Guns, Germs, & Babies?! Fun with polls
Okay, I'm NOT going to talk about The Flu in this blog and will barely mention guns. :-) It's just that Diamond's title came to mind as I wondered how to phrase the blog title.
Nate Silver has fun with graphs of polling data around the abortion choice issue. Look at all three graphs (small versions included below). He concludes, fairly I think, that the issue remains incredibly stable over the long term despite a recent poll from Pew which covered Guns and Abortion. Silver considers this poll might be "off" ("odds are the Pew result is a mild outlier") but he doesn't look into it much. For one thing, his blog doesn't touch the Gun part of the poll which might serve as a kind of reference check on the abortion issue if the poll is somehow skewed.
On both issues Pew finds a sharp recent trend towards the traditional conservative side. This is highly marked in the abortion issue by a very large shift among Moderate/Liberal Republicans, -24 change in 8 months in the % saying abortion should be legal. Such a shift cries out for explanation. Can it be a Palin effect? Is it data error? Unfortunately, Pew doesn't break out the Gun part the same way and the time frame is different too, so no simple comparison can be made from what Pew presents in the article..
The data also show that older folks (50+) also changed views to the tune of about -10% against abortion rights (change in % saying it should be legal), as did Protestants and [political] Independents, so it is not just that one slice.
But what I notice which also stood out is that the "no answer" fraction grew a lot. That is, the "pro + con" don't add up to 100, and the sum consistently is different for the two polls (April and Aug) by about 5%. For example, in the 65+ category, the poll went from 46/48 in August to 36/48 in April. The entire -10 shift was on one side of the issue as Seniors who thought abortion should be legal became uncertain in droves while those who thought it should be illegal stayed the same. Perhaps general economic uncertainty is making for a temporary conservative blip as moderates and liberals become uncertain in larger proportion. Since the trend is clear over all the slices, it doesn't seem to be merely ups and downs due to small sample size (but the Aug sample WAS much larger).
I'm taking the liberty of linking to Silver's graphs here, along with the Pew abortion graph. In the legal/illegal data (top left) you can see that there has been a lot of polling in the past 5 years, and that there is a lot of extreme scatter on the down side of the blue data but not on the upside (red reversed, ditto). I wonder how the graph would look if the top six and bottom six data points from 2004-2009 were thrown out as outliers. Are biased erratic poll results throwing off the trend line? Is the real trend even more favorable to legal abortions?
Nate Silver has fun with graphs of polling data around the abortion choice issue. Look at all three graphs (small versions included below). He concludes, fairly I think, that the issue remains incredibly stable over the long term despite a recent poll from Pew which covered Guns and Abortion. Silver considers this poll might be "off" ("odds are the Pew result is a mild outlier") but he doesn't look into it much. For one thing, his blog doesn't touch the Gun part of the poll which might serve as a kind of reference check on the abortion issue if the poll is somehow skewed.
On both issues Pew finds a sharp recent trend towards the traditional conservative side. This is highly marked in the abortion issue by a very large shift among Moderate/Liberal Republicans, -24 change in 8 months in the % saying abortion should be legal. Such a shift cries out for explanation. Can it be a Palin effect? Is it data error? Unfortunately, Pew doesn't break out the Gun part the same way and the time frame is different too, so no simple comparison can be made from what Pew presents in the article..
The data also show that older folks (50+) also changed views to the tune of about -10% against abortion rights (change in % saying it should be legal), as did Protestants and [political] Independents, so it is not just that one slice.
But what I notice which also stood out is that the "no answer" fraction grew a lot. That is, the "pro + con" don't add up to 100, and the sum consistently is different for the two polls (April and Aug) by about 5%. For example, in the 65+ category, the poll went from 46/48 in August to 36/48 in April. The entire -10 shift was on one side of the issue as Seniors who thought abortion should be legal became uncertain in droves while those who thought it should be illegal stayed the same. Perhaps general economic uncertainty is making for a temporary conservative blip as moderates and liberals become uncertain in larger proportion. Since the trend is clear over all the slices, it doesn't seem to be merely ups and downs due to small sample size (but the Aug sample WAS much larger).
I'm taking the liberty of linking to Silver's graphs here, along with the Pew abortion graph. In the legal/illegal data (top left) you can see that there has been a lot of polling in the past 5 years, and that there is a lot of extreme scatter on the down side of the blue data but not on the upside (red reversed, ditto). I wonder how the graph would look if the top six and bottom six data points from 2004-2009 were thrown out as outliers. Are biased erratic poll results throwing off the trend line? Is the real trend even more favorable to legal abortions?
Advertisement
Kind of encouraging that people are acting unpredictable---could they be thinking?
May 15, 2009 12:39 AM | Reply | Permalink
That's a hypothesis of sorts! But it seems more likely, based solely on the graphs, that the past few years have shown that some polls are unpredictably divergent rather than pollees becoming generally less predictable in their individual behavior. There are nice clumps of many polls with a handful of very different outliers. That looks like pollsters manipulating good data or chasing bad data in some cases.
When polls start thinking, I suspect we're in trouble.
May 15, 2009 3:10 AM | Reply | Permalink
Interesting! Wasn't there a Gallup poll a few days ago showing the majority of Americans are now against abortion? Would be interested in your thoughts.
May 15, 2009 9:04 AM | Reply | Permalink
"Gallup finds the source of the shift coming almost exclusively from Republicans, which includes independents that lean Republican. Those identifying themselves as pro-life rose by 10 points, from 60% to 70%, in one year. There has been essentially no change in views among Democrats, including independents that lean Democratic."
Looks like Gallup is leaning the same way as Pew. It's not quite the same question, the life/choice frame is different from the legal/illegal frame, as Silver notes. The trends he reports are in different directions. But Gallup's results do tend to support Pew, as I read them.
What do you think?
May 15, 2009 2:37 PM | Reply | Permalink
To be honest with you, I'm having a hard time believing this to be anything other than some kind of outlier anomaly.
We've been told for months now how the republican party ID has been declining, how the party is unpopular, etc, etc.
I would expect any positions that tend to be associated with republicans to decline too.
Could it be a counter-reaction to something like gay marriage? Similar to the 1960s counter-reaction to civil rights? Or do you believe this to be an anomaly too?
May 15, 2009 2:45 PM | Reply | Permalink
What was the 60s counter-reaction? You have
abortion is to gay marriage as _________ is to civil rights. I don't know enough about the 60s to guess accurately. Hippies??
Yes, Repo ID does seem to be falling. If so, the total numbers would be shrinking regardless of the percentages within that group shifting one way or another. And of course if the sub-group shrinks then you get fewer polled from the sub-group so the polling uncertainty goes up.
As I mentioned in the blog, I think it's interesting that Pew had a shift in undecideds which seemed very broad and also largely one-sided re abortion. That says that groups who are usually liberal are having second thoughts about abortion's legality.
May 15, 2009 2:58 PM | Reply | Permalink
The 60s counter-reaction was christian conservatives organizing and becoming politically active for the first time in US history.
Another hypothesis could be that the Democratic ID surge is the result of republicans being in so deep in the toilet, so the resistance to admit one's conservative leanings straight out, but still expressing them through issue ID?
May 15, 2009 3:57 PM | Reply | Permalink
You would think that those might go Independent first, but some might need a sense of party affiliation and thus jump to Demo status. I have not heard about actual party registration numbers and trends. That would be more "real" than just pollster results.
I've never voted for a Repo candidate at state or fed level, but I've been Indpendent for at least a decade after starting out being listed as a Demo.
I do tend to agree that polarizations can have incidental and consequential effects along the lines of what you suggest. I'm not clear on just how much the civil rights progress fomented a build-up of Christian conservatives... I have this naive view that true Christians would be natural supporters of civil rights.
May 15, 2009 4:35 PM | Reply | Permalink
Christian conservatives used to believe that being involved in politics is the opposite of the ideal way of life as a christian. Roe/Wade was the culmination of the 60s wave of civil rights movement and it woke them up. They saw it as State adopting an official policy of killing the unborn and (even worse) doing it via the Supreme Court versus normal policy debate and action and legislation.
Actually, the history of Roe/Wade is the real reason Obama "opposes" gay marriage. Dems now believe that grass-roots and state-by-state wave of gay marriage will be less divisive, hostile and polarizing that Roe/Wade was.
May 15, 2009 4:46 PM | Reply | Permalink
Roe v. Wade is more understandable than civil rights in general. But that's abortion again. So we have, in my formula:
abortion is to gay marriage as _christian conservative organizing_ is to abortion.
That does mean anything to me in that form. Did I get things in the wrong places?
May 15, 2009 4:51 PM | Reply | Permalink
does NOT...
oops
May 15, 2009 6:24 PM | Reply | Permalink
I'm not sure what kind of pattern you're trying to discern with this formula?
May 15, 2009 6:38 PM | Reply | Permalink
You made what I took to be an analogy. I put it into a formula and asked about the missing blank. You supplied it but then it turns out that "civil rights" was a misnomer. So I corrected the analogy and discovered it doesn't look like a meaningful analogy after all. So I'm asking if I got it right etc.
I'm trying to find out if you actually meant something or were just blowing smoke as you often do.
May 15, 2009 6:45 PM | Reply | Permalink
Thanks for your kind words and interesting conversation. Enjoy your evening.
May 15, 2009 6:55 PM | Reply | Permalink