Thursday, May 10, 2007

S 59 Debate: Replying to Anon

My April 25 post has launched what I take to be an interesting exchange between myself and one Anon. Anon's first comment is here. Anon's remarks are both penetrating and very well-expressed. I want to thank her for that, and I urge anyone who's still interested in thinking clearly about this issue to read what she has to say there.
This post is just my latest reply to Anon - one that grew and seemed silly left as a comment. I include all of Anon's most recent comment in green type and in order.
You state that, basically, there is no evidence that you know of that smacking causes positive outcomes for children but there is a bit of evidence that there is a correlation between smacking and negative things. Therefore, wouldn't the most appropriate thing to have happen be to ban smacking?
No. For three reasons.

1. Much more evidence is needed on the negative side. The (plausibly) causally interpretable negative stuff is relatively limited at this point: just a single study by Grogan-Kaylor (G-K). At the very least one would want to see that study followed up with more nuanced anti-social-behavior-score (ASBS) dependent variables that clearly separated inside and outside the home behaviors. Moreover, G-K's study is limited in ways that any one study tends to be. Consider that in the overall population from which G-K gets his sample data
the ASBSs are normed to a mean of 100 and a standard deviation of 15. G-K's basic result from his sample is that (controlling for an impressive amount of stuff) smacking boosts ASBSs by 2-3 points on this scale - i.e., by roughly a 1/4 of a standard deviation. That's then a conclusion he wants us to infer as holding about the whole population from which his sample is drawn and beyond that to kids everywhere. But then we want/need to know how representative his sample of 1800 kids is. Well, his sample had initial mean ASBS of 106.9. That's a little odd. First, the sample's skew from the population norm is double the final difference that's identified as smacking-related and then claimed to be inferable back to the whole population. I think that at the very least this confirms that the effect that G-K identified can't be particularly serious. Insofar as half-a-SD change in mean ASBS doesn't destroy representativeness of the sample to that extent a boost of 1/4-of-a-SD keeps one well within the normal variation range by the study's own lights. In this way, at least on one level, G-K's study partially undermines itself as a basis for any recommendations.
Second, and addressing representativeness directly, the standard error of the mean in a sample of 1800 is SD(population)/sq.root(1800) = .353. That means 95% of (N=1800) sample means should be in the range 100±.71 i.e., within two standard errors of the population mean. G-K's sample has a mean that's almost 20 standard errors away from the population mean - it's a hugely improbable, 1-in-a-bazillion, unrepresentative sample. That can be OK. There may be good reasons for having such an exceedingly un-random sample but it does mean that the otherwise very respectable sample size of 1800 doesn't mean what it would if the sample was random. Some quick calculations show that for G-K's sample mean of 106.9 to be within two standard errors of the population mean his sample would have to have roughly N=20. Wow. Does this sort of technical point matter? I'm not sure. I don't like it, but I'm hardly impartial. I do think, however, that this is the sort of technical point that usually only really gets settled by further studies with more representative samples. Conclusion: Needs more study.
For another example, G-K's sample of 1800 kids was coded as 80% no corporal punishment (cp) and only 20% having different positive numbers of cp events in a particular week. Is that a problem for the study or not? I don't know, but it concerns me that, after you control for a few things (e.g., child sex, age, and initial ASBS and family per-capita-income) then the number of individual smacked kids that are being compared at each controlled-for point gets very small (assume just 4 control factors and just 10 levels for each of those and you're down to, on average, 9 smacked kids in each comparison class. And all of that's with setting aside our earlier worries about unrepresentativeness. (Once I've studied up on a few statistical software matters a bit more, I intend to ask G-K for his raw data so I can investigate this properly.) Again, the big conclusion has to be: Needs More Study.

And, look, 1990's Medicine was full of cases in which exactly G-K's analytical tools – so-called fixed effects models – were applied to broadly observational data and they suggested conclusions that were nonetheless completely reversed when proper randomized trials were run. E.g., people eating more fruits and vegetables, which are rich in ß carotene, and people having higher serum ß carotene concentrations were observed to have lower rates of cardiovascular disease and cancer. Fixed effects models then identified the ß carotene as the causally relevant factor. But when people went out and tested efficacy of ß carotene supplements the results were horrifying. In trials where participants randomised to ß carotene supplements were compared with those randomised to placebo, ß carotene supplementation moderately boosted risks of cardiovascular death and carcinogenesis (e.g, here). At the time, this sort of nasty surpise led to a lot of handwringing in clinical medicine and epidemiology about fixed effects models and meta-analyses more generally. Were they valuable as guides to what randomized trials to run in future ("Meta-analysis, schmeta-analysis" was the memorable title of one review paper!)? And, heaven forbid that they were an autonomous basis for drawing serious conclusions and a substitute for actually doing the randomized trials! But in social science where randomized trials are impossible or unethical or both, we're supposed to take our marching orders from such studies and go full steam ahead, no questions asked! It ticks me off that Anne Smith et al. down at U.Otago take this line, and that Cindy Kiro would just parrot it. Sorry, that's unconscionable and I can't be party to it. The only serious conclusion remains: Needs more study.

2. Positive side needs more time to develop. G-K's study only came out a couple of years ago, and it's fair to say that, although it's nowhere near bulletproof as I've just argued, it has raised the bar in the area (which should tell you something about how ineffectual most of the other research on the topic is - see my discussions here and here), and there now needs to be a period during which that research is consolidated/confirmed and alternatives explored and if possible excluded. It would be fairly shocking if it were true that there are no good effects from smacking under any circumstances. I'm prepared to believe that that's so... but not yet! I can't stress enough how narrow the base of casually interpretable negative results is... those results might be able to be importantly reversed just by finding another set of outcomes to check for/measure as the dependent variable. The fairly nasty social and personal trends that have accompanied the decline of smacking over the last 50 years suggests that outside-the-home might be the place to look for (now much weakened!) positive effects (at least in some sub-populations). But I admit it's hard to know even what to measure.... These are big research challenges, I think.

3. Don't Ban. Inform. Even if the negative effects case went from strength to strength and the positive effects case never panned out at all then unless the negative effects in question are huge, which they aren't as far as we know (see point 1 above), I tend to think that the right move is to just give parents the relevant data.... Compare: some parents doubtless don't feed their kids especially well... still one doesn't want to criminalize giving kids coke or chips etc.. (Although perhaps that's something Sue Kedgeley will want to explore! Dear God let's hope not! But illiberalism abhors a vacuum of freedom...) Or if divorce or sole parenthood has predictably mildly negative effects on kids.... well, while people should be told about any data to this effect, systematically penalizing divorcees or sole parents through family law, let alone criminalizing them seems completely inappropriate. The criminal law and probably law more generally just seems to be the wrong instrument with which to address the sort of arguable, localized sub-optimality that all manner of parenting and life decisions tend to involve. The good news from psychology that I referred to in my last comment is that kids are remarkably resilient and almost distressingly impervious to our efforts to influence and nurture them. Babies seem to develop roughly the same personalities, IQs etc. regardless of what you do or don't do at home, and of who (e.g., biological or adoptive parents) raises them. Ndugu in his mud hut ends up just as visually acute and as smart as genetically comparable little Susie from Mirimar with her room full of allegedly brain-expanding mobiles and pretty pictures. (Most of what you can do to influence your kid's development is chose the neighborhood/school/out-of-home environments in which he or she grows up. Susie's ultimate advantage over Ndugu will largely consist in this. Judith Rich Harris's books are invaluable sources of references on this sort of stuff.) That's part of what makes official equanimity about diversity of family arrangements possible I think. But then tolerance of different parenting styles - styles that are as different as jazz is from rock or as classical is from blue-grass - with attitudes to discipline and rules and duties and consequences being a major variable generating that diversity, should I believe be part of that.
You also state something along the lines of, 'perhaps we're both too self-conscious and reflective to ever be fully successful-convincing smackers'. This implies that there are parents out there that are less self-conscious and reflective than us, which there is. I am not of the opinion that I am excessively self-conscious or reflective but am just trying to do the best by my child. Therefore, do you not think that there should be a ban on smacking until the vast majority of parents here in New Zealand are more self-conscious and reflective?
Parents, and people more generally differ quite a lot, both in their personalities and in their situations. I'm a hyper-verbal, high-stress, lots-of-regrets sort of person (e.g., I'm always amazed when I hear people say they have no regrets about anything - I always respond internally "Wow, then I guess I have nothing in common with you then!"). There are lots of advantages to being someone like me but many disadvantages too. Many people are much more practical and solid and down-to-earth, more "hands on", and doubtless more comfortable in their own bodies than me. I suspect that that's the kind of difference that probably does make a difference with respect to smacking and who it's gonna work best (or at all) for. And I simply don't see any reason why everyone should be expected to converge on my mindset and personality with its associated squeamishnesses and liabilities. Reflectiveness etc. at least to anything like the degree I have it is a slightly odd phenomenon and should not be a requirement for anything much! Vive la difference.
Note that one reason that randomized trials for behaviors are essentially impossible (as well as being unethical) is precisely because the behavior as an action of yours has to make sense from your perspective. If you as a non-smacker get assigned to the smacking group in our trial, say, then that's not just gruesome for you, you really don't believe in it and then there's absolutely no reason to believe that you can fake it convincingly or completely over-come your internal self-selection against that group and what it represents.
During the smacking debate I often heard hyper-verbal types saying "Wow, I'm amazed that anyone could ever smack a child, I guess I have nothing in common with such people!" In my view that's entirely analogous to my finding people who never have "regrets" somewhat unintelligible. The beginning of wisdom in both cases seems to be to force oneself to see that there could be advantages to not being like oneself and to having a society where not everyone had the same instincts or personality or way of being in the world. I'm an atheist and deep down find it very mysterious that theists should believe the things they do. But with a little effort I can see that there might be serious advantages to believing that there's a meaning behind everything (or whatever core theistic commitments boil down to). Too bad that that's apparently not an option for me then. (Of course if you're Richard Dawkins, you'll regard that kind of open-mindedness with suspicion, think all religious up-bringing is child-abuse pure and simple!) And so on. I think this basically answers your question, but maybe it just raises further questions...
A lot of people are of the opinion that the Bill is like the Government trying to tell parents how to raise their children. It's not really like that when it comes down to it, but it is just placing a limit on what parents can do to control their children, which is obviously completely understandable.
Well there was a limit before - reasonableness - and a range of clarifications and tightenings of that limit were offered and all were rejected.... so it's wrong to suggest that Bradford's Bill merely draws a limit, rather it's a very specific zero-tolerance policy/prohibition that (setting aside various indeterminacies) makes every smacker at least an unindicted criminal. In that sense it expresses the view that insofar as you correct your child by smacking them to that extent you're a bad parent, regardless of what the big picture is about your parenting, and indeed regardless of what good smacking may be doing the child (perhaps because the negative evidence is ultimately going to melt away or maybe just because your child is anomalous in some important respects, and with all your special detailed private knowledge of your child you can see that that is so). That's fairly highhanded and controlling in my view. Punishing parents who take their kids to McDonalds (zero-tolerance for coke and chips for kids) or who don't breast feed or who circumcise or who give their kids a religious up-bringing would seem to be able to justified in exactly the same ways. And that's a bit of a nightmare.
I must say that the insensitivity of the left on this sort of point simply astounded me. The kicker for me was the following passage from the Select Committee report that was supposed to address the practical worry (which never concerned me much) that parents might be prosecuted for assault willy-nilly if S 59 were repealed:
"We note that there are several potential offences directly related to the care of children that are rarely prosecuted. Such an example is if a caregiver sends a child to its room against its will, this technically constitutes kidnapping under section 209 of the Crimes Act. However, the police are not regularly prosecuting parents for this."
Even setting aside the bizarro world we've entered when one insists on thinking that kidnapping applies, even in principle, to paradigmatic exercises of parental authority, the basic move here is simply nightmarish for its intended audience. It's not at all reassuring to someone who thinks that losing S 59 amounts to an invasion of their home to hear the govt say, "Of course, technically, we could get you for kidnapping any time..." Compare: telling someone who's bugged by your latest hate-speech law, say, "Don't worry about the hate-speech law since, technically, we could do you for sedition any time..." In both cases, the basic picture is incredibly troubling: a government is arrogating to itself control in principle over every aspect of human life. Every aspect of your life is being gathered into govt hands/made indictable and only then granted back to you, by its leave. That stinks. I suspect that every time Clark or Bradford recycled the kidnapping point in interviews thereafter Bradford's Bill rightly lost another couple of points of public support. (The kidnapping point was actually legally incorrect as well with S 59 on the books. S59 makes correction no offense not just not-an-assault - see here for more on this point.) It was a classic case of the tone deaf left having absolutely no ability to understand (or interest in understanding) any sensibility other than its own.
Let me explain things from my side a bit more: without a ban on smacking, children are subjected to smacking at times when it is inappropriate, being smacked too much, being smacked for accidents like spilling milk, being smacked because the parent is angry or frustrated with other aspects of their life such as their work, those kinds of things.
I discuss this sort of worry here. I agree that parents are going to make mistakes of various sorts, but am not bugged by that too much. Angry, inconsistent parenting is another thing entirely. It is, I take it, paradigmatic bad parenting, and to be deplored. But that's a problem whether smacking is in the picture or not.
Until there is an appropriate checks-and-balances system put in place, not even light smacking should occur legally.
I'm not sure what you have in mind by "checks-and-balances". Neighborhood parent juries to which parents can appeal and ask to issue special time-limited smacking licences? (Anti-smacking researcher M. Straus once suggested that birth certificates should have "Warning: can be harmed by smacking" printed on them by analogy with govt. warnings printed on cigarette packets!) Or are you referring to utopias in which there are no more bad- or even any less-than-ideal or parents?

2 comments:

Anonymous said...

It's me again. I would just like to thank you for making my posts to you on the anti-smacking debate such an integral part of your website. That is something that I certainly didn't expect.

plague said...

You're very welcome Anon. Truly it's been a pleasure for me to be challenged to think through how to respond to your very well-made points.