The impact of marriage equality campaigns on stress: Did a Swiss public vote get under the skin?

Significance

The evolving legal landscape for lesbian, gay, bisexual, trans, intersex, and queer individuals remains a critical movement globally. While previous studies have explored the psychological impacts of societal changes, this research uniquely combines survey and biological data surrounding Switzerland’s marriage equality vote. Our findings reveal an increase in biological stress during this period, shedding light on the enduring health impacts of political campaigns. This highlights the crucial need to recognize and address the lasting consequences of such campaigns on individuals’ well-being.

Abstract

Lesbian, gay, bisexual, trans, intersex, and queer (LGBTIQ+) individuals encounter persistent structural inequalities and discrimination that can lead to detrimental psychological and physiological health outcomes. Amid evolving legal landscapes, little attention has been directed toward understanding the physiological health effects of societal shifts on these communities. This study aims to explore the impact of a national marriage equality vote and associated debates on psychological and biological stress among LGBTIQ+ individuals and cisgender, heterosexual, endosex individuals (termed cis-heterosexual) in Switzerland. We gathered longitudinal survey and biological data collected in hair samples among LGBTIQ+ and cis-heterosexual individuals before, during, and after the 2021 national vote (survey data: N_T1T2 = 954; N_T2T3 = 880; biological data: N_T1T2 = 393; N_T2T3 = 354). Preregistered analyses reveal a notable increase in biological stress levels (i.e., cortisol and cortisone levels), but not perceived stress, among both LGBTIQ+ as well as cis-heterosexual individuals who were close to them during the campaign. Results further point out the negative impacts of the campaign against marriage equality (i.e., no-campaign) on LGBTIQ+ individuals’ biological stress levels as well as on those of their allies. These effects were, however, moderated by exposure to the campaign for marriage equality (i.e., yes-campaign), indicating the powerful buffering effects of the yes-campaign on the impact of discrimination on individuals’ health. However, these positive effects appear to come at a cost, potentially impacting the well-being of individuals engaged in advocating for the yes-campaign. This research underscores the lasting impact of political campaigns on individuals’ health.

Throughout history, certain social groups have gained more power and status than others (1). More powerful groups have often influenced the lives of less powerful social groups by defining what is socially accepted, for instance, through popular votes. The impact of such group-based and societal processes on the health of minority group members, such as migrants (2), racial and ethnic minorities (e.g., BIPOC) (3, 4), and/or LGBTIQ+ people (5, 6), has been well documented. To date, however, little is known about the mechanisms through which these hegemonic, majority–imposed norms affect the cumulative physiological health of minority members. Investigating these mechanisms is crucial to uncover how these societal processes deeply impact minority health. A social-psychological model (7) proposes that these dynamics influence health by altering stress-responsive bodily systems. While this model has not yet been tested empirically with stress biomarkers, we suggest that a specific event like a marriage equality vote represents one such societal process that directly impacts health via stress-responsive biological systems as the key mechanism.

Campaigns against marriage equality (i.e., no-campaigns) can negatively impact the mental health of LGBTIQ+ individuals (5, 6, 8–12). For example, a longitudinal study (8) conducted during the Australian marriage equality plebiscite revealed that exposure to debate was associated with greater psychological distress among LGBTIQ+ individuals. Although rarely assessed, initial evidence suggests these effects might also extend to allies such as friends and families (13–15). Potential explanations for these detrimental societal effects can be found in the minority stress literature (16, 17). According to the minority stress model, stigmatized minorities are exposed to unique stressors during political campaigns against their rights (18), which bear the risk of increasing their sense of discrimination, rejection, and alienation from mainstream society. While the impact of campaigns on psychological stress has been assessed in the literature, little research has attempted to elucidate the physiological mechanisms translating the exposure to debates around LGBTIQ+ issues among LGBTIQ+ individuals and cis-heterosexual individuals.

Biologically, threats to the social self are among the most potent triggers of the hypothalamic-pituitary-adrenal hypothalamic–pituitary–adrenal (HPA) axis, the main system that activates the release of the “stress hormone” cortisol (19). Repeated and/or prolonged HPA-axis activation can lead to cumulative “wear and tear” over time (20) and foster stress-related conditions such as depression, pain, fatigue, and heart disease (21). Existing research on LGBTIQ+ individuals indicates that they show recalibrations in HPA-axis functioning (22, 23), likely linked to minority stress processes (22–24). For instance, structural stigma, experiences of discrimination in everyday life, and “coming out” processes represent unique minority stress processes. These stressors might even heighten during political campaigns targeting the LGBTIQ+ community. However, no research has yet examined stress biology during a national debate concerning the rights of this marginalized group, a gap this study aims to address.

The studies that have investigated the negative impact of anti-LGBTIQ+ campaigns on LGBTIQ+ individuals’ health (18) have often also ignored potential protective factors and the conditions under which they exert an influence. For example, emotional and instrumental forms of support (25–28) have been shown to buffer against detrimental health outcomes among minority group members. Thus, during these campaigns, exposure to social support, such as arguments favoring marriage equality (i.e., yes-campaign), might protect against the detrimental effects of minority stress on LGBTIQ+ individuals’ and their allies’ health. In parallel, the impact of exposure to the marriage equality campaign (yes-campaign) might differ based on individual engagement levels. Active involvement might heighten stress due to increased negativity or potential violence against LGBTIQ+ individuals. Consequently, individuals engaged in the yes-campaign might experience elevated stress levels (29). However, some studies (30) indicate mixed effects, showing both negative and positive outcomes.

To test our assumptions, we use the case of the Swiss marriage equality vote (including joint adoption, artificial insemination, and facilitated naturalization), which was held on September 26, 2021, and sparked widespread public debates. The results showed that 64.1% of the voters were in favor of marriage equality and were 35.9% against marriage equality. Our aim was to analyze the impact of this vote on the psychological and biological stress levels of both directly affected individuals (LGBTIQ+ individuals) and those indirectly impacted (cis-heterosexual people, including individuals who were close to LGBTIQ+ people).

To do so, we conducted a large-scale longitudinal study, which aggregated survey data among 954 individuals (Full Study), with a subcohort of 393 individuals in whom biological data were collected (Sub-Study). The data were collected three months before (Time 1: May/June 2021), immediately before and during (Time 2: Mid-September–End of September 2021), and 2 to 3 mo after the national vote (Time 3: December 2021 to 2nd of January 2022).

Drawing on the literature reviewed above, we preregistered^* the following hypotheses regarding the impact of the campaign on individuals (Time 1 to Time 2):

H1a: LGBTIQ+ individuals will exhibit a stronger increase in psychological (i.e., perceived stress) and biological stress (i.e., cortisol levels) from Time 1 to Time 2 compared to cis-heterosexual individuals.

H2a: In LGBTIQ+ individuals, the increase in stress from Time 1 to Time 2 will be predicted by the extent of exposure to the no-campaign. We had no specific predictions regarding cis-heterosexual individuals. On the one hand, cis-heterosexual individuals are not directly affected by the vote or the campaign. On the other hand, many of them consider themselves allies and might therefore show “vicarious” effects like the ones expected for LGBTIQ+ individuals.

H3a: The extent of exposure to the yes-campaign will moderate the effect of exposure to the no-campaign on the increase in stress from Time 1 to Time 2 among LGBTIQ+ individuals. Again, we had no clear expectations regarding cis-heterosexual individuals.

Regarding the impact of the campaign cessation on individuals between Time 2 and Time 3, we preregistered the following hypotheses:

H1b: LGBTIQ+ individuals will exhibit a larger decrease in psychological and biological stress from Time 2 to Time 3 compared to cis-heterosexual individuals.

H2b: Exposure to the no-campaign will lead to a smaller decrease in psychological and biological stress from Time 2 to Time 3.

Collectively, we did not have clear expectations as to whether and how level of engagement would impact perceived and biological stress. Finally, because the data collection coincided with the Covid-19 pandemic, we also controlled for the effect of the Covid-19 pandemic on changes in stress.

Results

The analyses reported below are based on a longitudinal survey and biological data collected from hair samples among LGBTIQ+ and cis-heterosexual individuals, including cis-heterosexual people close to LGBTIQ+ individuals, before, during, and after the marriage equality vote (survey data: N_T1T2 = 954; N_T2T3 = 880; biological data: N_T1T2 = 393; N_T2T3 = 354). Descriptive statistics can be found in SI Appendix, Table S9. Importantly, most cis-heterosexual participants were highly identified as supporters of LGBTIQ+ rights and were close to at least one LGBTIQ+ individual.

Time 1 to Time 2 Effects.

We first focused on differences between Time 1 and Time 2 to investigate the effect of the campaign on LGBTIQ+ and cis-heterosexual individuals’ psychological and biological stress levels.

Changes in Psychological Stress.

Psychological stress (i.e., perceived stress) was slightly higher shortly before the vote was held (T2: M = 2.77, SD = 0.76) compared to the baseline (T1: M = 2.72, SD = 0.76; b = –0.05, P = 0.010; Fig. 1). The effect was, however, small. Moreover, and in line with our expectations (H1a), there was a difference in changes in perceived stress between LGBTIQ+ and cis-heterosexual individuals (b = –0.16, P < 0.001). Only LGBTIQ+ individuals reported an increase in perceived stress at T2.

Fig. 1.

Psychological stress values (i.e., perceived stress scale) across time points.

Exposure to the Campaigns and Psychological Stress Accounting for Engagement.

We then calculated the full multiple regression model with psychological stress (i.e., perceived stress) as the dependent variable (Table 1). Group (i.e., LGBTIQ+ versus cis-heterosexual individuals; b = –0.17, P = 0.003) and perceived impact of Covid-19 at Time 2 explained changes in perceived stress (b = –0.06, P = 0.006). LGBTIQ+ individuals and individuals who felt that Covid-19 had a negative impact on them (regardless of their group membership) tended to report a higher level of perceived stress at T2 compared to T1.

Table 1.

Regression on the differences scores: Changes in perceived stress and cortisol

	T1 – T2		T2 – T3
	Perceived stress	Cortisol (Wins)	Perceived stress	Cortisol (Wins)
Intercept	0.07	–1.94***	–0.14**	1.79***
LGBTIQ+	–0.17**	–0.21	0.10†	–0.32
No-campaign	–0.09	0.58†	0.02	–0.40†
Yes-campaign	–0.01	–0.47†	0.01	0.32
Engagement	–0.04	1.61*	–0.03	–0.25
Covid	–0.06**	0.09	–0.00	–0.02
No-campaign*Yes-campaign	–0.04	0.97**	0.01	–1.08***
No-campaign*LGBTIQ+	0.09	–1.16*	0.01	0.50
Yes-campaign*LGBTIQ+	0.06	0.62†	–0.03	–0.22
Yes-campaign*Engagement	–0.11	–1.69**	0.06*	0.36*
No-campaign*Yes-camp.*LGBTIQ+	0.08	–0.58	–0.03	0.90**
Degrees of Freedom	903	363	834	328
F-statistic	3.09, P < 0.001	2.19, P = 0.018	1.08, P = 0.376	2.30, P = 0.013
R2	0.03	0.06	0.01	0.07

Note. ^***P < 0.001, ^**P < 0.01, ^*P < 0.05, ^†P < 0.10.

We did not, however, find an interaction effect between exposure to the yes- or no-campaigns and group (contrary to H2a). We also did not find an interaction effect between exposure to the yes- and no-campaigns on perceived stress (H3a).

Changes in Biological Stress.

Next, we investigated changes in biological stress (e.g., HPA-axis cortisol levels). In line with our expectations and partially in line with H1a, cortisol values were higher shortly before the vote was held (T2: M = 6.01, SD = 3.07) compared to the baseline (T1: M = 3.87, SD = 1.84; b = –2.13, P < 0.001, see Fig. 2). There were, however, no differences in changes in cortisol levels between LGBTIQ+ and cis-heterosexual individuals (b = 0.01, P = 0.955).

Fig. 2.

Biological (i.e., cortisol levels) stress values (i.e., cortisol levels) across time points. Please note that the cortisol values were winsorized (Materials and Methods).

Exposure to the Campaigns and Biological Stress Accounting for Engagement.

Results (Table 1) using biological stress as a dependent variable showed an interaction effect between exposure to the no-campaign and group (b = –1.16, P = 0.012; Fig. 3). Follow-up analyses indicated that the exposure to the no-campaign increased biological stress among LGBTIQ+ individuals but not among cis-heterosexual individuals (partially confirming H2a).

Fig. 3.

Time 1 to Time 2 interaction between no-campaign and group (cis-heterosexual vs. LGBTIQ+ individuals).

In line with our expectations (H3a), we found an interaction between the exposure to the yes- and no-campaigns (b = 0.97, P = 0.006; see Fig. 4). This suggests that exposure to the yes-campaign buffered against the negative effects of exposure to the no-campaign.

Fig. 4.

Time 1 to Time 2 interaction between no-campaign and yes-campaign.

Finally, we found an interaction between exposure to the yes-campaign and engagement (b = –1.69, P = 0.005; see Fig. 5). Exposure to the yes-campaign only had a positive effect on individuals’ stress levels if they did not (or minimally) engage in the campaign. This seems to indicate that there are—at least—two groups of individuals for whom the yes-campaign has different effects: Individuals who were highly engaged in the campaign (i.e., increased stress levels and no positive effect of exposure to the yes-campaign) and the others (i.e., buffering effect of exposure to the yes-campaign). In line with this interpretation, we found an interaction effect between exposure to the yes- and no-campaign (b = 0.83, P = 0.022)—without including engagement as an additional predictor—and after excluding individuals who were highly engaged in the campaign (n = 36).

Fig. 5.

Time 1 to Time 2 interaction between yes-campaign and engagement.

Time 2 to Time 3 Effects.

Next, we examined the differences between Time 2 and Time 3 to assess the impact of the campaign on the psychological and biological stress levels of both LGBTIQ+ and cis-heterosexual individuals. Time 2 data were collected during the campaign and just prior to the vote, while Time 3 data were gathered 2 to 3 mo after the vote and the conclusion of the campaign.

Changes in Psychological Stress.

Contrary to our expectations, perceived stress was slightly lower shortly before the vote (T2: M = 2.79, SD = 0.76) as compared to after the vote (T3: M = 2.84, SD = 0.76), b = –0.05, P = 0.012, see Fig. 1). There were no differences in changes in perceived stress between LGBTIQ+ and cis-heterosexual individuals (b = 0.06, P = 0.165; contradicting H1b).

Exposure to the Campaigns and Psychological Stress Accounting for Engagement.

We found an interaction between exposure to the yes-campaign and engagement (b = 0.06, P = 0.022; see Fig. 6). The effect was, however, small. Contrary to hypothesis H2b, we found no effect of exposure to the no-campaign on changes in perceived stress (b = 0.02, P = 0.686).

Fig. 6.

Time 2 to Time 3 interaction between yes-campaign and engagement.

Changes in Biological Stress.

In line with our expectations and partially in line with H1b, cortisol values were higher shortly before the vote (T2: M = 5.84, SD = 2.94) as compared to after the vote (T3: M = 4.24, SD = 1.65; b = 1.60, P < 0.001, see Fig. 2). There were, however, no differences in cortisol change between LGBTIQ+ and cis-heterosexual individuals (b = –0.17, P = 0.509; partially contradicting H1b).

Exposure to the Campaigns and Biological Stress Accounting for Engagement.

The model including cortisol as a dependent variable indicated a three-way interaction between the exposure to the yes- and no-campaigns and group (b = 0.90, P = 0.005). Follow-up analyses (Fig. 7) indicated that the effects found between Time 1 and Time 2 (e.g., the negative effect of the exposure to the no-campaign for individuals who were little/not exposed to the yes-campaign) were reversed between Time 2 and Time 3 (e.g., decrease in cortisol levels for individuals who were highly exposed to the no-campaign and were little/not exposed to the yes-campaign, partially in line with H2b). This was, however, only the case for cis-heterosexual individuals, while no differences were found among LGBTIQ+ individuals. There was also an interaction between exposure to the yes-campaign and engagement (b = 0.36, P = 0.042).

Fig. 7.

Time 2 to Time 3 interaction between no-campaign and yes-campaign exposure by group (i.e., cis-heterosexual = 0 vs. LGBTIQ+ individuals = 1).

Additional Sensitivity Analyses.

We conducted additional analyses to assess the robustness of our effects (SI Appendix). First, all analyses were also performed on cortisone levels, another measure of biological stress, yielding similar conclusions to those based on cortisol levels. Additionally, while we utilized the winsorizing technique to address extreme cases (i.e., outliers), we also conducted all analyses without outliers (i.e., by excluding outliers). The main findings remained consistent regardless of the outlier-handling method used. Furthermore, as a robustness check, we excluded individuals highly engaged in the campaign, and we observed an effect of exposure to both the no- and yes-campaigns without controlling for engagement in the yes-campaign. This finding aligns with our assumption that individuals highly engaged in the campaign were differently impacted by the vote than others. Finally, the exclusion (i.e., impact of Covid-19) and inclusion of additional covariates (i.e., age, smoking, drug intake, alcohol consumption, hormone intake, and hair treatment) did not alter our main findings.

Discussion

Despite legal changes toward greater equality, LGBTIQ+ individuals still face structural inequalities and discrimination that can be detrimental to their health and well-being. Societal changes, often reflected in campaigns and votes, reveal shifting attitudes. Unfortunately, these processes can expose individuals and their allies to discrimination, impacting their psychological and physiological health. Our preregistered study, centered around the Swiss marriage equality vote, observed heightened biological stress levels during the campaign among both LGBTIQ+ and cis-heterosexual individuals (see SI Appendix, Table S8 for a summary of the main findings and hypotheses). Importantly, most cis-heterosexual participants were close to at least one LGBTIQ+ person and were allies of LGBTIQ+ individuals, which might explain changes among cis-heterosexual individuals. Surprisingly, although perceived stress levels showed minimal change, HPA-axis stress hormones did not return to baseline postvote. Exposure to the yes-campaign mitigated the negative effects of the no-campaign, yet highly engaged individuals experienced heightened biological stress without this buffering effect. Following the vote, effects diminished among cis-heterosexual individuals, while lingering impacts persisted among the LGBTIQ+ community, influencing their biological stress levels in a cumulative manner.

Our study contributes significantly to existing research in several ways. First, it reveals the impact of political campaigns on HPA-axis functioning during a significant national referendum on marriage equality, highlighting how minority stress and support influence biological stress responses. This study provides evidence of how minority stress and support can impact biological stress values during a national vote or referendum that can change structural stigma; namely marriage equality, adoption laws, and reproductive support for LGBTIQ+ individuals. Second, it sheds light on the enduring health effects of short-term campaigns on individuals, uncovering potential negative health consequences of campaign engagement that warrant further exploration in psychology and public health. Finally, our use of hair cortisol, a unique approach in social policy research, provides unique insights distinct from psychological stress measures. Notably, this method offers additional information, and aligns with past findings showing the lack of correlation between subjective and objective stress measures (31–35).

While advancing the literature, our study has some limitations. First, our choice of hair cortisol as a retrospective marker for HPA-axis activity was constrained due to logistical challenges, preventing the collection of hair samples from the entire survey sample group. Despite these constraints, we considered hair samples the most appropriate approach for our research design as it fitted neatly with our time frame (i.e., allows to assess stress within the past month). Further, we could include individuals who could not take part in the Biomarker Sub–Study (i.e., including hair measures) in our Full Survey Study. Another limitation is the restricted findings concerning perceived stress, which remained relatively stable over time. This, coupled with the cortisol results, suggests that individuals may not be fully aware of the stress they experienced during the campaign. To date, there is only scarce evidence for a considerable association of subjective and objective stress measures, which might be attributable to method bias (e.g., confounding) and/or asynchrony of their temporal changes (34, 35). Finally, the study’s design with only three-time points made it challenging to differentiate between the campaign’s effects and the vote outcome’s impact. To address this, future “experiments of nature” could consider an additional fourth time point to untangle these effects.

Despite these limitations, our findings have notable strengths. Drawing from extensive community-based samples of both LGBTIQ+ and cis-heterosexual individuals, our findings underscore the detrimental impact of campaigns against minorities on their health and their allies. In particular, our research highlights the influential buffering effects of pro-equality campaigns mitigating the discrimination's impact on health. However, this effect came with a cost: namely increased biological stress for highly engaged individuals. In conclusion, a relevant and practical implication of this research is that individuals should consider the cumulative impacts of discrimination experienced through political campaigns on the health of individuals and social groups (i.e., those who engage in favor and against political changes). This revelation provides a clearer path for policymakers and practitioners to pursue social harmony and justice, particularly by prioritizing support for those actively involved in civil rights campaigns.

Materials and Methods

Design, Sample, and Data Collection.

We used Swiss data that were collected before (Time 1: May/June 2021), during (Time 2: Mid-September–end of September 2021), and after (Time 3: December 2021–2nd of January 2022) the marriage equality vote (Fig. 8). The research has been approved by the Ethics Committee of the Faculty of Arts and Social Sciences at the University of Zurich. Informed consent was obtained from all participants included in the study. We collected survey and biological data (i.e., cortisol in hair as a biological stress marker) among LGBTIQ+ and cis-heterosexual individuals from all Swiss cantons. For cost reasons, we collected the hair samples only among a subsample. Participants who were pregnant or breastfeeding and participants with less than 2 cm of scalp hair were not eligible to participate in the Sub-Study.

Fig. 8.

Study design and timeline

Before the start of the study, we recruited individuals and assessed their interest in participating in the Sub-Study, which involved providing hair samples during each wave of data collection. Those who expressed interest were then asked to provide their contact information for shipping the hair material. In total, 117 were excluded because they declined to provide contact information, and an additional 190 were excluded for not meeting eligibility criteria (e.g., pregnant or breastfeeding and participants with less than 2 cm of scalp hair). Despite being ineligible for the Sub-Study, these individuals were invited to complete the surveys. The hair samples were collected remotely and by the participants themselves. Studies have repeatedly shown that remote self-sampling yields results that are comparable with those obtained in laboratories (36). Participants who signed up for the Sub-Study were mailed the link to the online survey, a hair sampling kit (incl. written/video instructions, clamps, thread, envelopes), and a gift at each time point.

The LGBTIQ+ and cis-heterosexual participants were recruited via online platforms and through the Swiss LGBTIQ+ Panel (see www.swiss-lgbtiq-panel.ch) a yearly national longitudinal survey, first launched in 2019, following LGBTIQ+ and cis-heterosexual allies over time. Former participants of the Swiss LGBTIQ+ Panel were invited to participate in the study via email. Consequently, participants in the survey share a similar sample distribution as participants in the Swiss LGBTIQ+ Panel (37), with a tendency toward being younger and more educated. This demographic skew is a common bias in survey research. We further recruited cis-heterosexual participants via our LGBTIQ+ participants who signed up to the survey and were invited to bring their friends and families along. Additional participants from the general population were also recruited using flyers distributed in letter boxes. As a compensation for their participation, individuals took part in a lottery draw, received their personal cortisol/cortisone values (if provided), and were given follow-up information about the findings.

All participants were asked to complete the questionnaires (i.e., Full Survey Study). Only a portion of the sample was asked to provide hair sample(s) (i.e., Biomarker Sub-Study). For the first part of the analyses (i.e., effect of the upcoming vote), we included participants who participated at both Time 1 and Time 2. For the second part of the analyses (i.e., effect of the outcome of the vote), we included participants who participated at both Time 2 and Time 3. Individuals who had more than 20% missing data on relevant items in at least one completed time point were excluded.

All in all, 954 individuals (i.e., 671 LGBTIQ+ and 283 cis-heterosexual individuals) completed the online surveys both at Times 1 and 2. Among those, 393 individuals (i.e., subsample, 218 LGBTIQ+ and 175 cis-heterosexual individuals) provided hair cortisol. Further, 880 individuals (i.e., 616 LGBTIQ+ and 264 cis-heterosexual individuals) completed the online surveys at both Times 2 and 3, representing a 9.9% decrease in participation from Time 2 to Time 3 compared to Time 1 to Time 2. Among those, 354 (202 LGBTIQ+ and 152 cis-heterosexual individuals) provided their hair sample, reflecting a 7.7% decrease in participation from Time 2 to Time 3 compared to Time 1 to Time 2. Further descriptive statistics can be found in SI Appendix, Table S9. Importantly, both samples were partially matched in terms of age, gender, level of education, and geographical region, and participant characteristics remained consistent between Time 1 to Time 2 and Time 2 to Time 3.

Measures.

The online survey was available in German, French, Italian, and English. The following validated measures were used for the analyses. The full questionnaires, which also included additional items unrelated to the current study, can be found online.^†

Dependent variables.

Psychological stress (α_T1 = 0.86, α_T2 = 0.88, α_T3 = 0.89).

We assessed psychological stress in the past month at all three time points using the Perceived Stress Scale (38). For the full sample (i.e., survey and biological data), we used the full 10-items scale. For participants who only completed the survey, we used a shortened scale (i.e., four items) to reduce the length of the questionnaire. The items have been assessed on a 5-point Likert scale (1 Never to 5 Very Often). An example item is “In the past month, you felt that difficulties were piling up so high that you could not overcome them?”

Biological stress.

We measured biological stress at all three time points among a subsample of participants using hair cortisol, which is a highly reliable method to measure long-term HPA activity (31) at each of the three time points. Endogenous cortisol is incorporated into hair follicles via passive diffusion from the bloodstream. As hair is known to grow at an average rate of about 1 cm per month, a scalp-near collected hair segment of 1 cm provides a window into cortisol secretion over the past month (39). The hair samples were sent to Dresden LabService GmbH and analyzed using liquid chromatography and tandem mass spectrometry (LC–MS/MS). In addition to cortisol values, the hair samples were also used to assess cortisone (SI Appendix).

Independent variables.

Exposure to the no-campaign (α_T2 = 0.74).

Exposure to the no-campaign within the last month was assessed with seven items at Time 2 (as the campaign had not yet started at Time 1). Individuals replied to the following questions: “In the last month, how often have you been exposed to arguments against marriage equality from the following sources: 1) in a conversation, 2) group chat(s) (e.g., WhatsApp, Telegram, Signal), 3) social media platform(s), 4) flyer(s), 5) posters, billboards, stickers, or other public advertisements, 6) newspaper(s) or magazine(s), and 7) television or radio?” All items were assessed on a 7-point Likert scale (1 = Never, 2 = 1 to 2 times, 3 = 1 time per week, 4 = 2 to 3 times per week, 5 = 4 to 6 times per week, 6 = Everyday, 7 = Multiple times a day). Individuals could further choose “Other” and specify their answer, but this answering option was not included in the analyses.

Exposure to the yes-campaign (α_T2 = 0.78).

Exposure to the yes-campaign within the last month was assessed with seven items at Time 2 (as the campaign had not yet started at Time 1). Individuals replied to the following questions “In the last month, how often have you been exposed to arguments in favor of marriage equality from the following sources: 1) in a conversation, 2) group chat(s) (e.g., WhatsApp, Telegram, Signal), 3) social media platform(s), 4) flyer(s), 5) posters, billboards, stickers, or other public advertisements, 6) newspaper(s) or magazine(s), and 7) television or radio?” All items were assessed on a 7-point Likert scale (1 = Never, 2 = 1 to 2 times, 3 = 1 time per week, 4 = 2 to 3 times per week, 5 = 4 to 6 times per week, 6 = Everyday, 7 = Multiple times a day).

Engagement in collective actions (α_T2 = 0.74).

Engagement in collective action within the last month was assessed with eight items at Time 2. Individuals replied to the following questions: “We are interested in any actions that you have taken to express your opinion in the past month. Please select all actions from the below that you have taken: 1) encouraged my family and my friends to support marriage equality, 2) talked to cis-heterosexual individuals to motivate them to support marriage equality, 3) posted on social media (e.g., Facebook, Twitter, Instagram), 4) put a visible rainbow flag/ trans flag (e.g., on your balcony, your clothes, your bag), 5) donated to an organization that support marriage equality, 6) distributed flyers, 7) took part in an information booth at a public place, and 8) participated at a sit-in or demonstration.” The response categories were 1 = Yes, 0 = No.

Impact of Covid-19.

Participants indicated the extent to which their well-being had been affected by Covid-19 in the past month at Time 2, providing their answers on a 7-point Likert scale (1 = Very negatively affected, 4 = Not affected, 7 = Very positively affected).

Grouping variables.

Sexual orientation.

Participants were asked to indicate their sexual orientation. The response categories were Heterosexual/ Bisexual/ Pansexual/ Homosexual/ Asexual/ Other, namely.

Gender.

Participants were asked to indicate their gender. The response categories were Male/ Female/ Non-binary/ Other, namely.

Transgender status.

Transgender status was assessed using the following question: “Do you consider yourself to be a trans person?” The response categories were: Yes, trans woman/ Yes, trans man/ Yes, trans person with a non-binary gender identity/ Yes, trans person, other, namely:/ No, I do not identify as a trans person.

Intersex status.

Participants were asked to indicate whether they are intersex. The response categories were Yes/ No/ I don’t know.

Analytical Strategy.

All analyses were conducted with the R software (R Core Team, 2022). To answer our research questions, we differentiated between confirmatory and exploratory analyses to investigate the impact of the upcoming vote (Time 1 and 2) and the outcome of the vote (Time 2 and 3). All analyses were conducted using average scores for each construct with more than one item and no sample weighting was applied to the data.

It may further be noted that the data collection coincided with the outbreak of the Covid-19 pandemic in Switzerland. For this reason, it is possible that the outbreak of the Covid-19 pandemic impacted participants’ perceived and biological stress as well. Thus, we assessed to which degree Covid-19 impacted the life of the participants. We examined whether this impacted our results.

Descriptive statistics can be found in SI Appendix, Table S1. Because several participants had extreme cortisol values (e.g., range for cortisol values at T2: 0.13 to 253.12, for a median of 5.4), we applied different methods to deal with outliers. For all analyses reported in the main text, extreme cortisol values at each time point (1.5 MAD—deviations from the median) were winsorized. Analyses using other techniques can be found in supplementary material. Importantly, the main findings did not differ depending on the method used to deal with outliers.

First, we examined the effects of the campaign (Time 1 to 2). We created difference scores between perceived stress at Time 1 and Time 2, and between biological stress at Time 1 and Time 2. We then ran multiple regressions, which we extended step by step (see supplementary material). We then investigated the effect of the outcome of the vote (Time 2 to Time 3). We created difference scores between psychological and biological stress at Time 2 and 3 (Study 1). We used the same models as for Time 1 and Time 2 to test our hypotheses (see above).

Supplementary Material

Appendix 01 (PDF)

Dataset S01 (XLSX)

Dataset S02 (XLSX)

Dataset S03 (XLSX)

Data, Materials, and Software Availability

Anonymized (Data without identifying information) data have been deposited in Open Science Framework (https://osf.io/cn4mq/?view_only=d1969cf2c3e2482685188eee08a056d2) (40).