资源描述
外文文献 Whats Loyalty? Michael J. Withey 1 and William H. Cooper Loyalty in organizations has proved difficult to predict. One reason is that loyalty is complex and poorly understood. We report two studies that attempt to understand and predict loyalty by focusing on two components of the construct: active-constructive loyalty and passive-constructive loyalty. In the first study, we found that active acts of loyalty were predicted by variables quite different from those that predicted passive loyalty. The second study found that people identified by peers as high-loyalty employees performed many more active sets of loyalty than did those who were identified as low-loyalty employees. We conclude that loyalty consists of both active-constructive and passive-constructive behavior. KEY WORDS: loyalty ; commitment; active; passive.INTRODUCTION What is loyalty? In this article, we will suggest some ways to consider this question. Our starting point is Hirschmans (1970) treatment of exit, voice, and loyalty. Hirschman offers exit and voice as distinct responses when firms, organizations, and states are facing decline. What Hirschman means by loyalty is less clear. He first refers to loyalty as a form of attachment that makes voice more attractive when exit is available (Hirschman, 1970, p. 77). He describes loyalty as the product of (primarily economic) factors that wed the individual to the organization, making exit costly and reducing voice. At times, Hirschman describes loyalty as an attitude that affects the extent to which exit and voice are made use of. At other times he speaks of loyalty as a form of behavior in which the individual sup- ports the organization. Other treatments of loyalty have also been varied. They include, for example, Rusbult, Farrell, Rogers, and Mainouss (1988) characterization of loyalty as pas- sive-constructive behavior (e.g., being quietly supportive and being patient); Kolar- ska and Aldrichs (1980) work, which, referring to the response, silence rather than loyalty; and Grahams (1990) discussion of loyalty as an attitude without any behavioral component. These depictions of loyalty mirror the ambiguity of the construct in ordinary language. We sometimes speak of loyalty as an attitude, other times of loyalty as behavior. Within the loyalty behavior domain, there are both active elements (doing things that are supportive of someone or something) and passive elements (being quiet while exhibiting patient forbearance). In the present research, we will consider loyalty as a behavior and discuss the attitudinal elements of loyalty in terms of organizational commitment, defined as sharing the values of the firm (Porter, Steers, Mowday, & Boulian, 1974). Further, we will attempt to refine our under-standing of loyal behavior by distinguishing between the active and passive elements of loyalty. Summary of the Loyalty Literature Recently, several studies have considered Hirschmans loyalty construct. Far- reU (1983) classified exit, voice, loyalty, and neglect with the use of multidimensional scaling techniques. This study measured loyalty in passive terms, such as quietly doing my job and letting higher-ups make the decisions, and found that loyalty is indeed conceptually separable from the other responses. Loyalty, however, did not conform exactly to expectations, being shown to be passive (as expected) but slightly destructive to the organization (not constructive, as expected). This unexpected clas- sification of loyalty recurs in subsequent studies, raising questions about the true nature of the loyalty construct and how to measure it.Loyalty has been investigated in two recent studies of exit, voice and loyalty. In the first, Rusbult et al. (1988) found support for investment model predictions of loyalty. Specifically, loyalty was more likely to occur under conditions of high prior satisfaction, high investments, and relatively few alternatives. This view of loy- alty is consistent with a passive, constructive construct. In the second study, Withey and Cooper (1989) found quite different results. Loyalty was associated with low prior satisfaction, low investment, low organizational commitment, an external locus of control, and the belief that improvement in the situation was not likely. No re- lationship with alternatives was reported. Thus, loyalty takes on a clearly destructive appearance in this study. These results led Withey and Cooper to look to both measurement and conceptualization of loyalty as explanations. The measurement problem has been described by Cooper, Dyke, and Kay (1990) in terms of construct validity: loyalty has been operationalized in ways that do not match the loyalty construct. While loyalty is defined as supporting the or-ganization, items used to measure loyalty are too narrow and do not conform to most peoples notion of support. The Cooper, Dyke, and Kay study used the act frequency methodology (Buss & Craik, 1983) to assess the prototypicality of a num- ber of acts of loyalty, including the acts used by Farrell (1983) to measure loyalty. Interestingly, the three items used in both the Farrell and the Withey and Cooper studies cited above (e.g., say nothing to others and assume things will work out) ranked 99th, 101st, and 102nd among 103 acts of loyalty. More prototypical were acts such as give something extra when the organization needs it (lst) and do things above and beyond the call without being asked (4th). The Cooper et al. study supports the claim that there is a construct validity problem and suggests that previous research on loyalty has left much of the domain of loyalty unassessed. One promising way to approach this problem is to make a distinction between active and passive loyalty. This distinction was first raised by Farrell (1983) in categorizing the exit, voice, loyalty, and neglect responses, but it may also prove useful in interpreting loyalty. The distinction between active and passive is also evident when the items studied by Cooper et al. (1990) are examined. That is, a key difference between the prototypical loyalty items and the Farrell loyalty items is that the former are active while the latter are passive. In summary, more attention is needed at both the conceptual and the mea-surement level. On the conceptual level, loyalty needs to be investigated as both active and passive (and possibly as both constructive and destructive, again following Farrells categorization). On the measurement level, distinct measures of the dif- ferent forms of loyalty are necessary. The present research is designed to begin to answer these needs.Goals of the Present Research To begin the investigation of active and passive loyalty, the present research poses the following general questions. First, can differential predictors of active- constructive and passive-constructive loyalty be identified? Second, what is the re- lationship between active and passive loyalty and to what extent are they related over time? Third, in what ways are the actions of people who are defined as loyal different from actions of people who are defined as not being loyal? Two studies were conducted to address these questions. STUDY 1 The first study is a cross-sectional study which measures the different forms of loyalty and a set of independent variables that are thought to predict loyalty. Because Study 1 is a follow-up study on the sample reported by Withey and Cooper (1989), it is possible to assess relationships among active and passive loyalty over a six-year time lapse. It is also possible to conduct a six-year longitudinal study of the predictors of active and passive loyalty. Methodology Data were collected during the summer of 1990 through a survey mailed to those graduates of the Queens University Bachelor of Commerce program who had, in 1984, participated in the study reported by Withey and Cooper (1989). Of the 360 potential respondents, 210 returned their questionnaires representing a re- sponse rate of 58.0%. Sixty-eight percent of the respondents were male, all had undergraduate degrees, their average age was approximately 33 and their average work experience was approximately 10 years. Descriptive statistics for all measures appear in Table I. All, excepting locus of control which uses a 23-item forced choice format, are 5-point Likert scales. The independent variables were measured in terms of agreement with the statement; the loyalty measures were based on frequency of engaging in the behavior. With the exceptions noted below, all scales achieved a satisfactory level of interitem re-liability. Low internal consistency would be expected to attenuate correlation coef- ficients. The independent variables .and the measure of passive loyalty used were those reported by Withey and Cooper (1989). The measure of active loyalty was based on Cooper et al. (1990). A description of the variables follows. Examples of scale items are provided when the scale is new to the literature. A total of nine independent variables was included. Exit costs include skill specificity (e.g., My present job involves skills which would be useful in many other organizations, reversed) and sunk costs (e.g., If I left this job, my pension plan losses would be significant.), which are both aspects of Beckers (1960) side bets, and investments (Rusbult et aL, 1988). The sunk costs scale had low internal con-sistency. Voice costs asked about the effort required to bring about change and the likelihood of punitive responses to such efforts (e.g., It is risky to say too much about working conditions in this office). Belief in the possibility of improvement was measured by a 9-item combi- nation of measures of receptiveness to change (e.g., This office is organized so as to be receptive to employee input) and Graen, Liden, and Hoers (1982) measure of leader/member relations. Job satisfaction was measured using Brayfield and Rothes (1951) 18-item scale. Locus of control was measured using Rotters (1966) standard 23-item forced choice instrument. Organizational commitment was measured using Porter et al.s (1974) scale. The availability of alternative jobs was measured using the two items of Price and Bluedorns (1979) scale that refer to alternatives that are better than the present job. Analysis was conducted by using correlations and regression analysis to assess the ability of the independent variables to predict each form of loyalty. Further, correlations between independent variables and loyalty measured in 1984 and forms of loyalty measured in 1990 are reported. Several independent variables were associated with passive loyalty (see Table II). The passive loyalist emerged as being dissatisfied and uncommitted, having a relatively external locus of control, and facing high costs of change that seemed unlikely. This pattern of results replicates the findings of Withey and Cooper (1989). Quite different results were found for active loyalty. Active loyalty was asso- ciated with high commitment and satisfaction, a belief that improvement is possible and a set of costs that favor action and not passivity. Specifically, active loyalty was associated with high psychological exit costs (i.e., investment) but low economic exit costs (i.e., skill specificity and sunk costs) and low voice costs. It is noteworthy that, for most of the independent variables, the signs of the correlation coefficients are different for active and passive loyalty. Finally, the zero-order correlation between the two forms of loyalty is negative and significant .Because the present study is a follow-up of an earlier study, it is possible to present six-year longitudinal results. All measures described above (except active loyalty) were measured in the earlier study. In spite of the long time lag, a strong positive correlation was found between passive loyalty in 1984 and in 1990 , and a negative correlation was found between passive loyalty in 1984 and active loyalty in 1990 . Further, some modest correlations were found between the independent vari-ables measured in 1984 and both forms of loyalty measured in 1990. Specifically, passive loyalty was negatively related to belief in the possibility of improvement, organizational commitment , and locus of control , and positively related to skill specificity . Active loyalty was related only to skill specificity . Study 1 provides clear evidence for two distinct types of loyalty. First, the findings for passive loyalty that Withey and Cooper (1989) reported as counterin- tuitive were replicated. The earlier study had expected to find loyalty as a passive but essentially supportive behavior; instead loyalty emerged as something that re-sembled entrapment rather than support. Loyalty as entrapment was found in the present study as well, but a more constructive and active form was identified as an alternative behavior. The contribution of Study 1, then, is to distinguish between two types of loyalty and to demonstrate that each is associated with different in- dividual characteristics. These findings appear more robust when it is realized that at least some of them hold over a six-year time lag. The results suggest future directions for research. In particular, it becomes necessary to relate the different forms of loyalty to other behavioral constructs within the exit, voice, and loyalty framework. While passive loyalty resembles ne-glect, active loyalty resembles voice. Such relationships need clarification in future research.In addition, the loyalty typology could be broadened to investigate constructive and destructive forms of loyalty as well. Given that clear differences have been found between active and passive loyalty, there is reason to expect that loyalty is a multi-dimensional construct. The potential exists to identify four distinct types of loyalty, each with its own unique characteristics. For example, the active constructive type of loyalty might resemble voice while passive constructive loyalty would be similar to Hirschmans notion of quiet support. Passive destructive loyalty would seem to resemble neglect, and active destructive loyalty, while harder to conceptualize, may be considered in terms such as performing illegal acts suggested by the organization. STUDY 2 Study 2 is designed to assess whether people who are perceived as loyal en-gage in different acts than do people who are perceived as not being loyal. In essence, the study asks what it is that loyal people do that defines them as loyal. To assess differences between employees with high and low loyalty, 200 questionnaires were mailed to recent graduates of Queens Universitys School of Business. One hundred questionnaires asked respondents to think of a target person who, to them, typified high organizational loyalty (defined for respondents as pro- viding support for the organization, including active and passive support); the remaining 100 questionnaires asked respondents to think of a target person who, to them, typified low organizational loyalty. Respondents were then asked how frequently the target person engaged in a variety of different behaviors. Included in the behaviors were 100 loyalty acts and several additional behaviors. The analysis that is reported below is based on the initial 30 questionnaires that were returned (representing a response rate of 15%). Sixty-three per cent of the respondents were male. The average age of the respondents was 35.4 years and they had an average work experience of 11.8 years. The respondent had observed his or her target person for an average of 5.4 years. The average age and organizational tenure of the target persons were 39.6 and 10.0 years, respectively. Eighty-three percent of the target persons were male. The questionnaire contained, among other things, 105 behavioral acts. The acts included the 100 loyalty acts reported in Cooper et al.s act frequency study (1990), Farrelrs (1983) three items, and two test acts designed to detect careless respondents. The analysis was designed to allow the identification of loyalty acts that dif- ferentiated between high and low loyalty. Comparisons were made to determine which items had significantly different scores for high- and low-loyalty target persons. All 105 acts were included in the analysis. Thus, the analysis will determine which active and/or passive acts of loyalty distinguish people who are loyal from those who are not. The results indicate that a clear pattern was found regarding which of the 105 acts differentiated between high- and low-loyalty target persons. In particular, the acts that were best able to differentiate were those that were in the top quartile of the Cooper et al. (1990) prototypicality study, that is, the most prototypical loyalty acts. Of the 10 most prototypical acts, all are clearly active in nature (and three comprised the active loyalty measure in Study 1) and all differentiated between high- and low-loyalty target persons . Loyalty acts that placed in the middle prototypicality quartiles were less able, and acts in the bottom quartile were unable, to consistently differentiate between high- and low-loyalty target persons. None of the three acts used to measure passive loyalty in Study 1 (those adapted from Farrell, 1983) nor the two test acts were able to differentiate people described as loyal from those described as not loyal. The failure of the test acts to differentiate lends credibility to the methodology employed in Study 2; the failure of the passive loyalty acts to differentiate will be discussed below. The results of Study 2 must be considered in only a tentative way because of the small sample size. Nonetheless, the clear pattern of the results, given the small sample, lends them credibility. The results of Study 2 are consistent with Buss and Craiks (1983) idea of validity gradients. For Buss and Craik, the centrality of acts for a domain is indexed by the acts prototypicality: the higher the prototypicality, the more central the act is to the domain. When acts are central, they are more readily predicted by the known pre- dictors of behavior in the domain; the more peripheral the act, the harder it is to predict them. Put differently, the validity of an act for a domain is on a gradient from high (central or prototypical acts) to low (peripheral or low prototypical acts). An implication is that central acts will have a higher construct validity than will acts at the periphery. This is precisely what we found. The top 25 acts all differed in the predicted direction for high- and low-loyalty individuals. The middle quartiles differed less, and the acts in the bottom quartile were the least successful. In addition, the Farrell acts (which ranked very low in prototypicality) did not differentiate between high and low loyalists. Thus, the more central the acts were, the more likely they were to be per- formed more by high loyalists than low loyalists. This was not true for the peripheral acts of loyalty, including the three Farrell acts. These results are as predicted by Buss and Craiks validity gradient concept. They point to the usefulness of the act frequency method as a basis for developing measures of organizational behaviors, including loyalty. The results of Study 2 increase our confidence in both the measures of loyalty used in Study 1 and our ability to predict the different forms of loyalty. Further, the distinction between active and passive loyalty made in Study 1 ig supported. Finally, because the most prototypical acts are much more active than the least prototypical acts, the conclusion can be drawn that it is active and not passive loyalty that is used by observers to determine loyalty in co-workers. The two studies reported in this article begin to distinguish between active and passive forms of loyalty. In Study 1, it w
展开阅读全文