (r =.37, p <.05), and negative, albeit not significantly, with balanced contract scores (r =.16, ns). Factor 3 was defined by four feature ratings tangible, explicit, static, and collective features often associated with transactional contracts. However, scores on this factor correlated with all three contract type scores, but more strongly with relational (r =.37, p <.01), than with balanced (r =.27, p <.01) and transactional scores (r =.22, p <.01).

10 174 K.J. McInnis et al. / Journal of Vocational Behavior 74 (2009) Table 4 Factor loadings for psychological contract features in Study 2. Psychological contract feature Factors I II III Scope broad Time frame long-term Basis trust-based Symmetry equal Tangibility intangible Negotiated negotiated Flexibility flexible Symmetry unequal Time frame short-term Level individual Scope narrow Explicitness implicit Negotiated imposed Basis exchange-based Tangibility tangible Explicitness explicit Flexibility static Level collective Note: Loadings for features used to interpret the factors are in bold. In sum, the findings of this study reaffirm our earlier observation that contract features are best considered individually rather than combined to reflect dimension scores. They also confirm that there are associations among the features that might reflect different types of contract, although not necessarily the types that are most commonly discussed in the psychological contracts literature. We return to this issue in the general discussion Test of hypotheses We again tested our study hypotheses, as stated in Study 1, by examining the correlations between the individual feature ratings and both AC and NC. The means, standard deviations, and correlations are reported in Table 2, along with those of Study 1. As predicted, we found positive correlations between AC and the following contract feature ratings: equal, trustbased, long-term, collective, broad, and negotiated. We also predicted and found a negative correlation between AC ratings and the short-term, imposed, and unequal feature ratings. Although not predicted, we found that AC also correlated positively with ratings of the static and intangible features and negatively with implicit, narrow, and individual feature ratings. Contrary to prediction, we did not find a significant positive correlation between AC and the tangible feature ratings. Overall, however, our predictions with regard to AC were generally supported. Our findings were also consistent with those of Study 1. Of the hypothesized relations, 9 of the 10 features were confirmed in both studies (see Table 2). Also as predicted, NC correlated significantly positive with the equal, trust-based, negotiated, tangible, broad, and longterm feature ratings, and negatively with the short-term, imposed, and unequal features. Although not predicted, NC also correlated significantly positive with the static, collective, and intangible feature ratings and significantly negative with the narrow feature rating. Contrary to prediction, NC did not correlate significantly positive with the formal, explicit, and individual features. Similar to AC, relations between the features ratings and NC were generally consistent across the two studies. Indeed, 9 of the 12 predicted relations were supported in both studies (see Table 2). Table 5 Regression analyses for predicting affective and normative commitment in Study 2. Control variables Employer psychological contract fulfillment.27 *.17 Transactional psychological contract Relational psychological contract * Psychological contract factors Relational Organization-centered.24 *.15 Transactional DR 2.05 *.02 Note: Standardized beta coefficients, * p <.05, ** p <.01. AC b NC b