TNS BMRB: Yes 30%, No 51%

Today, pollster TNS-BMRB publishes its most recent poll on Scottish independence. Often as not, today's results are quoted in isolation. I thought it might be helpful and of interest to stage a beneath-the-topline retrospective on the company's findings since the start of 2013. One wee note of caution. In January, TNS BMRB was still using the old formulation about negotiating independence, adopting the Scottish Government's new question first only in February. The impact of the change seems negligible, so I've simply included the January figures in the charts below.  

I don't have much time to commit to the exegesis of this material, so today I'll mostly be presenting the data, and leaving it at that, with a couple of marginal notions, to aid understanding. Today's data is labelled as March '13 throughout. First up, TNS-BMRB's findings on the changing overall picture since January. 

Gender

Historically, the gender gap has remained a stubborn feature of independence polling. Interestingly, TNS-BMRB has consistently generated smaller discrepancies between the voting intentions of men and women in the referendum than many of its competitors.

Age

Polling data on the breakdown of voting intentions by age has been marked by significant volatility at the bottom, most youthful end of the spectrum, and a good deal of solidity as we approach the eldest cohort of respondents.  Opposition to independence amongst those aged over 55 continues, undented. By contrast, the sometimes more-pro-independence younger voters continues to leap about like a frog in a frying-pan.

35 - 54

55 - 65+

Social Grading

Like a number of other pollsters, TNS BMRB break down their data using "social grading" codes.  Respondents are classified based on the occupation of the "head of household". This information on the chief earner's profession is broken down into AB (upper) and C1 (lower) middle classes, with C2 representing the "skilled working class", and DE denoting the working class and those living at the lowest levels of subsistence.  To add a bit of important context, according to Ipsos-MORI, something like 27% of Britons would be classified as of AB social grade, 29% as C1s, 21% as C2s, and 23% as DE. 

Previous polls have tended to show that opposition to independence is substantially higher amongst AB voters than their poorest fellows.  While TNS-BMRB found that opposition to independence is 8% down amongst AB voters than at the start of the year, today's poll shows a spike in the number of the poorest Scots who oppose the measure, and an independence droop. That said, the poorest cohort of Scots remains the most supportive, while the richest remain unconvinced, with less a quarter of AB voters currently favouring independence.

Independence: AB and C1s.

Independence: C2DEs.

 

Those full tables here.

Scrooge McDuck votes No...

Yesterday, we took a look at Angus Reid's most recent poll findings on independence, which showed that just under a third of folk support the proposition, and just under half oppose it, with 20% of people still undecided. Broken down by gender, by age and by social grading, the numbers tell a now familiar story. Substantially lower support for independence among women than men (a 13% gap) and much higher levels of opposition to the idea from the middle classes and the oldest cohort of Scots (opposition to independence is 15% higher amongst those aged 55+ than those between 18 and 24 years of age, and 16% higher amongst ABC1 voters than C2DE voters). The pollster put another couple of questions to its respondents, the second of which was:

"Thinking of your own financial position, do you think independence will leave you better off, make no difference, or leave you worse off?"

Part of the mood music of the campaign thus far has been the assumption that the whole controversy will be determined, in the last instance, by voters' apprehensions, financial and economic. You'll remember the survey which asked folk if they'd be for or against independence if i) they'd be no better or worse off ii) if they would be £500 worse off and iii) £500 better off. It turns out that Scots say they'd follow the money.

Now, £500 may seem a paltry, even disloyal sum to justify succession, and a crabbit (even contemptible) reason to decide to become an independent state, but beggars can't be choosers. For my part, I'm a bit skeptical that folk really think and vote in this way - but the findings at least gesture towards the "thinness" of pro-Union sensibilities and the extent to which perceptions of the economic impact of independence on the average punter is going to play a substantial part in this campaign. Something any pro-independence sort has to be hearted by. So what did Angus Reid discover? Across all 1,003 respondents, the leading impression, but well shy of a majority, was that independence would pick their pockets.

In some respects, it's surprising that only 38% of folk think they'd be worse off with independence, given the bushfire rhetoric that we've been hearing of late from Alistair Darling, which has tended to run "Of course I believe that Scotland could go it alone, but [insert financial disaster narrative]". For the moment, however, 41% think that the wallets won't be sapped of their hard-earned lucre or might even plump up, given independent governance.

For YesScotland, still seriously to begin making the case for independence, these are encouraging signs. A post-apocalyptic image of a new Scotch peasantry, scratching out an attenuated life on thin gruel and lumpy mashed neeps without butter, seems not to have gained as much purchase as one might think. This is, in all probability, a register of the extent to which the campaigns have not really impinged on many people's consciousness, rather than evidence that a dismal pro-Union rhetoric of fire, brimstone and your evaporated doubloons won't work, but the poll at least suggests that the thrawn messages emanating from Better Together remain far from universal in their impact.

As ever, it is interesting to dip beneath the topline and tease out some of the differences. Let's start with gender.  As we saw yesterday, the same poll recorded big differences between men and women's attitudes towards independence.  What are their financial calculations?

Let's take a look at their answers disaggregated by age groups. General attitudes to independence tend to show a tapering rate of opposition to independence, increasing as respondent groups get more ancient. Again, we see a similar pattern in their answers about their financial situation, with diminishing uncertainty, and increasingly negative attitudes towards independence's impact on the crustiest cohort's swag bag.

And finally, by social grade, determined by the occupation of the "head of household", roughly dividing respondents into middle class (ABC1) and working class (C2DE) respondents. 

Yesterday we saw that while support for independence did not diverge widely between ABC1 and C2DE voters in this poll (separated by 6%), the far more substantial difference was in levels of determined opposition to independence, with opposition running 16% higher amongst ABC1 voters than C2DEs. Interestingly, we can see this difference more or less mirrored in their answers to the financial question, with 15% more ABC1 voters apprehending that they would be worse off with independence, than without it.  Scrooge McDuck votes no.

The message of all of this? Obviously, it is a complex thing, and a range of variables nudge the vote this way and that. There are folk, for example, who might accept that independence would have financial benefits for the country, but who would stick with the Union, out of a British Nationalist sensibility, or what have you.  Equally, there are people who might be uncertain about the financial impact of independence, but who remain nevertheless inflexible and committed backers of independence. There do seem to be some important correlations here, however.  The bourgeoisie present the most obvious challenge. How to begin to chip away at the large, largely negative assessment of independence's financial impact on them? It is worth bearing in mind that some 55% of the UK population would now be assessed as ABC1 voters. This skepticism, and the scale of the negative financial assessment amongst ABC1 voters, must be a big challenge for Yes Scotland.

With women voters, the picture seems more promising.  While a large wadge of women think independence will have a negative image of them and theirs, there's a substantial body of uncertainty, which at least roughly connotes persuadability.  On one interpretation, at the moment female voters are like the jury who take the burden of proof seriously, and reject the case adduced by the prosecution as insufficient to convict. That's not to say that they're dead certain that the villain in the dock is innocent of all charges, and couldn't be persuaded by different, more substantial argument from the procurator fiscal, to condemn the accused man. They've just not heard it yet.

To harp on an old string, finding ways to speak to and convince women to back independence is going to be absolutely vital if YesScotland is to get anywhere in this referendum campaign.  The weekend's Angus Reid poll contains motes of light on that score. It isn't all in vain. Yet.

Angus Reid, February: Yes 32%, No 47%.

"Should Scotland be an independent country?" A transparently biased question, do you think, sure to deliver a smashing pro-independence majority? Pollster Angus Reid put the claim to the question in a 1,003 person poll, conducted on the Mail on Sunday's shilling over the end of January and beginning of February. The first published poll using the new formulation agreed by the Electoral Commission, Angus Reid has tweaked its approach slightly.  The last referendum poll they conducted in January put support at Yes 32%, No 50%, but did not include a breakdown by social grade.  It also disaggregated responses according to six age brackets, from 18 to the over 65s.  This weekend's poll is a wee bit different, with disaggregation by age truncated to three rougher groups (18 - 34, 35 - 54, 55+). On the plus side, on this iteration, the pollster did include social grades (a detailed description of what these denote, here).

So what did they find? The overall totals with the new question were:

And broken down by gender...

Our old friend the gender gap, still very much in evidence here, with 13% point gap between support for independence between men and women, with a much larger percentage of female voters still undecided (12% higher than men).  Compared with January's findings, the male limb of the poll held pretty steady (indecision +1%). Indecision amongst female respondents is up (+7% on January's findings) with both support for and opposition to indepedence down (-1% and -6% respectively).  

A curious poll in terms of age, this. January's finer-grained poll showed the familiar taper in support for independence, and mounting opposition as folk get older.  While opposition to independence is at its highest in the oldest cohort of February's rougher, trisected poll, support for independence amongst those over 35 is six points above that of the most youthful third of respondents, approaching a full third of whom declare themselves undecided.

Lastly, Angus Reid also disaggregated by NRS social grades, which are based entirely on categorising the professional occupation of the head of the household.  In rough and ready terms, ABC1s are envisaged as the middle classes, from "higher managerial, administrative or professional" employees, through "intermediate managerial, administrative or professional jobs, to "supervisory or clerical, junior managerial, administrative or professional" workers.  C2DEs, by contrast, encompass "skilled, semi- and unskilled manual workers" and "those at the lowest levels of subsistence".

Past polls have consistently shown that opposition to independence is at its highest amongst better off respondents, and support at its lowest, the attitudes of poorer Scots its mirror image. Although levels of support for independence in this weekend's poll do not differ terrifically substantially between middle class and working class respondents (6%), a far more substantial gap separates opposition to independence from Angus Reid's ABC1 and C2DE voters (16%), with indecision amongst C2DE voters mostly mopping up the difference.

Angus Reid posed another couple of questions in this poll, perhaps the most interesting of which being:  

"Thinking of your own financial position, do you think independence will leave you better off, make no difference, or leave you worse off?"

For digestibility, I'll be breaking down respondents answers to that one, and how it plays along gendered, age and social lines, in another post later on today.  Polish off your abacuses, and stay tuned.

Those full tables.

Angus Reid, January: Yes 32%, No 50%

No devotee of the Express, I managed to miss the poll which the paper commissioned from Angus Reid last week on Scottish independence. The first Scottish constitutional headcount of 2013, the Express poll sampled a relatively small selection of folk (just 573 respondents) on the national question, and on a second (to my mind, rather poorly framed, muddled) series of preferred constitutional alternatives, from the status quo, to "some powers", "more powers", and "full independence". Since devo-something is realistically off the table until after 2014, I intend to focus solely on the substantive findings on independence.

The pollster's January findings more or less echo trends we've seen a number of times before in offerings from YouGov and Ipsos-MORI: evidence of a substantial gender gap, and an age-taper in support for independence, from young 'un to auld yins. Overall, Angus Reid found support for independence held steady at just under a third of the electorate, with half of their respondents preferring to remain in the United Kingdom, leaving 16% as yet undecided on the constitutional question.

No great shakes in the gendered column either. Polls have consistently shown a gender gap in support for independence of 10% or more. This Angus Reid poll is no exception. Voting intentions vary 10% between men and women questioned, some 37% to 27%, with a slightly larger cohort of Scottish damsels (+6%) declaring themselves undecided.

Angus Reid avoid the thorny domain of correlating voting intention to social class, but they do disaggregate their results by the antiquity of their respondents. Even more so than is usual, the pollster has sampled a tiny number of folk in each bracket.  Despite this small sample size, the company's findings aren't exactly a revelation.  With the odd bump and hump here and there, support for Scottish independence tapers off as respondents get older, with the oldest cohort of respondents (over 65s) recording the lowest level of support for independence (24%) and highest levels of opposition (65%).  Exhibiting the estimable contrariness of youth, the youngest group of respondents (18 - 24 year olds) again recorded the highest levels of support for (39%) and the lowest level of opposition (41%) to the idea that Scotland might be better off independent.

Those full tables.

Scottish Independence: those chancy Ipsos-MORI trends...

A somewhat uncomfortable juxtaposition this, given yesterday's jeremiad against commentary on Scottish independence which obsesses unduly over polling numbers, and which campaign is ahead by a nose, or as circumstances might have it, by a proboscis of Pinocchioesque scale.  What can I say? I'm going hazard the Americanisation, threaten complicity in the saturation of our politics with statistics, and take a closer took at Ipsos-MORI's latest independence poll, published this morning.  Commentary on these polls tends to be governed by how the relate to the study undertaken immediately previously, and if the analysis hitherto is anything to go by, then the good ship independence isn't exactly holed below the waterline, but this October poll shows it leaking support in a less than encouraging fashion. 

It strikes me, however, that it might be interesting, and certainly wiser, to cast our minds back a wee bit further than July this year, to see the ebbs and flows of these opinion polls in their proper aspect.  Accordingly, in slight amendment to my past approach to presenting polling data, this time I'll be contextualising the new findings, not with respect to the immediately prior poll, but to all five Ipsos-MORI independence surveys back to August 2011.  There is no particular reason of principle for my stopping at this point. That was merely when easily-accessible data stopped, and is I think far enough back, to offer perspective to today's findings, without exhausting us all with too much information.  

The lesson of this longer-term perspective? Firstly, you're really struck by the volatility of findings beneath the topline. While in poll after poll you find lower support for independence amongst women, and higher support for it amongst poorer than richer Scots, the rates of difference are all over the place, sliding hither and thon like a drunken centipede, giving rollerskates a try.  Opposition to independence amongst men, for instance, has ranged across fourteen percentage points, from a high of 58% opposition last August, to January 2012's low of 44%, increasing in July, and falling again by two pips this October. Women's support for independence has been consistently on the slide in Ipsos polls (down from 34% last August to 25% today).

Beyond the extremes, how support for independence might break down by age is mostly hunchwork.  We can consistently say that the oldest cohort of Scots is the most opposed to independence, but shy of that, all one can soberly say about those under fifty five is that they are consistently inconsistent in their constitutional preferences.  By way of an example, take the youngest group, 18 to 24 year olds.  Their support for independence in Ipsos polls has vacillated 12 percentage points from the highest to lowest level of support.  Indecision even more so, shifting 18% upwards today, compared to just 3% of the cohort questioned who said they were undecided about Scotland's constitutional future last December.  Over-confident analysis of these fluctuating findings looks decidedly chancy.  

One final observation or two on the latest poll, before the charts.  Across all but one of the categories we're looking at here - of gender, of social deprivation and of age, indecision is on the increase, and not solely at the expense of the pro-independence side of things. The percentage of respondents who are undecided in October's Ipsos-MORI is higher than any of the other four, whether they are men, women, old or young, and with only one exception, whether those questioned were wealthy or impoverished.  Divided up into the five percentiles of deprivation, only the second least deprived 20% felt less uncertain this October about how they might vote, than every other category of people.  This is at its starkest amongst the poorest 20% of Scots, usually independence polling's strongest supporters.  While independence remains the majority choice of the poorest 20%, levels of indecision have increased from a low of 9% in last autumn's poll, to a full quarter or respondents today. 

All in all, though, these aren't splendid-looking polls for those who support independence. Levels of support for independence today are at their lowest since August 2011 amongst men, amongst women, amongst the poorest, amongst 25 to 34 year olds, those aged 35 to 54, and amongst the inveterately opposed old codgers, counting more than fifty five years to their name. I suspect my own feelings mirror those of most nationalists surveying these results: they simply underline the challenge before us, rather than fostering despair. Salmond's buccaneer sensibility is the right one. Confound the bean-counters. Make the arguments. Strive to persuade your neighbours, your colleagues and friends. Keep the heid. It's not sewn up just yet.

And with that, to the pretty charts.  Let's start with gender, and their shifting Ipsos trends.

After which, to age.  For convenience, I've broken all of this down into Ipsos age bands.  I pondered a vast, mad, spider's web of a line graph, but it make for an impenetrable thatch of data to try and tease through.  If anyone has particular requests or preferences in terms of the presentation of the information, do please let me know. You'll notice, by the by, that Ipsos use slightly different, slightly fewer age brackets than, say, TNS-BMRB. Chronologically, the pollster's findings were as follows...

 

Unlike the social grading used by TNS-BMRB, which is based on the occupation of the "head of household", Ipsos favours distinguishing its respondents into one of five categories of affluence, running from the 20% who live in Scotland's most deprived areas, to the 20% who stay in the least deprived quarters of the land.  

One infelicity of this approach is that it is a bit tricky to entitle the data in a readily comprehensible way, save for the extremes of poverty, and extremes of wealth.  For accessibility, I've styled the arid categories 2, 3 and 4 as - second most deprived 20%, the middle 20%, and the second least deprive 20% respectively.  The charts run in order down the page, from poorest to richest to aid in their construction.  Already starkly hostile to independence, this month's poll records the lowest level for support for independence from the richest Scots yet, falling beneath 20%, while levels of indecision, as elsewhere, look to be on the rise.

 

Those full Ipsos-MORI October tables.

Scots drugs deaths increase by 563% & 623%, but for who?

Last week, the General Register Office for Scotland published its grim annual statistics on the number of drug-related deaths in Scotland in 2011. BBC Newsnicht took the findings for its topic last night, focussing on the much-increased incidence of the drug methadone being implicated in deaths.  I don't know about you, but I often find big data shots like this are difficult to digest.  On the telly, the demands of brevity make for fleeting graphics, glossed issues and compressed conversations.  The statistical releases themselves are a remorseless parade of cross-referenced columns, less than easily navigated.  I thought it might be of interest, therefore, to pull out some of the information published by the government statisticians, and re-render it in more easily intelligible forms. 

Obvious questions include: how many people are dying? Is the number increasing or decreasing? Chipping below the total number of deaths, are there changes in the gender and the age groups of those who are dying? Of the dizzying array of noxious and intoxicating substances out there for sampling, which ones are most implicated in most deaths, and how have these figures changed over time? I hope to provide sketchy, preliminary answers to all of these questions, but first, the total butcher's bill.  The General Register Office statistics cover the last fifteen years, during which period...

Like a great many of the statistics which have interested me in the past - imprisonment, homicide, suicide - Scottish men represent the overwhelming majority of recorded drug-related deaths.  Since 1996, the number of men dying has more than doubled (from a low of 179 deaths in 1997, to a high of 461 in 2008).  For women, by contrast, a look across the same period shows that a 1997 low of 45 deaths has increased to a high of 155 last year, with an upwards spike in 2008 which has seen over 100 women die drug-related deaths each year to date.

And showing increases between 1996 and 2011, by gender.

And women...

Feckless, raving, drug-addled youth. It's a familiar image, likely reinforced by the coverage given to some tragic instances of very young folk, perishing. The statistics, by contrast, show a rather different picture.  Fifteen years ago, the overwhelming majority (77%) of those who died drugs-related deaths were aged under 35, 35% of them before their twenty fifth birthdays.  Over fifteen years, however, that picture has rapidly altered.

Under 35s now make up just 41% of 2011 drug deaths.  Last year, fewer under 25 years olds died than 45 - 54 year olds.  To my eye, most startling over the decade and a half has been the huge increase in the number of 35 to 44 year olds dying, from just 32 deaths in 1996, to last year's high of 212: an increase of 563% in just fifteen years.  Although starting from a lower base, the increase in the number of 45 - 54 year olds was even greater, climbing over 623% during the period from just 13 deaths in 1996 to the current high of 94.  Despite some people's expectations, drug deaths amongst under 25 year olds are one of the few age strands which has been showing a consistent downward tendency these last four years. 

 

Seeing such substantial changes, the obvious question to ask is, "why"? Confounded tricky question it is too, and a few tentative cautions and hypotheses are probably beyond the scope of this (already substantial) blog. I may well come back to it anon, but until then, speculative discussion in the comments is positively encouraged. 

So what sorts of drugs are implicated in the ongoing rise in the number of Scottish drugs deaths, and how has this changed over time? On a statistical note, more than one drug may be reported per death, and should not be added to give total deaths. I'll be using the data generated by pathologists on the "drugs which were implicated in, or which potentially contributed to the cause of death".  The report also includes data on "other drugs which were present but which were not considered to have had any direct contribution to the death", for those interested in the cocktail of substances found in the systems of those who died.  

Like the changing chart on the age of those who died, the profile of the drugs which killed them has changed substantially over time, with volatile spikes, falls, troughs.  Statistics are not published on all of the sorts of substances which may rob a man of his life - the data is selective, and I'm sure you've heard of most of the drugs depicted.  

So that's how things have changed over time.  What of last year? What substances are proving a contemporary challenge? As the Newsnicht report focussed on, methadone looms large. 

Given the issues I've looked at so far - differences in gender, differences in age groups - I wondered whether the drugs implicated in deaths would differ substantially between men and women, and between older and younger victims.  First, let's compare genders.  It's worth emphasising again, these are all of the drugs which pathologists recorded as "implicated in, or which potentially contributed to the cause of death" appear in these charts. It would be possible, for example, for a fatally over-intoxicated unfortunate to have every single one of these substances in (probably his) system.  The Office publishes a separate chart on all those drugs deaths where only a single one of these substances (and "perhaps alcohol" too) were found in the deceased's system.

And women:

 

In reading these graphs, it is worth recalling that male deaths substantially outnumber female deaths, and so you can't read across from the percentages given.  While, for example, a greater percentage of women than men are dying with methodone knackering their systems, methodone is implicated in a greater number of male deaths (75 deaths to 200).  It's an elementary point, but worth keeping in mind.  Now.  What about age? The youngest cohort, whose under 25, constituting 9.9% of drugs related deaths recorded in 2011.  Substances "implicated in, or which potentially contributed to" their deaths were as follows:

Taking them in order of youthfulness, next, the 25 - 34 year olds, who made up 32.5% of those dying last year as a result of drug use.

35 to 44 year olds, 36.3% of deaths last year:

Second to last, the 45 to 54 year olds, who made up 16% of recorded deaths in 2011.

 

And finally, the oldest cohort, of over 55s, who made up just 6.2% of 2011's total.

So there we have it. Is it entirely what you expected? As we often find, quantitative data like this poses more riddles than it solves, even before we start asking thorny questions about causation, or trying to explain changes in the statistics and analysing what the devil we might do, to try to decrease the numbers of folk needlessly dying. I should also add, the report itself includes various other pieces of data which I've passed over for reasons of brevity.  Conspicuous by its absence is a geographical perspective on the national totals.  You can find further breakdowns by NHS Board area and council area respectively between 2001 and 2011. 

Read the full report here.