Hardcore for Nerds

The headline on this story in today’s Sunday Independent - “Support for suicide abortion down”:

“A SLIM majority (53 per cent) finds abortion acceptable in cases of a threat of suicide – down five points in three months, according to the latest Sunday Independent/ Millward Brown opinion poll.”

There are at least two issues with this that I can see (although the main overarching one is that the Irish media is not very good at transparent analysis of multiple polls over time).

The first is that, according to the RTÉ report on the same poll, a further 16% seemingly supported abortion in case of a threat of suicide in certain circumstances - which would be a total of 69% support in either some or all circumstances, with 23% opposed and 8% don’t know. That is much more in line with previous polls on the suicide question, which don’t appear to have split the question (or at least haven’t reported it if they have). After all, it’s not clear what the other circumstances are surrounding the threat of suicide, an extremely particular circumstance in itself - and how such circumstances would affect the X Case judgement itself. 

The second is that the (unnamed) poll being compared to doesn’t add up on either set of figures: despite saying that “the noticeable fall since February in support for abortion where there is a threat of suicide will be of concern to the Government”, the only poll I can find in that month was the Irish Times/MRBI poll which gave 71% support for abortion in case of suicide, either an 18-point drop or (if one compares the I believe more accurate 69% figure) a 2-point one. To further strengthen the comparability of the some-or-all circumstances figure, the combination of those opposed (11%) and no opinion (18%) - 29% altogether - in February’s poll is very close to the remainder of 31% in the current poll. But regardless of which figure is used, there is no five point drop. The closest to that is a January Sunday Times/Behaviour & Attitudes poll which found 59% support for the suicide option, although in the same month a RedC poll found 64% (presumably within the same margin of error, but not five points from 53% - indeed, between two polls with typical margins of error of +/-3%, a five point drop is not statistically significant).

If there really were only 53% support in total for abortion in the case of threat of suicide, that would be an even more dramatic drop of 6, 11, 18 or even 32 points over recent polls (the last according to the X Case question in RedC’s December poll, a month after the Savita case became public - but there are particular difficulties with using that poll as comparator, which I will discuss below). However, it seems to be to be an artificially low number created by splitting the question into two different levels of support. Reintegrating it to 69%* to compare with other questions based on the X Case criteria, and making certain other alterations (detailed below) to create comparable results, this is what the last six months of polling on abortion looks like:

(‘X Case’ means any question referring to the threat to life of the woman, including the threat of suicide; ‘Excluding suicide’ is the same with the obvious exception; and ‘On request’ is of course the true pro-choice position which remains a minority but a not insignificant one, certainly much larger than the 10% who consistently oppose even the most restrictive options offered - threat to life excluding suicide.)

Of course, in five polls from four firms, differences in question framing and sampling technique will have an impact on the extent of comparability - in fact, the biggest disparity is between the two RedC polls, which used two different forms of questions as discussed below.

The X Case And Excluding Suicide

The December RedC poll attracted some criticism for the seeming incompatibility of its results and the unusual way of framing the question: it asked for people’s responses to each option together, which caused some surprise when 85% supported the X Case criteria (including suicide) while 63% supported removing the suicide grounds. As the don’t knows remained constant (5-6%), that meant an extra 20% opposed the exclusion of suicide who most likely supported the X Case criteria in full. Although the result seemed counter-intuitive in the context of the poll, given that we have had two referenda in which the option of excluding suicide was defeated by a combination of liberal and conservative voters, it is not so surprising.

The following RedC poll in January was commissioned by a different organisation and used a different form of questioning again: respondents chose just one of the four options - X Case, X Case without suicide, on request, or never. On the assumption that anyone who supported the most liberal option of abortion on request would also support the X Case criteria if it was a standalone option, I have combined the two; and likewise for the less favourable again option of excluding suicide. Although this may seem to contradict the interpretation of the December poll, I think there is a difference between seeking people’s opinion of each option simultaneously (in which case more liberal voters are opposed to an option that excludes suicide) and effectively ranking them in terms of which gets the maximum support if they were the only option.

This latter approach seems to best reflect the Behaviour & Attitudes and MRBI polls, in which threat to life excluding suicide remained the option with the broadest support - and also the most restrictive one. On this basis, and including the Millward Brown result as 69%, support for this option has been declining gradually, although there is much less change in those opposing abortion in any circumstances, suggesting that a decline in positive support is translating into don’t knows instead. (For the Behaviour & Attitudes poll, outright opposition was gleaned by subtracting the number - 92% - supportive of any one of the four options given, from 100; don’t knows were not indicated, and would in any case make the actual figure smaller.) 

Other Limited Circumstances

The Behaviour & Attitudes poll lacked a question about abortion on request, although it did include an interesting figure for the number who supported each of the four options given - abortion in the case of risk to life, risk to life excluding suicide, as well as rape and fatal foetal abnormality - which was a bare majority of 51%. Since the least popular option overall, risk to life including suicide, still had 59% support it indicates that there is also still an element of non-linearity amongst the supposed liberalness of options - i.e. someone who supported suicide grounds may have opposed rape, as well as vice versa. It also indicates a difference between the proportion of the population shown in other polls to support abortion on request, on average about 30%, and the rather larger, but still barely a majority, number who support abortion in all of the (very) limited circumstances given.

More sympathetic views are evident when the trio of alternative circumstances (than risk to life) are considered individually. Across four polls there are three data points each for risk to health, rape, and fatal foetal abnormality:

For all three support is relatively constant in a band between 70-80%, although risk to health has dropped from a high in December (shortly after the Savita case) to a lower level in more recent polls. What is intriguing about the health option is that it closely tracks the X Case criteria question in each poll: 82% to 85% in the first RedC, 70% to 71% in the MRBI poll, and 69% to 69% in the latest Millward Brown (further support I feel for using the latter figure, not 53%, as the more comparable one).

This is despite the X Case criteria clearly and explicitly dealing with risk to life “as distinct from health”; but in the discourse surrounding ‘suicidal ideation’ and a suspicion that we will be following the broader UK mental health grounds for abortion by legislating for risk of life from suicide, the two seem to be converged in the public mind. In terms of ‘liberalness’, suicide and (implicitly physical) health grounds have been equated. 

The Trend in Excluding or Including Suicide?

I would think therefore that the initial high support for the X Case in the December RedC poll still stands as an accurate reflection of public opinion then, and although it is not representative of more recent polling on the issue it is so consistent with the also-declining health option that is remains a valid comparison. However, the stark change in preference from it being more popular than the exclusion of suicide, to being less popular, makes the two in combination hard to credit as comparators. Since each is out by about twenty points on the later four polls, removing them changes the average - making the X Case less popular and the exclusion of suicide more popular - by around five points in each case.

One can perhaps isolate the suicide question in the December poll as contrasting so much in its mechanics from the later ones. In the first, broadly liberal respondents chose to register their opposition to a proposal seen as either undesirable, unrealistic, or both**; whereas in later polls (after the political concerns over suicide grounds had become more vocal, and after the methodology had changed to produce less ‘confusing’ results) it has registered consistent support as the most broadly acceptable option, although hardly the most preferred.

Unfortunately, apart from the December poll the figures for ‘don’t knows’ on this question haven’t been published, so one can’t discern whether the gradual decline from 90% to 78% support for this option - rather than the less obvious drop, at least from January onwards, in the X Case including suicide - is due to increased opposition from whatever quarter or, I would think more likely, increased uncertainty over the issue. Indeed, if one uses the 69% figure for support for including suicide, the two options could be seen as converging.at somewhere around the 75% mark - with everyone beyond that either wholly opposed or just unsure about the issue. 

*An earlier version of the same chart with the Millward Brown figure at 53% is here

**Technically, since it represents a majority in favour of excluding suicide, if it were a question in a referendum it could alter the X Case judgement and overturn the two previous referendums; but there’s a long distance between a single poll question and a referendum result. The real question of such a referendum would be whether it would follow the trend between 1983 and 2002, where liberal and conservative constituencies remained polarised but a general social shift meant that a restrictive amendment passed in 1983 with 67% while a further one in 2002 was defeated by just 50.4%; or that of 1992 when an anomalous coalition of liberals and conservatives defeated an essentially identical restriction by 65% - could conservatives adopting a more pragmatic position today swing an exclusion by that amount? Well, it seem unlikely given the consistent majority in favour of the X Case as it stands - but the apparent shift in the polling questions.away from seeing the exclusion of suicide as an overly-restrictive option to be rejected, to an option that creates the least opposition, rather complicates such a prediction.

That’s the headline, but what’s the figure?

The report shows Ireland’s Gini coefficient stood at 30.7 in 2010, compared to an OECD average of 31.3. 

Not much of a difference, really. ‘Close to the average’ would be a more honest headline. And not only that, but the report itself points out that the average is over a wide spread:

“Differences in levels of income inequality across OECD countries remain large. The Gini coefficient ranges from 0.25 in Iceland to almost twice that value in Mexico and Chile. Nordic and central European countries have the lowest inequality of disposable income while inequality is high in Chile, Mexico, Turkey, the United States and Israel.”

But since the lower values are more bunched together, most countries are more equal than the average. In fact, we’re the 18th most equal country out of the 34, sharing the median position with the slightly lower Poland. ‘Ireland at the median of OECD income inequality’ would carry a rather different message than the actual headline (and presume slightly more knowledge of basic statistics). 

Furthermore, elsewhere on the OECD their own figures show that in 2009 Ireland’s Gini coefficient was (slightly) above average at 0.331, having increased sharply from 0.293 in 2008 - when it had been on a downwards trend for several years. In other words, Ireland had been becoming gradually more equal up until the recession hit. The 2010 figure in the latest report is noted as being provisional and based on the EU-SILC database; if accurate it is a good sign by comparison, but still a slight increase on pre-recession levels of income inequality.

The rest of the Irish Times article also features some partial interpretations of statistics:

“The report finds that the economic crisis had the effect of increasing inequality of earned income in most OECD countries, but that welfare systems offset the effects on inequality when total incomes (including social transfers) are considered.

In the three years to 2010, Irish households suffered the seventh-largest decline in market income (ie before welfare benefits) in the OECD.

The decline was just over 5 per cent, compared with an average decline of 2 per cent. Icelandic households suffered the largest decline in market incomes, at 12 per cent.

Polish households were at the other extreme, with market income growth of 3 per cent between 2007 and 2010.

The picture for household income changes considerably in most countries once the cushioning effects of social welfare systems are taken into account.

In Ireland average household disposable income, which includes social transfers, fell by just under 4 per cent in the three years to 2010.

Social transfers boosted disposable income by 2 per cent over the same period, the joint highest effect of transfers, along with Finland and New Zealand.”

Again, the relevant figures in the report show that while Ireland indeed had the seventh-largest decline in market incomes, it also had the fourth-largest decline in disposable incomes (-4%, behind only Greece, Mexico and Iceland) despite the effect of social transfers: apparently because, unlike Estonia, Spain and New Zealand which all suffered larger declines in market income, changes in tax had a negative effect on incomes in Ireland (-0.51%, the highest in the OECD). Or at least average incomes - to get the picture in relation to inequality, or the distribution of incomes, one has to turn to another chart.

Market income inequality rose by the highest amount, as measured by the change in the Gini coefficient, in the OECD bloc between 2007 and 2010 in Ireland: from 0.510 to 0.576. The next largest increase was in Spain, from 0.447 to 0.507. Yet the increase in disposable income inequality was much lower in Ireland over that period, from 0.299 to 0.307, compared to 0.309 to 0.338 in Spain. A general theme of the OECD report is that taxes and transfers - or “the mitigating effects of the welfare state” - have softened the impact of increased market income inequality. Yet only social transfers, not taxes, are mentioned in the article. Tax benefits had a positive effect on average incomes in many countries, although the effect was negative in Ireland; but in Ireland, the effect of market income inequality was lessened on disposable income by a notable degree - a sign perhaps that increases in income tax (in all but name) in Ireland’s low-tax society have given some protection to societal equality. 

Pretty good polemical review of the Savages album (which, as I’ve said before, I’m rather disappointed by) although I feel like the last third is unnecessary - or at least I don’t feel the need to judge their ideas, just that they don’t make a very interesting expression of them. That said, this is a great putdown:

“Everything about this record reeks of conservatism, right down to the last deviant sexual fetish and the drummer bringing a copy of The Fountainhead to the Pitchfork interview (+1 journalism points to the interviewer for mentioning it, but -25 journalism points for not actually, you know, asking her about it).”

And also, it’s almost too easy, but I’d already noticed that on re-reading their interesting-seeming ‘manifesto’ it’s actually rather wobbly. Or:

“Conservative with a lower case c, Savages are such bad poets they don’t (I’m assuming) realize they’ve written a manifesto that would make Thatcher proud. Seriously, it idealizes a golden age that never existed, bemoans the diversity of voices in the culture, and longs for rational simplification in the face of messiness.”

That’s certainly a literal interpretation of complaint about “too many voices” and “noise” as “a constant distraction”, but while’s there’s a certain post-punk astringency about it (which is again not only unoriginal but also undeveloped), it’s the “angry young tune” that confuses me now. What, exactly, is ‘young’ about a band so steeped in replicating the past?

There’s also this follow-up post which counters some expected criticisms that is worth reading. I came across Collapse Board through Jake Cleland when they were writing about Iceage, and similarly at that time providing a useful oppositional critique of the Pitchfork celebration of alternative post-punk that is, to me, more style than substance (and worryingly so). I don’t agree with everything they write, as it is in a strongly polemical style, but it’s good to see the effort being put in. There are some useful critical thoughts to come out of not being silent.