The Equal Pay Act of 1963 prohibited employers from paying male and female employees differentially for the same work, and Title VII of the Civil Rights Act of 1964 further supported this prohibition on pay discrimination. Despite these pieces of legislation, women’s earnings continue to lag behind men’s, a phenomenon termed the gender wage gap. Although women’s earnings relative to men’s have increased since the late 1970s (Blau and Kahn 2007), women’s median weekly earnings were estimated at only 81 % of men’s in 2009 (U. S. Department of Labor, Bureau of Labor Statistics 2010). Some of this disparity is likely due to occupational segregation (Stevenson 1984) and devaluing of women’s work (Alksnis, Desmarais, and Curtis (2008), as well as to gender differences in salary negotiations (Kulik and Olekalns 2012). However, the gap persists even when characteristics such as age, education, and tenure are controlled (Judge and Livingston 2008; Major 1994; Ostroff and Atwater 2003). The gap is present even when men and women have comparable levels of performance, suggesting that the translation of performance evaluations into compensation remains a problem (Auster 1988). Moreover, women earn less in many fields, including academia, professional and technical work, executive positions, and government (Blau and Ferber 1985), and the gap appears in both male-dominated and female-dominated organizations (Buchanan 2005).

Although women in general are aware of and concerned about the gender wage gap (Major 1994), individual women do not report significantly greater dissatisfaction with their own pay or perceive their pay as less fair than men do (see Brown 2001; Crosby 1982; Graham and Welbourne 1999; Jackson 1989; Major 1989). Crosby (1982) termed this phenomenon “the paradox of the contented female worker.” These findings are difficult to explain, considering that individual women recognize that women as a group are underpaid, and are concerned with this inequity (Crosby 1982). As Major (1994) stated, “Women should feel angry, resentful, and dissatisfied as a result of receiving fewer valued rewards than do men doing similar work” (p. 295). But studies find little evidence of such expected reactions (Major 1994). Instead, women appear satisfied with their lower compensation.

Why women remain satisfied with such inequity has been examined by various researchers (e.g., Crosby 1982; Jackson 1989; Major 1994; Phelan 1994). Some have used traditional theories of satisfaction such as equity theory (Adams 1965) or discrepancy theory (Lawler 1971) to explain women’s satisfaction with lower pay, whereas others have examined this issue from more sociological perspectives such as relative deprivation (e.g., Crosby 1982; Jackson 1989), system justification (e.g., Blanton, George, & Crocker, 2001), or personal entitlement (Major 1994). The current study reviews these various approaches and tests a model based primarily on Major’s model of personal entitlement, but with several unique aspects. First, this study tests a series of relationships identified in Major’s model, from the selection of referents, to measurement of referent standards, to feelings of entitlement and pay satisfaction, whereas previous research has typically only tested parts of her model (e.g., bivariate relationships). Second, the possibility that men and women differ in how much they value pay or money was investigated, and whether the value placed on money is related to pay satisfaction was tested. Importantly, this study used a measure of how much participants valued money itself, in contrast to previous research that has examined the importance of pay relative to other work attributes. In addition, this study uses a sample of working adults, providing an important extension to the field of the previous laboratory work on entitlement (e.g., Hogue and Yoder 2003; Major, McFarlin, and Gagnon 1984; Major and Testa 1989). Thus, in this study women’s and men’s actual referent choice, entitlement, and satisfaction were examined.

Theories of Pay Satisfaction

Generally, distributive justice-based theories of pay satisfaction (e.g., Adams’ equity theory 1965; Lawler’s discrepancy theory 1971) view satisfaction or fairness perceptions as the result of a comparison between an individual’s outcomes and inputs and the outcomes and inputs of another individual, or referent. Thus, from these perspectives women could be satisfied with their lower pay for various reasons, such as perceiving themselves to have lower inputs, choosing different referents, engaging in cognitive distortion, etc.

In addition to general theories of pay satisfaction, other theories or models have been proposed specifically to explain women’s satisfaction with lower pay. Crosby (1982) used the concept of relative deprivation together with discrepancy theory, to explain women’s satisfaction with lower pay. Relative deprivation theories attempt to explain what conditions are necessary for individuals to feel deprived or dissatisfied (Taylor and Moghaddam 1987). Two types of deprivation traditionally have been examined, fraternal (feelings of group deprivation relative to other groups) and egoistical (feeling that oneself is deprived). Women tend to exhibit fraternal deprivation, in which they feel that women as a group are paid less than men are, but not egoistical deprivation, in which individual women feel deprived (Major 1994). Using relative deprivation theory, Crosby (1982) emphasized that individuals will experience egoistical deprivation if they both want a particular outcome and feel that they deserve the outcome but do not receive it. In other words, they must perceive a discrepancy between the outcomes they want and the outcomes they receive, and between the outcomes they feel entitled to and the outcomes they receive. Thus, Crosby’s explanation of women’s pay satisfaction using relative deprivation is a discrepancy theory approach, focusing on the discrepancies between actual pay and both the amount of pay wanted as well as the amount of pay that should be received. Regarding women’s pay satisfaction, then, Crosby (1982) suggested that women could be contented with lower pay because they do not value pay as highly as men do (i.e., perceive no discrepancy between the pay they want and receive) or feel entitled to less pay than men do (i.e., perceive no discrepancy between the pay they feel they deserve and receive). Crosby’s use of relative deprivation theory has provided a basis for later approaches (Jackson 1989; Major 1987, 1994; Phelan 1994) to understanding this issue. In addition, several researchers (e.g., Mueller and Kim 2008; Phelan 1994) have proposed what will be termed here as “bivariate” explanations for the paradox, that offer competing hypotheses of how men and women differ to explain the paradox. (They are termed bivariate here as they generally look at relationships between gender and a variable in isolation, rather than as part of a more complex model.) For example, Phelan (1994) identified five alternative explanations for the paradox. Mueller and Kim (2008) review both specific theories of job satisfaction as well as a number of other explanations directed specifically at explaining the paradox. Both of these approaches have similarities to or overlap with aspects of both Jackson’s (1989) and Major’s (1987; 1994) models, which are more comprehensive models that propose how some of these competing hypotheses might operate causally. Jackson’s (1989) model bears some similarity to Major’s (1994) model, but Jackson focused more on gender differences in the value of pay as an influence on pay satisfaction, whereas Major (1987; 1994) focused on the antecedents to entitlement, which in turn leads to pay satisfaction. Each of these approaches to understanding the paradox is discussed in turn.

“Bivariate” Explanations

Based on various lines of research (e.g., equity theory, Adams 1965; work on self-pay, Major and Konar 1984), Phelan offered five possible explanations for the paradox. First, derived from equity theory (Adams 1965), she proposed that women have lower job inputs than men (differential job inputs hypothesis). Second, also derived from equity theory, she suggested that women choose same-sex referents (own-gender referents hypothesis). Third, based on work on self-pay (e.g., Major and Konar; Major, McFarlin, and Gagnon 1984), she suggested that women expect to receive less than men and consider a lower reward to be fair given the same inputs (differential entitlements hypothesis). Fourth, based on work on self-reported work values (e.g., Crosby 1982), she proposed that women place less importance on pay than men (differential job values hypothesis). Fifth, she suggested that subjective job characteristics, rather than objective job characteristics, affect satisfaction (e.g., Phelan, Bromet, Schwartz, Dew, and Curtis 1993), and that men and women differ less on these subjective characteristics (subjective rewards hypothesis).

In her study, Phelan (1994) failed to find evidence supporting the first four hypotheses, but did find support indicating that subjective job rewards influenced satisfaction. However, her study focused on organizational satisfaction rather than pay satisfaction. Also, she examined objective measures of justice rather than subjective measures of justice, which fails to address the question of whether perceptions of justice influence satisfaction. Because of such concerns, Mueller and Wallace (1996) replicated Phelan’s (1994) study using perceptual measures of justice, and also investigated pay satisfaction explicitly. As with Phelan’s (1994) study, Mueller and Wallace found no support for the differential job inputs hypothesis, the own-gender referents hypothesis, and the differential job values hypothesis. Unlike Phelan’s study, they found no support for the subjective rewards hypothesis, but they did find some support for the hypothesis that men and women have different entitlements.

Mueller and Kim (2008) reviewed the evidence for Phelan’s five hypotheses, and added several hypotheses of their own. Specifically, they suggest that gender differences in disposition might be an explanation for the paradox, as there is evidence that disposition affects job satisfaction (e.g., Judge and Larsen 2001). They also propose a selectivity hypothesis, which assumes that women who enter the labor force are more likely to be satisfied because they chose to enter employment (rather than had to, as with men in general). This hypothesis has not been tested. A differential norms hypothesis proposes that men are more likely to rely on equity norms whereas women rely on equality norms. Finally, a multiple standards framework suggests that the actual reward is compared to a just reward, and that men and women would differ on both actual and just rewards. In essence, this framework could subsume the differential inputs, differential norms, and own-gender referent explanations (Mueller and Kim 2008). In many respects, this latter framework is consistent with the principles of discrepancy theory (Lawler 1971), which looks at the discrepancy between what amount of pay is received and what one believes pay should be. Mueller and Kim investigated some of these propositions in an international survey including data from 30 countries, and failed to find support for the differential values, subjective rewards, or differential entitlements hypotheses. Thus, they suggest that justice-based explanations (e.g., differential job inputs, own-gender referents, differential norms) are likely to be most promising in explaining the paradox.

Based on Major and Konar’s (1984) work, Keaveny and Inderrieden (2000) also tested some of these same factors to assess their effect on the gender-pay paradox. Specifically, they examined the effect of differences in career paths, jobs inputs, career importance, family financial need, and spouse’s earnings on pay satisfaction. Generally, they found little support for the effect of these various factors as accounting for gender differences in pay satisfaction. A more recent study (Smith 2009), conducted among university employees in the United Kingdom, also examined satisfaction with pay among men and women, but included the staff grade (i.e., job level). Administrative women and men each had higher satisfaction with pay than academic women and men, suggesting that the job level may be an important factor to consider.

It is important to note here that the explanations reviewed and tested here are based on separate (i.e., bivariate) relationships, and it is possible that a more complex model with multiple explanations incorporated into it will provide not only a richer consideration of the causes of the paradox, but would also be a more appropriately specified model of the relationships (see James, Mulaik, and Brett 1982). Insofar as Jackson’s (1989) and Major’s (1987; 1994) approaches to understanding the paradox place such hypotheses in specific models of the factors leading to pay satisfaction, and also propose justice-based explanations, they are likely to provide more useful approaches to understanding the paradox.

Jackson’s (1989) Model

In Jackson’s (1989) model of women’s pay satisfaction, she proposes that gender differences in the meaning of money impact pay satisfaction by influencing choice of referents, by affecting pay expectations, and by impacting the value placed on pay. Thus, central to her work is the question of what money means to men and women, and she suggests that Crosby’s (1982) question whether women and men differ in the value of pay remains unanswered because of a failure to consider the meaning of money (Jackson 1989). In addressing gender differences in the value of pay, in Nieva and Gutek’s (1981) review they concluded that men tended to value pay and promotions more than women, whereas women placed higher value on social rewards and pleasant working conditions. However, they suggested that many of these differences might have reflected the reality faced by men and women. For example, women’s preference for good working conditions could be due to sex segregation, as women tend to be concentrated in jobs with such conditions (e.g., pink-collar rather than blue-collar jobs). It should be noted that research on self-pay allocations in laboratory studies indicates that when dividing a joint reward with a co-worker, women allocate less pay to themselves than men do, even for the same level of task performance, suggesting that women do value pay less (see Major and Deaux 1982).

However, whether such studies (i.e., self-reported value, self-pay allocation) really measure the value of pay is in question. A single item (e.g., “How important is pay to you?”) or ranking may not be sufficient to accurately assess the value of pay (Jackson 1989), and allocation may indicate deservingness rather than value of pay. Further, the question’s phrasing (e.g., asking about preferences vs. perceptions of others’ preferences) may also impact findings (Rynes, Gerhart, and Minette 2004). For example, in a study by Jurgensen (1978), when men and women were asked to rank job attributes in order of importance to themselves, pay was fifth for men and seventh for women. However, when the participants were asked to rank the job attributes in terms of importance to someone just like them, pay was first for both men and women, strongly suggesting a social desirability bias on the part of the participants.

Other approaches such as policy capturing (Judge and Bretz 1992) or embedded narratives (Krumboltz, Blando, Kim, and Reikowski 1994) may have potential for assessing work values. Alternatively, other measures of value of pay (or money in general) might be useful as suggested by Jackson (1989). In her work, Jackson (1989) concluded that the meaning of money had affective, cognitive, and behavioral components, and that the affective meaning of money might be useful for examining gender differences in value of pay. For example, measures such as Furnham’s (1984) Money Beliefs and Behavior Scale or Tang’s (1992) Money Ethic Scale, which focus on the attitudes and feelings about money, have potential to assess the value of money more effectively than traditional ranking or rating measures of the value of pay.

Thus, Jackson’s (1989) most important contribution to understanding the paradox of women’s pay satisfaction may lie in her focus on how to better conceptualize and measure the value of pay. Major (1987; 1994), in contrast, while acknowledging that an individual’s valuation of pay relative to other work characteristics may influence feelings of deprivation, has criticized the emphasis on the role of values in determining men’s and women’s satisfaction. She notes that such research assumes that (1) women and men differ in what they want from jobs, (2) women and men get what they want from their jobs, and (3) gender differences in what women and men want and get from their jobs explain job satisfaction. The problem lies in that men and women do not always get what they want from their jobs, even if they do value different work characteristics. Thus, Major’s work in this area has focused more on entitlement.

Major’s (1994) Model

Major’s (1987; 1994) model of entitlement or deservingness is based on equity theory (Adams 1965) and relative deprivation theory (Crosby 1982). Entitlement, as Major (1994) has defined it, is “an expectation that one should or ought to receive something” (i.e., outcomes; p. 299). This expectation is based on the individual’s fulfillment of certain preconditions (e.g., inputs), and is accompanied by a “moral imperative.” Research from various sources suggests that men and women do differ in their sense of entitlement (Major 1994). As noted above in Phelan’s (1994) study, findings that women allocate lower pay to themselves suggest that women feel lower entitlement, as do studies showing that women work longer than men for the same pay in laboratory tasks. Studies of full-time workers also show that women report feeling they deserve lower wages than men, even when factors such as age, education, and experience are controlled (Major 1994).

In Major’s (1994) model, she describes two direct antecedents of entitlement that may explain women’s lower entitlement. Specifically, she proposes that differences in evaluations of preconditions and differences in referent standards lead to differences in entitlement. Differences in the evaluation of preconditions refers to the possibility that women devalue their performance or inputs, and thus believe that they have not contributed sufficiently to deserve higher outcomes. For example, the tendency to undervalue women’s work could lead women to undervalue such work in general and their own work in specific (Major 1994). Research demonstrating women’s lower self-pay allocations (e.g., Hogue and Yoder 2003) is consistent with this proposition.

The second antecedent, group differences in referent standards, refers to the lower standards for comparison (e.g., referent salaries, pay expectations) that women hold. Women’s lower pay expectations may begin even before they embark on a career (Hogue, DuBois, and Fox-Cardamone 2010). These differences in standards are presumed to be a result of making ingroup social comparisons (i.e., with other women). Researchers such as Levine and Moreland (1987) and Major (1994) have suggested that when an individual makes a social comparison, the choice of target for comparison is chosen based on structural characteristics of the target (e.g., proximity, visibility), the perceived similarity of the target, and the goals of the individual making the comparison. Specifically, an individual is more likely to compare to another individual who is proximate, similar on the attribute under comparison, and who will enable the individual to fulfill his or her goals with regard to the comparison. Thus, men would be more likely to compare with men, and women compare with women, because of perceived similarity and proximity, given that gender segregation persists in the workplace (Greenberg and McCarty 1990). Work by Ostroff and Atwater (2003) supports this proposition, showing that female managers were more likely to have female supervisors, subordinates, and peers. Given that men and women are likely to make ingroup comparisons, they should have different referent standards. These different standards would lead to differences in entitlement and satisfaction.

The current study tests portions of Major’s (1994) model of personal entitlement, but extends her model based on Jackson’s (1989) work by also considering that men and women might differ in their valuation of pay or money, and that value of money might subsequently serve as an antecedent of satisfaction. The proposed model is depicted in Fig. 1.

Fig. 1
figure 1

Proposed model of pay satisfaction and hypotheses

Full size image

Proposed Model of Pay Satisfaction

As Fig. 1 shows, gender is proposed to influence referent choice. As Major (1994) has discussed, individuals are more likely to make ingroup comparisons because of similarity and proximity. This is consistent with Phelan’s (1994) own-gender referents hypothesis. Thus, women are more likely to select women as referents, and men are more likely to select men, given their similarity and proximity in the workplace.

Hypothesis 1:

Gender will be related to choice of referent, such that women will prefer female referents and men will prefer male referents.

Next, referent gender is proposed to affect the comparison or reference standards that provide information for evaluating one’s pay. These reference standards are likely based on social comparisons with the referents (Major 1987), and thus referent salary is expected to be a primary basis for referent standards. Insofar as women tend to earn less than men do, the salaries of female referents should be less than the salaries of male referents, and thus the reference standards of male and female referents are likely to differ.

Hypothesis 2:

There will be a relationship between referent gender and reference standards, such that female referents will have lower salaries than male referents.

Major (1994) suggests that because women have lower reference standards, underpaid women will nonetheless have a lower sense of entitlement because they will feel they receive what they deserve. Thus, in her model, these reference standards (based on social comparisons) are the precursor to beliefs about deservingness or entitlement.

Hypothesis 3:

Reference standards will be positively related to entitlement, such that higher reference standards will be associated with higher entitlement.

Finally, Major (1994) as well as Jackson (1989) propose that feelings of entitlement are a precursor to evaluations of outcomes, affect, and behavior. Thus, individuals who have a lower sense of entitlement are expected to be more satisfied with their pay, because they are getting what they feel they deserve.

Hypothesis 4:

Entitlement will be negatively related to pay satisfaction, such that higher entitlement will be associated with lower satisfaction.

The model presented here also goes above and beyond Major’s (1994) model by including the value of pay or money as a possible precursor of pay satisfaction, and is consistent with Phelan’s (1994) differential job values hypothesis. As discussed previously, Jackson (1989) suggested that value of pay also will affect satisfaction, such that individuals who place a higher value on pay will be less satisfied with their pay. Although whether men and women truly differ in the value they place on pay remains in question, and has been criticized as an explanation for women’s satisfaction with lower pay (see Major 1987, 1994), an investigation of this issue with an improved measurement of value of pay is warranted. Specifically, focusing on the affective meaning of money (see Jackson 1989) could illuminate gender differences, given that research (e.g., Furnham and Lewis 1986) has found gender differences in the meaning of money. Given past research (e.g., Nieva and Gutek 1981), it is expected that men will place a higher value on money, and that individuals who value money more will experience more pay dissatisfaction.

Hypothesis 5:

Gender will be related to value of money, such that men will value money more than women will.

Hypothesis 6:

Value of money will be negatively related to satisfaction, such that higher value of money will be associated with lower satisfaction.

Finally, discrepancy theories (e.g., Lawler 1971), as well as the multiple standards framework (Mueller and Wallace 2008), propose that the discrepancy between the actual outcomes and what is considered fair forms the basis for satisfaction and fairness evaluations. For example, an individual who is highly paid will likely select referents of a similar pay level, and thus have a high referent standard and high perceived fair pay. A highly paid individual who has high standards also should feel less entitled and more satisfied than an individual who is lower paid, given that the discrepancy between their outcomes and what they expect is less. For these reasons, actual pay level was included as a covariate in the model. Because age and experience also affect one’s pay level, these variables were included as controls in the model.

Method

Sample

The sample targeted for this research consisted of approximately 1,650 attorneys employed full-time in Texas in the year 2000. Using a sample of attorneys allows for the investigation of whether men and women in a non-female-sex-typed job differ in their referents, expectations, and satisfaction, and also allows education level to be controlled, as a J.D. is the standard required degree for attorneys. Of the approximately 1,650 questionnaires distributed, 208 were returned, for a response rate of 12.6 %. Thirty-eight participants were removed from the sample because they were employed part-time, not employed as attorneys, or failed to answer large portions of the questionnaire, resulting in a final sample size of 170. A total of 64 respondents were female (37.6 %), and 106 were male (62.4 %). Participants’ annual salary from their job ranged from $30,000 to $700,000, with a mean of $123,714 and a median of $90,000. Participants’ age ranged from 25 to 65, with a median of 40.0 years.

Although the response rate was somewhat low in the current study, it is not uncommon to obtain such low response rates in surveys, particularly in mailed surveys (see Roth and Bevier 1998). To examine whether the sample was unrepresentative of the population, the composition of the current sample was compared to the Texas Bar membership at the time of the data collection (see Cannon and Priestner 2001). The current sample is slightly more female in composition (37.6 % vs. 27 %), slightly younger (40 vs. 45 years old), and less experienced (9.95 vs. 15 years). However, the average salary in the sample (i.e., $123,714) is not inconsistent with the Texas Bar statistics (e.g., $124,776 for corporate attorneys, $109,282 for private practitioners, and $63,124 for government attorneys) (see State Bar of Texas 2001a, 2001b, 2001c). When considering pay differences between men and women, in the current sample, females earned 64 % of what males earned. Among the Texas Bar members, this ratio was 74 % for corporate attorneys, 85 % for private practitioners, and 98 % for government attorneys. Thus, it appears that women were somewhat more underpaid relative to men in the current sample as compared to the Texas Bar membership, which may be due in part to the fact that women in our sample tended to have less experience than the men (8.01 years vs. 11.04 years). Taken altogether, the current sample is several years younger and several years less experienced than the Texas State Bar membership as a whole, and has a higher percentage of females than the overall Texas Bar membership, which is consistent with the fact that more females have been entering the legal profession. Despite these differences, it does not appear that the current sample is particularly unrepresentative of the Texas Bar membership as a whole. Indeed, the sample may be more representative of the Texas Young Lawyers’ Association (TYLA; for lawyers under 36 years in age), whose membership at the time was 39 % female (see Cannon and Priestner 2001).

Procedure

Each participant was given a survey packet containing a cover letter about the study, an informed consent form, a copy of the survey, and an envelope for returning the survey. The data were collected in several phases. The first phase of data collection involved mailing questionnaires with cover letters and consent sheets to a random sample of 1,250 male and female attorneys employed full-time in Texas, from which 44 surveys (3.5 %) were returned. In the second phase of data collection, law firms, bar associations, and district attorneys’ offices in the North Texas region were contacted. At those organizations willing to participate, questionnaires were provided to a contact person (e.g., a recruiting coordinator, office manager) for distribution to participants, and questionnaires were returned by mail to the researcher. A total of 112 surveys were distributed, and 27 (24.1 %) were returned. In the third phase of data collection, questionnaires were distributed in person to attorneys attending Continuing Legal Education (CLE) events hosted by the State Bar of Texas. Data were collected on three separate occasions. Attendees were approached as they registered for the events, and were asked if they would be willing to complete a questionnaire and were offered a candy bar. Approximately 275 questionnaires were distributed at these events, and 134 (54 %) were returned to the researcher in person or by mail.

Measures

Participant demographics.

Participants provided basic information, including gender, age, job title, job experience, level of education, and salary. Participant gender was coded as 0 = female and 1 = male. Participants’ self-reported salary, converted into annual salary figures and log-transformed to reduce positive skew, served as the measure of actual pay for the hypothesis tests. Age and experience were included in the model as control variables for actual pay, and were allowed to covary with participant gender.

Referent demographics.

Participants were asked to select a referent individual in their organization with whom they would be likely to compare their pay, and then to provide basic information about the referent, including gender, approximate age, education, job title, organizational level, years of experience in the job, and salary. Referent gender was coded as 0 = female and 1 = male. Referent salary, converted into annual salary figures and log-transformed, served as the measure of reference standards.

Feelings of entitlement.

The extent to which participants experienced feelings of entitlement to higher pay was assessed by three items. Two of these items were adapted from Crosby (1982): “Would you say that your pay is more than you deserve, similar to what you deserve, or less than you deserve?” and “In view of your training and abilities, is your present pay more than it ought to be, similar to what it ought to be, or less than it ought to be?” Both items were measured using a seven-point Likert-type scale ranging from 1 = Much less to 7 = Much more. A third item asked, “Do you think you are underpaid, overpaid, or fairly paid?” and was measured on a seven-point Likert-type scale, ranging from 1 = Very underpaid to 7 = Very overpaid. These three items were reverse-scored, such that higher scores indicated greater feelings of entitlement. The internal consistency reliability (alpha) was 0.95.

Satisfaction with pay.

Pay satisfaction was measured using the Pay Satisfaction Questionnaire (PSQ; Heneman and Schwab 1985). The PSQ consists of 18 items measuring satisfaction with pay level, benefits, pay raises, and pay structure/administration. Also of particular importance is the fact that the PSQ has also been found to exhibit measurement invariance across men and women (De Gieter, Hofmans, De Cooman, and Pepermans 2009). Each item was rated on a five-point Likert-type scale ranging from 1 = Very dissatisfied to 5 = Very satisfied. An overall score for pay level satisfaction was calculated by averaging the ratings for the four items on that scale. The internal consistency reliability for pay level satisfaction was 0.97.

Value of money.

Because traditional measures that assess multiple work values using a ranking or rating approach may be subject to various problems (see Brief and Aldag 1995), the 18 items on the Obsession dimension of Furnham’s (1984) Money Beliefs and Behavior Scale were used to assess the value of money (i.e., affective meaning of money). Some examples of items on this dimension are “Money is how we compare each other” and “I would do practically anything for money if it were enough.” The items were scored on a seven-point Likert-type scale ranging from 1 = Strongly disagree to 7 = Strongly agree. Item responses were averaged to obtain scale scores such that higher scores represented greater value placed on money. The internal consistency reliability of this scale was 0.83.

Results

Table 1 provides descriptive statistics and intercorrelations for the study variables. As in previous research showing the existence of a gender wage gap, men and women differed significantly in terms of actual salary with women reporting significantly lower actual salaries (t = -3.34, p < .01) than men did. The wage gap in the sample was actually 64 % (i.e., women earned 64 % of what men earned), which is quite a bit lower than the 81 % estimate from the Bureau of Labor Statistics mentioned earlier. However, the 64 % found here is somewhat closer to the within-occupation gender-wage gaps reported for lawyers (73.3 %) and for judges, magistrates, and other judicial workers (56.8 %), in the 2000 U.S. Census (see Weinberg 2004). Also, the width of the gap in the current sample is consistent with other research. For example, Noonan, Corcoran, and Courant (2005) found that 15 years after graduation women earned only 61-63 % of men in their cohort. Consequently, the gap observed here of 64 % is not inconsistent with other research showing a more substantial gender wage gap in professions such as law.

Table 1 Descriptive statistics and intercorrelations for study variables
Full size table

To further investigate the wage gap in this sample, an ANCOVA was conducted to examine the effect of controlling for age and experience on the gender wage gap. Controlling for these factors reduced the gap substantially (female M = $104, 988; male M = $134,614), for a ratio of 78 %. However, despite controlling for these factors, a portion of the wage gap remains unexplained.

Men and women also differed significantly in terms of referent salary, with women reporting significantly lower referent salaries (t = -2.47, p < .05) than men did, for a gender-wage gap for referents of 73 %. However, women and men did not differ significantly in terms of entitlement (t = 0.68, n.s.), nor did women indicate greater dissatisfaction with their salary levels than men (t = -0.18, n.s.), consistent with previous research showing that women are underpaid relative to men and yet do not feel less satisfied (e.g., Crosby 1976). There were also no gender differences in the value of money (t = -0.22, n.s.).

Model Test

The proposed path model was tested using AMOS 16.0. The initial model achieved a good fit (χ2 = 29.09, df = 21, p = .11; χ2/df = 1.39; CFI = .99; NFI = .95; RMSEA = .048). However, the path coefficient from gender to value of money of .02 was nonsignificant, suggesting that this path might be removed. A second model was tested with this path removed, and also had a good fit (χ2 = 29.14, df = 22, p = .14; χ2/df = 1.33; CFI = .99; NFI = .95; RMSEA .044). This more parsimonious model also did not have a significantly worse fit than the proposed model, and thus is the preferred model (see Fig. 2).

Fig. 2
figure 2

Final model showing standardized estimates of relationships among study variables

Full size image

As Figure 2 shows, participant gender was significantly but rather weakly related to referent gender. A chi-square test of association was also conducted to further elucidate this relationship (χ2 = 4.10, p < .05). Specifically, 78 % of males chose male referents, and 36 % of females chose female referents. Thus, although women were more likely than men to select female referents, both genders exhibited a tendency to prefer male referents. Given that the occupation of attorney is traditionally a male sex-typed job, these findings are not surprising, as male referents in a law firm would certainly be proximate and similar on a variety of dimensions. Crosby (1982) found similar results, in that high prestige women were likely to choose referents of the opposite sex. Thus, moderate support was obtained for Hypothesis 1.

Hypothesis 2 was also supported, as referent gender was significantly and positively related to reference standards, such that female referents had lower salaries than male referents. However, it is worth noting that the covariate actual pay had a much stronger relationship with reference standards. This suggests that factors other than referent gender (e.g., job level) are likely to be more influential in determining reference standards.

Reference standards (i.e., referent salaries) in turn were positively, albeit marginally significantly, related to feelings of entitlement. However, again, actual pay again had a stronger (but negative) relationship with feelings of entitlement. Importantly, although the path coefficient between reference standards and entitlement was positive, the zero-order correlation between these variables was negative, indicating that reference standards likely functions as a classical suppressor variable for the relationship between actual pay and entitlement. Thus, Hypothesis 3 was only weakly supported.

Finally, entitlement was strongly and negatively related to pay level satisfaction, supporting Hypothesis 4. Actual salary also had a positive relationship with pay level satisfaction above and beyond entitlement, although this relationship was somewhat weak. However, this does suggest that actual pay influences pay satisfaction directly and beyond the social comparisons and evaluations of fairness made by individuals.

With respect to Hypothesis 5, that men would value money more than women would, the path from gender to value of money was nonsignificant, and thus this hypothesis was not supported. This finding suggests that at least some groups of women may value money or pay as highly as men do, contrary to the conclusions in Nieva and Gutek’s (1981) review. However, value of money was significantly and negatively related to satisfaction, as predicted, albeit weakly. Thus, modest support was obtained for Hypothesis 6.

Discussion

This study tested an extension of Major’s (1994) model of pay satisfaction designed to examine why women are satisfied with lower pay than men. Importantly, this study tests a complex model of the factors leading to pay satisfaction, rather than testing parts of a model or “bivariate” relationships. Thus, the results presented here place the paradox of women’s pay satisfaction in a richer context than previous work typically has done. Additionally, these relationships are tested with a working adult sample, from whom measures of actual referent choice, entitlement, and satisfaction were obtained. The value of pay to the individual was also measured in a unique way, by focusing on the meaning of money (i.e., money obsession) rather than on traditional ranking or rating measures of job values. Thus, this study extends and complements previous research on women’s paradoxical pay satisfaction.

First, it should be noted that men and women in the sample did differ significantly in their actual level of pay, thus providing further evidence of the continuing pay gap. Some of this difference is accounted for by their differences in age and experience (see Table 1). However, even when controlling for these factors, gender remained a significant predictor of actual pay, suggesting that a portion of the wage gap is due to other factors, such as gender differences in negotiation, discrimination, etc.

The current study hypothesized that pay satisfaction results from women’s lesser feelings of entitlement stemming from differences in social comparisons, and from differences in the valuation of pay. The results of the analyses provide mixed support for the hypothesized model of pay satisfaction, and offer potential avenues for future research on gender differences in pay satisfaction. First, men and women were expected to differ in the value placed on pay or money, but they did not. As previous research on this issue has produced inconsistent findings (Nieva and Gutek 1981) this finding is not surprising. Moreover, Golding, Resnick, and Crosby (1983) found that male and female attorneys (i.e., high-status jobs) did not differ significantly in the value they placed on pay. Thus, the lack of gender differences in value placed on money may be due in part to the unique nature of this sample (i.e., attorneys, a traditionally male sex-typed job). Alternatively, men’s and women’s value of money or pay may be converging over time given women’s increased participation in the workforce and at higher-status jobs. It is worth noting that in Powell and Eddleston’s (2008) study of the “paradox of the contented female business owner,” they found that female business owners valued the achievement of business success (which included earning a lot of money) less than male business owners did. Thus, whether men and women differ in their work values, especially with respect to money, remains in question.

Importantly, this study measured the value of money using multiple items designed to measure obsession with money (e.g., affective meaning of money), rather than importance ratings or rankings of job characteristics including pay. The use of these items was intended to allow for a more in-depth investigation of the value of pay or money to individuals. The question of whether men and women in other jobs (e.g., lower-status) would differ on this scale remains unanswered. Further, although men and women did not differ on value of money, value of money was related to pay level satisfaction. The more that money was valued, the less satisfied individuals were with their pay. This effect was not very strong, however, and was much smaller than the effect of entitlement in predicting satisfaction. Thus, in the current study, gender differences in the value of money failed to explain women’s paradoxical satisfaction with pay, and value of money also appears to play only a small part in determining satisfaction.

With respect to the social comparisons that lead to pay satisfaction, modest support was obtained. First, both men and women in this sample largely selected male referents, but women were more likely than men to select female referents. This finding likely reflects the reality of their organizations, at least in part, insofar as attorney has been traditionally a male sex-typed job, and more male referents would likely be available to both men and women. This finding is also consistent with Felicio and Miller’s (1994) study, in which they found that although female medical students made more comparisons with women than men did, both tended to prefer male referents, which may reflect the relative lack of female role models available.

Moreover, the law profession is becoming more gender-neutral (e.g., 46 % of law firm associates are female; National Association for Law Placement 2009; the percentage of attorneys who are women increased from 8 % in 1980 to 27 % by 2000; American Bar Association 2009), and the increasing participation of women in the profession is likely reflected in the fact that women were more likely to select female referents than men were. Specifically, given that women’s participation in the legal profession has been increasing over time, it is likely that women are underrepresented among the older and more experienced attorneys. As women in the sample tended to be younger and less experienced, they would be less likely to select older and more experienced attorneys (who would be more likely to be male) as referents, for reasons of similarity and proximity. For women in the sample, then, younger and less experienced attorneys are more likely to be similar and proximate. As there are more females proportionately among this subset of the profession, women should have more female referents to choose from in this subset, which should produce a tendency for women to select female attorneys, beyond simply the commonality of their gender. Men in the sample, in contrast, being slightly older and more experienced, would be more likely to select the older and more experienced attorneys (who would be more likely to be male) as referents, also for reasons of similarity and proximity. Thus, there should also be a tendency for men to select male attorneys, as there are more males who are similar and proximate in their age/experience subset.

Together, these tendencies should lead to tendencies for men and women to select referents of their own gender, based on reasons of similarity and proximity, rather than just gender. In fact, when participants provided free-response reasons for their choice of referent, similarity (e.g., similar level, experience, education, age, tenure, hours worked, pay, skills) and proximity (e.g., worked together or near) were the primary reasons mentioned. These reasons are consistent with the surrounding dimensions (i.e., occupation, age, education, seniority, and productivity) relevant to choosing a comparison other (Martin 1981). The referent’s gender did not typically arise as a reason for referent choice. Thus, for women in this study, their choice of referent appears to have been based primarily on similarity in job-relevant characteristics instead of similarity in gender. A possible reason for this is that gender is less informative for social comparisons when other, more useful information (e.g., about performance) is available. Indeed, Gibson and Lawrence (2010) suggest that choice of referent based primarily on gender may be an artifact of the laboratory setting, where little is known about the performance-relevant characteristics of the referents.

The literature on gender differences in referent choice provides some support for the speculation that gender is a less useful characteristic for comparison in organizational settings. For example, Gibson and Lawrence (2010) found that referent career level was more important in determining career salary expectations than was referent gender for a managerial sample. Felicio and Miller (1994) also suggested that choice of referent was based on similarity in terms of performance, rather than gender similarity per se. Thus, for higher-level professions that are less sex-segregated, such as the ones in these studies as well as the current study, the choice of referent is likely based on similarity in factors that will allow the employee to improve, advance, etc., rather than on gender similarity, which is relatively unrelated to career progression. Indeed, Kulik and Ambrose (1992) suggest that more cross-sex comparisons are likely as women enter male-dominated fields.

However, it should also be noted that the failure of both men and women to mention gender as a reason for selecting referents might also be due to socially desirable responding (see Felicio and Miller 1994), such that neither males nor females might be willing to admit that their choice of referent was based on gender. Given that attorneys are quite likely sensitive to issues of discrimination, it is possible that they might avoid mentioning gender as a reason for referent choice, Future research on the role of socially desirable responding in referent choice would be valuable to address this possibility.

Referent gender was a significant predictor of reference standards (i.e., referent salary), indicating that female referents tended to have lower salaries. This may be due to differences in negotiation, occupational choice, discrimination, etc. However, actual salary had a stronger relationship with reference standards, suggesting that actual pay is a stronger covariate or determinant of reference standards than gender. For example, the year of entry into a firm would likely be a common influence on both the referent’s salary and the participant’s actual salary, and could partially explain the relationship between these variables.

Reference standards, in turn, were positively but weakly related to feelings of entitlement, and only when actual pay was included as a predictor of entitlement. The reversal in sign on the relationship between reference standards and entitlement is indicative of a case of suppression. Again, however, actual pay was a stronger predictor of entitlement, suggesting that participants were relying more on their current salaries than on their referent others’ salaries in determining the fairness of their pay.

Finally, as expected, some support was found for the relationship between entitlement and satisfaction, such that greater feelings of entitlement were associated with lower satisfaction. Thus, as women had lower reference standards than men and lower entitlement, they did not feel more entitled to higher pay than men, and were also not more dissatisfied with their pay, consistent with previous research (see Crosby 1982; Major 1994). Importantly, actual pay was also related to pay level satisfaction, albeit somewhat weakly, above and beyond entitlement. As a great deal of literature has assumed that pay should have a direct effect on pay satisfaction (Heneman and Judge 2000; Judge, Piccolo, Podsakoff, Shaw, and Rich 2010) this finding has implications for the study of pay satisfaction. Specifically, the finding that actual pay is a strong negative predictor of feelings of entitlement (i.e., higher pay, lower entitlement) is consistent with discrepancy theory-based approaches (e.g., Crosby 1982; Mueller and Wallace 2008) to understanding the paradox. Such approaches suggest that women are satisfied because they do not see a discrepancy between the pay they feel they deserve and the pay they receive, and thus women’s satisfaction could be due to failing to perceive a discrepancy between what they deserve and what they receive. However, the relatively weak relationship of reference standards and feelings of entitlement suggests that more research needs to be directed toward understanding how women determine the pay they deserve.

One possibility for addressing how women determine their deservingness lies in examining their job inputs (see Major 1994; Mueller and Kim 2008). Given that women in the current study were younger with less experience on average than men, it is possible that they were satisfied with their lower pay because they had lower inputsFootnote 1. To test this possibility post hoc, a hierarchical regression analysis was conducted with pay satisfaction as the criterion. Actual salary was entered on the first step, age and experience on the second step, and gender on the third step. Salary (β = .58, p < .01) and experience (β = -.21, p < .01) were significant predictors of satisfaction, although age (β = -.07, p > .10) was not a significant predictor. Importantly, when gender was added in the third step, it was a significant predictor of satisfaction (β = -.14, p = .05), such that women still had higher satisfaction even after controlling for the other factors. Thus it is possible that inputs such as experience account for part of women’s satisfaction with pay, but other inputs need to be examined in order to further investigate this possible explanation of the paradox.

Limitations

Several limitations affect the generalizability of the results of this study. First, the sample selected was a traditionally male sex-typed (though possibly becoming gender-neutral) job. Thus, the findings from this study might differ when examining men and women in other jobs, particularly female-sex-typed jobs. For example, a substantial proportion of women in this sample selected male referents, which likely reflects the fact that the profession has a large male population. Further, the selection of attorneys may be a limitation. It is possible, given the legal field’s emphasis on issues of “justice,” that attorneys might be particularly sensitive to issues of fairness and discrimination, and thus these findings might not generalize to other samples of professionals as well as nonprofessionals. Nonetheless, women in this occupation likely face the same obstacles as other professionals (e.g., lower pay, “glass ceilings”). Further, women in this sample did not feel more dissatisfied with their lower pay, consistent with findings for other occupations, particularly female-dominated ones.

As discussed in the Methods, the relatively low response rate obtained in this study is another limitation, although the sample does not seem to be particularly unrepresentative of the Texas Bar membership at the time. One concern, however, is that a substantial portion of the sample came from the Continuing Legal Education events, which could bias the sample. An examination of the demographics of the sub-sample from the CLE events revealed it to be older (M = 41.61) and more experienced (M = 11.72) than the other samples, and also having a lower proportion of females (28 %). Thus, the CLE subsample was more like the Texas Bar membership than the other sub-samples. However, as these sub-samples were not particularly unrepresentative of the overall membership, the decision was made to retain them, to maximize sample size.

Another limitation is the possibility of common method bias, given that all measures were collected at the same time and self-reported. However, as many of the variables in the study are perceptions (e.g., feelings of entitlement, value of money, satisfaction), substantial use of self-report was necessary, and perceptual measures are likely to be more appropriate in this type of research than many objective measures (see Adams 1965; Lawler 1971). Nonetheless, it would be useful to obtain objective organizational measures of both predictors and criteria (e.g., participant and referent salary from personnel records, consequences of satisfaction such as turnover) and control variables (e.g., other job inputs such as performance evaluations). Further, it would be advantageous to collect these in a longitudinal design, to make stronger causal inferences about the precursors of pay satisfaction.

A related issue lies in the use of participants’ estimates of referent salary. As these were estimates, rather than actuarial information, little is known about how accurate participants’ estimates really were. Research has shown that employees underestimate the pay of those at higher levels, and overestimate the pay of those at lower or the same levels in their organizations (Lawler 1965; Milkovich and Anderson 1972). In the current sample, 69 % of referents chosen were at the same or a lower level. This finding, in conjunction with Lawler’s (1965) and Milkovich and Anderson’s (1972) findings, suggests that the referent pay was probably somewhat overestimated on the average. This tendency was somewhat more exaggerated for men, as 73 % selected referents at the same or a lower level (vs. 61 % for women). As men then would tend to perceive their referents as being paid more than they actually are, these perceptions should lead to a wider gender wage gap for the perceived referent salaries (relative to the actual referent salaries). However, without knowing the actual salaries of referents, it is impossible to know just how inaccurate men’s and women’s estimates of referent salaries were. Nonetheless, given that this research dealt with attitudes, which are influenced not only by objective facts but by (mis-)perceptions, such perceptual data, accurate or not, may be more appropriate for this kind of research (see Lawler 1971). However, future research, particularly within organizations that embrace pay openness, would be valuable for elucidating the effect of accuracy of referent salary perceptions on pay attitudes.

The measurement of the value of money was a potential limitation. The obsession with money scale used to assess the value of money might not have been entirely appropriate for this work. It is possible that other measures of the value of pay or money developed specifically for the workplace could represent the construct more effectively. However, this study was unique in focusing on the value of money itself, rather than a rank-ordering of pay relative to other values, and thus perhaps came closer to a conceptualization of Jackson’s (1989) “meaning of money.” Future research should examine other individual differences that influence equity comparisons and pay satisfaction, including socioeconomic status and personality. For example, individuals who are breadwinners might apply a need-based rather than equity-based rule of distributive justice when determining their pay satisfaction. Personality characteristics such as positive affectivity (Shaw, Duffy, Jenkins, and Gupta 1999) and negative affectivity (Maraist, Davison, Brief, Dietz, and O’Shea 1999), which have been shown to be related to pay satisfaction, might also be relevant, as might narcissism, which contains elements of entitlement (Emmons 1984).

Implications and Conclusion

The research discussed here has a number of implications for both theoretical and applied research. First, this study examined the comparative and evaluative processes that were expected to impact pay satisfaction in a field study using employed individuals. Because much of the previous research in this area has examined these comparative processes in laboratory settings, this study provided insight into the actual comparisons that individuals make. As noted earlier, women were more likely than men to select female referents. However, why they tended to choose female referents more than men did is still in question. As discussed earlier, women in the sample were younger and less experienced than men, on the average, and it is likely that they were exposed to more potential female referents for reasons of similarity (in age and experience) and proximity (e.g., being at similar organizational levels). Indeed, when participants were asked (in a free-response format) why they chose the referent they did for comparison, reasons of similarity or proximity came up frequently. Some participants noted other personal characteristics of the referent (see Kulik and Ambrose 1992), such as the referent was someone they respected or someone with whom they had a personal relationship (e.g., friend or family), but rarely did they mention the referent’s gender. The question remains, however, whether referent gender plays an important (possibly implicit) role in referent selection but is just not mentioned by participants, or whether it is really less important than the other characteristics. Further elaboration of why individuals choose particular referents, from among the various referents available, would be valuable for understanding more about referent choice, and in particular the role of referent choice in affecting women’s entitlement and pay satisfaction.

Another issue with respect to referent choice is that individuals often consider multiple referents. In this study, participants were asked to select a single referent. However, in reality, when determining pay fairness and satisfaction, individuals may consider multiple referent others who are similar and proximate, but also may consider themselves in the past, their friends and family, etc. (see Goodman 1974). The lack of information about other possible referents that the current study participants may have had in mind when evaluating their pay satisfaction is likely one reason that weaker relationships were obtained for reference standards. Clearly, it would be valuable to obtain more insight into the multiple different referents considered by the participants.

This study also has important implications for research into the value of pay. Previous research has been mixed regarding whether men and women differ in their work values (see Nieva and Gutek 1981). As most studies have examined the importance of money relative to other work values, thus failing to limit socially desirable responding, an important advance in the current study was its specific focus on the meaning of money (i.e., the value of money). This study showed that men and women might not differ in terms of how much they value money, a finding that may have been obscured in other studies by using rank-order methodologies. It would also be interesting to examine how men’s and women’s valuation of money might change depending on the economic situation; for example, in a recession, it might be possible that men and women under similar economic strain would tend to report similar value of pay. A poor economic situation might also affect other aspects of pay satisfaction. Indeed, some research suggests that individuals who enter the workforce during a recession have lower expectations and higher satisfaction than others who entered during more prosperous times (see Bianchi 2012; De Hauw and De Vos 2010).

Finally, not only were women in the study significantly underpaid relative to men, women’s referents were also significantly underpaid compared to men’s. This suggests that women either selected lower-paid targets, or were exposed primarily to lower-paid coworkers by working in lower-paying jobs. Altogether, it appears that women may be exposed consistently to a pattern of lower salaries, not only their own but also coworkers’ salaries. This pattern may lead women to internalize the lower pay standards, and thus to have lower expectations than men. This result is of concern for researchers attempting to uncover ways to reduce the gender wage gap. Specifically, Major (1994) suggested that providing women with high comparison standards and feedback about their contributions could help to increase women’s feelings of entitlement. However, women in this study were employed in high-paying jobs and had fairly high contributions, in terms of education, age, and experience, yet still perceived lower pay as fair. Clearly, more work needs to be done to examine whether providing information about reference standards and contributions will change women’s fairness perceptions and satisfaction.

Blanton et al.’s (2001) work on system justification may provide an important direction for such research. Specifically, they found that when pay was framed as compensation for past work, female college students compared their pay with other women, but when the pay was framed as part of a future job offer, they compared their pay with men. Thus, the context or form of the pay under evaluation (e.g., current salary vs. future salary, merit raises based on past performance vs. incentives for future performance) may also affect referent choice and thus referent standards. Indeed, Graham and Welbourne (1999) found that women had lower pay satisfaction after a gainsharing system was introduced, suggesting that their referents may have changed as result of the new system, or that women were more attentive to the fairness of their pay under the new system. Thus, examining the effect of pay interventions on gender differences in pay satisfaction may provide additional insights into how women (and men) regard their pay.

In conclusion, this study found that the gender-wage gap persists, but women are nonetheless satisfied with their pay. This research has attempted to serve two purposes: to investigate the nature and causes of pay satisfaction, and to provide illumination of the reasons for women’s paradoxical satisfaction with underpayment. Regarding the first purpose, this study showed that both entitlement and value of money relate to satisfaction, thus providing some support for both Major’s (1994) and Jackson’s (1989) models of pay satisfaction. This research also examined the causes of women’s paradoxical satisfaction with underpayment, which appears to stem in part from differences in their social comparisons, as well as in their lower feelings of entitlement. Women selected lower-paid referents, felt less entitled to higher pay, and were satisfied with their lower pay, even though they did not value pay less than men did. Thus, this study shows that 50 years after passage of the Equal Pay Act, and over 30 years since Crosby (1982) identified “the paradox of the contented female worker,” even professional women are satisfied with their lower pay.