PDA

View Full Version : Research Needed - Percent of Voters to Donors




Fyretrohl
12-18-2007, 02:00 PM
So, anyone here know of any statistics on how many people vote vs donate in any given campaign? If, for example, 1200 Huckabee supporters donate, that can be projected into X poll percentage and Y votes, more or less. Then, we can extrapolate what our polls should be based on those ratios.

bbartlog
12-18-2007, 02:29 PM
It's not as simple as that. Having one piece of data on the level of support doesn't by itself allow you to extrapolate to level of participation in another activity (like voting), because it doesn't tell you whether donations represent a very low level of participation from a large but undermotivated group of supporters, or a very high level of participation from a smaller but highly motivated group.
Ideally, you would want to get a lot of data about supporters of the candidates and their willingness to do various activities on behalf of their candidates. Then, if you could find some way to compare the effort or difficulty of these activities, you could model the 'enthusiasm distribution' or what-have-you for the various supporters.
In reality, you normally get two or three datapoints and have to estimate the overall distribution just from that, which is why we end up with a lot of assumptions (e.g. we assume some sort of normal distribution, or we assume the variance in enthusiasm is the same for all supporters) when we make our models.
In the end, you come up with something complicated that will probably tell you the same thing that you arrive at by a rough guess: we will probably turn out at about twice the frequency others do (80% vs 40%, or 50% vs 25% in Iowa and other places where participation is a pain).
But if you're looking for a way to totally second-guess the polls (as opposed to a way to adjust the percentages they show when estimating vote totals), then you won't find it just by some formula based on donations.