2.1.1 Participants

We recruited for this study via student email lists and social media posts and targeted as many responses as possible before an agreed stopping date. The study was completed by 1537 participants with a mean age of 26.05 years (SD = 9.22). There were no exclusion criteria, beyond being over 18 years old. Participants were mostly (60%) women (n = 920), with 565 men, 25 who reported their gender as other, and 27 who preferred not to say. Most participants (86%) were Irish nationals, and the remaining participants were from another EU country (9%) or outside of the EU (5%). When asked if they identified as a feminist, 58% agreed or strongly agreed, 21% disagreed or strongly disagreed, and 21% neither agreed nor disagreed. 87% of the sample said they used social media a few times a day or more, over the past year. Participants reported interest in the #MeToo movement, with 75% saying they were somewhat likely, likely or very likely to read an online article about the #MeToo movement if they came across it.

2.1.2 Materials

Feminism Attitudes. A scale to measure support for the feminist movement was developed for this study. Participants were asked to rate their agreement with five statements on a scale of (1) Strongly Agree to (5) Strongly Disagree; I identify as a feminist, I believe the feminist movement has gone too far (reverse scored), I support the feminist movement, I believe the feminist movement is necessary, I support the #MeToo movement. These were summed to create a Feminism Attitudes scale. Cronbach’s alpha indicated high reliability (α = .91). We purposefully did not define feminism for participants before they undertook the study, as we were interested in their existing perceptions of feminism (as they perceived it) and did not wish to influence responses.

News Stories. Participants were presented with 8 news stories in a random order. All stories were presented as an image followed by 2–3 lines of text.

True Stories. All participants saw the same six true stories; Bill Cosby convicted of sexual assault, Controversial comments about rape victims by Irish radio host George Hook, “House of Cards” cancelled after Kevin Spacey allegations emerge, Controversial comments made by actor Liam Neeson about the #MeToo movement, Google employee James Damore fired over an internal memo about discrimination, Donald Trump comments about a war on men in America. All true stories were presented with a general image of the subject of the story (e.g., a photograph of Liam Neeson giving an interview).

Fake Stories. All participants saw two fake stories; one concerning estimates of fabricated rape claims and one concerning a riot that broke out at a protest. As can be seen in Table 1, two versions of each story were created – one that aligned with feminist views (e.g., that there are low levels of fabricated rape claims, that men’s rights protestors were in the wrong) and one that did not align with feminist views (e.g., that there are higher levels of fabricated rape claims, that feminist protestors were in the wrong). All participants saw the feminism-aligned version of one story and the feminism-misaligned version of the other story. Both stories were entirely false. There have never been any studies to estimate rates of fabricated rape claims in Ireland. There was an “I Believe Her” protest in Dublin, as described in the riot story, but there was no violence, no injuries and no damage to property. Both versions of the riot story were accompanied by the same image of a man and woman shouting at each other at a protest and both versions of the rape statistics story were accompanied by the same image of an Irish courthouse.

For each story, participants were asked if they remembered the event and could choose from I remember seeing/hearing about this, I do not remember seeing/hearing this but I remember it happening, I do not remember this event but I believe it happened, I remember this event differently, I do not remember this event. As in previous work (Murphy et al., 2019), participants were classed as remembering an event if they indicated that they specifically remembered seeing it or they generally remembered it happening. They were classed as not remembering an event if they said they remembered the event differently or not at all. Unless otherwise stated, participants who stated that they merely believed an event had happened were excluded from the analysis (as in Murphy et al., 2019). This is to allow an assessment of false memories for an event, separate from belief in that event, given that these are distinct processes which do not always mutually occur (Otgaar et al., 2014; Wade et al., 2018).

After each story, participants were asked “How ultimately good/bad do you believe this event was for the feminist movement?” and could answer on a slider from 0 (very bad) to 100 (very good).

After viewing all eight stories, participants were told “Some people who completed this survey were shown fabricated news stories (news stories that didn’t happen, they were completely fabricated by the researchers). If you think any of the stories you saw were fake, please select them below” and were shown all stories again.

2.1.3 Procedure

The study was conducted entirely online, using the Qualtrics survey platform. Participants first completed the demographics and feminism attitudes questions. Then each participant viewed eight news stories (six true, two fake) presented in random order. Participants were then asked to pick out any stories that they believed were fabricated before they were debriefed. We used a debriefing procedure shown to be effective in a similar fake news study (Murphy et al., 2020). The study received ethical approval from the School of Applied Psychology Ethics Committee, University College Cork.

2.2.1 Do attitudes towards feminism predict reported memories for fabricated feminism-related events?

Responses to the two feminist-aligned stories were collapsed and binary logistic regressions were conducted to assess the effect of feminist attitudes on false memories. Analyses were conducted separately for those who interpreted the feminism-aligned story as positive for the feminist movement (n = 438) and those who viewed the story as negative (n = 594). A further 184 participants declined to rate the story and so are not included in either analysis. Those who reported a mere belief in the fabricated event are also excluded here.

For those who interpreted the story as expected, the model was statistically significant, χ2(1, N = 438) = 16.49, p < .001, R² (Cox & Snell) = .04, R² (Nagelkerke) = .05, and correctly classified 59% of cases. As shown in Table 2, there was a significant effect of feminism attitudes, such that for every one-point increase (i.e., more negative attitudes), the odds of reporting a memory for the feminism-aligned story were 8% lower. For those who did not interpret the story as expected, the model was not statistically significant, χ2(1, N = 594) = 1.60, p = .206, R² (Cox & Snell) = .00, R² (Nagelkerke) = .00. As shown in Table 2, there was no significant effect of feminism attitudes on memories for this group.

Identical analyses were conducted on the rate of falsely reported memories for the feminism-misaligned stories. Analyses were again conducted separately for those who interpreted the feminism-misaligned story as negative for the feminist movement (n = 817) and those who viewed the story as positive (n = 233). A further 199 participants declined to rate the story and so are not included in either analysis. Those who reported a mere belief in the fabricated event were also excluded.

For those who interpreted the story as expected, the model was statistically significant, χ2(1, N = 817) = 11.88, p = .001, R² (Cox & Snell) = .01, R² (Nagelkerke) = .02, and correctly classified 62% of cases. As shown in Table 3, there was a significant effect of feminism attitudes, such that for every one-point increase (i.e., more negative attitudes), the odds of falsely reporting a memory for the feminism-misaligned story were 5% greater. For those who did not interpret the story as expected, the model was not statistically significant, χ2(1, N = 233) = 0.02, p = .886, R² (Cox & Snell) = .00, R² (Nagelkerke) = .00. As shown in Table 2, there was no significant effect of feminism attitudes on memories for this group.

Note that participants who reported a memory of hearing about the event are classed as “remembering” in the analyses shown in Table 2 (with mere beliefs excluded). If we used a more liberal classification of memory and included those who reported merely believing the event had occurred, the false memory rate would increase to 70% overall, with 35% reporting one false memory or belief and 35% reporting two false memories or beliefs. The regression results do not change when those who reported a memory or belief are included. For those who interpreted the story as expected, more negative feminist attitudes predicted significantly greater likelihood of a false memory or belief for the feminism-misaligned story (OR = 1.04, [95% CI: 1.01–1.07]) and reduced likelihood of a false memory or belief for the feminism-aligned story (OR = 0.91, [95% CI: 0.88–0.94]). This was not the case for those who did not interpret the event as expected.

For illustrative purposes, we grouped participants according to their score on the feminism attitudes scale (minimum 5, maximum 25); Support Feminism (5–12, n = 917), Neutral (13–17, n = 379), Oppose Feminism (18–25, n = 216). Figure 1 shows rates of reported memories for each fabricated event across these groups. Note that only those who interpreted the event as expected (e.g., positive events as positive for feminism) are included here, those who did not interpret the event as expected were excluded.

2.2.2 Do attitudes towards feminism predict ability to identify fabricated feminism-related events?

After responding to all the news stories, participants were told they may have been exposed to fake news and were asked to select any stories they thought were fabricated. The fabricated stories were selected as fake by the majority of participants; feminist riot 47%, men’s rights riot 47%, high false rape claims 52%, low false rape claims 58%. The true stories were selected as fake at a lower rate overall; George Hook 16%, Bill Cosby 4%, Donald Trump 12%, Google 43%, Liam Neeson 22%, Kevin Spacey 7%. Two binary logistic regressions were conducted to assess the impact of feminism attitudes on ability to identify the story as fake. As before, these were conducted separately for those who interpreted the story as expected and those who did not.

For the feminism-aligned events, the model for those who interpreted the story as expected was statistically significant, χ2(1, N = 577) = 28.22, p = <.001, R² (Cox & Snell) = .05, R² (Nagelkerke) = .06, and correctly classified 61% of cases. As shown in Table 3, there was a significant effect of feminism attitudes, such that for every one-point increase (i.e., more negative attitudes), the odds of participants identifying the feminism-aligned story as fake were 9% greater. For those who did not interpret the story as expected, the model was not statistically significant, χ2(1, N = 724) = 0.07, p = .787, R² (Cox & Snell) = .00, R² (Nagelkerke) = .00.

For the feminism-misaligned events, the model for those who interpreted the story as expected was statistically significant, χ2(1, N = 998) = 5.54, p = .019, R² (Cox & Snell) = .01, R² (Nagelkerke) = .01, and correctly classified 52% of cases. As shown in Table 3, there was a significant effect of feminism attitudes, such that for every one-point increase (i.e., more negative attitudes), the odds of participants identifying the feminism-misaligned story as fake were 3% lower. For those who did not interpret the story as expected, the model was not statistically significant, χ2(1, N = 276) = 1.38, p = .241, R² (Cox & Snell) = .01, R² (Nagelkerke) = .01.

Participants were again grouped according to their score on the feminism attitudes scale, for illustrative purposes. Figure 1 shows the rates of correct identification for each fabricated event across these groups. Note that only those who interpreted the event as expected (e.g., positive events as positive for feminism) are included here, those who did not interpret the event as expected were excluded.