The Baby Bump

Abstract

In light of the prevalent experience, theoretical importance, and underexamination of the intersection of pregnancy and work, the current study explores how pregnant employees manage their concealable stigmatized identities at work over the course of pregnancy. Using a weekly survey methodology, we were able to examine within-person changes in identity management and physical health. Results suggested a reciprocal relationship between revealing and physical health wherein revealing led to more frequent physical health symptoms and more frequent physical health symptoms led to decreased revealing. Furthermore, concealing exerted a unidirectional impact on physical health wherein concealing predicted subsequent decreases in physical health symptoms. Finally, supportive work–family cultures and supervisor support were linked to lower concealing, higher revealing, and less frequent physical health symptoms at the initial measurement occasion (i.e., earlier stages of pregnancy); however, these benefits appeared to diminish over time. The implications of these findings for theory and practice are discussed.

Keywords

diversity/gender individual decision making identity work–family conflict/management well-being

My boss’s reaction was aloof, disingenuous, and borderline indifferent. . . . I had to deal with telling her before my annual review where she basically guilted me into working longer hours. . . . Two months after disclosing my pregnancy, I feel distant from my work and worried that I could be laid off at any time if I don’t work longer hours regardless of how sick I am feeling.

—Study Participant

Many women struggle to balance the roles of hard-working employee and dedicated mother; indeed, 80% to 85% of women become mothers while employed (Schwartz, 1992). Accordingly, a large body of research on the work–family interface has emerged with a particular focus on the experiences of women and mothers (see Allen, 2001). One critical context in which motherhood and work intersect that has received surprisingly little research attention is pregnancy. While pregnant women are generally revered for fulfilling the expectations of the traditional female role (e.g., motherhood), pregnant workers may be viewed as neglecting both work and family roles (King & Botsford, 2009). Indeed, empirical research has demonstrated that pregnant employees experience many of the same forms of negative backlash that are associated with characteristics that are more commonly considered to be stigmatized (e.g., racial minority status, a lesbian or gay identity, obesity; Cuddy, Fiske, & Glick, 2004; Hebl, King, Glick, Kazama, & Singletary, 2007). According to the U.S. Equal Employment Opportunity Commission (2011), pregnancy discrimination claims filed in 2011 increased by almost 50% since 1997. Together, this evidence suggests that pregnancy represents a devalued social identity—a stigma—in the workplace.

Because pregnancy may be stigmatized and is potentially concealable for a substantial amount of time, pregnant workers likely face complex disclosure decisions. Pregnant workers may be hesitant to tell others about their pregnancy because of the stigma associated with that status, but may also be compelled to reveal their pregnancy to take advantage of valuable resources, a predicament often referred to as the “disclosure dilemma,” whereby targets attempt to balance two competing motives, authenticity and self-protection (King & Botsford, 2009). Thus, pregnant workers face critical decisions about how, when, and to whom they disclose their pregnancies at work.

Previous research suggests that identity management and disclosure decisions are stressful, particularly with regard to physical health, for employees burdened with managing a concealable stigmatized identity (Ragins & Cornwell, 2001; Ragins, Singh, & Cornwell, 2007). However, conclusions regarding causality are limited given the focus on static stigmas and correlational nature of extant data. Though it is reasonable to suggest that identity management decisions trigger changes in physical health, it is also possible that changes in physical health drive subsequent identity management behaviors.

The present research advances extant theory and research in three ways. First, our study builds on limited previous qualitative and cross-sectional research (Greenberg, Ladge, & Clair, 2009; Ladge, 2008; Major, 2004; Millward, 2006) examining the work–pregnancy interface by investigating identity management and disclosure decisions longitudinally over the course of pregnancy using a weekly survey methodology, thereby allowing for more accurate reporting of experiences as they actually unfold. Second, by exploring these processes through a within-person lens, we examine how identity management decisions affect pregnant employees’ physical health well-being, providing a critical extension of previous cross-sectional research linking disclosure and well-being (Chaudoir & Quinn, 2010; Ragins & Cornwell, 2001). In doing so, we empirically test the plausibility of two opposing theoretical explanations regarding the impact of identity management and disclosure decisions on physical health well-being. Whereas stigma theory focuses on disclosure as a potential risk to well-being via stigmatization, self-verification theory advances disclosure as a process that provides an opportunity to be authentic and thus beneficial to well-being. Third, and finally, we explore how contextual aspects of the organization may affect identity management behavior and physical health well-being during pregnancy.

The remainder of the article unfolds as follows. First, we clarify the identity management framework we use and outline the behaviors examined in the present study: concealing, revealing, and signaling. Next, we describe how the dynamic nature of pregnancy can affect the use of identity management strategies over the course of the pregnancy. Drawing from stigma and self-verification theories, we advance primary as well as alternative hypotheses regarding the influence of identity management behaviors on physical health symptoms during pregnancy. Finally, we generate hypotheses regarding the influence of contextual support on identity management behaviors.

Identity Management of a Concealable Stigma

While identity disclosure has often been conceptualized as a dichotomous phenomenon (Croteau & Lark, 1995; Levine & Leonard, 1984), recent scholarship suggests identity management and disclosure decisions are complex and multifaceted (Button, 1996, 2001; Chrobot-Mason, Button, & DiClementi, 2001; Clair, Beatty, & Maclean, 2005). Preliminary research indicates that individuals manage their invisible identities using a variety of strategies including concealing, revealing, and signaling, which likely result from the psychological calculus of weighing the benefits of personal authenticity against fears of potential discrimination (e.g., Button, 1996, 2001; Chrobot-Mason et al., 2001).

Concealing behaviors are conscious, active attempts to hide a true identity and “pass” as a member of the majority group and may include behaviors such as fabricating stories, dodging personal questions, speaking in generalities, or actively preventing others from learning information that may lead them to affirm an invisible identity (Button, 2001; Clair et al., 2005). These can be contrasted with revealing, which entails actively and purposefully informing others of one’s stigmatized identity via explicit verbal communication or conspicuous behavioral actions (Button, 2001). In the context of pregnancy, examples of revealing might include talking about the baby, openly discussing doctors’ appointments, or referring to maternity leave. The defining feature of revealing behavior is that it undeniably confirms the invisible identity by the conclusion of the behavior. A third type of identity management behavior is signaling, which involves “testing the waters” and may include dropping hints, providing clues, and sending implicit messages (Clair et al., 2005). The indirect nature of signaling allows individuals to gauge the confidant’s reaction before actually disclosing and “backtrack” if it becomes apparent negative reactions are probable, but signaling may also “invite speculation . . . encouraging . . . peers to read between the lines” (Woods, 1994: 176).

The Dynamics of Identity Management and Physical Health During Pregnancy

One unique aspect of pregnancy as a concealable stigmatized identity is its dynamic nature, which distinguishes pregnancy from other concealable stigmas that can be “hidden” indefinitely if the target chooses. Two dynamic aspects of pregnancy, its changing visibility and the experience of changing physical health symptoms, likely affect the progressive use of identity management strategies throughout the pregnancy. Indeed, previous research suggests in the case of a progressively visible stigma, timing of disclosure is largely dictated by the stigma’s increasingly obvious nature (Hays et al., 1993; Marks et al., 1992). Therefore, as pregnancy becomes physically apparent, pregnant workers are likely to be more expressive of the pregnancy through revealing behaviors since concealing the identity becomes difficult and unrealistic. Furthermore, signaling behavior is most likely to occur when the pregnancy is hidden given there is less and less to signal as pregnancy status becomes more ostensible. Formally,

Hypothesis 1: Concealing (a) and signaling (b) will have negative slopes over time, whereas revealing (c) will have a positive slope over time.

Concealing, Revealing, and Physical Health: Competing Perspectives

We have argued that individuals with concealable stigmas are burdened by decisions regarding how, when, and to whom to disclose their identity that often occur in the context of two competing motives: authenticity and self-protection. We further contend that concealing and revealing present a varying degree of opportunity for authenticity in interpersonal interactions, which ultimately influences physical health well-being over and above the trajectories of change described above. To the extent authenticity is suppressed, the individual experiences a discrepancy between her own understanding of self and the self she expresses to others.

Previous scholars have conceptualized such inconsistent understandings as meaning violations. According to the Heine, Proulx, and Vohs (2006) meaning maintenance model, meaning can be understood as “the mental representations that allow us to understand our experiences,” whereas meaning violations reflect “experiences that are inconsistent with the expectations that follow from [these] understandings” (Proulx & Inzlicht, 2012: 318). Such violations of meaning result in a state of aversive arousal, a biologically based state often characterized by changes in skin conductance, blood vessel constriction, and cardiac activity (Proulx & Inzlicht, 2012).

Aversive arousal is theorized to be particularly damaging to physical health well-being via the cognitive dissonance that is assumed to emerge when familiar understandings or expectations of the self are challenged in some way, thereby producing a discrepancy between an individual’s previous, comfortable understanding of a situation and the actual existing situation (Proulx & Inzlicht, 2012). Indeed, the negative health effects that occur as a function of cognitive dissonance have been widely documented in social psychological research (e.g., Strauman, Lemieux, & Coe, 1993; Townsend, Major, Sawyer, & Mendes, 2010). Drawing from this work, we argue that concealing behaviors represent a meaning violation whereby an individual’s understanding of her self-concept is violated, which produces a state of aversive arousal and corresponding negative physical health consequences. Indeed, extant empirical research supports the notion that concealing behavior leads to damaging physical health consequences including increased depression (Gross & John, 2003), increased minor ailments (Pennebaker, 1990), accelerated cancer progression (Fawzy et al., 1993), and decreased somatic activity (Gross & Levenson, 1993, 1997). Thus, we argue the link between concealing and physical health is likely explained, in part, by the aversive arousal that emerges when being inauthentic in interpersonal interactions.

In contrast, revealing behaviors allow individuals to remain authentic and maintain meaning in their familiar understandings of their self-concepts and thus are less likely to produce aversive arousal via cognitive dissonance. Consequently, if a pregnant employee feels she can be expressive of her pregnant identity at work, physical health well-being should be enhanced via reduced stress associated with authenticity in interpersonal interactions. Indeed, extant empirical evidence supports the notion that revealing behaviors may function to protect against a variety of negative physical health consequences (Strachan, Bennett, Russo, & Roy-Byrne, 2007).

Taken together, the above reasoning and evidence suggest that in practice, concealing behaviors hinder personal authenticity in interactions, which violates individuals’ familiar understanding of self and produces cognitive dissonance. These conflicting views of self, in turn, produce a state of aversive arousal and subsequent negative physical health effects. In contrast, revealing behaviors produce opportunities for personal authenticity, which in turn benefit physical well-being via reduced psychological stress associated with the expression of a valued identity (Swann, Polzer, Seyle, & Ko, 2004; Swann, Rentfrow, & Guinn, 2002). Consistent with self-verification theory (Swann, 1987), this explanation suggests identity management behaviors that produce dissonance in one’s actual and presented self (i.e., concealing) trigger declines in physical health, whereas identity management behaviors reflecting authenticity (i.e., revealing) lead to improved physical health.

Expressing a stigmatized identity may not always lead to improved well-being, however. Indeed, empirical research has demonstrated pregnant working women incur devaluation and stigmatization—as evidenced by negative stereotypes, social rejection, discrimination, and economic disadvantage (Budig & England, 2001; Cuddy et al., 2004; Hebl et al., 2007; Williams & Segal, 2004). As a result, revealing one’s pregnant status or pregnancy-related information may put pregnant workers at increased risk for discrimination, which recent meta-analytic evidence has shown is physically and psychologically damaging (Pascoe & Smart Richman, 2009). Further supporting this notion is evidence suggesting that exposing a stigmatized identity can be detrimental to physical health (D’Augelli & Grossman, 2001; Ragins & Cornwell, 2001).

In addition to avoiding overt forms of stigmatization, pregnant women may also avoid revealing their pregnant status or pregnancy-related information to protect their professional status from implicit assumptions about decreased commitment to the job or increased likelihood of turnover. Pregnant workers maintain two potentially competing identities—pregnant woman and employee—that are important to and valued by expectant workers (Bailey, 1999; Greenberg et al., 2009; Major, 2004). Since the image of the “ideal professional” is stereotypically White, masculine, heterosexual, and well educated (Duehr & Bono, 2006; Ridgeway & Correll, 2004; Rosette, Leonardelli, & Phillips, 2008; Rosette & Tost, 2010), a pregnant employee may struggle with the fact that her pregnant identity conflicts with her “ideal professional” image. Indeed, recent qualitative evidence suggests pregnant employees downplay or conceal pregnancy-related information in their interactions at work, especially with their supervisors, out of fear that making the pregnancy salient would reduce their power in the situation (Greenberg et al., 2009). Consistent with stigma theory, this explanation suggests pregnant women who conceal their pregnant status to protect themselves and their professional images from both blatant and subtle stigmatization may experience better physical health well-being relative to their pregnant counterparts who incur stigmatization as a result of revealing their pregnant identities. Stigma theorists have advanced the idea that individuals with concealable stigmas are at an advantage relative to those with visible stigmas (i.e., race, gender) since concealing, and thereby avoiding stigmatization, is a viable option for these individuals (Jones et al., 1984; Ragins, 2008).

We have presented two opposing perspectives explaining how identity management and disclosure decisions relate to physical health well-being among pregnant employees. On the one hand, self-verification theory would suggest that expressing a stigmatized identity would improve physical health outcomes via the opportunity for personal authenticity; however, stigma theory argues expression of a stigma likely leads to devaluation of the individual, which is stressful and negatively affects mental and physical health. Given these contrasting perspectives, we test alternative hypotheses regarding the relationship between identity management and physical health. Consistent with a self-verification perspective, concealing will lead to negative health outcomes, whereas revealing will lead to positive health outcomes. Formally,

Hypothesis 2₀: Concealing at time t-1 will predict an increase in frequency of physical symptoms at time t.

Hypothesis 3₀: Revealing at time t-1 will predict a decrease in frequency of physical symptoms at time t.

Consistent with a stigma perspective, concealing will lead to positive health outcomes, whereas revealing will lead to negative health outcomes. Formally,

Hypothesis 2₁: Concealing at time t-1 will predict a decrease in frequency of physical symptoms at time t.

Hypothesis 3₁: Revealing at time t-1 will predict an increase in frequency of physical symptoms at time t.

Signaling and Physical Health

Given extant research has focused much less on the identity management behavior signaling, we rely primarily on conceptual, logic-based rationale in arguing signaling will be particularly stressful for targets, and thus negatively affect health. First, the strategic nature of signaling is likely to produce high cognitive load; the target must be constantly in tune with the confidante’s actions, attempting to determine the best course of action without being obvious or explicit. Second, signaling in and of itself is likely emotionally taxing, cueing the target’s awareness of the potentially stigmatized nature of pregnancy, thereby increasing the salience of the stigma. That is, the simple need to signal one’s pregnant identity serves as a reminder that others may, in fact, not be accepting of the pregnancy. Previous research has explored the notion that degree of stigma salience triggers distress to some extent with regard to psychological but not physiological distress (Pachankis, 2007; Quinn & Chaudoir, 2009); however, we expand on this idea by incorporating physiological effects of stigma salience. Formally,

Hypothesis 4: Signaling at time t-1 will predict an increase in frequency of physical symptoms at time t.

Contextual Influences on Identity Management Behaviors

In the next section, we explain how a supportive organizational context will influence the anticipated growth trajectories of each identity management strategy described earlier. Specifically, we contend that the extent to which employees perceive their organization and supervisor support and value the interface between work and family will encourage more identity-expressive behaviors including revealing and signaling and will reduce concealing. Though it is less clear that signaling is an identity-expressive behavior (as compared to revealing), we argue that signaling reflects a desired openness about the pregnancy since signals are generally conveyed with the intent and desire to ultimately express an invisible identity.

Indeed, supportive work–family cultures communicate it is safe for pregnant employees to take advantage of available resources without negative repercussions, provide expectant mothers with more opportunities for mentorship, and may attenuate fears pregnant workers hold about negative stereotyping and stigmatization (King & Botsford, 2009). Furthermore, supervisors can play a key role in determining the extent to which employees feel comfortable utilizing family-friendly benefits by either encouraging or undermining employees’ attempts to balance work and family (Perlow, 1995; Starrels, 1992; Thompson, Thomas, & Maier, 1992).

Hypothesis 5: Contextual support will be negatively related to average levels of concealing and positively related to average levels of revealing and signaling.

We contend that two dominant forces are at play in predicting identity management behaviors across the course of pregnancy—the freedoms conveyed by supportive context and the constraints conveyed by the increasing visibility of pregnancy—and that each of these factors exerts more influence at different stages of pregnancy. In the early stages of pregnancy when the pregnancy is not yet obvious, a more supportive context permits more expression (and less concealing) of pregnancy as compared to less supportive contexts. However, as the pregnancy becomes more visible, everyone’s behaviors are constrained to higher levels of revealing and lower levels of concealing and signaling in light of the physical evidence of pregnancy. Thus, the revealing and concealing identity management trajectories of women who work in more supportive contexts will be more stable (i.e., have flatter slopes), whereas the revealing and concealing trajectories of those in unsupportive contexts will be characterized by more discernible change over time (i.e., steeper slopes). In contrast, the signaling trajectories of those in supportive working contexts will be characterized by more marked change over time since supportive contexts encourage more identity expression (signaling) in the earlier stages of pregnancy but signaling drops to low levels in the later stages of pregnancy regardless of the supportiveness of the context in light of the pregnancy’s visibility. Formally,

Hypothesis 6: Contextual support will positively predict the slope of concealing (a) and will negatively predict the slope of revealing (b) and signaling (c) such that to the extent the context is supportive, the concealing slope will become less negative, the signaling slope will become more negative, and the revealing slope will become less positive.

Method

Participants

Pregnant women who were employed and in their second trimester (at which time miscarriage rates are substantially reduced) were eligible to participate. Participants had an average of 20 weeks remaining in their pregnancy. Expectant mothers were recruited for participation through advertisements in obstetrician offices, day care centers, and newspapers in the greater Washington, D.C., metropolitan area as well as listservers and social networking sites nationwide. Respondents included 116 pregnant employees. The sample identified as 83% White, 8% Asian, 8% Hispanic, 2% African American, 1% Native American, and 3% Other. Note that ethnic category percentage totals do not sum to 100% because participants were permitted to check all applicable options, resulting in a few instances in which participants indicated membership in two or more ethnic groups. The average age of the sample was 30 years old, and 88% of participants were married. Participants had spent an average of 3.7 years at their current organization. Furthermore, the women in our sample worked in a wide variety of industries including health care, law, education, service, engineering, accounting, finance, administration, and transportation. Participants were compensated a total of $100, granted at approximately evenly distributed points over the course of the data collection, to encourage retention, consistent with the best practices in experience sampling research (Bolger, Davis, & Rafaeli, 2003; Christensen, Barrett, & Bliss-Moreau, 2003).

Procedure

Pregnant working women completed a web-based baseline survey including basic demographic variables, individual characteristics (e.g., identity management), contextual support variables (e.g., work–family culture, perceived supervisor support), and information regarding the disclosure of pregnancy at work. Each week thereafter, participants completed a web-based survey assessing identity management and physical health symptoms until the participant ceased working before childbirth.

Measures

Prior to data collection for the current study, we collected cross-sectional data from pregnant employees (N = 135), which served as pilot data for the current study. One of the primary purposes for the pilot data was to ensure that our identity management measure was functioning in a manner that was consistent with our expectations based on prior theoretical and empirical research (Button, 1996, 2001; Woods, 1994). The pilot study assessed identity management “over the course of the past month” using the same 12 items that were used in the current study. The resulting factor structure was identical to the factor structure found at the weekly level in the current study; that is, the same items loaded onto concealing, revealing, and signaling.

Identity management

To assess revealing, concealing, and signaling behaviors, we used 12 items from Button’s (1996, 2001) identity management strategies scale. Each week, participants were asked to indicate the extent to which they had engaged in each of 12 identity management behaviors (1 = never, 5 = always) during the previous week while at work. Since the scale was originally developed to reflect workplace identity management of sexual orientation, we adapted the items to reflect pregnancy. A sample item from the revealing item is “I said something that indicated I was pregnant,” a sample concealing item is “I tried to behave or dress in a way that would keep others from guessing that I am pregnant,” and a sample signaling item is “I did something to try to see if others might be accepting of my pregnancy.”

To ensure the items loaded similarly onto the three factors (i.e., revealing, concealing, signaling) across time, we specified a multilevel confirmatory factor analysis wherein we constrained the factor loadings to be equal across persons as well as across time. We assessed model fit using the comparative fit index (CFI), for which a value of .95 or higher indicates acceptable model fit (Bentler & Bonett, 1980), and the root mean square error of approximation (RMSEA), for which a value of .05 or lower suggests a well-fitting model (Hu & Bentler, 1999). The absolute fit indices indicated good model fit (RMSEA = .038, CFI = .95), providing support for metric invariance of our items. In addition, a longitudinal factor model was specified with data from each participant’s first and final measurement occasion to test measurement invariance. The configural invariance model showed adequate fit, χ²(225) = 362, RMSEA = .073, CFI = .924. Metric invariance was then tested by constraining the factor loadings to be equal across time. The metric invariance also showed adequate fit, χ²(234) = 397, RMSEA = .077, CFI = .911, although the change in fit was significant, Δχ²(9) = 35, p < .01. The scalar invariance model was also tenable, χ²(243) = 416, RMSEA = .078, CFI = .905; however, the change in fit from the metric to the scalar invariance model was also significant, Δχ²(9) = 19, p < .05. Although direct model comparisons using the chi-square difference tests do not support scalar invariance, the scalar invariance model fits adequately and the global fit indices did not show much change when moving from configural to metric to scalar invariance.

It is important to note that the longitudinal—rather than interaction-level—focus of this study means that our measures captured average weekly identity management behaviors as opposed to how these behaviors function in conjunction and separately within specific interactions. The patterns among the revealing, concealing, and signaling subscales (see Table 1) suggest that they are indeed separate dimensions instead of revealing and concealing being opposite ends of a single spectrum, as their intercorrelations do not exceed .30 at the within-person level. Thus, our measure tells us on average that revealing and concealing behaviors can occur simultaneously within the same week.

Table 1

Descriptive Statistics and Zero-Order Correlations Among Study Variables

Variable	M	SD	α_BETWEEN	Conceal	Reveal	Signal	Symptoms	Culture	α_WITHIN
Conceal	1.39	0.60	.96	__	−.24*	.09	.05	__	.83
Reveal	3.26	1.00	.96	−.31**	__	.27**	.12*	__	.86
Signal	1.64	1.28	.70	.22	.35**	__	.09*	__	.50
Symptoms	2.38	0.47	.78	.14	.05	.14	__	__	.50
Culture	4.55	1.21	.94	−.23*	.17	.01	−.19	__	__
Support	3.71	0.86	.93	−.22*	.19*	.03	−.05	.61**	__

Note: N = 116. α_BETWEEN = between-person-level scale score reliabilities; α_WITHIN = within-person-level scale score reliabilities. Means and standard deviations provided in original scales at the between-person level (1-4 for physical symptoms, 1-7 for work–family culture, 1-5 for all other variables). Between-person-level correlations located on diagonal; within-person-level correlations located on off diagonal.

p < .05.

p < .01.

Physical health symptoms

Each week, participants were asked to indicate the extent to which they had experienced each of five symptoms during the previous week (1 = rarely or none of the time, 4 = all of the time). The scale included symptoms such as “my sleep was restless,” “felt sick,” and “experienced physical pain.” Note that higher ratings reflect more frequent symptoms and thus worse physical health.

Perceived supervisor support

Perceived supervisor support was assessed at baseline using an eight-item scale and was treated as a person-level variable. Sample items are “my supervisor really cares about my well-being” and “my supervisor cares about my opinions” (Eisenberger, Huntington, Hutchinson, & Sowa, 1986).

Work–family culture

Work–family culture of the organization was assessed at baseline by asking participants to indicate their agreement with a set of statements regarding their organization (1 = strongly disagree, 7 = strongly agree), including items such as “in general, managers in this organization are quite accommodating of family-related needs” and “this organization is supportive of employees who wanted to switch to less demanding jobs for family reasons” (Thompson, Beauvais, & Lyness, 1999).

Data Analysis Strategy

To examine the directionality of the relationship between identity management and physical health symptoms, we analyzed the data using a bivariate latent change score (BLCS) approach (for a review, see McArdle, 2009) using Mplus Version 6 (Muthén & Muthén, 2010; see Figure 1). An extension on autoregressive and cross-lagged time series models, the BLCS approach assumes repeated measures data of two variables and imposes a series of restrictions to identify leading and lagged indicators of change across time. In a BLCS model, the latent change in a variable over time is the outcome of interest and comprises three components: the slope parameter (α), which reflects constant change in a variable, the proportional change parameter (β), which reflects change in a variable that is proportional to the previous state of the variable, and the coupling parameter (γ), which represents the change in the variable that is dependent on the previous state of the other variable (Grimm, 2007). These coupling parameters are key to understanding the causal effects of stigma management strategies on physical health detailed in H2, H3, and H4. Before continuing, it is important to note that although one participant completed surveys over the course of 27 weeks, only the first 22 time points yielded variance in the variables. Thus, our analyses reflect the patterns of change over the course of 22 time points.

Figure 1

Simplified Path Diagram of Bivariate Latent Change Score Model With Four Occasions of Measurement

Three series of four BLCS models were estimated to examine the relationship between identity management and physical health symptoms, one series per identity management strategy (i.e., revealing, concealing, signaling). Here, we detail the series of four BLCS models reflecting the relationship between concealing and physical health as an example. The first of the four estimated models, in which no coupling parameters were estimated, served as the most restrictive model (i.e., fewest number of parameters estimated) from which to compare the fit of subsequent models. First, we modeled true scores for concealing as measured by composite scores for concealing at each point in time (concealT1 − concealT22). Next, we modeled autoregressive effects for concealing by estimating the paths between concealT1 and concealT2, concealT2 and concealT3, and so on, which were fixed at 1. Finally, we modeled latent change scores (i.e., difference scores) at each point in time (excluding T1) as measured by the change in concealing relative to the previous time point beyond autoregressive effects. For example, the difference between concealT1 and concealT2 represented the latent change score for concealing at T2.

The intercept term for concealing was estimated as the true score for concealing at T1 and denoted with a subscript 0 (e.g., con₀). Note that the first measurement occasion occurred at 12 weeks pregnant (the earliest a participant would be eligible to participate), so intercept values reflect average levels at this point in pregnancy. The slope term (constant change parameter) for concealing was estimated as the mean of the difference scores for concealing and denoted with a subscript 1 (e.g., con₁). Building from this, the proportional change parameter for concealing (β_con), which reflects change in concealing over and above constant change, was estimated by regressing the concealing difference score at T2 onto the true score of concealing at T1, regressing the concealing difference score at T3 onto the true score of concealing at T2, and so on. We then repeated this estimation of all relationships described above with the physical health symptoms composite variable. Finally, we regressed the slope and intercept coefficients for both concealing (con₀, con₁) and physical health (symp₀, symp₁) onto our person-level composite variables, work–family culture, and supervisor support. Means for all exogenous variables and means and variances for all latent variables were set to zero. Residuals of exogenous variables were constrained to equality, and time-specific covariances among variables were constrained to equality across time. Finally, covariance paths between all slope and intercept factors were estimated.

The next two models in the series are unidirectional in nature and include all effects specified in the no coupling model described above plus one coupling parameter. The first of the two unidirectional models estimates concealing as the leading indicator of subsequent changes in physical health symptoms by specifying a path from the true score for conceal at T1 to the latent change score for physical health symptoms at T2, from conceal at T2 to the latent change score for physical health symptoms at T3, and so on (γ_symp). The second of the two unidirectional models estimates physical health symptoms as the leading indicator of subsequent changes in concealing by specifying a path from the true score for physical health symptoms at T1 to the latent change score for conceal at T2, and so on (γ_con).

Finally, in the last of the four models in the series, both coupling parameters described above (γ_symp and γ_con) are included in the model estimation, yielding a model in which both variables exert influence on one another over time. Progression through this series of four models was repeated with revealing and signaling. It is important to note that we can only compare fit indices for models that are nested within one another. Thus, we can compare model fit of the no coupling model to either of the unidirectional models and to the bidirectional model, we can compare either of the unidirectional models to the bicoupling model, but we cannot compare model fit of the two unidirectional models to each other. Figure 1 depicts the estimated paths in the bidirectional model, as the other three models are nested within this model. Initially, we allowed the proportional and coupling change parameters to be freely estimated at each point in time but found no evidence for time-varying effects. Thus, we constrained these paths to equality across time within variables to facilitate interpretation of the effects (Halbesleben & Wheeler, 2015; McArdle, 2009).

To achieve model convergence, we suppressed the calculation of the chi-square since this calculation often causes convergence issues with incomplete longitudinal data (Grimm, An, McArdle, Zonderman, & Resnick, 2012). Thus, we compare model fit using the −2 log likelihood (–2LL), examining the change in the −2LL values among nested models. This resulting difference between −2LL values is chi-square distributed with degrees of freedom reflected by the difference in number of estimated parameters from one model to the next. Thus, this value can be compared to a critical value to ascertain whether the less restricted model fits the data significantly better than the more restricted model.

Results

Means, standard deviations, reliabilities, and zero-order correlations among study variables are reported in Table 1. Prior to analyzing our main hypotheses, we analyzed the unconditional means model for each identity management strategy to assess the appropriateness of using a multilevel modeling technique on this data. Results indicated intraclass correlations of .57, .66, and .55 for concealing, revealing, and signaling, respectively. Thus, approximately one third of the variability in each strategy can be attributed to within-person differences, justifying our use of multilevel modeling.

Concealing

Compared to the no coupling model (LL = −1365.36, parameters = 32), the unidirectional model estimating concealing as a leading indicator of changes in physical health symptoms (LL = −1359.52, parameters = 33) yielded a significant improvement in model fit (LR = 11.68, df = 1, p < .01). However, the bicoupling model (LL = −1359.47, parameters = 34) added no additional improvement in model fit above the concealing to physical health symptoms unidirectional model. Finally, in comparison to the no coupling model, model fit did not improve when physical health symptoms was estimated as a leading indicator of changes in concealing (LL = −1364.694, parameters = 33). These fit indices suggest support for the notion that concealing exerts an impact on physical health, but not vice versa. Thus, we report the results of the best fitting model, the unidirectional model that estimated concealing as a leading indicator of changes in physical health symptoms, in detail below. In addition, the parameter estimates and fit statistics for this model are reported in Table 2. Note that we report unstandardized slope and intercept values as well as unstandardized change parameters since these values are less meaningful and more difficult to interpret in the standardized metric (McArdle, 2001).

Table 2

Fit Statistics and Parameter Estimates for the Bivariate Latent Change Score Model With Coupling From Concealing to Physical Symptoms

	Concealing	Physical Symptoms
Means (μ)
Intercept	1.61**	2.33**
Slope	0.22**	0.03
Dynamic parameters
Proportional change parameter	−0.19**	0.03
Symptoms → Δconcealing coupling parameter	—
Concealing → Δsymptoms coupling parameter		−0.06**
Random effects
σ intercept	0.89**	0.92**
σ slope	0.93**	0.99**
Time-invariant covariates
Work–family culture → intercept	−0.11	−0.31*
Work–family culture → slope	−0.31**	−0.11
Supervisor support → intercept	−0.26*	0.05
Supervisor support → slope	0.30	0.13
Fit statistic
Log likelihood	–1359.52
Parameters	33

Note: Intercept values, slope values, and dynamic parameters are unstandardized; random effects and time-invariant covariate values are standardized.

p < .05.

p < .01.

The slope estimate for concealing was positive and significant (con₁ = .21, p < .01), suggesting that concealing tended to change at a constant positive rate throughout pregnancy, failing to support H1a. The proportional change parameter for concealing was significant and negative (β_con = −.19, p < .01), suggesting that prior levels of concealing served to reduce the increases in concealing that occurred subsequently. For example, the increase in concealing that follows from a prior concealing score of 3 (on the original 5-point scale) is .19 units smaller than the increase in concealing that follows from a prior concealing score of 2.

The slope estimate for physical health symptoms was not significant, suggesting that on average, physical health symptoms did not appear to change at a constant rate. Furthermore, the proportional change parameter for physical health symptoms was not significant, indicating the magnitude of weekly changes in health symptoms were not dependent on physical health levels during the prior week.

The negative coupling parameter from concealing to subsequent changes in physical health symptoms (γ_symp = −.06, p < .01) suggests that engaging in concealing was associated with subsequent declines in the frequency of experienced physical health symptoms in support of H2₁. Therefore, H2₀ was not supported. Specifically, a one-unit increase in concealing (on the original 5-point scale) is associated with a subsequent decrease in the frequency of physical health symptoms of .06 units (on its original 4-point scale). Although this change parameter is seemingly small in magnitude, it represents the change that occurs after controlling for the slope and the prior status of physical health symptoms. To demonstrate the meaningful nature of these effects as they accumulated over time, we plotted the predicted values for physical health symptoms across the entire period of the study as a function of varying initial concealing scores (see Figure 2).

Figure 2

Predicted Values of Physical Health Symptoms Across the Span of the Study as a Function of Initial Concealing

Assuming all individuals report an average level of physical health symptoms (M = 2.33) at the beginning of the study, a low concealing employee’s predicted score on physical health symptoms by the end of the study (3.18) is considerably higher as compared to an average concealing employee (2.77) or a high concealing employee (2.37). That is, the individual who initially concealed at the highest rate experienced relatively little change in the physical health symptoms, whereas those who initially concealed at lower rates reported greater increases in the frequency of physical health symptoms over the course of the study. This illustration demonstrates that while changes may be relatively smaller on a weekly basis, the accumulation of these smaller changes can be quite meaningful over time.

Contextual support variables were included in the model as time-invariant predictors of the intercepts and slopes for concealing and physical health symptoms. Interestingly, work–family culture negatively predicted the slope for concealing (–.31, p < .01) and the intercept for physical symptoms (–.31, p < .05) such that women in more supportive work–family cultures tended to exhibit less positive changes in concealing over the course of pregnancy and reported a lower frequency of physical health symptoms 12 weeks into their pregnancies. Similarly, perceptions of supervisor support negatively predicted the intercept for concealing (–.26, p < .05) such that women with more supportive supervisors reported lower average levels of concealing. Thus, H5a was partially supported. Work–family culture did not affect the slope of physical health symptoms or the intercept for concealing, and perceived supervisor support was not predictive of the intercept for concealing or the slopes for both concealing and physical health symptoms. Overall, the results failed to support H6a. Taken together, these results suggest pregnant employees who work in more supportive contexts tend to engage in less concealing, experience less frequent physical symptoms on average, and exhibit less positive changes in concealing over the course of pregnancy.

Indeed, these results leave us with an unanswered question regarding the impact of a supportive environment on physical health. On the one hand, contextual support was associated with lower average levels of physical health symptoms. On the other hand, contextual support was also associated with lower average levels of concealing and larger decreases in concealing over time, which was associated with larger progressive reductions in physical health symptoms. This begs the question of whether a supportive environment may actually lead to declines in physical health via reductions in concealing.

To build further understanding of overall impact of contextual support on concealing and physical health symptoms, we plotted the concealing growth trajectories (Figure 3) at high and low levels of work–family culture (given this contextual support variable was significantly associated with the slope of concealing). Figure 3 suggests concealing actually increases over time for those in less supportive work–family cultures, whereas it decreases over time for employees in more supportive work–family cultures. Thus, the benefits of contextual support early on (i.e., at the intercept or the initial period of measurement) appear to diminish over time, in part due to the accompanying decreases in concealing behavior. Ironically, this pattern suggests that pregnant employees in more unsupportive work–family cultures may actually experience physical health benefits as a function of increased concealing over time. It is important to note, however, that tests of the indirect effects of contextual support on physical health symptoms over time indicated these effects were not significantly mediated by concealing behaviors. We will further elaborate on these counterintuitive findings in the Discussion section.

Figure 3

Latent Factor Scores for Concealing Over Time at High and Low Levels of Work–Family Culture

Revealing

As compared to the no coupling model (LL = −1934.239, parameters = 32), the unidirectional model that estimated revealing as the leading indicator of changes in physical health (LL = −1930.67, parameters = 33) and the unidirectional model that estimated physical health as a leading indicator of changes in revealing (LL = −1930.06, parameters = 33) showed significant increases in model fit, (LR = 7.14, df = 1, p < .01 and LR = 8.37, df = 1, p < .01, respectively). However, the bicoupling model (LL = −1926.10, parameters = 34) reflected significantly better fit as compared with the no coupling model (LR = 16.27, df = 2, p < .01), the revealing to changes in physical health unidirectional model (LR = 9.13, df = 1, p < .01), and the physical health to changes in revealing unidirectional model (LR = 7.904, df = 1, p < .01). Thus, the results of the bicoupling model are discussed in detail below (see Table 3).

Table 3

Fit Statistics and Parameter Estimates for the Bivariate Latent Change Score Model With Bicoupling From Revealing to Physical Symptoms and From Physical Symptoms to Revealing

	Revealing	Physical Symptoms
Means (μ)
Intercept	2.97**	2.33**
Slope	0.96**	−0.24^**
Dynamic parameters
Proportional change parameter	−0.08**	0.06
Symptoms → Δrevealing coupling parameter	−0.26**
Revealing → Δsymptoms coupling parameter		0.03**
Random effects
σ intercept	0.92**	0.92**
σ slope	1.00*	0.99**
Time-invariant covariates
Work–family culture → intercept	0.05	−0.31^*
Work–family culture → slope	−0.09	0.14
Supervisor support → intercept	0.25*	0.07
Supervisor support → slope	0.05	−0.02
Fit statistic
Log likelihood	–1926.10
Parameters	34

Note: Intercept values, slope values, and dynamic parameters are unstandardized; random effects and time-invariant covariate values are standardized.

p < .05.

p < .01.

The slope value for revealing was significant and positive (rev₁ = .97, p < .01), indicating that participants tended to change positively in revealing at a constant rate over the course of pregnancy in support of H1c. Furthermore, the negative proportional change parameter (β_con = −.08, p < .01) indicated that prior levels of revealing served to reduce the increases in revealing that subsequently occurred. For example, the increase in revealing that follows from a prior revealing score of 3 (on the original 5-point scale) is .08 units smaller than the increase in revealing that follows from a prior revealing score of 2.

For physical health symptoms, the slope was significant and negative (symp₁ = −.24, p < .01), suggesting participants tended to experience a constant decline in the frequency of physical health symptoms over the course of pregnancy. In the Discussion, we revisit these discrepancies between the physical health trajectories across the different models. Finally, the proportional change parameter for physical health symptoms was not significant, indicating the magnitude of change in symptoms is not proportional to the previous level of physical health symptoms.

The positive coupling parameter from revealing to subsequent changes in physical health symptoms (γ_symp = .03, p < .01) suggests that to the degree women engaged in revealing, they subsequently reported larger increases in the frequency of experienced physical health symptoms, supporting H3₁. Thus, H3₀ was not supported. As Figure 4 illustrates, assuming all individuals report an average level of physical health symptoms (M = 2.33) at the beginning of the study, a low revealing employee’s predicted physical health symptoms score by the end of the study (2.15) is considerably lower as compared to an average revealing employee (2.86) or a high revealing employee (3.58). In other words, individuals who initially revealed at the lowest rates actually experienced a slight decrease in physical health symptoms over the course of the study as compared to employees who initially revealed at higher rates and experienced progressive increases in the frequency of physical health symptoms over time. Finally, the negative coupling parameter from physical health symptoms to subsequent changes in revealing (γ_rev = −.26, p < .01) indicates a higher frequency of physical health symptoms led to subsequently larger decreases in revealing behavior.

Figure 4

Predicted Values of Physical Health Symptoms Across the Span of the Study as a Function of Initial Revealing

Supervisor support positively predicted the revealing intercept (.25, p < .05), providing partial support for H5b, and work–family culture negatively predicted physical health symptoms (–.31, p < .05). Neither contextual support variable predicted the slope values for revealing or physical health, failing to support H6b. Overall, these findings suggested that pregnant employees with more supportive supervisors tended to reveal more in the early stages of pregnancy, and those who worked in more work–family supportive organizations tended to report less frequent physical health symptoms earlier on in the pregnancy. However, as we discussed above, this pattern of findings highlights the possibility that the early benefits of a supportive working environment (lower symptoms) may be attenuated over time in part as a function of subsequent increases in revealing throughout the pregnancy, though as with concealing, tests of the indirect effects of contextual support on physical health symptoms over time suggested these effects were not significantly mediated by revealing behaviors.

Signaling

In comparing model fit statistics among the estimated BLCS models reflecting the relationship between signaling and physical health symptoms, the results suggested neither the unidirectional model that estimated signaling as the leading indicator of changes in physical health (LL = −1747.70, parameters = 33) nor the unidirectional model that estimated physical health as a leading indicator of changes in signaling (LL = −1747.85, parameters = 33) nor the bicoupling model (LL = −1747.7, parameters = 34) yielded significant improvements in model fit over and above the no coupling model (LL = −1748.05, parameters = 32). As a result, H4 was not supported. We report the results of the no coupling model in Table 4.

Table 4

Fit Statistics and Parameter Estimates for the Bivariate Latent Change Score Model With No Coupling From Signaling to Physical Symptoms or From Physical Symptoms to Signaling

	Signaling	Physical Symptoms
Means (μ)
Intercept	1.63**	2.31**
Slope	0.08	−0.16*
Dynamic parameters
Proportional change parameter	−0.05	0.07*
Symptoms → Δsignaling coupling parameter	—
Signaling → Δsymptoms coupling parameter		—
Random effects
σ intercept	0.99**	0.93**
σ slope	0.94**	0.89**
Time-invariant covariates
Work–family culture → intercept	0.12	−0.33*
Work–family culture → slope	−0.20	0.33*
Supervisor support → intercept	−0.11	0.11
Supervisor support → slope	0.30	−0.002
Fit statistic
Log likelihood	–1748.05
Parameters	32

Note: Intercept values, slope values, and dynamic parameters are unstandardized; random effects and time-invariant covariate values are standardized.

p < .05.

p < .01.

Neither the slope coefficient nor the proportional change parameter for signaling was significant, suggesting signaling tends to remain relatively stable throughout the course of pregnancy. Thus, H1b was not supported.

For physical health symptoms, the slope was significant and negative (symp₁ = −.16, p < .05), suggesting again that participants tended to experience constant negative rate of change in physical health symptoms over the course of pregnancy. Finally, the proportional change parameter for physical health symptoms was significant and positive (β_symp = .07, p < .05), suggesting that prior levels of physical health served to reduce the decreases in physical health symptoms that subsequently occurred. For example, the decrease in physical health symptoms that follows from a prior symptoms score of 3 (on the original 4-point scale) is .07 units smaller than the decrease in physical health symptoms that follows from a prior symptoms score of 2.

Neither of the contextual support variables predicted the slope or intercept for signaling, failing to provide support for H5c and H6c. However, to the extent that women reported working in more work–family supportive cultures, they tended to report less frequent physical health symptoms on average (–.33, p < .05) as well as experience a less negative constant rate of change in the frequency of physical health symptoms experienced over the course of pregnancy (.33, p < .05).

Discussion

The primary purpose of this article was to explore how pregnant women manage their concealable stigmatized identities at work over the course of pregnancy and the implications of those processes on physical health well-being. In addition, we explored the possibility that physical health symptoms influence subsequent identity management behaviors. Finally, we examined how sources of contextual support affected the progressive use of identity management behaviors over the course of pregnancy. Taken together, the results of the current study yield important conclusions about how pregnant employees manage their identities at work.

First, the results highlight the unique interplay among concealing, revealing, and physical health symptoms. Specifically, our results suggested a unidirectional relationship between concealing and physical health wherein concealing led to improved physical health. In contrast, the relationship between revealing and physical health appeared to be reciprocal in nature wherein revealing led to declines in physical health and declines in physical health triggered decreases in revealing. The fact that concealing and revealing exhibited empirically distinct patterns with physical health rather than exactly the same results but in the opposite direction provides support for the notion that both strategies are important in their own right and yield unique insights over and above the other. Overall, our results suggest that pregnant workers’ physical health is indeed affected by how they choose to manage their pregnant identities; however, decisions about revealing pregnancy are also based in part on physical health.

Second, our findings demonstrate how a supportive organizational context may potentially function to improve pregnant employees’ experiences at work, but may also ultimately produce unintended adverse physical health consequences. We found that women who worked in supportive contexts reported lower levels of concealing, higher levels of revealing, and less frequent physical health symptoms at the beginning of the study. This suggests that pregnant employees in more supportive contexts may benefit early on from lower physical health symptoms and may also feel more comfortable in their environments, facilitating increased openness about the pregnancy. However, our findings also suggest these early benefits of supportive working environments may be undermined via progressive decreases in concealing and increases in revealing, which our results suggest lead to larger declines in physical health. Therefore, a supportive working environment did not ultimately confer health benefits (though it may have conferred other, unmeasured, psychological benefits more closely tied to the reduction in concealing).

Theoretical Implications

Our findings shed light on key issues for development of theory on the identity management of concealable stigmas. First, it is clear that the identity management process not only varies between individuals but also within individuals across time, consistent with extant scholarship suggesting the importance of situational cues in determining disclosure decisions (Cain, 1991; Chrobot-Mason et al., 2001; Woods, 1994). Thus, it is clear from these findings that examinations of the experiences of stigmatized workers must take a more nuanced view to understanding identity management—it cannot be enough to ask whether someone is “out” or “not out.” Instead, we must think of identity management and disclosure as a complex process involving ongoing decisions and behavioral episodes.

Second, our findings shed light on the competing motives pregnant workers face when making decisions about to whom, when, and how to disclosure their pregnant status. Consistent with a stigma theory explanation (Goffman, 1963), our results suggest pregnant women experience improved physical well-being when they conceal their pregnant status from others at work and declines in physical health following revealing behavior, supporting the notion that pregnant employees hide their pregnancies as a protective strategy from negative stigmatization such as stereotyping, assumptions regarding decreased commitment, and loss of power.

Finally, our findings highlight the complex nature of the relationship between identity management behaviors and physical health over the course of pregnancy. Our results suggest this relationship is not as simple as identity management influencing physical health or physical health influencing identity management; rather, concealing emerged as the leading indicator of physical health, whereas revealing and physical health exhibited a more dynamic relationship wherein they influenced one another across time. These findings align with previous qualitative evidence suggesting that pregnant workers fear physical changes and challenges associated with pregnancy will detract from their professional image (Greenberg et al., 2009).

Practical Implications

From a practical perspective, the results of the current study yield several important implications. First, poor physical health during pregnancy leads to a higher risk of health problems for both mother and child (Glover & O’Connor, 2002; Strauss, 2000). Furthermore, poor employee health has direct costs for organizations including increased medical costs (Goetzel, Hawkins, Ozminkowski, & Shaohung, 2003; Wang et al., 2003) and decreased productivity (Burton, Conti, Chen, Schultz, & Edington, 1999). Given these high-stakes outcomes, factors that affect physical health during pregnancy such as identity management and contextual support warrant sincere consideration.

A second practical implication of these findings is the need for organizations to consider how increased physical complications for pregnant employees might affect individuals’ relationships with their supervisor(s), coworkers, and the job itself, particularly if they feel pressure from these constituencies to hide or minimize the physical effects of their pregnancies. Our results shed light on a potential solution: More supportive contexts were associated with better physical health in the early stages of pregnancy. However, as we discussed above, supportive contexts may also produce adverse unintended health consequences that accumulate over time.

From the perspective of organizational support theory, expectant mothers who perceive high support right before departing for maternity leave may feel obligated to contribute to the organization in return as a function of the reciprocity norm (Rhoades & Eisenberger, 2002). As a result, they may feel guilty about leaving and experience adverse physical health consequences as a byproduct of the stress associated with such guilt. This is not to suggest that supervisors not be supportive of their employees who are on the verge of maternity leave. Rather, supervisors should consider explicitly communicating to pregnant employees that they should not feel guilty about taking time away from work for maternity leave.

This is particularly important given that the way in which expectant employees experience work during pregnancy likely affects their intention to return to work post–maternity leave as well as their attitudes and experiences at work upon reentry, as suggested by previous research demonstrating enduring effects of work–pregnancy experiences on job attitudes postpartum (Holtzman & Glass, 1999; Killien, Habermann, & Jarrett, 2001). Thus, even though in some sense the “stigma” of pregnancy disappears after the child is born, a pregnant employee’s workplace experiences likely have important long-term outcomes.

To this end, supervisory training may be a beneficial means of communicating the importance of showing support toward and building trust with subordinates. The unique challenges experienced by pregnant employees and the pressure they feel to manage their identities at work could be integrated into the training. Ultimately, pregnant employees have little control over the extent to which their supervisors and organizations are supportive of work–family integration, but they do have control over how they choose to manage their pregnant identities, and the notion that certain strategies may be more or less advantageous under certain conditions is empowering to individuals facing such decisions. Thus, our findings shed light on the active and strategic role pregnant women may take in shaping their own work–pregnancy experiences.

Limitations and Future Research Directions

The current findings should be interpreted in light of the study’s limitations. First, contrary to our predictions, our results yielded no evidence of a relationship between signaling and physical health well-being. One reason for this may have been the relatively low internal consistencies of our signaling items. Given signaling has been relatively underexamined in the identity management literature as compared to revealing and concealing, further exploration and clarification of the nature and measurement of signaling reflect an important and fruitful direction for future research.

In addition, counter to our initial expectations, concealing appeared to change positively rather than negatively over time after controlling for autoregressive effects. Post hoc reasoning suggests one potential explanation for this counterintuitive finding may be that some mothers decide early on whether to conceal or not to conceal. For those who decide to conceal, it is easy at first as there are no outward signs or symptoms to worry about; over time, however, these individuals have to put forth more effort to conceal as the visibility of the pregnancy increases. Furthermore, this possibility might also account for the significant amount of variability in the concealing growth trajectories that we observed.

Second, the complexities associated with estimating these types of models inevitably create some difficulty with the interpretation of results. For instance, the BLCS results examining concealing did not provide support for the notion that physical health symptoms changed at a constant rate over the course of pregnancy; however, the results of the other two BLCS analyses (examining revealing and signaling) suggested that physical health symptoms declined significantly over time. One potential explanation for these inconsistencies may stem from the possibility that physical health symptoms vary in a nonlinear fashion over the course of pregnancy. For example, symptoms may occur more frequently during first trimester due to morning sickness, occur less frequently during the second trimester, and increase again during the third trimester as childbirth approaches.

Furthermore, the relatively low internal reliability of the physical health symptoms measure at the within-person level may have contributed to the variant findings across models. Indeed, the physical health symptoms measure was more formative than reflective in nature (see Edwards & Bagozzi, 2000). In other words, rather than the covariance of specific symptoms being the underlying construct of interest, the measure was more of a count variable reflecting the sheer number of symptoms experienced in a given week. Thus, we may not actually expect the physical health symptoms measure to have high internal consistency at the within-person level since on any given day some people may not exhibit the same exact pattern of symptoms as the day before or the week before. However, it may be more realistic to expect higher internal consistency of the physical health symptoms measure at the between-person level. Regardless, in light of this divergence in our results, future research should indeed further examine the nature of the physical health trajectory during pregnancy.

Third, ideally our sample would have included pregnant women still in their first trimester; however, the high rate of miscarriage during that period led us to exclude such participants. Furthermore, our sample was relatively homogenous in terms of ethnicity and martial status, as most of our participants were married and White. However, drawing from previous research suggesting that minority women tend to be “doubly stigmatized” for their membership in multiple disadvantaged groups (Berdahl & Moore, 2006), it seems likely that the impact of identity management on physical health may be even more exacerbated for ethnic minority pregnant employees as compared to the effects found in our sample.

Fourth, we have no way of knowing for sure what specifically was motivating these women to reveal and conceal as we did not ask them directly. Future research should aim to identify and assess the full range of reasons that motivate pregnant employees’ disclosure decisions. Furthermore, another fruitful direction for future research is the examination of potentially beneficial stereotypes about working mothers, including beliefs about mothers as more reliable and dependable as compared to younger, unmarried, childless women.

Fifth, when using a multiple measurement method like experience sampling, a common concern is that repeatedly asking participants about the same things could prime them to behave in ways that are different from how they would naturally behave (Scollon, Kim-Prieto, & Diener, 2003). However, we believe reactivity was less of a concern in our study given an entire week separated each measurement occasion as opposed to traditional experience sampling designs that generally ask participants to respond to surveys several times a day over the course of a few weeks (Beal & Weiss, 2003).

Finally, our weekly measure of identity management did not capture the more nuanced characteristics of identity management at the interaction level (i.e., who the women are managing identities with). Future research should take a more fine-grained view of identity management to more fully understanding these processes. For instance, concealing to a close friend may be damaging to well-being, whereas concealing to a client may result in relatively little impact on well-being.

Given the findings and limitations, future research on identity management experiences of pregnant employees should examine other predictors (i.e., both time-invariant and time-varying) that may account for additional slope and intercept variance in the identity management growth trajectories (e.g., personality, perceived discrimination, job stress, gender composition of the organization) and other possible consequences of identity management behaviors (e.g., turnover, satisfaction, organizational commitment).

Conclusion

To conclude, we have provided unique empirical data illustrating how expectant employees manage their pregnant identities and their reports of physical health over the course of pregnancy. From our investigation, it is clear that the nature of the relationship between identity management and physical health is complex. Specifically, our results suggest the relationship is unidirectional for concealing such that concealing led to subsequent improvements in physical health. However, the relationship was more complex for revealing such that revealing led to declines in physical health and declines in physical health led to subsequent decreases in revealing. Taken together, our results support a stigma theory explanation wherein the link between pregnant workers’ identity management decisions and their physical health well-being is explained, in part, by self-protection from and increased vulnerability to stigmatization. Finally, supportive contexts were associated with better physical health, higher levels of revealing, and lower levels of concealing in the early stages of pregnancy; however, these physical health benefits appeared to diminish over time. Future research should seek to examine identity management within a framework that conceptualizes it as a complex process involving ongoing decisions and behavioral episodes.

Footnotes

Acknowledgements

This article was accepted under the editorship of Deborah E. Rupp.

References

Allen

T. D.

2001. Family-supportive work environments: The role of organizational perceptions. Journal of Vocational Behavior, 58: 414-435.

Bailey

1999. Refracted selves? A study of changes in self-identity in the transition to motherhood. Sociology, 33: 335-352.

Beal

D. J.

Weiss

H. M.

2003. Methods of ecological momentary assessment in organizational research. Organizational Research Methods, 6: 440-464.

Bentler

P. M.

Bonett

D. G.

1980. Significance tests and goodness of fit in analysis of covariance structures. Psychological Bulletin, 88: 588-606.

Berdahl

J. L.

Moore

2006. Workplace harassment: Double jeopardy for minority women. Journal of Applied Psychology, 91: 426-436.

Bolger

Davis

Rafaeli

2003. Diary methods: Capturing life as it is lived. Annual Review of Psychology, 54: 579-616.

Budig

M. J.

England

2001. The wage penalty for motherhood. American Sociological Review, 66: 204-225.

Burton

W. M.

Conti

D. J.

Chen

Schultz

A. B.

Edington

D. W.

1999. The role of health risk factors and disease on worker productivity. Journal of Occupational and Environmental Medicine, 41: 863-877.

Button

S. B.

1996. Organizational efforts to affirm sexual diversity: A multi-level examination. Unpublished doctoral dissertation, Pennsylvania State University, University Park.

10.

Button

S. B.

2001. Organizational efforts to affirm sexual diversity: A cross-level examination. Journal of Applied Psychology, 86: 17-28.

11.

Cain

1991. Personal stories: Identity acquisition and self-understanding in Alcoholics Anonymous. Ethos, 19: 210-253.

12.

Chaudoir

S. R.

Quinn

D. M.

2010. Bystander sexism in the intergroup context: The impact of cat-calls on women’s reactions towards men. Sex Roles, 62: 623-634.

13.

Christensen

T. C.

Barrett

L. F.

Bliss-Moreau

2003. A practical guide to experience-sampling procedures. Journal of Happiness Studies, 4: 53-78.

14.

Chrobot-Mason

Button

S. B.

DiClementi

J. D.

2001. Sexual identity management strategies: An exploration of antecedents and consequences. Sex Roles, 45: 321-336.

15.

Clair

J. A.

Beatty

J. E.

Maclean

T. L.

2005. Out of sight but not out of mind: Managing invisible social identities in the workplace. Academy of Management Review, 30: 8-95.

16.

Croteau

J. M.

Lark

J. S.

1995. On being lesbian, gay, or bisexual in student affairs: A national survey of experiences on the job. NASPA Journal, 32: 189-197.

17.

Cuddy

A. J. C.

Fiske

S. T.

Glick

2004. When professionals become mothers, warmth doesn’t cut the ice. Journal of Social Issues, 60: 701-718.

18.

D’Augelli

A. R.

Grossman

A. H.

2001. Disclosure of sexual orientation, victimization, and mental health among lesbian, gay, and bisexual older adults. Journal of Interpersonal Violence, 16: 1008-1027.

19.

Duehr

Bono

J. E.

2006. Men, women, and managers: Are stereotypes finally changing? Personnel Psychology, 59: 815-846.

20.

Edwards

J. R.

Bagozzi

R. P.

2000. On the nature and direction of relationships between constructs and measures. Psychological Methods, 5: 155-174.

21.

Eisenberger

Huntington

Hutchinson

Sowa

1986. Perceived organizational support. Journal of Applied Psychology, 71: 500-507.

22.

Fawzy

F. I.

Fawzy

N. W.

Hyun

C. S.

Elashoff

Guthrie

Fahey

J. L.

Morton

D. L.

1993. Malignant melanoma: Effects of an early structured psychiatric intervention, coping, and affective state on recurrence and survival 6 years later. Archives of General Psychiatry, 50: 681-689.

23.

Glover

O’Connor

T. G.

2002. Effects of antenatal stress and anxiety: Implications for development and psychiatry. British Journal of Psychiatry, 180: 389-391.

24.

Goetzel

R. Z.

Hawkins

Ozminkowski

R. J.

Shaohung

2003. The health and productivity cost burden of the “top 10” physical and mental health conditions affecting six large U.S. employers in 1999. Journal of Occupational and Environmental Medicine, 45: 5-14.

25.

Goffman

1963. Stigma: Notes on the management of spoiled identity. New York: Simon & Schuster.

26.

Greenberg

Ladge

J. J.

Clair

J. A.

2009. Negotiating pregnancy at work: Public and private conflicts. Negotiation and Conflict Management Research, 2: 42-56.

27.

Grimm

K. J.

2007. Multivariate longitudinal methods for studying developmental relationships between depression and academic achievement. International Journal of Behavioral Development, 31: 328-339.

28.

Grimm

K. J.

McArdle

J. J.

Zonderman

A. B.

Resnick

S. M.

2012. Recent changes leading to subsequent changes: Extensions of multivariate latent difference score models. Structural Equation Modeling, 19: 268-292.

29.

Gross

J. J.

John

O. P.

2003. Individual differences in two emotion regulation processes: Implications for affect, relationships, and well-being. Journal of Personality and Social Psychology, 85: 348-362.

30.

Gross

J. J.

Levenson

R. W.

1993. Emotional suppression: Physiology, self-report, and expressive behavior. Journal of Personality and Social Psychology, 64: 970-986.

31.

Gross

J. J.

Levenson

R. W.

1997. Hiding feelings: The acute effects of inhibiting negative and positive emotion. Journal of Abnormal Psychology, 106: 95-103.

32.

Halbesleben

J. R. B.

Wheeler

A. R.

2015. To invest or not? The role of coworker support and trust in daily reciprocal gain spirals of helping behavior. Journal of Management, 41: 1628-1650.

33.

Hays

R. B.

McKusick

Pollack

Hilliard

Hoff

Coates

T. J.

1993. Disclosing HIV seropositivity to significant others. AIDS, 7: 425-431.

34.

Hebl

M. R.

King

E. B.

Glick

Kazama

Singletary

2007. Hostile and benevolent reactions toward pregnant women: Complementary interpersonal punishments and rewards that maintain traditional roles. Journal of Applied Psychology, 92: 1499-1511.

35.

Heine

S. J.

Proulx

Vohs

K. D.

2006. The meaning maintenance model: On the coherence of social motivations. Personality and Social Psychological Review, 10: 88-111.

36.

Holtzman

Glass

1999. Explaining changes in mothers’ job satisfaction following childbirth. Work and Occupations, 23: 365-404.

37.

L. T.

Bentler

P. M.

1999. Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling, 6: 1-55.

38.

Jones

E. E.

Farina

Hastorf

Markus

Miller

Scott

R. A.

1984. Social stigma: The psychology of marked relationships. New York: Freeman.

39.

Killien

M. G.

Habermann

Jarrett

2001. Influence of employee characteristics on postpartum mothers’ health. Women and Health, 33: 63-81.

40.

King

E. B.

Botsford

2009. Managing pregnancy disclosures: Understanding and overcoming the challenges of expectant motherhood at work. Human Resource Management Review, 19: 314-323.

41.

Ladge

J. J.

2008. Becoming a working mother: Identity, efficacy and re-socialization following re-entry. Unpublished doctoral dissertation, Boston College, Chestnut Hill, MA.

42.

Levine

M. P.

Leonard

1984. Discrimination against lesbians in the work force. Signs: The Lesbian Issue, 9: 700-710.

43.

Major

V. S.

2004. Pregnancy in the workplace: Stigmatization and work identity management among pregnant employees. Unpublished doctoral dissertation, University of Maryland, College Park.

44.

Marks

Bundek

N. I.

Richardson

J. L.

Ruiz

M. S.

Maldonado

Mason

H. R. C.

1992. Self disclosure of HIV infection: Preliminary results for a sample of Hispanic men. Health Psychology, 11: 300-306.

45.

McArdle

J. J.

2001. A latent difference score approach to longitudinal dynamic structural analysis. In Cudeck

du Toit

Sorbom

(Eds.), Structural equation modeling: Present and future: 342-380. Lincolnwood, IL: Scientific Software International.

46.

McArdle

J. J.

2009. Latent variable modeling of differences and changes with longitudinal data. Annual Review of Psychology, 60: 577-605.

47.

Millward

L. J.

2006. The transition to motherhood in an organizational context: An interpretative phenomenological analysis. Journal of Occupational and Organizational Psychology, 79: 315-333.

48.

Muthén

L. K.

Muthén

B. O.

2010. Mplus: Statistical analysis with latent variables: User’s guide. Los Angeles: Muthén & Muthén.

49.

Pachankis

J. E.

2007. The psychological implications of concealing a stigma: A cognitive-affective-behavioral model. Psychological Bulletin, 133: 328-345.

50.

Pascoe

E. A.

Smart Richman

2009. Perceived discrimination and health: A meta-analytic review. Psychological Bulletin, 135: 531-554.

51.

Pennebaker

J. W.

1990. Opening up: The healing power of confiding in others. New York: William Morrow.

52.

Perlow

L. A.

1995. Putting work back into work–family. Group & Organization Management, 20: 227-239.

53.

Proulx

Inzlicht

2012. The five “A”s of meaning maintenance: Finding meaning in the theories of sense-making. Psychological Inquiry, 23: 317-335.

54.

Quinn

D. M.

Chaudoir

S. R.

2009. Living with a concealable stigmatized identity: The impact of anticipated stigma, centrality, salience, and cultural stigma on psychological distress and health. Journal of Personality and Social Psychology, 97: 634-651.

55.

Ragins

B. R.

2008. Disclosure disconnects: Antecedents and consequences of disclosing invisible stigmas across life domains. Academy of Management Review, 33: 194-215.

56.

Ragins

B. R.

Cornwell

J. M.

2001. Pink triangles: Antecedents and consequences of perceived workplace discrimination against gay and lesbian employees. Journal of Applied Psychology, 86: 1244-1261.

57.

Ragins

B. R.

Singh

Cornwell

J. M.

2007. Making the invisible visible: Fear and disclosure of sexual orientation at work. Journal of Applied Psychology, 92: 1103-1118.

58.

Rhoades

Eisenberger

2002. Perceived organizational support: A review of the literature. Journal of Applied Psychology, 87: 698-714.

59.

Ridgeway

Correll

S. J.

2004. Motherhood as a status characteristic. The Journal of Social Issues, 60: 683-700.

60.

Rosette

A. S.

Leonardelli

G. J.

Phillips

K. W.

2008. The white standard: Racial bias in leader categorization. Journal of Applied Psychology, 93: 758-777.

61.

Rosette

A. S.

Tost

L. P.

2010. Agentic women and communal leadership: How role prescriptions confer advantage to top women leaders. Journal of Applied Psychology, 95: 221-235.

62.

Schwartz

F. N.

1992. Women as a business imperative. Harvard Business Review, 70(2): 105-113.

63.

Scollon

C. N.

Kim-Prieto

Diener

2003. Experience sampling: Promises and pitfalls, strengths and weaknesses. Journal of Happiness Studies, 4: 5-34.

64.

Starrels

M. E.

1992. The evolution of workplace policy research. Journal of Family Issues, 13: 259-278.

65.

Strachan

E. D.

Bennett

W. R. M.

Russo

Roy-Byrne

P. P.

2007. Disclosure of HIV status and sexual orientation independently predicts increased absolute CD4 cell counts over time for psychiatric patients. Psychosomatic Medicine, 69: 74-80.

66.

Strauman

T. J.

Lemieux

A. M.

Coe

C. L.

1993. Self-discrepancy and natural killer cell activity: Immunological consequences of negative self-evaluation. Journal of Personality and Social Psychology, 64: 1042-1052.

67.

Strauss

R. S.

2000. Adult functional outcome of those born small for gestational age: Twenty-six-year follow-up of the 1970 British Birth Cohort. Journal of the American Medical Association, 283: 625-632.

68.

Swann

W. B.

1987. Identity negotiation: Where two roads meet. Journal of Personality and Social Psychology, 53: 1038-1051.

69.

Swann

W. B.

Jr. Polzer

J. T.

Seyle

D. C.

S. J.

2004. Finding value in diversity: Verification of personal and social self-views in diverse groups. Academy of Management Review, 29: 9-27.

70.

Swann

W. B.

Jr. Rentfrow

P. J.

Guinn

J. S.

2002. Self-verification: The search for coherence. In Leary

Tagney

(Eds.), Handbook of self and identity: 367-383. New York: Guilford.

71.

Thompson

C. A.

Beauvais

L. L.

Lyness

K. S.

1999. When work–family benefits are not enough: The influence of work–family culture on benefit utilization, organizational attachment, and work–family conflict. Journal of Vocational Behavior, 54: 392-415.

72.

Thompson

C. A.

Thomas

C. C.

Maier

1992. Work–family conflict: Reassigning corporate policies and initiatives. In Sekaran

Leong

F. T.

(Eds.), Womanpower: Managing in times of demographic turbulence: 59-84. Newbury Park, CA: Sage.

73.

Townsend

S. M.

Major

Sawyer

P. J.

Mendes

2010. Can the absence of prejudice be more threatening than its presence? It depends on one’s worldview. Journal of Personality and Social Psychology, 99: 933-947.

74.

U.S. Equal Employment Opportunity Commission. 2011. Pregnancy discrimination charges; EEOC & FEPAs combined: FY 1997–FY 2011. Retrieved from http://www.eeoc.gov/eeoc/statistics/enforcement/pregnancy.cfm

75.

Wang

Schultz

A. B.

Musich

McDonlad

Hirschland

Edington

2003. The relationship between National Heart, Lung, and Blood Institute weight guidelines and concurrent medical costs in a manufacturing population. American Journal of Health Promotion, 17: 183-189.

76.

Williams

J. C.

Segal

2004. Beyond the maternal wall: Relief for family caregivers who are discriminated against on the job. Harvard Women’s Law Journal, 26: 77-162.

77.

Woods

J. D.

1994. The corporate closet. New York: Free Press.