Fundraising is not statistically significant? I find that hard to believe. Also, how current are your polls, and what is their margin of error? Garbage in, garbage out. If your underlying data is less meaningful than your final prediction, then one has to cast doubt on the whole enterprise.
Put in more conventional statistical language, at the 95% confidence level, you have no statistically significant prediction regarding the outcome of the Referrendum C vote.
O,
I remember having a discussion just like this over at polstate when I made my predictions for the Salazar vs. Coors Senate race, click here to look at the original MHD posting on that.
You couldn't believe that the pollsters had a regular and consistent error (e).
When I showed that to Matt he didn't believe it either, but it works. Matt and Joe are both working on improving the model, they think that funding should have a greater role, just like you.
The model would have predicted the outcome of every other off year election that we found polling data on. Matt tells me that the data set isn't large enough to predict if it will pass or fail with the 95% conficence level met. But, and to directly answer your question, Matt says that he can say with 95% certainity that the model predicts the PROBABILITY of passage with a 95% confidence level.
In 1992 polls showed TABOR, Amendment 1 loosing in the week before the election......
Thanks Jim.
Commenting by HaloScan
