Economic Consequences of Kinship: Evidence from US Bans on Cousin Marriage

Economic Consequences of Kinship: Evidence from US Bans on Cousin Marriage. Arkadev Ghosh, Sam Il Myoung Hwang, Munir Squires. January 18, 2023. https://www.dropbox.com/s/gy15bofmrtejdcz/Cousin_Marriage_2023.pdf

Abstract—Close-kin marriage, by sustaining tightly-knit family structures, may impede development. We find support for this hypothesis using US state bans on cousin marriage. Our measure of cousin marriage comes from the excess frequency of same-surname marriages, a method borrowed from population genetics that we apply to millions of marriage records from the 18th to the 20th century. Using census data, we first show that married cousins are more rural and have lower-paying occupations. We then turn to an event study analysis to understand how cousin marriage bans affected outcomes for treated birth cohorts. We find that these bans led individuals from families with high rates of cousin marriage to migrate off farms and into urban areas. They also gradually shift to higher-paying occupations. We also observe increased dispersion, with individuals from these families living in a wider range of locations and adopting more diverse occupations. Our findings suggest that these changes were driven by the social and cultural effects of dispersed family ties rather than genetics. Notably, the bans also caused more people to live in institutional settings for the elderly, infirm or destitute, suggesting weaker support from kin.

JEL: D00, N00.

Growing evidence suggests that intergroup contact, psychology’s most-researched paradigm for reducing prejudice, has the “ironic” effect of reducing support for social change in disadvantaged groups

Reimer, N. K., & Sengupta, N. K. (2023). Meta-analysis of the “ironic” effects of intergroup contact. Journal of Personality and Social Psychology, 124(2), 362–380. Jan 2023. https://doi.org/10.1037/pspi0000404

Abstract—Growing evidence suggests that intergroup contact, psychology’s most-researched paradigm for reducing prejudice, has the “ironic” effect of reducing support for social change in disadvantaged groups. We conducted a preregistered meta-analytic test of this effect across 98 studies with 140 samples of 213,085 disadvantaged-group members. As predicted, intergroup contact was, on average, associated with less perceived injustice (r = −.07), collective action (r = −.06), and support for reparative policies (r = −.07). However, these associations were small, variable, and consistent with alternative explanations. Across outcomes, 25%–36% of studies found positive associations with intergroup contact. Moderator analyses explained about a third of the between-sample variance, showing that, at least for perceived injustice, associations with intergroup contact were most consistently negative in studies that measured direct, qualitatively positive contact among adults. We also found evidence for an alternative explanation for the apparent “ironic” effects of intergroup contact as, after controlling for the positive association of negative contact with support for social change, positive contact was no longer associated with any of the outcomes. We close by discussing the strengths and limitations of the available evidence and by highlighting open questions about the relationship between intergroup contact and support for social change in disadvantaged groups. 

How and Why We Want to Be More Moral; as known, the bigger predictor of moral motivation was the extent to which we believed that making the change would have positive consequences for our own well-being

How and Why People Want to Be More Moral. Jessie Sun, Joshua Wilt, Peter Meindl, Hanne M. Watkins, Geoffrey P. Goodwin. Journal of Personality, January 18 2023. https://doi.org/10.1111/jopy.12812

Abstract—What types of moral improvements do people wish to make? Do they hope to become more good, or less bad? Do they wish to be more caring? More honest? More loyal? And why exactly do they want to become more moral? Presumably, most people want to improve their morality because this would benefit others, but is this in fact their primary motivation? Here, we begin to investigate these questions. Across two large, preregistered studies (N = 1,818), participants provided open-ended descriptions of one change they could make in order to become more moral; they then reported their beliefs about and motives for this change. In both studies, people most frequently expressed desires to improve their compassion and more often framed their moral improvement goals in terms of amplifying good behaviors than curbing bad ones. The strongest predictor of moral motivation was the extent to which people believed that making the change would have positive consequences for their own well-being. Together, these studies provide rich descriptive insights into how ordinary people want to be more moral, and show that they are particularly motivated to do so for their own sake.

From the ungated version:

General Discussion

How do people conceptualize moral improvement, and when do they feel more motivated 

to undertake such improvements? Table 7 shows a summary of the key findings. Across two 

large studies that featured two very different samples (YourMorals.org vs. CloudResearch) and 

methods (expert codings vs. self-codings), we find that people conceptualize moral improvement 

in diverse ways. A wide range of traits was represented, but the most common goal was to 

become more compassionate, followed by goals to become less reactive and more honest. People 

were somewhat more inclined to conceptualize moral improvement in terms of starting or 

increasing positive tendencies (rather than stopping or decreasing negative ones). People were 

less motivated to make agreeableness-related improvements. They were more motivated to make 

improvements that they perceived as being within their control. They were also more motivated 

to make improvements that they believed would benefit themselves, and to a less pronounced

extent, others.

Diverse Concepts of Moral Improvement

Moral relevance norms from past studies show that compassion, respectfulness, honesty, 

fairness, bravery, and loyalty are among the most prototypical moral traits (Goodwin et al., 2014; 

Strohminger & Nichols, 2015; Sun & Goodwin, 2020). Do people think of moral improvements 

in terms of these prototypical moral virtues? Demonstrating the diversity of people’s concepts of 

moral improvement, moral improvement goals spanned more than 20 trait categories in both 

studies. Goals to become more compassionate, more honest, and more respectful, however, were 

among the five most-frequently mentioned moral improvements in at least one of the two 

samples. In contrast, goals to become more loyal and more fair were mentioned less than 5% of 

the time in both samples.

[...]

It is particularly striking that goals to become more compassionate were, by far, the most 

frequently mentioned moral improvements. Indeed, in Study 2, compassion was mentioned 

almost three times more frequently than the second most common trait category (goals to 

become less reactive). Why might this be the case? There are at least two non-mutually-

exclusive interpretations for why some trait categories (e.g., becoming more compassionate)

were more frequently mentioned than others (e.g., becoming more honest, fair, or loyal). First, 

perhaps these are the traits that are most relevant to the moral situations people encounter in 

everyday life. Providing some support for the relevance hypothesis, an experience sampling 

study of moral events in daily life suggests that Care/Harm (which corresponds most closely with 

the trait of compassion) was the most frequently mentioned moral domain (Hofmann et al., 

2014). Thus, perhaps people most frequently mentioned goals to become more compassionate in 

part because this is the most relevant moral domain in everyday life.

A second possibility is that people tend to perceive more opportunities for improvement 

on the traits that they more frequently mentioned. Compassion may be distinct from traits such as 

honesty, fairness, and loyalty, in at least two ways. First, judgments of honesty, fairness, and 

loyalty may be based more on behavior than are judgments of compassion. That is, whereas 

honest behavior (vs. “honest thoughts” or “honest feelings”) may be the primary basis for 

judging oneself as being an honest person, people might consider themselves to be truly

compassionate only if they think, feel, and act compassionately. Second, honesty, fairness, and 

loyalty may have consequences for a smaller set of potential targets compared to compassion. 

For example, one could be loyal to a romantic partner, friends, company, or a nation, but not to 

everyone in the world. In contrast, the scope of one’s compassion could be extended beyond 

close others to not only strangers in one’s community (e.g., homeless people), but also those in

foreign countries (e.g., people who are dying from preventable diseases), nonhuman animals, and 

future generations (Singer, 1981). For these reasons, it may be easier to perfectly fulfil one’s 

duties to be honest (by being completely truthful), fair (by being entirely unbiased), and loyal (by 

never betraying others), whereas the opportunities for becoming more compassionate seem 

almost endless by comparison (Trammell, 1975; see also Trafimow & Trafimow, 1999).

Interestingly, improvements to several traits that are not inherently moral traits—

decreased reactivity and increased open-mindedness, productiveness, and self-control—were 

also among the most frequently-mentioned moral improvements. These results suggest that even 

if certain traits are not inherently or specifically moral (e.g., self-control can be deployed towards 

moral, immoral, or morally neutral ends; Landy et al., 2016; Hofmann et al., 2018), people may 

still see such traits as being additional means to achieving moral ends.

Self-Interest is a Key Motivation for Moral Improvement

What motivates people to be more moral? From the perspective that the function of 

morality is to suppress selfishness for the benefit of others (Haidt & Kesebir, 2010; Wolf, 1982), 

we might expect people to believe that moral improvements would primarily benefit others 

(rather than themselves). By a similar logic, people should also primarily want to be more moral 

for the sake of others (rather than for their own sake).

Surprisingly, however, this was not overwhelmingly the case. Instead, across both 

studies, participants were approximately equally split between those who believed that others 

would benefit the most and those who believed that they themselves would benefit the most

(with the exception of compassion; see Figure S2). The finding that people perceive some 

personal benefits to becoming more moral has been demonstrated in recent research (Sun & 

Berman, in prep). In light of evidence that moral people tend to be happier (Sun et al., in prep)

and that the presence of moral struggles predicts symptoms of depression and anxiety (Exline et 

al., 2014), such beliefs might also be somewhat accurate. However, it is unclear why people 

believe that becoming more moral would benefit themselves more than it would others. 

Speculatively, one possibility is that people can more vividly imagine the impacts of their own 

actions on their own well-being, whereas they are much more uncertain about how their actions 

would affect others—especially when the impacts might be spread across many beneficiaries.

However, it is also possible that this finding only applies to self-selected moral 

improvements, rather than the universe of all possible moral improvements. That is, when asked 

what they could do to become more moral, people might more readily think of improvements 

that would improve their own well-being to a greater extent than the well-being of others. But, if 

we were to ask people to predict who would benefit the most from various moral improvements 

that were selected by researchers, people may be less likely to believe that it would be

themselves. Future research should systematically study people’s evaluations of how various 

moral improvements would impact their own and others’ well-being.

Similarly, when explicitly asked for whose sake they were most motivated to make their 

moral improvement, almost half of the participants admitted that they were most motivated to 

change for their own sake (rather than for the sake of others). However, when predicting 

motivation from both the expected well-being consequences for the self and the well-being 

consequences for others, we found that people’s perceptions of personal well-being 

consequences was a significantly stronger predictor in both studies. In other words, if anything,

people are relatively more motivated to make moral improvements for their own sake than for 

the sake of others. This is consistent with the findings of another study which examined people’s 

interest in changing a variety of moral and nonmoral traits, and showed that people are 

particularly interested in improving the traits that they believed would make them relatively 

happier (Sun & Berman, in prep). Here, it is striking that personal fulfilment remains the most

important motivator of personal improvement even exclusively in the moral domain

Hatred of the opposing political camp fuels susceptibility to misinformation that favors one's own side—both among Democrats and Republicans—and particularly among the politically sophisticated

Affective Polarization and Misinformation Belief. Libby Jenke. Political Behavior, Jan 18 2023. https://link.springer.com/article/10.1007/s11109-022-09851-w

Abstract: While affective polarization has been shown to have serious social consequences, there is little evidence regarding its effects on political attitudes and behavior such as policy preferences, voting, or political information accrual. This paper provides evidence that affective polarization impacts misinformation belief, arguing that citizens with higher levels of affective polarization are more likely to believe in-party-congruent misinformation and less likely to believe out-party-congruent misinformation. The argument is supported by data from the ANES 2020 Social Media Study and the ANES 2020 Time Series Study, which speaks to the generalizability of the relationship. Additionally, a survey experiment provides evidence that the relationship is causal. The results hold among Democrats and Republicans and are independent of the effects of partisan strength and ideological extremity. Furthermore, the relationship between affective polarization and misinformation belief is exacerbated by political sophistication rather than tempered by it, implying that education will not solve the issue. The results speak to the need for work on reducing affective polarization.

Discussion

This paper tests and confirms four hypotheses across three data sets: the ANES Social Media Study, the ANES Time Series Study, and a survey experiment. The data show that an increase in affective polarization caused an increase in acceptance of in-party-congruent misinformation. But the opposite was true of out-party-congruent misinformation: an increase in affective polarization caused a decrease in acceptance of out-party-congruent misinformation. These associations held for both Democrats and Republicans and were independent of the effects of partisan strength and ideological extremity. The effect of political sophistication on the relationship between affective polarization and misinformation belief was to exacerbate the belief in in-party-congruent misinformation and further reduce the belief in out-party-congruent misinformation. The use of observational data demonstrates the generalizability of the results, and the experimental data demonstrate the causal validity of these claims.

These results affirm that affective polarization has implications for political attitudes and behavior as well as the social consequences that have been well documented (Panagopoulos et al., 2020; Chen & Rohla, 2018; Nicholson et al., 2016; Huber & Malhotra, 2017; Iyengar et al., 2012). They join and expand upon the recent findings of Druckman et al. (2021) and Wagner (2021): in addition to policy stances and turnout likelihood, affective polarization also has a relationship with levels of misinformation belief.

The findings suggest that the increase in affective polarization over the past few years is matched by polarization in terms of partisans’ beliefs. This in turn has both social and political implications. Regarding the social implications, it is difficult to engage in political conversation with someone whose fundamental ideas about what is true differ. If Democrats and Republicans believe the misinformation congruent with their own party and disbelieve the misinformation congruent with the other party, then they will be working from different sets of “facts.” We should expect more rancor in discussions between affectively polarized partisans.

More important are the political implications. Affectively polarized people are more likely to turn out to vote than those less affectively polarized (Wagner, 2021). Since affectively polarized people are more likely to believe in-party-congruent misinformation, their issue and candidate opinions are also more likely to be dependent on false information. Kuklinski et al. (2000) showed that misinformed voters do indeed use the misinformation in forming their political opinions, potentially causing different collective preferences than would exist had they been well informed. Together, these findings imply that those most likely to participate in our democratic process are more likely to have preferences that have been developed through the use of misinformation. Thus, the increase in affective polarization portends trouble for our representative democracy.

There are many avenues for future research on affective polarization and misinformation. First, though this paper has shown that affective polarization causes misinformation belief, it is also possible that misinformation belief increases affective polarization. Believing negative false information about the out-party may increase one’s feelings of anger towards the out party. The two concepts may be mutually reinforcing. This possibility should be examined by future papers. Second, there is an alternative explanation to motivated reasoning for misinformation belief, cheerleading. It is not possible in this study to tell if respondents answered the misinformation questions incorrectly on purpose, cheerleading for their parties rather than giving honest answers. Studies have confirmed that respondents answer misinformation questions truthfully (Peterson & Iyengar, 2021; Berinsky, 2018), and yet this has not been tested in the context of these surveys. A study that did so would offer even more compelling evidence that the specific misinformation questions were answered genuinely.

Work should also focus on validating the mechanisms of the theory presented in this paper. Confirming that motivated reasoning is the mechanism connecting affective polarization and misinformation belief would be useful. Taber and Lodge (2006) suggested that motivated reasoners are more likely to think intuitively when receiving information congruent with their prior beliefs and more likely to think analytically when receiving incongruent information. If this is the case regarding the information processing of affectively polarized citizens, this could suggest solutions besides reducing affective polarization. For example, if using quick or intuitive processing system when receiving in-party-congruent information is the mechanism that causes acceptance of this misinformation, then focusing attention to accuracy could mitigate belief. Indeed, Pennycook et al. (2021) found that shifting people’s attention towards accuracy increased the quality of news that they subsequently shared. It would be useful to know if this shift in focus works for affectively polarized citizens.

Milton Lodge and Charles Taber highlighted an important goal of social science research: finding when affect has negative implications for cognition (Groenendyk & Krupnikov, 2021). What should be added to this goal is the discernment of which of these negative implications have the greatest impact on society. The relationship between affective polarization and misinformation belief shows that affective polarization is concerning not only for the well-being of our society socially but also politically and should renew efforts to identify mechanisms by which it can be reduced.

Effects of a 7-Day Pornography Abstinence Period on Withdrawal-Related Symptoms in Regular Pornography Users

Effects of a 7-Day Pornography Abstinence Period on Withdrawal-Related Symptoms in Regular Pornography Users: A Randomized Controlled Study. David P. Fernandez, Daria J. Kuss, Lucy V. Justice, Elaine F. Fernandez & Mark D. Griffiths. Archives of Sexual Behavior, Jan 18 2023. https://link.springer.com/article/10.1007/s10508-022-02519-w

Abstract: Little is known about whether withdrawal-like symptoms manifest when regular pornography users attempt to abstain from pornography. The present study used a randomized controlled design to examine whether (1) negative abstinence effects that may be potentially reflective of withdrawal-related symptoms manifest when a non-clinical sample of regular pornography users attempt to abstain from pornography for a 7-day period and (2) these negative abstinence effects would only manifest (or manifest more strongly) for those with higher levels of problematic pornography use (PPU). A total of 176 undergraduate students (64.2% female) who were regular pornography users (defined as having used pornography ≥ three times a week in the past 4 weeks) were randomly assigned to an abstinence group (instructed to attempt abstinence from pornography for 7 days, n = 86) or a control group (free to watch pornography as usual, n = 90). Participants completed measures of craving, positive and negative affect, and withdrawal symptoms at baseline and each night of the 7-day period. Contrary to the confirmatory hypotheses, there were no significant main effects of group (abstinence vs. control) or group × PPU interaction effects on any of the outcome measures, controlling for baseline scores. These findings indicate that no evidence of withdrawal-related symptoms was found for abstaining participants, and this was not dependent on level of PPU. However, exploratory analyses showed a significant three-way interaction (group × PPU × past 4-week frequency of pornography use [FPU]) on craving, where an abstinence effect on craving was found at high levels of PPU only once past 4-week FPU reached the threshold of daily use. While these exploratory findings should be interpreted with caution, they suggest that abstinence effects could potentially manifest when there is a combination of high PPU and high FPU—a hypothesis that warrants investigation in future prospective abstinence studies.

Discussion

The present study used a randomized controlled design to examine whether (1) negative abstinence effects potentially reflective of withdrawal-related symptoms manifest when regular pornography users (defined in the present study as having used pornography ≥ three times a week in the past four weeks) try to abstain from pornography for a 7-day period and (2) these negative abstinence effects only manifest (or manifest more strongly) for those with higher levels of PPU.

Confirmatory Analyses

The results showed that both confirmatory hypotheses (H₁ and H₂) were not supported. Contrary to the first hypothesis, there were no significant main effects of group (abstinence vs. control) on craving, negative affect, positive affect or withdrawal symptoms during the experimental period, controlling for baseline scores. This indicates that when assessed prospectively in comparison with a non-abstaining control group, no evidence of withdrawal-related symptoms was found for the abstaining participants in this sample. This finding is in contrast to previous cross-sectional research relying on retrospective self-report that found 72.2% of pornography users who had at least one past pornography abstinence attempt recalled experiencing at least one withdrawal-like symptom upon cessation (Dwulit & Rzymski, 2019).

The second hypothesis that negative abstinence effects during the experimental period would only manifest (or manifest more strongly) for those with higher levels of PPU was also not supported, as there were no significant group × PPU interaction effects on all outcome measures. Nonetheless, it is important to interpret this finding in light of the fact that because this was a non-clinical sample of regular pornography users, the range of scores on the PPCS was relatively low [M = 54.72, SD = 18.60, with only 15.3% of the sample having PPCS scores that met or exceeded the suggested clinical cutoff of ≥ 76 in Bőthe et al. (2018)]. This finding does not rule out the possibility that in a sample with higher levels of PPU, moderating effects of PPU might be observed.

An unexpected finding was that both negative affect and withdrawal symptoms decreased over the experimental period for both abstinence and control groups. This decrease appeared to be indicative of the “downward drift” phenomenon (see Gilbert et al., 2019, p. 538), which is the tendency for participants to report progressively fewer negative affect-related symptoms when tested repeatedly, regardless of assignment to treatment or control. This phenomenon has been observed in studies of abstinence from cigarette smoking (Gilbert et al., 2002) and gaming (Evans et al., 2018) and also non-abstinence-related studies (e.g., Sharpe & Gilbert, 1998). While multiple explanations have been proposed for the phenomenon (Arrindell, 2001; Sharpe & Gilbert, 1998), one plausible explanation is that repeated testing (in the present study, the daily surveys) acts as an intervention that facilitates self-monitoring of negative affect and possibly coping mechanisms to deal with the negative affect. This finding further underscores the importance of including a non-abstaining control group in future prospective abstinence studies, as decreases in negative affective symptoms over time by abstaining participants could easily be misattributed to abstinence.

Exploratory Analyses

Importantly, exploratory analyses provided partial support for the possibility that the null findings observed in the confirmatory analyses could have been the result of not accounting for the moderating effects of a third variable (i.e., past 4-week FPU). The majority of the sample (61.4%) reported past 4-week FPU on the lower end of the spectrum (i.e., three or four times a week), in large part contributed to by the large proportion of female participants in the sample (64.2%) most of whom also reported past 4-week FPU of three or four times a week (77.0%). Consistent with past research (Chen et al., 2018; Grubbs et al., 2019c; Weinstein et al., 2015), female participants in this sample had lower baseline rates of FPU, PPU and craving than male participants. Therefore, the possibility that abstinence effects might only manifest at higher levels of FPU was examined in exploratory models with past 4-week FPU added as a moderator and gender controlled for as a covariate.

A significant three-way group × PPU × past 4-week FPU interaction on craving was found, but past 4-week FPU did not play a significant moderating role in the negative affect, positive affect and withdrawal symptoms models. At high levels of PPU (+ 1 SD), the control group had higher craving scores than the abstinence group over the experimental period when past 4-week FPU was “three times a week,” but inversely the abstinence group had higher craving scores than the control group over the experimental period when past 4-week FPU was “once a day” or “more than once a day.” Notably, this indicates that there was an abstinence effect on craving when PPU was high, but only once past 4-week FPU reached the threshold of daily use. It is noteworthy that this effect was found even though high PPU at + 1 SD (i.e., M = 72.82) in this sample did not reach the clinical cutoff of ≥ 76 specified in Bőthe et al. (2018). This finding is similar to a previous cross-sectional survey of undergraduates in Canada that found an increase in addiction-related symptomatology once FPU reached the threshold of daily use (Harper & Hodgins, 2016). It is uncertain as to why the abstinence group reported significantly lower craving than the control group at high levels of PPU and past 4-week FPU of ‘three times a week,’ but one speculative explanation could be that abstaining participants below a specific threshold of FPU (i.e., three times a week) due to lower habit strength (Sirriani & Vishwanath, 2016) had enough self-regulation such that they were able to successfully implement strategies to regulate their experience of craving (e.g., prevent its occurrence or reduce its intensity whenever it did emerge) to help them achieve the abstinence goal.

There are two possible interpretations of the abstinence effect on craving under high PPU and high FPU conditions that are not necessarily mutually exclusive. First, if negative abstinence effects are interpreted within an addiction framework as withdrawal-related symptoms, then craving could potentially be a withdrawal symptom in PPU. High FPU on its own (without high PPU) may be indicative of preoccupation with pornography that may not be associated with negative life consequences (Bőthe et al., 2020a, 2020b), but the fact that craving manifested for those with daily or more FPU only when PPU was high reinforces the idea that abstinence-induced craving could be reflective of a dependency on pornography and potentially a withdrawal symptom (if a pornography withdrawal syndrome exists). This interpretation would be consistent with a finding from a recent systematic review that craving was the most common abstinence effect across multiple potential behavioral addictions (Fernandez et al., 2020). Second, craving could also have been a manifestation of sexual desire and/or arousal during the experimental period. Because higher baseline rates of FPU could have to some extent been reflective of higher sexual drive (Leonhardt et al., 2021), the fact that craving only manifested for those with high PPU when FPU was daily or more supports the sexual desire explanation. If participants’ primary sexual outlet was masturbating to pornography, then urges to use pornography could be a natural manifestation of sexual desire and/or arousal throughout the deprivation period (Castro-Calvo et al., 2022). The fact that an abstinence effect was found for craving and not the other outcome variables (i.e., positive affect, negative affect and withdrawal symptoms) further lends support to this non-pathological interpretation, because affect-related disturbances would arguably be expected to also be present if an actual withdrawal syndrome exists. At the same time, the lack of abstinence effects on these other outcome variables needs to be interpreted with caution because given the limited statistical power in the sample to detect three-way interaction effects, similar but smaller effects may not have been detected for these variables. Overall, all findings in these exploratory analyses should be interpreted with caution because of their post hoc exploratory nature and limited statistical power to detect effects of interest. While these exploratory findings are noteworthy, they should be regarded as hypothesis-generating for future studies.

Implications

In sum, the findings of the present study have two main implications for understanding the potential manifestation of withdrawal-related symptoms during pornography abstinence. First, confirmatory analyses showed that there was no evidence of negative abstinence effects (i.e., withdrawal-related symptoms) among a sample of pornography users who were using pornography at least three times a week, and this was not dependent on level of PPU (but with the caveat that the sample had relatively low levels of PPU). These null findings are important to emphasize as they provide preliminary evidence that the average regular pornography user who uses pornography somewhat regularly (i.e., a few times a week) generally does not experience withdrawal-like symptoms while trying to abstain from pornography for a 7-day period.

Second, exploratory findings raise the possibility that negative abstinence effects might manifest only when both baseline FPU and PPU are high—a hypothesis that needs to be tested in future prospective studies. Exploratory analyses found that craving was an abstinence effect when PPU was high, but only once past 4-week FPU was at least daily or more. This tentatively suggests that craving could potentially be a PPU withdrawal symptom if a pornography withdrawal syndrome exists, but future adequately powered studies need to verify this finding a priori and rule out alternative theoretical explanations (e.g., craving being solely a manifestation of sexual arousal). While craving was the only significant abstinence effect under these conditions, it cannot be ruled out that adequately powered studies might find other abstinence effects.

Overall, it is crucial that the findings of the present study be considered in terms of its specific sample characteristics (i.e., non-clinical, majority female sample of undergraduates from a sexually conservative country, most of whom were using pornography 3–4 times a week [61.4%], had PPCS scores below the clinical cutoff of 76 [84.7%] and had no intrinsic desire to quit their pornography use [89.8%]). These findings may not generalize to clinical samples, non-clinical samples with higher FPU or PPU, predominantly male samples, samples from more sexually liberal countries or samples composed solely of pornography users intrinsically motivated to quit their pornography use. Future studies using similar prospective designs and diverse samples with varying gender ratios and levels of FPU, PPU and intrinsic motivation to abstain from pornography use are needed to replicate and extend these findings.

It is also important to consider when interpreting the present study’s findings that the study period (February–March 2021) was during the third wave of the COVID-19 pandemic in Malaysia, when movement control and social distancing measures were enforced throughout the country (Rampal & Liew, 2021; Zamri et al., 2021). This could have potentially influenced the present study’s findings in two ways. First, participants’ self-reported affective states during the study period could potentially have been impacted by elevated psychological distress levels brought about by the pandemic (Marzo et al., 2021; Necho et al., 2021). Second, participants’ sexual lives could have also been affected by the pandemic. Frequency of partnered sex among participants may have decreased due to reduced access to potential sexual partners (Herbenick et al., 2022). While some longitudinal data have shown that as a general trend, FPU and PPU did not increase over time during the pandemic, this might not necessarily have been the case for all pornography users (Grubbs et al., 2022; Koós et al., 2022). For example, there has been evidence of retrospective self-reports, particularly among male participants, of increased FPU and frequency of masturbation during the pandemic when compared to pre-pandemic frequencies (Gleason et al., 2021; Sallie et al., 2021). It is plausible that some participants in the present study may have developed an increased reliance on masturbating to pornography as a sexual outlet in the absence of opportunities for partnered sex during stay-at-home orders, which could have made abstaining from pornography more challenging than usual. In sum, it is important to take into account that the present study’s findings may not necessarily generalize to a post-pandemic setting.

Limitations and Directions for Future Research

The present study has several limitations that need to be highlighted. First, because the study’s aim was to investigate abstinence effects irrespective of gender, there were no gender restrictions in the inclusion criteria, resulting in a sample that was close to two-thirds (64.2%) female. Compared to females, males tend to use pornography more for sexual pleasure (Bőthe et al., 2021a; Grubbs et al., 2019c) and have a stronger sex drive in general (Baumeister et al., 2001). As such, males may be more reliant on pornography as a sexual outlet than females. Therefore, it may be reasonable to speculate that males could have greater difficulty in abstaining from pornography than females even if they have similar rates of baseline FPU. However, because females had lower past 4-week FPU than males in the present sample, exploratory analyses with gender as moderator were not run because they would have been difficult to disentangle from exploratory analyses with past 4-week FPU as moderator. Future studies could consider restricting inclusion criteria to just male participants to examine whether abstinence effects manifest in a male-only sample or run adequately powered studies with both genders included (but with narrower inclusion criteria for baseline FPU [e.g., ≥ six times a week]) to investigate potential varying abstinence effects by gender, if any. Differing contexts of pornography use across genders (e.g., women in relationships are more likely than men to use pornography primarily or only with their partner; Carroll et al., 2017) could also be accounted for as a potential moderator of effects in future studies.

Second, nearly half (45.4%) of participants in the abstinence group reported using pornography at least once during the experimental period, which could have led to a mitigation in the frequency and/or intensity of any withdrawal-related symptoms experienced for these participants. While excluding participants who lapse from analyses may introduce bias because these participants are likely to experience the greatest amount of withdrawal in the first place (Hughes, 2007b; Piasecki et al., 2003a, 2003b; Shiffman et al., 2004), it is important to keep in mind that withdrawal-related symptom scores of the abstinence group as a whole may have been influenced by the considerable proportion of participants who did lapse in the present study.

Third, because the focus of the present study was to examine effects of abstinence from pornography specifically, participants in the abstinence group were allowed to masturbate without pornography or engage in non-pornography-related sexual activity. An advantage of allowing non-pornography-related sexual outlets is that potential manifestations of withdrawal-like symptoms can be more clearly attributed to a dependency on pornography use instead of dependency on masturbation or sexual activity (e.g., if withdrawal-like symptoms persist even after masturbating without pornography). However, a drawback of this is that it may not resemble many real-world abstinence attempts, given that many individuals with PPU may decide to abstain from pornography and masturbation or abstain from sexual activity altogether as a short-term recovery strategy (Fernandez et al., 2021). Future studies could consider modifying abstinence instructions so that they also include masturbation or sexual activity altogether.

Fourth, while the experimental design used in the present study contributed to rigorous internal validity due to randomization and use of a control group, experimental designs are inherently limited in the extent to which the studies’ results are generalizable to the wider population (because of the use of a relatively small, homogenous sample) and applicable to real-world situations (because of the emphasis on controlled conditions that prioritize internal validity over ecological validity). Importantly, abstinence in the present study may not be representative of real-world abstinence attempts because participants knew that abstinence would only last for a temporary 7-day period, they could lapse without consequence, and they could terminate their participation at any moment. Non-experimental longitudinal studies of intrinsic abstinence attempts are needed to supplement data provided by studies that experimentally manipulate abstinence.

Fifth, because it is plausible that a pornography withdrawal syndrome, akin to the withdrawal syndromes of some substances (cf. Hughes et al., 1994), might only begin or worsen after a significant period of abstinence (e.g., one or two weeks), the fairly brief duration of the present study (i.e., 7 days) meant that any potential withdrawal symptoms that might have emerged beyond an initial 7-day period could not be captured. Future studies could consider increasing the duration of the abstinence period to investigate this possibility.

Sixth, the outcome measures used to assess potential pornography withdrawal-related symptoms in the present study were a limitation because they were modified from existing scales constructed to assess alcohol craving (Flannery et al., 1999), smoking withdrawal symptoms (Welsch et al., 1999) and general affect (Thompson, 2007). Measures of pornography-specific withdrawal-related symptoms need to be developed and validated for use in future abstinence studies.

Seventh, despite having a significantly shorter recall period compared to a single end-of-week assessment, end-of-day surveys still rely on retrospection and as such remain susceptible to some recall bias (Newman & Stone, 2019). Research comparing aggregated momentary affect ratings throughout the day to retrospective end-of-day affect ratings demonstrates that end-of-day ratings of negative affect tend to be slightly biased toward peak and recent affect (Neubauer et al., 2020). Future studies can use ecological momentary assessment (EMA) instead of daily surveys for greater sensitivity to fluctuations of affect throughout the day.

Eighth, because the baseline measures for all outcome variables had a bigger recall period (past 7 days) compared to the daily measures (past day), changes in daily scores relative to baseline could not be examined. Incorporating a pre-intervention period where baseline data are also collected using the same daily surveys would allow for standardization of measures and examination of changes from baseline.

Finally, the present study did not account for the type or genre of pornography that participants typically used. It is possible that specific types of pornography (e.g., non-mainstream content; Hald et al., 2018) or greater variability of pornography content consumed (Lewczuk et al., 2021) may be associated with increased difficulty in abstaining from pornography use and can be accounted for in future abstinence studies.

The influence of sexual activity on sleep

The influence of sexual activity on sleep: A diary study. Carlotta Florentine Oesterling, Charmaine Borg, Elina Juhola, Marike Lancel. Journal of Sleep Research, January 16 2023. https://doi.org/10.1111/jsr.13814

Summary: Aiming to promote overall health and well-being through sleep, the present studies examine to what extent sexual activity serves as a behavioural mechanism to improve sleep. The relation between sexual activity, i.e., partnered sex and masturbation with or without orgasm, and subjective sleep latency and sleep quality is examined by means of a cross-sectional and a longitudinal (diary) study. Two hundred fifty-six male and female participants, mainly students, completed a pre-test set of questionnaires and, thereafter, a diary during 14 consecutive days. The cross-sectional study was analysed using analysis of covariance and demonstrated that both men and women perceive partnered sex and masturbation with orgasm to improve sleep latency and sleep quality, while sexual activity without orgasm is perceived to exert negative effects on these sleep parameters, most strongly by men. Accounting for the repeated measurements being nested within participants, the diary data were analysed using multilevel linear modelling (MLM). Separate models for subjective sleep latency and sleep quality were constructed, which included 2076 cases at level 1, nested within 159 participants at level 2. The analyses revealed that only partnered sex with orgasm was associated with a significantly reduced sleep latency (b = −0.08, p < 0.002) and increased sleep quality (b = 0.19, p < 0.046). Sexual activity without orgasm and masturbation with and without orgasm were not associated with changes in sleep. Further, no gender differences emerged. The present studies confirm and significantly substantiate findings indicating that sexual activity and intimacy may improve sleep and overall well-being in both men and women and serve as a directive for future research.

4 DISCUSSION

The present cross-sectional and longitudinal (diary) studies aimed to investigate whether specific sexual activities (i.e., partnered sex and masturbation with or without orgasm) affect subjective sleep latency and sleep quality. The cross-sectional study showed that partnered sex with orgasm as well as masturbation with orgasm are perceived to reduce sleep latency while increasing sleep quality in men and women. Both men and women found that partnered sex and masturbation without orgasm increased sleep latency and decreased sleep quality, albeit men perceived stronger negative effects. The longitudinal study yielded diverging results. Specifically, while partnered sex with orgasm significantly shortened sleep latency and improved sleep quality, masturbation with orgasm did not affect the respective sleep variables. The effects of partnered sex without orgasm and masturbation without orgasm did not have strong enough effects on sleep latency or sleep quality to be detectable in the present design. Although women indicated a higher average sleep quality than men, gender was not found to significantly moderate the relation between sexual activity and sleep.

The present results largely support the hypothesis that sexual activity with orgasm results in reduced subjective sleep latency and increased subjective sleep quality in both men and women. While both studies found significant effects of partnered sex, masturbation – though retrospectively perceived as sleep-promoting – did not exert detectable effects in the longitudinal study. As masturbation with orgasm was indeed perceived to effectively promote sleep when assessed in retrospect, the results may suggest that, in fact, both sex with a partner and masturbation impact sleep latency and sleep quality, while the effect of partnered sex may be stronger and thus more salient. This postulate aligns with findings by Brody and Krüger (2006), who have shown that orgasm following sexual intercourse results in a 400% higher post-coital prolactin surge than masturbation-induced orgasm. As prolactin promotes sleep and is part of a feedback loop communicating sexual satiety, the increased post-coital surge of prolactin may explain why partnered sex is often perceived as more satisfying than masturbation and why the sleep-facilitating effect of sexual activity with orgasm is more salient when a partner is involved, as also found by Pallesen et al. (2020) and Gallop et al. (2021).

The finding that both partnered sex and masturbation without orgasm yielded no – or even negative – effects on sleep point to the relevance of orgasm and its concomitant psychophysiological effects. As orgasm is established to increase the heart rate and blood pressure and results in the release of oxytocin and prolactin – hormones, both postulated to influence sleep (Brody & Krüger, 2006; Fekete et al., 2014; Gianotten et al., 2021; Lipschitz et al., 2015) – neuroendocrine changes following orgasm may contribute to the reduction in sleep latency and increase in sleep quality following partnered sex with orgasm. The present observations are in accordance with earlier findings by Pallesen et al. (2020). Although Lastella et al. (2019) found that sexual activity is also reported to affect sleep when orgasm is not taken into consideration, the percentage of men and women reporting improved sleep latency and sleep quality increased when specifically asked about sex with orgasm. Further, reported gender differences indicating that perceived effects of sex on sleep are stronger in men were non-apparent when orgasm was assessed and may therefore emanate from a gap in orgasm frequency between men and women. Therefore, the discrepancy in results may also stem from less nuanced wording of items applied by Lastella et al. (2019). In a more recent study, Sprajcer et al. (2022) found that orgasm frequency explained 3.1% of the variance in subjective sleep latency, as participants reporting an orgasm “every time” sexual activity occurs fell asleep on average 12 min faster than those who less frequently or never report orgasm.

Given that masturbation with orgasm did not produce significant changes in longitudinally assessed sleep, orgasm per se does not sufficiently explain the reduction of sleep latency and increased sleep quality following partnered sex with orgasm. Other factors accompanying partnered sex with orgasm may also contribute to its positive effects on sleep, such as the mere experience of intimacy with one's partner promoting couple bonding (Kruger & Hughes, 2011), well-being, and emotion regulation (Gianotten et al., 2021) and may thereby improve sleep. Germane to this, non-sexual touch and cuddling have been shown to have calming, sleep-promoting effects, especially for women (Dueren et al., 2022). Compared with masturbation, partnered sex is often associated with more intense and longer-lasting physical activity – resulting in a heightened relaxed state afterwards – which may explain why partnered sex without orgasm resulted in a borderline-significant effect on sleep quality in the diary study despite being reported significantly less frequently than partnered sex with orgasm (n = 85 vs. n = 173). Lastly, the psychological effects of relationship satisfaction, loving and feeling loved, as well as having a sense of belonging or security also warrant consideration and have been shown to impact sleep (Kent et al., 2015; Troxel et al., 2007). Sprajcer et al. (2022) found that individuals who are emotionally satisfied fall asleep on average 10–12 min faster than emotionally unsatisfied individuals, and that orgasm frequency and emotional satisfaction are higher if sexual activity occurrs with a long-term partner, compared with casual sexual relationships. These findings highlight the importance of considering emotional and relationship factors when deriving inferences on the effects of sexual activity on sleep. Anyhow, if penetration has occurred, the positive effects of intimacy on sleep may be undermined if women, and even more so men, do not achieve orgasm. Both women and men retrospectively reported negative effects of sexual activity without orgasm on sleep. This negative perception, although not supported by the longitudinal findings, may be attributed to adjuvant emotions such as frustration, dissatisfaction, uncomfortable bodily sensations resulting from sexual arousal without orgasm, or confounding events that prevented sexual activity from resulting in orgasm.

The hypothesised gender difference suggesting that the effects of sexual activity on sleep are stronger in men than in women was not supported, as changes in subjective sleep latency and sleep quality following partnered sex with orgasm did not differ between men and women. This finding corresponds to results of Kruger and Hughes (2011), who also did not find any gender differences in the influence of sexual activity on sleep, and of Lastella et al. (2019), who did not find a gender difference when sex with orgasm had occurred. The absence of gender differences in the sleep effects of sexual activity with orgasm may be due to comparable endocrine processes following orgasm in men and women (Georgiadis et al., 2009; Mah & Binik, 2002). The widely held notion that men fall asleep faster than women after sexual activity may have emanated from the existing gender gap in achieving orgasm, i.e., women are less likely to reach orgasm during heteronormative sexual activity than men (Blair et al., 2018). Case numbers of the present study corroborate this notion, as although the sample consists of more than twice as many women as men, men reported a higher number of occurrences of both partnered sex and masturbation with orgasm. While following heteronormative scripts, women tend to engage in sexual activities that frequently result in orgasm for men but less often for women (e.g., vaginal penetration only, which does not suffice to achieve orgasm for most women; Lloyd, 2022). Research has further shown that the male orgasm frequently signifies the end of sexual intercourse (Opperman et al., 2014), which decreases the opportunities to achieve an orgasm for women. Women might simply reach orgasm less often and, therefore, less frequently benefit from the sleep-promoting effects of orgasm, which, in turn, may explain why society and cross-sectional research relying on self-report data postulate that men fall asleep faster following sexual intercourse with orgasm.

4.1 Limitations and future directions

While the results of the present study underpin the positive effect of sexual activity on sleep, several aspects may limit the interpretability of the findings. The convenience sampling procedure that also made use of a university-student participant pool of a Dutch University possibly limits the generalisability of the results, as most of the participants are young adults from western countries. As the understanding of sexuality concepts varies greatly across cultures (Hall & Graham, 2012), using a more diverse, inclusive sample is encouraged in future replications to increase external validity. Furthermore, suggestive wording in the pre-test items and normative responding based on the general widespread opinion of the sleep effects of sexual activity possibly resulted in recall and acquiescence bias and might contribute to the variability in results between the cross-sectional and longitudinal analysis. The longitudinal study was a preliminary attempt to bring a more thorough and objective insight into the impact of sexual activity on sleep. Yet, the borderline-significant effect of sexual activity without orgasm on sleep quality warrants further investigation, as it may point to a possible effect in the population that was underpowered (power = 0.51) and not detectable due to a small number of cases of partnered sex without orgasm (n = 85), of which the majority (n = 71) was reported by women. Thereby, examining this effect with a higher number of cases may be valuable, especially for female samples in which sexual activity without orgasm is particularly frequent, compared with male samples.

Due to the purely observational nature of the present study, future research might benefit from investigating the relationship between partnered and solo sexual activity, genital responses, the endocrine processes possibly underlying the effect sex has on sleep, and the psychophysiological markers of sleep in an experimental setting, thereby furthering what has only been done by Brisette et al. (1985). Another pathway could be to use a multi-modal machine learning approach implementing wearable devices to detect whether subjective and objective relations of sex and sleep patterns correlate. Self-monitoring the beneficial effect of sexual activity on sleep may have positive psychological effects and promote self-awareness.

Future research may consider differences between the effects in hetero- and homosexual couples and non-binary individuals, which were underrepresented in the present sample, as well as the effects different types of sexual activity have on sleep. Further, it is important to highlight that, by reducing sexual activity to partnered sex and masturbation, the present study applied a comprehensive – but restricted – definition of sexual activity. This served the extension of previous research in order to establish an underlying relationship between sexual activity and sleep. For future research, it would be important to apply a more inclusive, integrative conceptualisation of sexual activity by including a wide variety of sexual practices. Moreover, circumstantial factors such as having a new-born or small children, which possibly require frequent night-time engagement, may also be considered in future work, as both sex and sleep endure significant challenges and changes following the birth of a child (Kahn et al., 2022).

4.2 Strengths and implications

As the first to build upon previously conducted cross-sectional studies while also including a longitudinal design, the present study corroborated and extended the evidence for a sleep-promoting effect of sexual activity on sleep. By conducting an analysis in which the data are not aggregated but analysed with respect to their nested structure using MLM, the present study offered the opportunity to clarify diverging results regarding gender differences, type of sexual activity (masturbation vs. partnered sex), and the role of orgasm appraised by prior research. Moreover, controlling for relevant covariates, especially alcohol consumption which appeared to obscure the relationship between sexual activity and sleep, was valuable in the present study and is recommended for future research. The 14-day duration of the diary study, that includes weekdays as well as weekends, demands increased commitment of participants and further increases the value of inferences.

By using a cross-sectional design resembling the study conducted by Pallesen et al. (2020), their main findings could be replicated. The present study shows that both men and women perceive sexual activity followed by orgasm to reduce sleep latency and increase sleep quality (Gallop Jr. et al., 2021; Lastella et al., 2019; Pallesen et al., 2020). The results of the diary study corroborate the finding that sexual activity improves sleep while highlighting the effect partnered sex has on sleep, compared with masturbation. The heightened effect of partnered sex may partly be explained by the increased neuroendocrine changes following intercourse-induced orgasm, in combination with the valuable effects of experiencing intimacy with one's partner. Penetration and sexual intercourse aligned to heteronormative scripts may not necessarily be required to experience the beneficial effects of sexual activity on sleep. This notion is supported by the borderline-significant effect of partnered sex without orgasm on sleep quality, which is frequently reported by women and shows that intimacy alone may be sufficient to experience positive effects on sleep.

The present discordance of results between the cross-sectional study measuring the perception of the effects and the longitudinal study measuring the actual experience underlines the importance of applying objective measures and prospective measures appraising the perceived effect to the concepts of interest, as the subjective experience of sleep was shown to be a strong predictor of physical and mental well-being and cross-sectional methods are prone to be influenced by expectations and norms surrounding sexuality. In general, the same heteronormative implications of sexuality that underlie the orgasm gap between women and men may influence conceptions about “normal” sexuality – thereby resulting in confounded popular notions, such as men falling asleep first following sexual intercourse. Therefore, culture-specific norms and beliefs surrounding sexuality warrant consideration when interpreting the results of subjective research on sexuality.

The outcomes of the present research have important implications for sleep- and sexual medicine, as they highlight the value of considering partnered sex, masturbation, orgasm, and intimacy as a means to promote good sleep. The establishment of a relationship between sexual activity and sleep serves as a directive for future research to identify possible underlying mechanisms, such as endocrine or social-psychological processes, and to attempt an establishment of the effect using objective measures.

What if Diversity Trainings Are Doing More Harm Than Good?

What if Diversity Trainings Are Doing More Harm Than Good? Jesse Singal. The New York Times, Jan 17 2023. https://www.nytimes.com/2023/01/17/opinion/dei-trainings-effective.html

Diversity trainings have been around for decades, long before the country’s latest round of racial reckoning. But after George Floyd’s murder — as companies faced pressure to demonstrate a commitment to racial justice — interest in the diversity, equity and inclusion (D.E.I.) industry exploded. The American market reached an estimated $3.4 billion in 2020.

D.E.I. trainings are designed to help organizations become more welcoming to members of traditionally marginalized groups. Advocates make bold promises: Diversity workshops can foster better intergroup relations, improve the retention of minority employees, close recruitment gaps and so on. The only problem? There’s little evidence that many of these initiatives work. And the specific type of diversity training that is currently in vogue — mandatory trainings that blame dominant groups for D.E.I. problems — may well have a net-negative effect on the outcomes managers claim to care about.

Over the years, social scientists who have conducted careful reviews of the evidence base for diversity trainings have frequently come to discouraging conclusions. Though diversity trainings have been around in one form or another since at least the 1960s, few of them are ever subjected to rigorous evaluation, and those that are mostly appear to have little or no positive long-term effects. The lack of evidence is “disappointing,” wrote Elizabeth Levy Paluck of Princeton and her co-authors in a 2021 Annual Review of Psychology article, “considering the frequency with which calls for diversity training emerge in the wake of widely publicized instances of discriminatory conduct.”

Dr. Paluck’s team found just two large experimental studies in the previous decade that attempted to evaluate the effects of diversity trainings and met basic quality benchmarks. Other researchers have been similarly unimpressed. “We have been speaking to employers about this research for more than a decade,” wrote the sociologists Frank Dobbin and Alexandra Kalev in 2018, “with the message that diversity training is likely the most expensive, and least effective, diversity program around.” (To be fair, not all of these critiques apply as sharply to voluntary diversity trainings.)

If diversity trainings have no impact whatsoever, that would mean that perhaps billions of dollars are being wasted annually in the United States on these efforts. But there’s a darker possibility: Some diversity initiatives might actually worsen the D.E.I. climates of the organizations that pay for them.

That’s partly because any psychological intervention may turn out to do more harm than good. The late psychologist Scott Lilienfeld made this point in an influential 2007 article [https://journals.sagepub.com/doi/10.1111/j.1745-6916.2007.00029.x] where he argued that certain interventions — including ones geared at fighting youth substance use, youth delinquency and PTSD — likely fell into that category. In the case of D.E.I., Dr. Dobbin and Dr. Kalev warn that diversity trainings that are mandatory, or that threaten dominant groups’ sense of belonging or make them feel blamed, may elicit negative backlash or exacerbate pre-existing biases.

Many popular contemporary D.E.I. approaches meet these criteria. They often seem geared more toward sparking a revolutionary re-understanding of race relations than solving organizations’ specific problems. And they often blame white people — or their culture — for harming people of color. For example, the activist Tema Okun’s work cites concepts like “objectivity” and “worship of the written word” as characteristics of “white supremacy culture.” Robin DiAngelo’s “white fragility” trainings are intentionally designed to make white participants uncomfortable. And microaggression trainings are based on an area of academic literature that claims, without quality evidence [https://www.thecut.com/2017/01/how-microaggression-training-could-harm-minority-students.html], that common utterances like “America is a melting pot” harm the mental health of people of color. Many of these trainings run counter to the views of most Americans — of any color — on race and equality. And they’re generating exactly the sort of backlash that research predicts.

Just ask employees at the Smithsonian’s National Museum of African American History and Culture, which had to issue an apology after it posted an Okunesque graphic that presented rational thought, hard work and “emphasis on scientific method” as attributes of “white culture.”

Then there are the lawsuits. As The New York Times Magazine noted in 2020, at least half a dozen people who had been employed by the New York City Department of Education filed lawsuits or won settlements in cases relating to mandatory D.E.I. trainings. Racial affinity groups, a popular intervention in which participants are temporarily separated by race so they can talk about, well, race, have perhaps proved even more problematic. They’ve sparked complaints in places like Jacksonville, Fla. (where a principal was temporarily reassigned after she attempted to separate white students from students of color to discuss “cultural issues”), and Wellesley, Mass. (where the creation of racial affinity groups for students provoked a now-settled lawsuit from a conservative group).

Not every complaint is valid, not every lawsuit has merit and backlash to conversations about racial justice is nothing new. Martin Luther King Jr. had an unfavorable rating of 63 percent before his assassination. If common diversity trainings definitively made institutions fairer or more inclusive in measurable ways, then one could argue they are worth it, backlash and mounting legal fees notwithstanding. But there’s little evidence that they do.

So what does work? Robert Livingston, a lecturer at the Harvard Kennedy School who works as both a bias researcher and a diversity consultant, has a simple proposal: “Focus on actions and behaviors rather than hearts and minds.”

Dr. Livingston suggests that it’s more important to accurately diagnose an organization’s specific problems with D.E.I. and to come up with concrete strategies for solving them than it is to attempt to change the attitudes of individual employees. And D.E.I. challenges vary widely from organization to organization: Sometimes the problem has to do with the relationship between white and nonwhite employees, sometimes it has to do with the recruitment or retention of new employees and sometimes it has to do with disparate treatment of customers (think of Black patients prescribed less pain medication than white ones).

The legwork it takes to actually understand and solve these problems isn’t necessarily glamorous. If you want more Black and Latino people in management roles at your large company, that might require gathering data on what percentage of applicants come from these groups, interviewing current Black and Latino managers on whether there are climate issues that could be contributing to the problem and possibly beefing up recruitment efforts at, say, business schools with high percentages of Black and Latino graduates. Even solving this one problem — and it’s a fairly common one — could take hundreds of hours of labor.

The truth, as Dr. Livingston pointed out, is that not every organization is up to this sort of task. Ticking a box and moving on can be the more attractive option. “Some organizations want to do window dressing,” he said. “And if so, then, OK, bring in a white fragility workshop and know you’ve accomplished your goal.”

The history of diversity trainings is, in a sense, a history of fads. Maybe the current crop will wither over time, new ones will sprout that are stunted by the same lack of evidence, and a decade from now someone else will write a version of this article. But it’s also possible that organizations will grow tired of throwing time and money at trainings where the upside is mostly theoretical and the potential downsides include unhappy employees, public embarrassment and even lawsuits. It’s possible they will realize that a true commitment to D.E.I. does not lend itself to easy solutions.

Jesse Singal (@jessesingal) is the author of “The Quick Fix: Why Fad Psychology Can’t Cure Our Social Ills” and a co-host of the podcast “Blocked and Reported.” He writes the newsletter Singal-Minded.