On correlations

In trying to figure out exactly what factors play the greatest role in determining a party's electoral prospects, one of the thorniest problems is the difficulty in separating outcomes from each other. Virtually any positive result (more money, more members, more votes, more public attention) can plausibly be painted as either a cause or an effect of any other - making it difficult to determine what a party should focus on first in trying to build itself up.

Which makes bluegreenblogger's efforts to separate out the correlations between various factors with voting outcomes into a highly interesting read, even if they're based on a limited sample (riding results for Ontario Greens in 2007):

Correlation between spending and % Vote: 0.76573958
Correlation Between 2007 EDA Assets and %vote: 0.53411402
Correlation between Transfers into campaign and % vote: 0.72692593
Correlation between campaign contributions and % vote: 0.27744132
Correlation between spending and total votes: 0.75095156
Correlation between EDA assets and total votes: 0.57536407
Correlation between transfers into campaign and total votes: 0.6979957
Correlation between campaign contributions and total votes: 0.3293264

What strikes me as noteworthy about bluegreenblogger's findings is the difference in correlation between the factors which seem to reflect strength within a riding, and those which are based solely on party spending choices.

After all, two of the factors discussed would seem to speak to the support within a riding association: the existing asset base and the contributions received during the campaign. But these are the two factors with a lower correlation to vote outcomes - suggesting a relative disconnect between the strength of a riding association and electoral outcomes.

In contrast, the strongest correlation is between the raw amount spent by a riding association and the votes received. And it's particularly interesting that even "transfers" from outside a riding into campaign coffers appear to relate more closely to the votes received than do contributions from a riding's members.

Of course, it's worth wondering to what extent the effect might differ between parties or election types - or indeed whether the cause and effect may lie outside the transfer numbers. But to the extent the Ontario Greens' experience is representative of political parties in general, it could be that anybody wishing to maximize the bang for their donation buck might be best advised to fund a party's central coffers to be distributed where the money can do the most good.