Need suggestions for class project (poli. sci)

[mcah5]

Hello,

I have to do a project for my political science course. It should be in the area of "design, implementation, and analysis of survey data; design, development, and analysis of voting technology." The professor said that he preferred that the projects that had quantitative data and quantitative results. An example project of what somebody did last year was how to detect election fraud through use of statistics.

I have no background in political science, and was wondering if somebody could suggest an appropriate topic? The biggest problem I'm having is coming up with a topic that can fill up 15 pages. The ideas that I have so far from googling is:

1) Derive the solution to the Ballot Problem, but this won't fill up the 15 pages I need and somebody has already done

2) Fourier transform president popularity vs. time to see if there are any wave-like trends in how well-liked presidents are. Again, I can't stretch this to 15 pages.

Any ideas? Thanks.

 

[gravenewworld]

i'm just throwing out an idea, but you could possibly research voter theory. supposedly whoever appeals the most to the middle voter should win an election. how often does this hold true? when has it held true? what effect does this have on politics? you could even throw in some game theory in there. if you were a politician and strongly believed in the median voter theory, wouldn't you go out of your way to have almost the same exact platform as your opponent, but oh so slightly different so as to win over that median voter?

 

[jhe1984]

Yeah, I've got a few thoughts. Please clarify the parameters of the project. Can you use voting statistics for non-US countries?

1) Examine the relationships between GDP and voter turnout. The US, with the highest GDP, has a remarkably low voter turnout. Then you look at, well, Iraq or Afghanistan (not your ideal voting locations) and they've had incredibly high voter turnout. What's more, the Soviet Union, Cuba (I believe), and Iraq under Hussein all had near perfect voter turnout. Maybe the West never had such good candidates. Fraud in its own class, the relationship would likely be rather suprising since I bet - but can't prove - that relatively new, poorer democracies have comparatively high voter turnout compared to established, rich nations.

2) Examine the relationship between type of voting system - electoral college in US, referendum in UK, etc - and voter turnout. Is one system higher than another? Why?

3) Compare voter turnout in countries where either voter registration is mandatory (I think maybe New Zealand, Australia) and incentives [or punishments] provided - tax credits, cash payouts, possible punishment(?) to systems where voting is completely unprovoked.

4) Look at history of get-out-the-vote campaigns. I bet broad campaigns in all demographic groups are rare if ever in US. Even MTV one is not necessarily supported equally by both sides. Simply put, it's probably just as appealing for a party to have a small number of almost all accounted for votes to contest that to be faced with a larger number of "who knows" votes. Thus, I'd bet get-out-the-vote campaigns tend to be organized demographically, organized politically and opposed to broad bipartisan national campaigns.

5) Examine geography and voter turnout for different areas of the US. Is voter turnout markedly lower in muddy, rainy rural areas than sunny flat areas? in deserts versus cities? in snow versus beach? Does having the election be in November favor the Republican party (with today's current political map) over the Democratic party given that many NE states usually vote Democrat and those states may experience harsh weather at that time, etc?

 

[mcah5]

All right! Thanks for both of your help.

I actually don't have any more information about the parameters of the project. I'll throw these ideas at the professor and see whether they are within project guidelines.

Thanks :)

 

[JierenChen]

Hello,

I have to do a project for my political science course. It should be in the area of "design, implementation, and analysis of survey data; design, development, and analysis of voting technology." The professor said that he preferred that the projects that had quantitative data and quantitative results. An example project of what somebody did last year was how to detect election fraud through use of statistics.

I have no background in political science, and was wondering if somebody could suggest an appropriate topic? The biggest problem I'm having is coming up with a topic that can fill up 15 pages. The ideas that I have so far from googling is:

1) Derive the solution to the Ballot Problem, but this won't fill up the 15 pages I need and somebody has already done

2) Fourier transform president popularity vs. time to see if there are any wave-like trends in how well-liked presidents are. Again, I can't stretch this to 15 pages.

Any ideas? Thanks.

How do you Fourier transform data?

jhe threw out some really good ideas. I'd go with those kinds of things: comparing two sets of data and finding very simple, linear relationships. I'm almost positive that's exactly what the professor is looking for.

 

[selfAdjoint]

If you do want to detect wave like or quasi-cyclic componenets the way to do it is not the they FFT but an autoregressive approach. Your software ought to provide something along those lines.