Are one third of Australian men emotionally abusive? We don't really know

The Australian Institute of Family Studies recently put out a press release:

One in three men report using intimate partner violence

Understandably, this report got wide uptake by politicians and media figures:

The problems begin

In the release itself, we find:

“Emotional-type abuse was the most common form of intimate partner violence, with 32% of men in 2022 reporting they ever made an intimate partner feel ‘frightened or anxious’, with 9% ever ‘hitting, slapping, kicking or otherwise physically hurting’ an intimate partner when they were angry.”

The wording: “ever made an intimate partner feel ‘frightened or anxious’” didn’t seem right. But this was only the media release, so I decided to look at the question wording in the survey itself:

Have you ever behaved in a manner that has made a partner feel frightened or anxious?

(Some people online wrongly thought that the reader added the words “emotional abuse” after this, but this is a mistake- that was just a dataset tag.)

At face value, this question is not a good measure of emotional abuse. Are people taking this question at face value, or inferring from context that the framers are talking about emotional abuse, or something in between? This is unclear.

But suppose you do take it at face value, or semi-face value. I’d posit that if you’ve never done anything that made your partner anxious, something might be wrong with your relationship- either you’re letting your partners will control your own far too much, or your partner doesn’t care about you enough. I don’t currently have a partner, but I am a neurotic wreck, and I get anxious about my friends:

1. Smoking.

2. Drinking to drunkenness even at a party.

3. Criticising me- even in honest and completely appropriate ways.

4. Leaving me on seen- even though it’s completely normal.

5. Going out with destructive, erratic people.

6. Not taking care of their health.

7. Making career decisions I wouldn’t make in their place

Criticism is particularly germane here. If you love someone, there is a good chance you will feel a spike of anxiety if they criticise you. However, sometimes it is important to criticise- even firmly- our friends, relations and lovers, e.g., to help them grow, or to assert our boundaries.

Now you could say: “Come on, Philosophy Bear, get with the program, they clearly don’t mean any of that.” But when writing surveys, you can’t assume the audience will be with the program, and even as an extremely educated reader, I would not know quite what to make of the question when encountering it.

So I think the question gives us minimal evidence on the real rate of emotional abuse. We have reason to think it might be an overreport (see above) and reason to think it might be an underreport (people lie, both to surveys and to themselves, about their past behaviour).

In fairness, the drafters had an impossible task. Capturing emotional abuse in a question that doesn’t require savvy respondents is not easy. If I were drafting the question, I would have: A) Pulled my hair out and gotten mad at myself, then B) Written out the question they already have, added the word “deliberately,” and called it a day.

Framing things in terms of deliberateness is not without cost- intention is a very shaky concept even for a lawyer or philosopher, and most people aren’t lawyers or philosophers. Also, it still doesn’t capture what is meant, since sometimes we have a right to make others feel anxious deliberately, at least in some sense of “deliberate”. I might know that quitting my job with a shitty boss who screams at me will make my partner anxious, and do it despite that knowledge. If we use the foreknowledge test of deliberateness, I have then deliberately made my partner anxious. Still, on balance, I think “deliberately” would have been an improvement over the existing wording.

Part of the problem here is that emotional abuse is inherently a matter of degree. It has to cross a certain threshold to count as abuse. Above, I gave the example of getting anxious about my friends drinking too much, this is clearly not emotional abuse. Suppose, though, that I had a friend who drinks enormous amounts of alcohol specifically to make his friends and family worried that he might die- that would be abusive [and a sign of pretty serious mental illness]. Likewise, criticism isn’t abuse, but criticising someone numerous times a day, every day, can be, even if the individual criticisms considered in isolation are fair enough. Frequency, premeditation, will-to-harm- all of these factors should be taken into its evaluation, and so I suspect no single survey question can ever measure it. To decide whether or not something is emotional abuse requires judgment, and even if we trust respondents to try their best to be honest (dubious), it is doubtful we can trust their judgment about their own case.

So I don’t think AIFS avoidably stuffed up the survey design- I’m not sure there was a much better question available. However, in light of these difficulties, their press release should have been much more circumspect.

Interlude: We need to ask for exactly what we want

I think this is linked to a more general problem. There’s a dance where people set absurdly high bars in public announcements about avoiding gender, sexual, and domestic violence because they want to catch all the bad behaviours, but hope people will understand what’s really being said- and take it seriously, not literally. This is a bad way to explain rules, as I learnt to my detriment:

Guy: So, do you want to have sex?

Philosophy Bear: Yeah, sure.

Guy: You don’t sound enthusiastic

Philosophy Bear: Let’s have sex

Guy: But you’re not enthusiastic

We talked it over. It emerged that he’d taken the concept of “enthusiastic consent” seriously, and so didn’t regard a clear, but not excited, “yes” as a “yes”. I genuinely did want to have sex, but I can come across as a bit blasé about things. By the time we figured it out, the moment was gone. It sounds like a skit, but it happened, and I was annoyed about it.

Some matters unavoidably require judgment, but as far as possible, we shouldn’t expect, assume, or rely upon people to pick up on subtext or complexity. Equally, we shouldn’t ask people to do things we don’t really expect or want them to do.

An oddity in the provenance of the survey questions

I’m really not sure what’s going on with the origin of these questions.

I note that the Ten to Men Insights #3 Report: Chapter 1 report states:

The questions used in Ten to Men were adapted from the COHSAR (Comparing Heterosexual and Same Sex Abuse in Relationships; Donovan et al., 2005) measure. At the time of developing the Wave 1 survey, for 2013–14, the COHSAR was one of the few measures of intimate partner violence that had been validated among a sample of men (Donovan et al., 2005).

But when we go to the paper that describes the development of the instrument (Donovan et al. 2006), we find:

In order to provide a detailed picture of same sex domestic abuse, while at the same time being able to compare same sex and heterosexual experiences of such abuse, the research used a multi-method approach, involving:

1. A UK-wide survey of domestic abuse in same sex relationships (total respondents=800, with 746 usable questionnaires).

2. Five focus groups with lesbians, gay men and heterosexual women and men of different ages and ethnicities to examine perceptions of love and domestic abuse (total of 21 individuals).

3. Semi-structured interviews with 67 individuals identifying as lesbian (19), gay male (19), heterosexual (14 women, 9 men), bisexual (3) or queer (3), to compare experiences and meanings regarding violence and abuse in adult relationships.

In other words, only a handful of heterosexual men were involved in the development of the instrument, possibly as few as a bit more than 10. Now I don’t think that particularly matters, but I do find it very odd that “the COHSAR was one of the few measures of intimate partner violence that had been validated among a sample of men” is being cited as a reason to use the measure, in light of that. A sample of men almost exclusively in same sex relationships is very different to a sample of men generally with respect to IPV.

Even more seriously, the COHSAR is a measure of being a victim/survivor of domestic violence; it is not designed to measure offending.

But most seriously of all, while the authors claim that they adapted the questions from the COHSAR, there is no question like: Has your partner ever behaved in a manner that has made a partner feel frightened or anxious? In the COHSAR. There just isn’t. There’s a question that measures the impact of abuse if it has occurred that reads: “Felt anxious/panic/lost concentration”, but that’s it.

So, at least with regards to this question 1. It isn’t in the COHSAR 2. Even if it was in the COHSAR, it would be a question asking if the respondent had experienced it, not if they had perpetrated it, and 3. Even if it were in the COHSAR, it would have been validated almost exclusively with men in same sex couples.

The politics of big numbers versus small numbers

Finally, what’s the import for action of a bigger or smaller number? Other than pedantry, why even bother to debate this?

Our first aim here should be the truth. We should not decide what the rate of emotionally abusive men is based on political convenience, because if we are dishonest in that sort of way, we will never win.

Nevertheless, I think there is a sense in some of the public discourse about this that wants to defend results indicating that a larger number of men are abusive because they think it will be a wake-up call or a positive contribution to the discourse. I don’t think that’s right.

But I think it goes the other way- I think it is politically advantageous if the number is smaller. This is not a reason to pretend the number is smaller- we need to look for the truth. However, it does put the issue in perspective.

Why think lower numbers are better for political action? Remember:

1. A hard nosed view of the world where there are evil sickos everywhere has never favoured the left, and has frequently been used as a cudgel against feminism. Optimism about the (bounded) decency of humanity has tended to drive a lot more progress than cynicism.

2. At a certain point, if we conceptualise wrongdoing in a way that makes it extremely common for people to perpetrate it, people start to see it as less serious than it is. We don’t see common villainy as true villainy.

There’s another political angle here. AIFS released this infographic:

I am extremely concerned about these claims in particular because it seems to me that, precisely due to their depression, depressed men may be more likely to agree with statements like: Have you ever behaved in a manner that has made a partner feel frightened or anxious? Now, it certainly could be that depressed men are more likely to engage in IPV, but equally, it could reflect a tendency to widely and in a literal manner accept such negative self-ascriptions. The last thing we want to do is increase the stigma around depression.