Past Election Trends

[ShapeShifter]

Does it mean anything that for the past 6 consecutive presidential election the number of Democratic votes have increased?

According to this trend, I predict a Kerry win in 2004.

But, something tells me that Bush is going to win.

[ShapeShifter]

That doesn't mean much. It went up in 1984 because Anderson was not running. In 1988 it increased because Dukakis worked better than Mondale. Turnout was high in 1992, but that fact is surprising. Gore got nearly as high a percentage as Clinton in 1996, because Perot was not in the race. I don't think this fact means too much. In the same way the Republican turnout has increased in the last three presidential elections.

Okay, taking the past 6 election for the Democrats and past 3 election for the Republicans - I predict a 57 million to 54 million defeat of Bush. This is not scientific but I love playing with numbers.

[ShapeShifter]

Shapeshifter, that's good.  Well, not good as in good for Kerry, but good as in interesting.  It might be more neutral if you do that best-fit line on a plot of votes for party X and extrapolate to 2004.  That's probably what you did for the democrats for the past six contests.  Do the same thing for the republicans for the same number of electoral contests to make it consistent.

I used the average percentage increase for both party - one with the past 6 election and one with the past 6 election for democrats and 3 election for republicans - I got pretty much the same results. But, can you exaplain more what you are talking about? Thanks

[ShapeShifter]

dammit!  Shapeshifter I had a really detailed answer for you and posted and got the old 'shame-on-you-silly-bastard-for-not-leaving-enough-time' message!  Basically, go back 6 (or some number of contests) and look at the total number of votes for the Democratic candidate each year.  Plot that number as a function of year.  So there's one point every 4 years, right?  The points will show some scatter (up and down fluctuations) and probably not lie on a straight line, but you can use the linear regression tool in a spreadsheet program such as SigmaPlot or MS Excel to find the line that makes the best fit through that data.  Or just do it with graphing paper and a ruler.  Then extend that line to 2004.  Do the same thing with the republican candidate and see what you get.  Now, mind you, it may not have any relevance, but it might be a fun mathematical exercise.  And it would certainly have greater basis in reality than the straw polls regularly conducted on this forum.

rwn:  well, I never looked it up, but always took its etymology as self-evident.

Thanks. I will get back at you when I get it done.

[ShapeShifter]

I did my average percentage method for each state. My result was same as 2000 but with NH, FL, and NV going to Kerry. once again, I know this is not scientific but I love playing with numbers.

[ShapeShifter]

I did the best fit line for past 6 election - Kerry by a landslide. That is unlikely. But, with a best fit line for past 3 election - Bush wins by a couple million - more probable.

[ShapeShifter]

I used the same number provided in this lovely website.

The method I used for the best fit line was with excel.

http://www.digitalcoding.com/physics/excel_chart.htm

This website shows you how to do that with excel.

Double check my findings and see if I did a mistake.