Shonky Research

My feeling after having read the report is that when it was ready for the printer, the authors had in fact reached the point where they were about ready to consult with people experienced in such research, as a necessary preliminary to the main investigation. I would have suggested a smaller pilot sample. This should have disclosed the pitfalls that lay in wait for them. By taking such steps they could have avoided the traps that they later fell into.

In my view the study was very seriously, if not fatally, flawed by choosing either defective definitions, or no definition at all. Poor definitions results in banging the table and knifing the wife being placed in the same continuum of physical abuse. Absence of definition seems apparent where they compared their results with results from other studies. There is no guarantee that apples were being compared with apples, and not pumpkins. Similarly, to continue the metaphor, they studied only half the apple — no comparable study of female on male abuse/violence/insult a great many of the comparisons and conclusions were inappropriately drawn, though often there can be no certainty about this.

I found much of the writing rather confused, so that I cannot interpret it with certainty — e.g. p.35

“It has been shown that when asked direct, general questions about abuse, people [?men] say that they disapprove, but when the questions become less direct and more specific [sic!], a kind of underlying condoning begins to emerge.”

What does that passage mean? It is not highly charged, but the meaning is obscure to me.

There also appears to have been an unwillingness to face possible outcomes squarely, so rather than risk discovering that abuse could have a racial component [ethnic is only Greek for race], the possibility of having to face such data was excluded from the outset. It was neither honest nor wise. I found much the same problem with the discussion of education and income levels etc. as potentially associated factors in abuse quite unconvincing, and in fact very confused because different results appeared to be reported on different pages.

I thought it unfortunate that the authors discounted a valuable notion from an Australian study (p 61) of “justifiable” abuse and “okay actions“. There seems to be no recognition that different circumstances may justifiably alter perspectives. For example, it could well be justifiable to use some force to avoid a greater evil, such as real harm to a child. Restraint of a spouse intent on mayhem with a French cook’s knife could well be necessary rather than merely justifiable or okay. I found it unreal that there was no description of the circumstances of the questioned “abuse” or “violence”.

It seems to me that the researchers were just not up to the task. This showed in a variety of ways. They had no hesitation in citing data on non-molestation orders, but failed to note that most of these are granted on ex parte actions, where husbands have no chance to challenge the initial application. How many are struck out after the husbands are heard?

Again, on p. 25 I read “…nearly all research has looked at the abuse of women by men, presumably because it is regarded as the most common form of domestic abuse.” Is that presumption justified? Fide Newbold, in the US where handguns are freely available, the likelihood of being shot by a spouse is about equal for both sexes! What price this pre-eminence of male on female violence in the absence of any data?

Perhaps the most surprising outcome of this study, I deliberately refrain from dignifying it with the name of “research”, is that the authors do not make an urgent demand for the complementary sexually directed study.

The subject is of real importance. The investigation deserves to be carried out in a workmanlike manner. Sadly this was not competent.

If the Bent Spoon is for shonky research, which is more likely to mislead than to illumine future action, there can be no doubt that Hitting Home richly deserved such categorisation. It was ill-considered and of no value in the pursuit of reliable information upon which to categorise a problem, and to plan remedial action.

The Skeptics should not, in my view, change the award of the Bent Spoon. It seems that the investigators knew the results they wanted, and set out to get them. Their reportage achieved that end.

Walter C. Clark, Woodend

Missed Opportunities

The Bent Spoon Award going to the authors of Hitting Home is what attracted myself and others to the conference. I was very impressed at the quick response of a group to this report after the lack of response to misleading studies done on sexual abuse.

I am concerned that changing the award at this time would not only give credence to this sloppy piece of research but would diminish the credibility of the New Zealand Skeptics.

I see little in value in reiterating the point about adding apples and oranges, that, Denis Dutton has done very well. What concerns me as much as the misleading statistics is that the authors of the report missed opportunities to find some of the causes of spousal abuse.

Firstly, there was no distinction between de facto and legally married couples. The incidence of sexual abuse of children is higher when they are living in a de facto household. It is likely that the cases of extreme physical abuse would also be higher in de facto relationships. I believe that the question was left out deliberately as the result would not suit the agenda of the authors of the report.

Secondly, the men should have been asked if they were victims of the various categories of abuse. It would have been interesting to see the correlation between those abusing and those receiving abuse. Overseas studies for multiple forms of violence between intimate partners indicate that men and women do use about the same amount of violence in relationships. Please note enclosed list of references.

In short, the authors asked questions that would give the desired answer. Namely that men are primarily responsible for domestic violence.

The fact is that this report is clearly flawed. It made no attempt to take a scientific approach to a serious social problem which can only lead to further polarisation of the sexes. This is why the executive of the Skeptics collectively awarded the Bent Spoon to the report.

Even if it could be demonstrated that some other publication is more deserving of the Bent Spoon, it is simply too late.

Firstly, that would be taken as endorsement of the report. It would have been better not to have even commented on the report in the first place.

Secondly, and more important for the long term future of the Skeptics, future awards would only go to publications that would not upset members of the Skeptics.

Next year ASH might get the award for using flawed data in regard to the effects of passive smoking. I personally support ASH, although I am not a member. However, I would not support ASH, Greenpeace, or any other organisation deliberately or even accidentally distorting the facts in the presentation of a study.

Perhaps in future awards could be selected in a different manner. A short list could be published in a newsletter. Members would then have an opportunity to express their view. A postal vote could possibly be arranged.

These matters are very much secondary to the question of giving the Bent Spoon Award to someone else. In hindsight the “Hitting Home” report might not have been the best choice. I and many others thought it was. To give the award to someone else now, would permanently affect all future award choices and no doubt restrict the topics printed in the newsletters.

Chuck Bird, Auckland

Bent Spoon Valid

At the Auckland confab when the dispute over Hitting Home burst over the AGM, I felt that there was little that I could contribute since I had no idea of the content of the report and only recollections of other information, so rather than add ad hominem remarks to an argument that was already involving strong emotions, I stayed quiet. But now that I have read the report, I feel that the award of the Bent Spoon was valid and deserved. Whether it was the most deserving of the possible candidates I can’t say, as I watch very little TV, rarely listen to the radio (when chancing on a talk-back show while in search of music, I’m tempted to flee to a cave and ignore any shadows on the walls), and don’t read the infamous Women’s Weekly.

To be worthy of a Bent Spoon, or criticism generally, the candidate should be important enough to deserve attention, be a stupid or bungled or confused effort by people who should know better, and worst, be wilfully misleading with intent to profit. Although the report contains work by persons and organisations that has been properly done, nevertheless the report is dishonest. Honest research requires conclusions that best represent the evidence collected, and the evidence collected is that which most directly bears upon the issues of interest. (Spare me remarks about circles.)

The authors fail to follow principle. They mention discussions over the form of the surveys yet give no details, the only time they access issues of internal consistency is when they consider whether the sex of the interviewer had any affect (it didn’t). All they have to say about their decision to study only man’s abuse of only women, and only by interviewing men is that it breaks new ground. They make brief mentions of other research (providing that it can be found in computer catalogues) but no attempt to cross reference men’s self-assessment with their wives’ opinion, although one can easily imagine difficulties.

The authors do not seem to have noticed that they are all three women, and that going by the names in the acknowledgement, nearly everyone else in the Justice Department involved with the project is likewise a woman. Yet there is a hint that they could have noticed when comparing their results with those of other groups: strangely enough, the group that notes the highest prevalence of men’s abuse of women is the Women’s Shelter (p 87).

Another comparison is just brushed aside. Their sample group has a different pattern of income from that of the nearest NZ census. This is due to inflation, they suggest, except that inflation has been relatively low. What is less often admitted is that lower incomes have been falling, while high incomes have risen still higher. But this is a quibble.

What I regard as the most serious failing is the flaw in the basic data collection in that the questions are ambiguous. They wish to investigate men’s abuse of women, which I take to mean unjustifiable acts, since obviously, if there was a good reason for doing something, then it could not be regarded as abuse even if nasty because its motive was not a desire to inflict nastiness on a woman but something else again. To give a specific example, suppose that mum was chastising a child and dad shoves her away. (Ah, but what if the child had been hitting the cat, for catching a bird.) Whatever the complications, surely this is quite different from him shoving her aside as he walks by, just to show who’s boss, and the difference in the quality of the act bears directly on the question of abuse and abusive attitudes. This seems to me to be so obvious that it must have come up in discussion, yet the authors say nothing on the issue

Even if it didn’t occur to the researchers it must be likely that some at least of the two thousand interviewees had similar thoughts and responded accordingly to their question “not OK in any circumstances”, indeed some may even have lived such scenarios. So of what use is the remark “Not a single behaviour, even using a knife or gun, was judged as unacceptable by all New Zealand men” (p 144). Instead of judging their attitudes to women, it may be their imagination that is being put to the test.

Yet the authors remain oblivious to this issue when on page 61 they compare some Australian research that used the word “Justifiable” rather than “OK” in some circumstances. They merely recite the differences, with no sign of thought as to possible reasons. Thus, 7% thought that lethal violence was possibly justifiable (Oz) but only 1% possibly OK (NZ). Could this not suggest a difference in the meaning of the two words, as exemplified by the arguments over just wars even though all agree that wars are not OK? Data is however data, and fascinating patterns await your notice, as the comparison shows. As the severity of the act increases (shoves, slaps, throws object at, assaults), the Oz figures run 15,14,10,7, while for NZ 19,11,8,1. What might this mean, if anything? But instead, silence. The authors simply view women as irreproachable in all circumstances so that irrespective of motive, any act against them is fully and purely abuse, by definition.

Much of this could have been avoided by the simple means of instructing the interviewers to exclude thoughts of self-defence or protection of third parties. What is at issue is men’s attitudes towards women as revealed by their own choice between alternatives, not as driven by some factor external to this relationship. You could argue that some men might decline an act on absolute principle, whereas others might require severe provocation, some less and another group no provocation at all, and that this does indeed reflect their attitudes to women, but it also reflects attitudes to humanity at large. Few people are Jains.

Thus the questions asked, especially those concerning physical acts, fail to address the research issue square on and allow needless confusions that could have been excluded.

The authors’ treatment of the results they do obtain is grotesque. They are determined to push them into showing that abuse is common and serious amongst all men to such a degree that I feel that this was their fixed view from the beginning, and they ignore all interpretations other than the one they want upheld. This requires some crude steps.

They list eleven acts of physical abuse that can be ranked by the seriousness of the likely physical injury. For acts of psychological abuse, ranking is more a subjective matter but nevertheless the number of persons accepting such acts in some circumstances has the same strongly skewed distribution, from 19% down to less than 1% and 33% to 1% respectively. Well and good. But the factor they choose for analysis is “Number of types accepted”. This is lunatic, for it equates “uses or threatens with weapon” (two counts) with “shoves or slaps”, also two counts. No explanation is given as to why this choice was made or others rejected, instead they leap at once to shout (in bold type, p60) “one in four said yes to at least one type of physical abuse” and also “six in ten…psychological”, plus further drivel about the average number accepted and standard deviations.

Lunatic is the wrong word, and stupidity is out of place. This is deliberate misrepresentation, and I don’t mean that they got their arithmetic wrong, or that the interviewees didn’t make those responses. There is the story of the beggar with a sign reading “Wars 2, Legs 1, Wives 2, Children 4, Wounds 2; Total 11”. No doubt if a survey was made, statistics on this factor could be generated. What should have been used is “Worst type approved”. There may have been some thought of this as it is noted that if a particular abuse is approved, lesser ranked acts are likely to be approved also, but although the approval rates vary by a factor of about thirty with a clear connection to severity, this is ignored in favour of shouting (in bold) “A significant number say that physical and psychological abuse is OK in some circumstances” without noticing that 75% deny approval to any physical act, and 42% of psychological acts (such as harsh words), even as the authors have defined them.

It could be argued that reducing a wide range of acts to a single severity scale and scoring men’s views through their acceptance of various acts with a possible further weighting according to frequency is a hopeless task and to be avoided. Yet exactly that happens in courts when a judge decides on the number of years of imprisonment to impose, a practice with millennia of history, and operated by personnel in a not too distant department.

Yes there would be problems (which haven’t dissuaded the IQ testers!) and there would be arguments over rankings, but Weight of Abuse Approved or Worst Type Approved are at least attempting to assess abusiveness, unlike Number of Types, which involves declaring that all acts have equal weight, an absurdity built in from the beginning and which flies in the face of the prevalence data that by their existence demonstrate that everyone else views murder as more serious than shoving, rather than both being worth one demerit each.

One act, hitting, is considered in various contexts, and on p 65 there is an interesting list of circumstances ordered by the proportion of men who apportion blame to the woman being hit. At the top with 48% is “He catches her in bed with another man” (I suppose that everyone interprets this common euphemism the same way) down to 1% for “He can’t find a job”. As before, any thought about the possible meaning of this ordering or what men might be thinking is not mentioned. Instead, all are immediately equivalenced so that we have another Number of Circumstances report thereby enabling a leap to the obvious and only conclusion. Combining the trivial few percent who assigned blame to neither allows the authors to say “In one circumstance half the men say the man is not at all responsible”. Yes, but it is also true that the other half think otherwise. The next paragraph notes some acceptance of blame by men but ends with “there are no circumstances in which every man says that the man alone is responsible”.

The authors simply deem an act abuse and therefore wrong (which is anyway what the word means) no matter what motive lies behind the act. There are no distinctions between malicious hurt, wrong, justifiable, or what else do you expect? All acts are solely expressions of a wish to abuse. There seems to be no recognition of the fact that a relationship involves two people in close interaction. Men are assessed by an impossible standard whereas women are held to no standard at all for they are perfect in act and thought.

The conclusion states that men must accept responsibility for their actions; well enough, so also must women. Women must know that being caught in bed with another man is likely to provoke extravagant behaviour, as is endlessly reported in the news media, depicted in films and plays, and found in novels. In fact they do, and are not above flirting with someone else in a pub precisely so as to stir up the boyfriend (who is expected to go for the other bloke!). He of course should remain calm (dear friend, let us reason together) but may well not. If she abhors violence, she has a free choice not to precipitate it but sometimes prefers drama and risk. One can imagine circumstances when infidelity is his fault (he is boring, infertile, inadequate, obnoxious, unfaithful…) but her choices have their likely outcomes. Human behaviour is not often driven by one factor acting in isolation.

But the authors admit no complications. Scrutiny of their list of abusive actions shows what a surprised husband can’t do without being declared abusive. Violence is out, so are threats. Destroying something belonging to her is displaced violence, and abusive. So are words as they will almost certainly involve putting down family and friends (especially the close friend), and anyway, he is seeking to stop her from doing something she wanted to do and so is abusing her.

It appears that he must simply apologise for the intrusion, pack and leave. Only that would be approved of by the authors, who note on p 28 that 47% of the 193 female homicide victims in NZ (78-87) were killed by an existing or former male partner. Well, no-one approves of murder, but what has happened to the statistics on the other obvious combinations of who kills whom? Is it even surprising, considering that few people would have strong feelings about strangers?

In short, the ideals of research have not been met. The information that has been collected is of poor quality, and its analysis deformed. The conclusions offered were predetermined, and other possible conclusions ignored. Indeed, I doubt that the authors have a good understanding of statistical analysis, given that on p 181 they complain that their analysis computer programme allows only fifteen predictor variables for a logistic regression, and that they regularly used fifty predictor variables in their linear regressions. (See F. S. Acton’s Numerical Methods That Work, the section “What not to compute”.)

And while this “research” and argument continues, unambiguous abuse that is unambiguously serious also continues.


I hope that no-one imagines that I deny that there is a problem, or that things could be better. I am saying that this report isn’t going to help. I hope that we can stay clear of escalating displays of superiority in caringness, concernedness, and righteousness.

I agree that the procedure for selecting the recipient of the Bent Spoon award involves four persons other than the Executive so that our collective responsibility falls rather heavily on Dr Dutton, but I don’t see that there is much of an alternative for so scattered a group as the NZ Skeptics, especially when the members of the Executive happen to live in the same city. I was as usual surprised by the choice of recipient, but only in the sense that I hadn’t heard of it beforehand. I regard the award as appropriate, and retain confidence in the ability of the Executive to select the worthy in the future.

Nicky McLean Lower Hutt

Recommended Posts