Distancing or drawing together: Sexism and organisational tolerance of sexism impact women's social relationships at work

Research shows that women often face sexism in the workplace (Davison & Burke, 2000; Stamarski & Son Hing, 2015; Swim, Hyers, Cohen, & Ferguson, 2001). Sexism can be defined as unequal evaluations and treatment of men and women based on their sex (Swim & Hyers, 2009) and includes a range of daily hassles and negative life events (Klonoff & Landrine, 1995). Sexism affects women negatively in different ways, such as by leading to unfair payment (Petersen & Morgan, 1995) and lack of leadership opportunities (Barreto, Ryan, & Schmitt, 2009; Eagly & Carli, 2007). Moreover, disadvantage in the workplace causes stress and generally negatively affects women's Received: 22 October 2019 | Accepted: 26 May 2020 DOI: 10.1002/ejsp.2695

psychological wellbeing (Barreto & Ellemers, 2013;Borrell et al., 2010;Schmader, Johns, & Forbes, 2008). For instance, women's reports of personal experiences with discrimination are positively correlated with self-reported depression (Kobrynowicz & Branscombe, 1997) and inversely correlated with personal self-esteem (Schmitt, Branscombe, Kobrynowicz, & Owen, 2002). However, not much is known about the impact of discrimination on interpersonal relationships within the group that is targeted by discrimination. Prior research in this area does provide some indications, but these are largely inconsistent. This article aims to advance understanding of how sexism affects women's interpersonal relationships with other women by taking into account the role of tolerance of sexism in the organisational environment.

| The importance of social relationships at work
The quality of one's social relationships is an important aspect of individual wellbeing (Baumeister & Leary, 1995;Ryff & Singer, 2000). In the workplace, workers with high quality work relationships with their peers and managers report more positive emotions (Colbert, Bono, & Purvanova, 2016) and greater job satisfaction (Ragins & Dutton, 2007;Einarsen, Hoel, Zapf, & Cooper, 2011). In addition, social relationships at work also affect organisations. High quality workplace relationships with peers and managers are positively associated with productivity (Ragins & Dutton, 2007;Einarsen et al., 2011) and organisational commitment (Hanpachern, Morgan, & Griego, 1998).
Research on the role of social relationships in health and wellbeing indicates that social relationships constitute an important resource to cope with negative or stressful experiences (Cohen & Wills, 1985;Frisch, Hausser, Van Dick, & Mojzisch, 2014;Jetten, Haslam, & Haslam, 2012). In other words, those who have positive social relationships with others are better able to cope with a variety of negative life events (and some argue this is especially the case for women compared to men; Taylor et al., 2000). Social relationships with other members of the same in-group have particularly important benefits for health and wellbeing (Frable, Platt, & Hoey, 1998;Sanchez & Garcia, 2009). Studies focusing specifically on workplace friendships, or informal relationships with peers at the same level in the hierarchy (Bridge & Baxter, 1992;Myers & Johnson, 2004), have also highlighted the importance of peer relationships, especially among members of minority or disadvantaged groups (Hays, 1989;Jones, 1991). In sum, social relationships with other women are an important resource for women's wellbeing. Here, we examine whether experiences of sexism might interfere with these relationships, hindering or facilitating women's access to this important resource.

| The impact of sexism on women's social relationships with other women
There is a developing line of research on how devaluation affects social relationships, but findings in this area are inconsistent and sometimes contradictory. On the one hand, research has shown that being targeted by prejudice and discrimination can negatively affect close interpersonal relationships, including those with friends and family (Doyle & Molix, 2014a, 2014b, 2015b. However, that work does not specifically address the effects of devaluation on relationships with other members of the devalued group (i.e., in-group members). In the context of gender, some research has shown that women can respond to gender discrimination by distancing themselves from other women (Derks, Van Laar, & Ellemers, 2016;Ely, 1994;Faniko, Ellemers, & Derks, 2016), especially if being a woman is not an important part of their identity to begin with . At the same time, however, it has often been suggested that members of devalued groups tend to "draw together" when threatened (Haslam & Reicher, 2006). For example, the extent to which individuals perceive themselves and their group to be targets of discrimination is positively associated with in-group identification (Branscombe, Schmitt, & Harvey, 1999;Jetten, Branscombe, Schmitt, & Spears, 2001).
One source of confusion in this area is that what can broadly qualify as self-group distancing is very diverse and varies in critical ways. For example, researchers studying self-group distancing have examined how experiences of sexism impact how women relate to women as a whole (e.g., how similar they see themselves to the typical woman,  to what extent they identify with other women, Branscombe et al., 1999), as well as how experiences of sexism impact the extent to which women use stereotypically feminine versus masculine traits to describe themselves Faniko et al., 2016;Faniko, Ellemers, Derks, & Lorenzi-Cioldi, 2017), and by looking at how experiences of sexism impact women's evaluations of their subordinates Faniko et al., 2016Faniko et al., , 2017. Our aim in this article is to contribute to this literature by focusing specifically on women's interpersonal relationships with in-group members, that is, female co-workers. In addition, while the majority of the self-group distancing work was particularly focused on uncovering the conditions under which women distance themselves from each other, we are additionally interested in when they might draw together in response to sexism (see also . Indeed, self-group distancing is problematic both because it can be interpreted as discrimination from in-group members and because it inhibits the social support (seeking and provision), which is most directly evidenced by directly examining when women draw together.
Prior research on interpersonal relationships (and in-group relationships) therefore suggests that women might respond to sexism at work by distancing themselves from other women, but there is also evidence that the opposite can happen, whereby experiences of sexism instead lead women to draw together. In this article, we hope to clarify these somewhat contradictory findings by considering the role played by organisational climate in shaping women's relationships with other women at work. We argue that when women experience sexism at work, the organisational climate in which that sexism occurs can hinder or facilitate women's access to this important coping response. Specifically, when women experience sexism at work, the perception that the organisational climate is intolerant of sexism will lead women to report more positive social relationships with other women in the workplace. However, the perception that organisational climate is tolerant of sexism may hinder drawing towards other women as a response to sexism.

| The role of organisational climate
Research has shown that organisational climates that tolerate sexism tend to be associated with more sexism and sexual harassment compared to organisational climates that do not tolerate sexism (Parker, Griffin, Sprigg, & Wall, 2002;Willness, Steel, & Lee, 2007). However, an organisational climate that does not tolerate sexism does not automatically eliminate sexism altogether (Kaiser et al., 2013). We argue that perceived organisational tolerance of sexism is also likely to shape health and wellbeing outcomes when sexism does occur in an organisation. When sexism occurs in an organisation where there is a climate of tolerance of sexism, this might affect how pervasive sexism is expected to be.
Sexism that is perceived to be pervasive has been shown to have more damaging effects on wellbeing (an important correlate and precursor of social affiliative behaviour) than sexism that is perceived to be rare Stroebe, Dovidio, Barreto, Ellemers, & John, 2011). In addition, when organisations tolerate sexism, victims are more likely to expect that complaining or seeking social support is likely to be very costly-and it is already known that the anticipated social costs of complaining decrease actual complaints (Shelton & Stewart, 2004). In accordance with this reasoning, we argue that the organisational climate within which sexism occurs will impact women's social relationships following experiences of sexism. We specifically expect that when women experience sexism at work, the perception that the organisational climate does not tolerate sexism will have a beneficial effect on their relationships with female colleagues. However, we expect that this beneficial effect will fade away when the organisational climate is perceived to tolerate sexism.
Organisational climates include practices, procedures, and norms about behaviour that an organisation expects from its employees (Ostroff, Kinicki, & Tamkins, 2003;Schein, 2010;Zohar & Hofmann, 2012). The particular focus of this article is on organisational tolerance of sexism, which therefore refers to the absences of practices, procedures, and norms about the inappropriateness of sexism in the workplace. Importantly, organisational climates can also be expressed at the interpersonal level (Paustian-Underdahl, King, Rogelberg, Kulich, & Gentry, 2017), for instance by managers and by peers (Madlock & Booth-Butterfield, 2011), and having a high-level organisational policy that clarifies intolerance of discrimination does not mean that this necessarily translates into similar levels of intolerance of discrimination at lower levels in the organisation (Brady, Kaiser, Major, & Kirby, 2015;Hebl, Foster, Mannix, & Dovidio, 2002). In this article, then, organisational climate is operationalised at three levels, namely the extent to which (a) peers, (b) managers, and (3) organisational policies do or do not reject sexism and support those targeted by sexist treatment.
Organisational policies around equality and diversity clarify that discrimination is unacceptable and facilitate disciplinary action when it occurs-and therefore may reduce expectations of discriminatory or unfair treatment (Button, 2001;Ragins & Cornwell, 2001;Ruggs, Martinez, Hebl, & Law, 2015). Previous studies support the idea that managers also strongly influence the extent to which the workplace is perceived to tolerate discrimination (Bulutlar & Oz, 2009;Salin, 2003;Samnani, 2012). One reason for this is that managers are expected to have the power to determine how much intolerance of discrimination is emphasised at the local level, as well as how discriminatory events are handled (Edelman, 2005;Martinez, Ruggs, Sabat, Hebl, & Binggeli, 2013), and to provide support to victims of discrimination (Beehr, King, & King, 1990;Ely, 1994;Ragins & Scandura, 1994). However, the most proximal determinant of organisational climate is one's peers. As is the case for managers, it has been shown that peer networks play an important role in establishing social norms within the work context (James & Sells, 1981;James, James, & Ashe, 1990), and this is also likely to be true with regard to norms around sexism. However, it is not known how this might affect social relationships among women after they have experienced sexism, which is the focus of the current work. One previous study (Doyle & Molix, 2015a) with sexual minorities in the United States showed that discrimination was associated with greater friendship strain, but that this effect was reversed for those who lived in states with laws and policies that supported sexual minorities against discrimination. However, this study examined interpersonal relationships broadly (rather than social relationships with members of the in-group) and it was not conducted within an organisational setting as is the current research.

| The present research
In this research, we examined how experiences of sexism at work affect women's social relationships with other women. We assessed the role played by tolerance of sexism in the workplace, including perceptions of tolerance of sexism at the peer-, manager-, and policylevel. We expected to find that women respond to sexism by drawing together when they perceive the organisational climate as less tolerant of sexism. However, we expect that drawing together behaviour is not triggered when they perceive the organisational climate as more tolerant of sexism. Additionally, we aim to provide insight into which source (or sources) of tolerance of sexism (i.e., peer, manager, policy) have a more substantial impact on women's social relationships with other women following experiences of sexism at work. Studies in this article conform to APA ethical guidelines and have been approved by the ethical review committee at the University of Exeter. Data associated with these studies can be viewed at https:// osf.io/af95p/ ?view_only=a9dd2 6d879 834a3 0bd49 652cd 4195db0.

| S TUDY 1: CORREL ATIONAL S TUDY
In Study 1, we surveyed women working in the United Kingdom (UK), asking about their experiences of sexism in the workplace during the previous 6 months as well as perceived tolerance of sexism among their peers, their managers, and policies within their organisations. In addition, we assessed social relationships between participants and their female co-workers using a measures of co-worker friendship.
In line with previous literature examining interpersonal relationships as a function of exposure to discrimination (e.g., Doyle & Molix, 2014a) we hypothesised that, overall, experiences of sexism would have detrimental effects on social relationships among women (H1). However, we hypothesised an interaction between experiences of sexism and perceived tolerance of sexism within the organisation (H2), such that experiences of sexism would be positively associated with social relationships with female colleagues, but only when the organisational climate was perceived to be less tolerant of sexism. We hypothesised that this effect would disappear when the organisational climate was perceived to be more tolerant of sexism.

| Design and participants
This was a cross-sectional study among working women resident in the UK. All participants were sampled from Prolific Academic and were compensated £2 for their time and effort, which is on par with payments for tasks of comparable length (approximately 15 min).
Using the effect size obtained in a pilot study 1 (f 2 = 0.02; power = 0.80, α = .05), power analyses conducted in G*Power version 3.1 (Erdfelder, Faul, & Buchner, 1996) indicated a sample size of 395 was necessary to detect a comparable effect. A total of 406 participants were recruited online. However, one participant was excluded from the study as they reported they worked alone with no peers or managers in their daily work environment. The age of the final 405 participants More than half of the participants (64%) indicated having a female manager. In terms of percentage of women in the branch or immediate work group, 232 participants (57.3%) indicated that more than 60% of their colleagues were female and 145 (35.8%) stated that they worked with between 20% and 60% female co-workers.

| Procedure and measures
Participants were invited to take part in an online study about workplace experiences. The measures were presented in the order described here.

Perceived experiences with sexism at work
Participants completed an 11-item sexism scale developed in prior work, focusing on personal experiences with sexism (Van Breen, Barreto, Darden, & Dimitriou, manuscript in preperation). Example items are: "During the last 6 months, have you been in a situation where anyone at work did not take what you said seriously because you are a woman?" and "During the last 6 months, have you been in a situation where anyone at work assumed you had inferior ability (e.g., in maths or science) because you are a woman?" Participants indicated how frequently in the past 6 months they had such an experience, ranging from 1 (never) to 5 (most of the time). The scale demonstrated good internal consistency in the present research, α = .92.

Perceived organisational tolerance of sexism
Two items were taken from Bingham and Scherer (1993) focused on how participants perceived sexist incidents were dealt with within the organisation: "Unequal treatment of men and women is clearly discouraged in my workplace (including stereotypical comments or jokes)" and "People at my work ignore the unequal treatment of women when it happens". Two other items were developed for the purpose of this study: "There are formal procedures to address the unequal treatment of women at my workplace", and "When women are treated unequally in my workplace this is corrected". These items were repeated three times, referring to perceived peer-, manager-, and policy-level tolerance separately. For example, to measure peer tolerance, the item "Unequal treatment of men and women is clearly discouraged in my workplace" was adjusted to "My co-workers clearly discourage the unequal treatment of men and women in my workplace", while for manager tolerance it was adjusted to "My manager clearly discourages the unequal treatment of men and women in my workplace", and for policy tolerance it was adjusted to "Policies at my work clearly aim to discourage the unequal treatment of men and women in my workplace". Participants rated their level of agreement with each item on a scale ranging from 1 (strongly disagree) to 5 (strongly agree). Relevant items were reverse scored. Responses to these 12 items were subjected to a principal components analysis with direct oblimin rotation, which confirmed that the items clustered into three factors reflecting peer-(α = .97), manager-(α = .90), and policy-level tolerance (α = .86). 1 The pilot study consists of secondary analyses of a data set collected in 2015-2016 by undergraduate students working under the supervision of Dr Safi Darden and in collaboration with Prof Manuela Barreto. The data was collected to compare the effects of non-sexualised sexism and sexual harassment in the workplace. Importantly for our purposes, the data included reported personal exposure to sexism and self-reported organisational tolerance of sexism at work. While this did not allow us to examine social relationships, it allowed us to take initial steps towards understanding the role of organisational tolerance of sexism.

Workplace friendships
The central dependent variable in this study was a measure of women's friendship with other women at work. To measure this, we adjusted the workplace friendship scale developed by Nielsen, Jex, and Adams (2000) by adding "female" before co-workers for each sentence. Six items measured friendship opportunity (e.g., "I have the opportunity to develop close friendships with my female co-workers at my workplace", α = .90) and six items measured friendship prevalence (e.g., "I have formed strong friendships with my female co-workers at work", α = .90). Participants responded to these items on a scale from 1 (strongly disagree) to 5 (strongly agree). These two scales were highly correlated, r(405) = .62, N = 405, p ≤ .001, and were thus combined for further analyses (α = .92).
It is important to emphasise that the measure of sexism experiences focused on personal experiences with sexism in the workplace, whereas the measure of tolerance of sexism focused on perceptions of how peers, managers, or the organisation as a whole responded to sexism, irrespective of who the target is. Also, while the peer tolerance measure focused broadly on "co-workers" and "people", the measure of workplace friendships clearly and specifically referred to "female co-workers".

Demographics
We included two items adapted from Glomb et al. (1997) to assess the gender composition of the immediate work environment.
The items were: "What is your manager's gender?" and "Please estimate the percentage of women in your branch or immediate work group" (ranging from 1 "0%-20%" to 5 "81%-100%"). Finally, participants indicated their age, highest educational attainment, occupation, how long they had been at their current job, the size of the organisation, the size of their immediate work group, how many people they supervised, place of birth, and their employment status.

| Analytic strategy
We utilised hierarchical linear regression, and entered managers' gender (dummy coded as 0 = male, 1 = female), age, and reported percentage of women in the branch or immediate work group (meancentred) as covariates. Age was entered as a covariate due to prior research with non-college samples demonstrating a negative association between age and experiences with sexism (e.g., Lott, Asquith, & Doyon, 2001). We then entered perceived sexism, and peer-, manager-, and policy-level tolerance of sexism (all mean centred). Finally, we added the interactions between sexism and each tolerance of sexism variable (i.e., three separate interaction terms). Means, standard deviations, and correlations for these variables are provided in Table 1.

| Workplace friendship with women
Overall, the main effect of sexism on workplace friendship with women was not statistically significant, b = 0.05, t(397) = 0.85, p = .40, 95% CI [−0.07, 0.16], indicating that exposure to sexism was not associated with social distancing from other women.
The main effect of peer tolerance of sexism was statistically sig- indicating that women who perceived their peers as less tolerant of sexism also reported greater workplace friendship with women. As

| Discussion
Summarising the results of Study 1, we did not observe evidence in support of H1, or social distancing following exposure to sexism at work. However, we did observe some evidence in support of H2.
That is, sexism was associated with stronger social relationships among women, but only in climates where peers were perceived to be relatively less tolerant of sexism; this drawing towards other women was disrupted in climates where peers were perceived to be relatively more tolerant of sexism. In contrast, sexism was associated with stronger social relationships among women in climates where managers were perceived to be relatively more tolerant of sexism.
Policy-level tolerance only had a main effect on women's interpersonal relationships in the workplace but did not appear to modify the effect of sexism.

| S TUDY 2
Having found initial evidence supporting the idea that sexism and organisational climate together might affect women's interpersonal relationships at work, in Study 2 we sought to replicate and expand on this work by investigating our hypotheses in a quasiexperimental design. Specifically, we measured women's perceptions of tolerance of sexism at their workplace (peer-, manager-, and policy-level), and then introduced a manipulation of experiences of sexism by asking women to think back to a recent experience of sexism in their own lives. The outcome variable of central focus was once again friendships with female co-workers.
For exploratory purposes, we also included a new outcome variable: closeness to female co-workers. Specifically, we aimed to explore whether the central effects observed for friendship among women also appear on other indicators of social relationships (i.e., closeness between women).
Our hypotheses remained largely the same as in Study 1. First, as before, we hypothesised that experiences of sexism would have a detrimental effect on women's friendships with other women in the workplace. Even though we did not find evidence for distancing in Study 1, we believed it would be beneficial to re-examine this idea in an experimental design. Second, we hypothesised that peer tolerance of sexism would interact with experiences of sexism to affect women's friendships with their female co-workers.
Specifically, we expected that being reminded of experiences of sexism would increase reported friendship with female co-workers, but only for those who perceived that peers do not tolerate sexism. Support for this hypothesis would replicate the effect found in Study 1. Third, we hypothesised that managers' tolerance of sexism would mirror the effect described for peers above. 3 The alternative breakdown showed that women who perceived their managers to be less tolerant of sexism reported stronger workplace friendship with other women. This effect was stronger for women whose actual experiences of sexism were low b = −0.30, While this was not what we observed in Study 1, we sought to test this hypothesis further with an experimental manipulation of sexism before reconsidering hypotheses regarding the pattern of effects for manager tolerance. Given that policy-level tolerance did not modify the effect of sexism in Study 1, we raised no hypotheses here, and instead included it as an exploratory variable.
This study was pre-registered on the Open Science Framework at https://osf.io/82mgq.

| Design and participants
This quasi-experimental study was a two cell (sexism vs. no sexism) between-participants design, with continuous scales measuring perceptions of peer, manager, and policy tolerance of sexism in the workplace.
All participants were sampled through Prolific Academic. Using the effect size (f 2 = 0.019) detected in Study 1 for the interaction between sexism and peer tolerance, power analysis in G*Power version 3.1 (Erdfelder et al., 1996) indicated that a sample size of 416 was necessary to achieve power of 0.80. Since participants who do not report having at least one experience of sexism in session 1 will not be invited to session 2, to be able to reach this number we recruited 700 participants for the first session. Of these 700, 462 participants reported that they had experienced at least one instance of workplace sexism and gave consent for the researchers to contact them a week later, again through Prolific Academic. Of these 462 participants who were invited to the second session, 392 participated; however, 15 were excluded from the study later as they did not provide the information required to manipulate sexism, or indicated that they had never experienced sexism at work (even though they indicated that they had previously experienced sexism during session 1). Therefore, the final sample comprised 377 participants. All 377 participants in this study were women in employment and residence in the UK.
Participants ranged in age from 18 to 64 years old with a mean of 33.31 years (SD = 10.10 years).
Of the 377 participants, 251 (66.6%) were in full time employment, while 126 (33.4%) were working part time when the data was collected. More than half of the participants (50.4%) indicated having a female manager; 153 participants (40.6%) indicated that more than 60% of their colleagues were female and 165 (43.8%) stated that between 20% and 60% of their co-workers were female.

| Procedure and measures
The study received ethical approval from the Psychology Ethics Committee at the University of Exeter. The study included two sessions a week apart. Participants provided informed consent in each session and were compensated £0.50 for session 1 (approximately 5 min in duration) and £1.00 for session 2 (approximately 10 min in duration), in line with Prolific Academic guidelines.
In session 1, participants provided the same demographic information as in Study 1 and responded to the same 12 questions to assess tolerance of sexism. As before, we conducted a principal components analysis with direct oblimin rotation to examine whether these 12 items clustered into the three hypothesised levels (peer-, manager-, and policy-level tolerance). The analysis revealed that the three scales loaded in separate factors, but one of the four items measuring peer and manager tolerance did not load with the respective items. This item was "My co-workers/my man- Participants also reported whether or not they had ever personally experienced sexism in their current workplace in the same way as in Study 1, and this was used to filter participation in the second session. Only participants who reported that they had experienced at least one instance of workplace sexism were asked to participate in the second session. This was done because the manipulation (in the second session) asked participants to recall a time when they had experienced sexism at work. Participants who were eligible for the second session were asked to give consent to be contacted a week later. Those who were not eligible (or indicated that they did not want to be contacted) were directed to the end of the study and debriefed.
To disguise the goals of the study, we also included a few filler items. Four items measured the perceived discrimination of sexual minorities (Doyle & Molix, 2016) and nine items measured workplace age discrimination (Marchiondo, Gonzales, & Ran, 2016). These scales were not included in any analyses.
In the second session, participants were randomly allocated to one of two experimental conditions (sexism, no sexism) and responded to the dependent measures. In the sexism condition, participants were asked to recall and describe an experience when they felt they had been treated unfairly at work because they were women. To ensure that participants recalled this situation in sufficient depth, several prompts were used: "Describe, for exam- and six items measured friendship prevalence (α = .88). As in Study 1, these two scales were highly correlated, r(391) = .70, N = 377, p < .001, and were thus combined for further analyses (α = .91).
New to Study 2, to measure closeness to female co-workers, participants were presented with a "name generator", in which they were asked to list up to five people to whom they felt close at work.
For each individual, participants provided initials, gender, and indicated how close they felt to them on a 5 points scale from 1 (not at all) to 5 (very much). We summed those who were female.

| Analytic strategy
To test our hypotheses, we conducted moderation analyses in hierarchical linear regressions, entering managers' gender (dummy coded as 0 = male, 1 = female), age, and reported percentage of women in the branch or immediate work group (mean-centred) as covariates.
We then entered sexism (dummy coded as no sexism = 0, sexism = 1) and peer-, manager-, and policy-level tolerance of sexism (mean centred). Finally, we added each of the three interactions between sexism and each of the tolerance variables. Means, standard deviations, and correlations for these variables are provided in Table 2.

| Workplace friendship with women
As in Study 1, but contradicting Hypothesis 1, the main effect of

| Exploratory analyses: Closeness to female co-workers
We next examined whether similar effects to those observed for workplace friendship with women were also evident on the new measure, closeness to female co-workers. Participants' closeness to male co-workers was added as covariate in these analyses because These effects were not apparent for women who perceived greater tolerance of sexism at the policy level, b = −0.17, t(359) = −1.11, p = .27, 95% CI [−0.49, 0.14]. However, at this stage these effects should be considered preliminary as they were not directly hypothesised.

| Discussion
As in Study 1, in Study 2 we failed to find evidence for social dis-

| S TUDY 3
This study aimed to replicate Study 2, with a new sample. It was virtually identical to Study 2 with one exception: In this study, we focused on closeness to female co-workers, and accordingly chose to measure it before measuring workplace friendship (that is, directly after the manipulation of sexism) in order to account for potential order effects. Therefore, we first asked participants to list their friends and then rate closeness with them before moving on to rate their general friendship with their female colleagues.
The hypotheses of this study were identical to those of Study 2, although we now focus on closeness to female co-workers as well as friendship with female co-workers as the key outcomes.

Workplace friendships with female co -workers Personal experience with workplace sexism
More peer tolernace of sexism Less peer tolerance of sexism with female co-workers, but only for those who perceived that peers were less tolerant of sexism in the workplace. However, we expected that this effect would disappear for those who perceived that peers were more tolerant of sexism. Given that manager-and policy-level tolerance produced inconsistent results in Studies 1 and 2, we included them here without making firm hypotheses regarding their effects.

| Method
Means, standard deviations, and correlations for these variables are provided in Table 3.

| Design, participants, procedure, and measures
The design and procedure of this study were identical to those of Study 2 with the exception of the order of outcome measures.
Using the effect size (f 2 = 0.019) detected in Study 1, power analysis in G*Power version 3.1 (Erdfelder et al., 1996) indicated that a sample size of 416 was necessary to achieve power of 0.80. To be able to reach this number at the end of the second session, we again recruited 700 participants for the first session of the study.
Of these 700, 465 participants reported that they had experienced at least one instance of workplace sexism and gave consent for the researchers to contact them a week later, again through Prolific Academic. Of these 465 participants who were invited to the second session, 405 participated; however, 14 were excluded from the study as they did not complete the questions involved in the manipulation of sexism or indicated that they had never experienced sexism at work (even though they had indicated previously experiencing sexism at work during session 1). The final sample comprised 391 participants. stated that between 20% and 60% of their co-workers were female.
Participants completed the same measures (in the same sessions) as in Study 2 (peer tolerance α = .74, manager tolerance α = .84, policy tolerance α = .87). As in Study 2, in the second session of Study 3, participants were randomly allocated to the sexism or no sexism conditions. Then, participants completed the same measures of closeness to female co-workers and workplace friendships (α = .91; correlation between friendship opportunity and prevalence: r = .73, N = 391, p < .001), but in the reverse order.

| Closeness to female co-workers
Using the same analytic strategy as in Study 2, we first examined whether the effects observed for workplace friendships in Study 1 and 2 were also evident on our measure of closeness to female co-workers in Study 3. We added women's closeness to male coworkers as a covariate in the model as we did in Study 2. Results showed that women who perceived their peers to be less tolerant of sexism reported greater closeness to female co-workers. Note: M and SD are used to represent means and standard deviation, respectively. The answer categories for percentage of women are: 1 = 0%-20%, 2 = 21%-40%, 3 = 41%-60%, 4 = 61%-80%, 5 = 81%-100%. **p < .01.

TA B L E 3 Means, standard deviations, and correlations for Study 3
However, these significant main effects were not qualified by a statistically significant interaction between sexism and peer toler-

| Workplace friendships with women
As in Study 2, the main effect of sexism on workplace friendships

| Discussion
Here, we found suggestive evidence for Hypothesis 1 on the measure of closeness; experiences of sexism led women to report reduced closeness with their female colleagues in this study.
Moreover, we again found evidence in support of our second hypothesis, that organisational tolerance of sexism moderates the effect of sexism on women's relationships with other women.
Specifically, we found that women reported greater closeness to their female co-workers after recalling sexism, but only for those who perceived their peers to be relatively low in tolerance of sexism.

Although this pattern was revealed in different measures in
Studies 1 and 2 relative to Study 3, these measures are conceptually similar, and the patterns are the same across studies. The order in which these variables were presented might explain why the effect appeared on different measures in Studies 2 and 3. In Study 2, participants first completed the measure of workplace friendship, then closeness to female co-workers; but this order was reversed in Study 3.
With regard to the other two levels of organisational tolerance, the results of Study 1 were not replicated in Studies 2 and 3.
Specifically, in Study 1 experiences of sexism were associated with greater friendship with female colleagues, but only among those who reported greater tolerance of sexism from their managers. However, this effect was not replicated in Studies 2 and 3. With regard to perceived policy tolerance, in Study 1, policy-level tolerance did not 6 The alternative breakdown showed that among women who were reminded of sexism

| G ENER AL DISCUSSION
Across three studies, we demonstrated that women's social relationships with their female colleagues might be affected by personal experiences with sexism and the organisational climate in which that sexism occurs. When experiences of sexism occur within an organisational climate that is perceived to be intolerant of sexism, women might strengthen social bonds with their female co-workers as a way of coping with sexism. That is, after experiences with sexism, the perception that peers (and potentially organisational policy) do not tolerate sexism can draw women to each other. However, the perception that one's peers tolerate sexism hinders this effect, thereby making it harder for women to engage in such a positive coping response. This shows that peer tolerance of sexism is key in shaping the social consequences of experiences of sexism. In these studies, affiliative tendencies took the form of stronger workplace friendships with other women as well as greater reported closeness to female co-workers.
Although evidence for distancing was not consistently found, there was some evidence that experiences with sexism led women to report less closeness with other women in Study 3. This finding is consistent with some past work Doyle & Molix, 2014b;Parks-Stamm, Heilman, & Hearns, 2007). Previous work has shown distancing effects, but this was in the context of male-dominated work environments Ely, 1994). The current work was not specifically focused on male-dominated contexts, but still we find some evidence that women may distance themselves from others following experiences of sexism.
While the majority of prior work in this area demonstrates patterns of self-group distancing, our results mainly highlight when women draw together (or fail to do so). Here it is important to stress that this "other side of the coin" is not unrelated to self-group distancing and can in fact help shed light on when this emerges, as well as when it does not. This has in fact already been acknowledged in prior work, as when Derks and her colleagues  examined support for collective action and showed that senior women who were highly identified with their gender responded to sexism by reporting more support for collective action in favour of women. Again, we contribute to this analysis by focusing on when women draw to each other and when they do not, which does not provide direct evidence of self-group distancing, but does contribute to broader understandings of when women are able to support each other in response to sexism.
Taken together, then, this work sheds light on the question of when women draw together with other women (Branscombe et al., 1999;Haslam & Reicher, 2006;Jetten et al., 2001).  (Zapf, Knorz, & Kulla, 1996) and sexism in particular (Cihangir, Barreto, & Ellemers, 2014;Major, Kaiser, & McCoy, 2003). However, past research also shows that other in-group members are not necessarily supportive, especially when it comes to supporting discrimination claims (Kahn, Barreto, Kaiser, & Rego, 2015;Kaiser, Dyrenforth, & Hagiwara, 2006). Taking this literature together with the findings from the current work, we suggest that experiences of sexism lead women to pursue closeness with women who are intolerant of sexism as a way of accessing social support from others whom they expect to be supportive-a well-considered coping response. That is, this study extends previous literature on the benefits of social support by demonstrating that women actively pursue closeness/draw together with those who might provide social support, but may not be able to do so when this type of support is not expected or available (in the case of greater tolerance of sexism).
With regard to tolerance, it was peer tolerance of sexism, in particular, that most consistently emerged as a moderator of responses to sexism (although policy tolerance also demonstrated suggestive evidence in the same direction). However, we did not find a similar interaction between sexism and manager tolerance of sexism. In Study 1, the main effect of manager tolerance of sexism suggested an overall positive association with friendship with female co-workers. This was qualified by an interaction with sexism, in which experiences of sexism were related to stronger friendship with female co-workers, but only among those who perceived their managers to be more tolerant of sexism. Although this finding was not as hypothesised, we consider it interesting to discuss and potentially worthy of further investigation.
Previous research has examined how experiences of sexism affect managers' attitudes towards their subordinates Derks et al., 2016), and subordinates' attitudes towards their managers (Sterk, Meeussen, & Van Laar, 2018 however, some interesting patterns emerged with regard to closeness between women, suggesting that in cases of less policy-level tolerance of sexism, sexism was significantly associated with greater closeness to female colleagues (Study 2). There was also a marginally significant interaction showing that sexism is associated with greater workplace friendship with female co-workers in cases of less policy level tolerance of sexism (Study 3). These effects mirror the central findings of this study with regard to peer tolerance, although patterns were not as strong. One reason why they were less strong might be that policy-level tolerance of sexism is relatively abstract (or distal) for women. In other words, policy-level tolerance might be quite removed from women's daily experiences because interactions and experiences with peers are more tangible and concrete for women.
In a similar vein, previous research suggests that having an organisational policy that clarifies intolerance of discrimination does not mean that this necessarily translates into less tolerance of discrimination at lower levels in the organisation (Brady et al., 2015;Hebl et al., 2002), suggesting some degree of dissociation between these levels.

| Limitations and future directions
One limitation of this work is that our participants were primarily employed in female-dominated work environments (although this was not deliberately due to our sampling strategies). It would be interesting to see if the results regarding experiences of sexism and tolerance of sexism differ for women in male-dominated work environments. One reason why it is important to work with women in a male-dominated setting is that men often show that they are unaware of, and disconnected from, sexist treatment of women (Tougas & Beaton, 2002).
Therefore, in a male-dominated environment, women might feel greater peer tolerance of sexism, as well as have fewer female co-workers with whom to affiliate. That is to say, the findings of the current studies can only emerge when other women are around. It might be interesting to examine, in addition, what is the minimum proportion of women in a work environment necessary to encourage women to draw to each other in response to sexist treatment.
In addition, as the studies in this article did not manipulate, but measured, perceptions of tolerance and interpersonal relationships, it is not possible to know precisely whom participants were thinking about when responding to these measures and whether or not these were the same people. That said, the measure of tolerance we used specifically refers to "my co-workers" and "people" (which does not exclude men), while the measures of social relationships explicitly mention "my female co-workers" or investigate social closeness to "female co-workers". However, we acknowledge that this does not completely disambiguate the results and therefore future research might try to exert more control over these interpretations. Additionally, we tried to separate three levels of tolerance of sexism (peer-, manager-, and policy-level tolerance), but these three levels are not necessarily entirely distinct from one another. Specifically, manager tolerance and policy-level tolerance of sexism may, in real-life contexts, not be independent. One reason for this is that managers' actions may play a role in shaping organisational policies. For example, Edelman (2005) points to the fact that many organisational managers create their own organisational policies against discrimination. In addition, Martinez and colleagues (Martinez et al., 2013) argue that organisational-level policies are not always clear for employees; for this reason, sometimes managers need to filter these policies through their own beliefs and actions.
The role of manager tolerance of sexism should also be investigated further in future work. The suggestive evidence observed here needs to be replicated, but research examining motivational underpinnings of both seeking support from other women and potentially filing formal complaints following sexism could help to clarify discrepant findings in the current studies. Overall, these results underscore the importance of continuing to try to distinguish the various levels of organisational climate that might convey tolerance of sexism.

| CONCLUSIONS
The goal of the present research was to understand when women draw together (vs. distance from each other) as a response to sexism.
Specifically, we tested whether three levels of organisational tolerance of sexism (peer, manager, and policy) have an impact on women's tendency to draw together at work in response to sexism. The combination of cross-sectional and quasi-experimental studies showed that the effect of peer tolerance of sexism was the most consistent among the three levels of organisational climate, demonstrating that when women perceived that their peers were less tolerant of sexism, experiences of sexism led to increased affiliation with other women, but this did not occur when women perceived their peers to be more tolerant of sexism. This insight may have important practical implications for organisations' efforts to improve organisational climate.

CO N FLI C T O F I NTE R E S T
The authors report no conflicts of interest.

E THI C AL APPROVAL
We confirm that studies reported in this manuscript conform to APA ethical guidelines and have been approved by the ethical review committee at the University of Exeter.