17 December 2010

Oh, my! PAPA, are you so wonderful?

Mike Smithson’s PoliticalBetting.com describes itself as “Britain’s most-read political blog – and the best online resource for betting on politics.” Punters, as well trying to read the runes from opinion pieces, take as keen an interest in opinion polls of voting intentions as politicians claim not to. So, to assist its readers, PoliticalBetting has recently introduced PAPA - The Politicalbetting All Pollsters’ Average:
[PAPA is] calculated by taking the average shares for the three main parties from the latest poll of from each of the polling firms in the preceding month.
Unlike the UKPR average PAPA does not discriminate for or against any pollster and only includes the latest survey from each of the firms. The aim is to recalculate it as each new poll comes out or as old polls drop out after passing the month-old point.
I think it would be fair to describe PAPA as an unweighted (but see below) rolling average (a rolling polling average!) of the most recent (ie in the last four weeks, rather than “preceding month”) polls by the major pollsters.  “UKPR” refers to another popular political website, Anthony Wells’ UK Polling Report (no opinion pieces, just polling data), which advises:
The best way to judge the polls is to take the broad picture, not an average, look at the general trends and if they are showing a contrasting picture look at the possible reasons why. I hope that is something that UK Polling Report does for readers. If you don’t have time for that though, and just want a simple overall figure that tells you how the parties are doing, then here is the UKPollingReport Polling Average.
The UKPR Polling Average takes in polls from the last 20 days and gives them weightings based on various factors, including how recently they were conducted, the past record of the pollster producing the figures, the methodology used, the sample size and how many polls have been produced by a single pollster.
UKPR goes on to explain the weightings it uses. But how do UKPR and PB crunch their numbers? I took data from the two websites at 12:00 GMT on 16 December 2010 (UK notation so 16/12/10 etc). They draw on non-identical but overlapping sets of recent polling results, as shown in the table below. To bring them into alignment I’ve worked out what I think PAPA would have been on 10 December, but it doesn’t have much effect on what follows.

Certain points can be drawn from this table. Firstly, PAPA is not “unweighted”, rather it gives an equal weighting to those polls PB select and a zero weighting to all others. Secondly, the only reason for following polls is to observe changing public opinion, so, all other things being equal, the more recent a poll, the greater its significance. Because PAPA polls can span a month and are equally weighted, stale results are not given less emphasis. Therefore PAPA is inherently a lagging indicator by comparison with UKPR’s figures. Finally, there is the issue of accuracy, a big subject (see the answers to "FAQs By Members Of The Public” from the British Polling Council). While combining polls carried out at the same time should reduce random errors and give a more accurate result, I am doubtful as to whether PAPA should be presenting data to three significant figures (ie introducing a decimal).

Lord Parkinson (a Cabinet minister under Mrs Thatcher) was asked about political betting on BBC Radio4 PM on 15 December. He said that he never gambled, summarising bookmaking as: “You bet, we win”. I’m inclined to agree with him, but if I were a gambling man and wanted a summary figure for recent polling, I would take UKPR’s average more seriously than PAPA.

POST TITLE: Younger readers may be unaware of a 1954 number one hit recording by Eddie Fisher, “Oh! My Pa-Pa", translated from the song, “O mein Papa”, in a 1939 German musical:
Oh, my pa-pa, to me he was so wonderful
Oh, my pa-pa, to me he was so good
No one could be, so gentle and so lovable
Oh, my pa-pa, he always understood.
etc, etc

No comments:

Post a Comment