Princeton Election Consortium

A first draft of electoral history. Since 2004

Will momentum encounter the Queen?

September 16th, 2014, 4:00am by Sam Wang


Last week I pointed out that most surveys indicated that the Scottish independence referendum is unlikely to pass. Nonetheless, a close look suggests that Thursday’s election will be extremely close, thanks to the elusive quality of political momentum.

Shown above are the results of opinion surveys conducted in Scotland on this question. Each data point shows the median of 2 to 6 surveys, and the gray zone indicates the 1-sigma confidence band.

The word “momentum” gets thrown around loosely in politics. To get back to its meaning in physics, one definition might be a change in opinion that looks like it will continue in the same direction. In that sense, “yes” has had the momentum.

The question may be decided by as little as one or two percentage points in either direction. It looks like “no” is still narrowly favored…but considering the movement since August, “yes” has an outside chance, maybe 1 out of 4, of winning. This will be a nailbiter.

Now, in the home stretch, Queen Elizabeth has spoken, however mildly, against the referendum. On Sunday, Her Majesty told a well-wisher, “Well, I hope people will think very carefully about the future.” A recent YouGov survey indicates that perhaps alone among public figures, the Queen is well thought-of in Scotland. From a political standpoint, she couldn’t speak unless it appeared that the referendum was close: if it was fated to pass, she’d have to go along; if it was fated to fail, best to appear above it all.

Can the Queen stop the momentum? If the polls simply stop moving, we won’t know if she was the decisive factor. But if they reverse, I’d suggest her words as a possible cause.

Tags: Politics

14 Comments so far ↓

    • 538 Refugee

      I’m glad he put this up. A link needs to go in his sidebar. I looked for this kind of thing the other day and simply found nothing on the site. I didn’t make it through all of the article. Post election I might have to. For now I’m just hoping likely voters are the ones that pay attention and that is why polls work in time frames of months. The only ‘swan’ so far is Orman.

      One fairly major nit to pick with the weightings I did read. Leaving out polls that are identified by party. Just to tickle an itch I had about PPP being a reliable pollster I did some reading. Interesting stuff. Bottom line, are the facts more important that partisan cheer leading to your cause? He said he will release his pollster ratings later so I don’t know that PPP is excluded for sure.

      Now, Sam and Nate just need to go out for a beer and life will be good.

  • xian

    Nate has now laid bare his attack on Sam’s approach…

  • 538 Refugee

    How difficult is it for a pollster to be assured they are getting a representative sample in a ‘one off’ situation like this? I think confidence level would have to be fairly low?

    • Matt McIrvin

      I suppose the best you can do is look for analogous situations in other countries, like the Quebec independence referenda. But there are going to be large cultural differences in those cases.

  • Aaron Booth

    Oh my… 538 just tweeted that GOP chances just dropped to 53.3% and that is STILL not a “toss-up.” Seriously?

  • Bill

    The important question is: is the apparent upward trend in August a true trend towards a “yes” majority, or is it merely random variation about a stable mean or median. There are not enough data in this time series to model the variability.

    • Craigo

      I think the movement is clear. Here are the Yes-No medians for the twenty two polls conducted over last six weeks:

      September
      44-47

      August
      39-49

      All but one of the September polls have a Yes vote higher than the previous median of 38%, and all but two have a lower No vote than the August median of 49%.

      The seven pollsters involved have an average sample size of ~1,000, which at 95% confidence level yields a margin of error of 3.2%. The movement of the No median is outside this range, while the Yes vote is within it and the Undecideds straddle it.

      44% is a far cry from the media’s breathless “too close to call” narrative. But there’s such a thing as too skeptical. The movement is real – it’s just not enough.

    • John

      Couldn’t it be a true trend which is nonetheless not enough to reach a “yes” majority?

  • Matt McIrvin

    I’m not sure I’ve ever believed in political momentum. If you want explanations it’s better to look for underlying mechanisms, like the incredible, insulting clumsiness of the No campaign.

    (At this point, I think if I were Scottish I’d be sorely tempted to vote Yes regardless of the long-term merits just to stick a thumb in Cameron’s eye. If I were English I’d be begging them to vote No, because of the shambles the UK is likely to become without them. I am neither (or both, depending on how you reckon these things), so it’s not my fight; it does amaze me how many Americans seem to have an attachment to the Scottish independence cause that is essentially romantic.)

    • Craigo

      Worth nothing that during the two previous British referenda for which we have polling data (alternative vote and Welsh devolution), public polling underestimated the No vote slightly.

    • David D.

      Likewise polling for the 1995 Quebec independence referendum had fewer No votes than the final result.

  • JerseyLiberal

    He is presenting a view and possible explanation, but is keeping the data clean.

  • wendy fleet

    omg omg SAM!! Is this not the Dreaded Finger on the Scale quasi-mystic possibly claptrap weighing factors that pec eschews???

    Is this not exactly the kind of thing that of which pec is putatively pure?? PLS don’t rock my fragile bark, man!

Leave a Comment