First Section: Post paper proposals here
Your paper proposal should follow this format:
1) Your name
2) Your main thesis or what you want to write about
3) The main relationship that you want to examine
Here is an example:
I am interested in whether poor people turned out during the Depression.
My main relationship I want to examine is whether poverty is positively related to turnout in a Depression year. I will compare this to poverty and turnout in a non-Depression year. I will also control for how much a respondent liked FDR.
1) Your name
2) Your main thesis or what you want to write about
3) The main relationship that you want to examine
Here is an example:
I am interested in whether poor people turned out during the Depression.
My main relationship I want to examine is whether poverty is positively related to turnout in a Depression year. I will compare this to poverty and turnout in a non-Depression year. I will also control for how much a respondent liked FDR.
posted by Paul Gronke at 11/06/2006 10:51:00 PM
40 Comments:
Topic idea: Do hispanic voters who favor immigration vote more often than those who do not? Does it matter if the hispanic voter is of Spanish descent vs. South/Central American?
~McNitt
**Addition to above post**
My thesis is that hispanic voters who favor immigration vote more than those who do not. Furthermore, I want to examine if those who are of Central American or South American descent are more likely to vote than those of Spanish descent, and whether or not this affects the individuals feelings about immigration.
There aren't enough Hispanic voters in the survey for you to distinguish them.
Is there any measures related to atttidues toward immigration?
The level of income is an important element that has great impacts on turnout. According to 2004 survey, the voting percentage of citizens with high income is almost double times higher than of those with low income. The wealthy ones vote, write, campaign and petition more than the poor. SDA shows that the poor with income lower than $20,000 has the voting percentage 48%, and the rich with income higher than $50,000 has the voting percentage 77%.
The measures for attitudes towards immigration are under v5115 and v5116. As for the problem of N being too small, it looks like I need a new topic =).
Do those with strong personal or familial ties to the military turn out at a greater rate than the rest of the population? Does this demographic experience significantly higher turnout in wartime? My thesis is that those who serve or have served in the military, as well as their kin, voted at a higher rate in 2004 (during the Iraq War) than in 2000. Because military issues are highly politicized, affecting both party identification and views of George W. Bush, I will also control for these factors.
Name: Michelle Lopez
Null Hypothesis: People who are white, highly educated, and believe that the government pays attention to what people think are more likely to be contacted by a political party.
Too elementary?
Dependent Variable:
- if a party contacted voter X (v5008 in Post-Election Survey)
- other possible variables to consider (v5009 in Post-Election Survey, if other contacted voter X)
Independent Variable:
1. Demographical Variables:
-Race: I want to make this a dummy variable: 1 = white, 0 = non-white. Possible?
-Number of years of education:
Possible variables:
- highest grade/year of college completed (v3252 in Pre-Election Survey)*
- did you get a high school diploma (v3252a in Pre-Election Survey)
- highest degree earned (v3253b in Pre-Election Survey)
2. Attitudinal Variable (from the Post-Election Survey):
- How much attention does the government pay to what people think (v5203)
- Other possible attitudinal variables to use
o V5201: public officials don’t care what people think
o V5202: people like me don’t have any say in what the government does
o V5204: elections make governments pay attention to what people think
1) Ryan Loney
2) I'm interested in writing about how levels of political activity effects turnout from different backgrounds. In other words, how growing up wealthy/poor or race influences levels of activism and voting turnout.
3)Comparison of turnout based on level of activism, socio-economic background, and race.
Hypothesis: Those who vote in both presidential and midterm elections are more educated, have higher income and a greater sense of efficacy than those who only vote in presidential elections; as a result, those who only vote in presidential elections vote because they have been mobilized by a political party or other group.
A couple things: I want to compare 2004 and 2002 midterm elections, but I need the midterm election data. Can I find that? If so, where?
If midterm election data is not available is there a way to use the 2004 data and compare my variables (education, income, efficacy) with a national average to show discrepencies? Or should I find a new topic?
1) Andrew Betson
2) I am interested in whether religous people (different religions) turned out in 2004 and the midterm election of 2002.
3) Main relationship is whether religious people voted in 2004. Compare this is to religious turnout in 2002. I will control for how much the respondent liked Bush (religous fellow).
So I want to compare whether or not religious people turnout in numbers for both on and off year elections.
Name: Dano Wall
My proposal is to examine the relationship between turnout and attention to the media in times of relative peace and times of war, taking into account the respondents’ approval/disapproval of the war. The null hypothesis is that there is no relationship here. This paper, being a time-series type study, will examine the 1952, 1968, 1992 and 2004 elections. For the bivariate analysis I will use crosstabs to compare p210-didvote with v3012 (attention to national news). The multivariate analysis will be a comparison of correlations to control for v3133/v3134 (feeling about the Iraq/Vietnam war). My hypothesis is that turnout will increase in times of war, and I will investigate the effects of increased/decreased attention to media on this, as well as like/dislike of the war and the way it is being handled. In times of peace I hypothesize that these factors will decrease and have less impact or correlation with turnout.
My name is Xire, Xirejiancuo
I am interested to investigate different levels of income and how that affects turnout. The level of income is an important element that has great impacts on turnout. The people with higher income are more easily mobilized and they show more interest in politics than the poor. According to Rosenstone and Hansen, the wealthy ones vote, write, campaign and petition more than the poor. According to 2004 survey, the voting percentage of citizens with high income is almost double times higher than of those with low income. SDA shows that the poor with income lower than $20,000 has the voting percentage 48%, and the rich with income higher than $50,000 has the voting percentage 77%. (p210-didvote, income = v3293x)
Based on understanding this part, I want to deal with the levels of education between the wealthy and the poor, because usually the wealthy ones have higher education than the poor ones. People with high education have more interests in politics than those with lower education. The SDA survey shows that in 2004 election, the voting percentage of people with high school or low diplomas is 65% and the voting percentage of people with college and higher degrees is about 86%. (p210-didvote, education = v3254)
So, I will deal the dependent variable, “turnout” with two independent variables, “income” and education.”
So I'm dealing with these backwards.
XIRE.
Ok, this is fine but you are required to use at least one "attitudinal" (or non demographic) variable, not just demographics.
You implicitly include this in your theory. You claim that higher income and educated people are more interested in politics. So it is interest or is it education/income?
Figure out a way to control for BOTH to see if all have an effect.
DANO,
I think you can examine whether people who care more/less about Iraq would turn out.
But I don't think you can compare the size of the effect the way you suggest. Keep in mind that the variable numbers you list are only from the 2004 study. Are there comparable variables in 1968? In order to answer this question, you have to look at the cumulative survey.
If you look at the cumulative study, you could use V848, which deals with the danger of a nuclear war, and compare that effect to the Iraq effect.
But I'd suggest limiting yourself just to 2004 and think about how you can analyze the impact of attitudes abotu Iraq on turnout in 2004.
If you want to compare across years like that, you
Andrew B.
This looks straightforward. The difficulty you are going to have is operationalizing religion. Do you mean denomination? Frequency of attending services? Belief in biblical inerrancy?
You can't compare to 2002 but you can to 2000. There will be religion questions in that survey as well.
Bailey,
Unfortunately, the only way to compare midterm to presidential years is to use the cumulative study.
But from that study, look at these variables:
- V9030 and V9031 (contacts)
- V702 (did vote)
- V114 (income) and V140 (Education)
Make sure you control for year (v4). I'd suggest looking at 1998 and 2000, and comparing to 2004.
Ryan,
This looks just fine. I think we have all the measures already coded that you are interested in. It sounds like you want to compare activisim (non electoral) to turnout, and relate SES and race to both. If you have activism in another mode and turnout both coded 0,1, you can run regressions on both, using SES race, and one more variable as independent variables, and compare the size of the coefficients to assess relative effects.
Please try to consider a non-demographic variable as well.
Michelle,
I wanted you to consider turnout as a dependent variable. Would you think about putting all of that into your model?
So race, education, income = turnout.
Now what happens if you additionally consider contacts?
And you can still look at your qustions, which is how are contacts related to education, income,race.
Will Brown,
Good idea. How are you going to measure military involvement (what variables)? Can you think of a theory (using the language in the course) that tells us why military involvement would lead to higher turnout (hint: think of what is going on this year and how that impacts costs/benefits of the election outcome).
1) Zack Friedman
2) To what extent did the campaign issue of same-sex marriage affect turnout?
3) Although v3210, R position on gay marriage, is not very nuanced, as it just asks whether the respondent supports or opposes gay marriage without asking about the intensity of their opinion, v5074, Feeling Thermometer: Gay Men and Lesbians, correlates strongly with v3210, particularly among opponents of gay marriage. (I'd probably have to do some of that fun multivariate stuff to minimize error there, though, because there are enough instances where the two diverge.) There are a few routes I could take from there, potentially examining effects on turnout in states with gay marriage amendments on the ballot, comparing gay marriage supporters and opponents, or just how much the issue affects turnout if alternatives seem too complicated.
1. Alex Gersovitz
2. I'm generally interested in whether group membership, or rather an individual's identification with a certain group causes that individual to not only vote but also to believe that their vote matters. This examination of rational choice and political efficacy as they pertain to the formation of group identity probably can't be accomplished using purely SDA. However, a quantitative element of the result of these processes by looking at the voting patterns of a certain group on issues that, by the uniting characteristic of the group, the group would care about. Did group members vote based on a conscience defined by group membership or defined by higher principles of democracy and civic duty? Why did they believe that they're vote counted?
3. Thus, the embodiment of my research interest for the purpose of this paper will be an examination of turnout of Jewish voters in 2000 and 2004. In 2000, Lieberman (JEW) ran for VP, while in 2004 no Jews ran for president or vice president. I'll find some way to measure whether Jewish voting for Democrat or Republican was in some way influenced by the religious affiliation of the candidates. And also maybe some section on their policies in the Middle East, or whatever issues happen to be pertinent to Jews. Bivariate regression: didvote and religious affiliation in 2000 and 2004 as well as who the vote was cast for. Multivariate: I have no idea yet.
Alex,
May not be enough jewish voters in the survey to do this. Can you look at the numbers and find out?
Would be interesting to see if jewish id and turnout relationship differed in 2000 and 2004.
MV is easy: figure out a decent turnout model (say income, education, region, pol interst), then toss in your jewish dummy variable.
I like the general thrust (group ID + turnout). There may be other ways to get at this question.
zack
great topic. You are thinking about the measures precisely correctly. The problem will be if these are intercorrelated (you can check this) then we need to make some sort of scale or index out of them. I can help you with this.
LIke with Alex g., you need a well specified model of turnout, then add in the gay attitudes variables.
2) Do voters who consider themselves patriotic vote more often than those who don't?
3) dependent: didvote
independent:
-how good does seeing the flag make R feel?
-how strong is love for country?
-how important is being an american?
I also want to compare the results from 1996, 2000 and 2004 to see if there is a significant difference.
Brian Urrutia
Topic idea: I'd like to look at Patriotism and Income. Does a higher income status result in a stronger sense of patriotism? More importantly, do high income people turnout more than low income people, and do more patriotic people turn out more than less patriotic people? If these are combined, does turnout increase significantly?
My thesis is that people of higher income have a stronger sense of patriotism and thus, turnout more. I would like to investigate Partisanship as a control variable.
2] Given the positive effect of differences in "feelings" between two candidates on turnout, I would like to consider the effect of perceived differences in candidates' preferences on voter turnout.
3] Specifically, I would like to estimate a model of the effect of perceived differences between candidates' view on the optimal size of government (using the government services/spending scale variables v3138-9) on turnout while controlling for things that may proxy for/contribute to experience/information (such as education, age, and income). Moreover, I propose that, because a given voter's feelings regarding two candidates is likely influenced by the voter's perception of the candidates' preferences (i.e. a greater perceived preference difference may lead to a greater feeling difference), inclusion of the "p210-abs-ftdiff" will reduce the effect of both difference variables on voter turnout.
Do people actually vote strategically?
How common is it?
How do relationships between turnout and opinions of a vote's effectiveness, like v3096 (will pres. race be close in my state) compare to relationships between turnout and variables that reveal general feelings of voting's importance.
I am Sherry, I want to study the relationship between the state of the economy and voter turnout. I want to see whether voter turnout increases or decreases depending upon whether the state of the economy is considered good or bad. And I want to compare this for the years of 1996 and 2004.
Topic idea: Did people who belive the 2000 vote was unfair, v3004, have a lower turnout in 2004 than those who did not. My thesis is that people who believed that the 2000 vote was unfair turned out at a lower rate than those who did not. Also, would look at v3005 to see if how much they thought it was unfair mattered.
Name: Doug Barron
Does media exposure have a demonstrable effect on feelings of efficacy? Do media users more likely vote? Does media encourage participation or discourage participation by voting?
I will use p210-mediuse, p210-efficacy,l and p210-didvote.
According to readings, media exposure encourages participation by subsidizing part of the costs of participation. I am predicting to find that people with greater feelings of efficacy are more likely to vote.
Devin Judge-Lord
Null: Participation in local activism does not affect faith in national election.
Hypothesis:
People who participate through non-intuitional or local means (5267 protest, 5268 share concern, 5171 volunteer, 5166 worked on community problem, 5168 community meeting, 5169 protest, <5169member#orgs?>) also believe in their efficacy (5243 vote makes difference, 5247 voters represented) in national institutions like voting.
I think I will be running a multivariate with quite a few independent variables the describe faith in non-intuitional participation, and dependents of efficacy and then turnout. I may use efficacy as a independent when I do turnout.
I may also examine people's opinions of traditionally “activist” groups like environmentalists (5072) and feminists. (5059)
I want to change my topic slightly since I cannot find good data for patriotism in other surveys
2) Did voters who consider themselves patriotic vote more than those who don't?
3) dependent: didvote
independent:
-how good does seeing the flag make R feel?
-how strong is love for country?
-how important is being an american?
4) I want to then see if there is a coorelation between the patriotic factors and how partisaned the voter was to see if it was an indirect effect on turnout.
Bryson,
All looks good. Keep in mind when you do your "final" model, you should control for "standard" variables (income, education).
If you want to see if patriotism and partisanship are correlated, then do that exactly (that is, correlate the two). That doesn't exactly do the kind of causal ordering you want, but it gets you part of the way.
What will your bivariate breakdown be?
Devin,
What is your dependent variable? This sounds reasonable, but if the ultimate DV is voting turnout, this does not sound like what you are testing.
Perhaps what you mean to suggest is that faith in the outcome is positively associated with turnout? That seems to be that it woudl work OK.
As with many other posters, you are positing a two stage model: participation-->Faith-->turnout, which is feasible to estimate, but not with the tools we have.
Why not:
activism x turnout (biv)
faith x turnout (biv)
now faith + activism + otherstuff = turnout.
Doug,
We talked at the office hours. you are pretty much set, I think. Let me know if I need to do more.
Calenzani,
Looking good. Biv. breakdown - the fairness variable x turnout. Make sure you multivariate model (if you do a regression) includes some of the "standard" indicators (income, educ, etc).
Sherry,
I think this is a good idea BUT you need to respecify a bit. You don't have data on the state of the national economy, you have data on PERCEPTIONS of the economy.
So you'd want to break people down into three categories, economy doing bad, economy neither good or bad, economy good.
Then compare turnout among these three groups. Can do this for 1996 and 2004.
Then run a turnout model, where turnout = age, income, educ, + your economic variable.
Xeno,
That looks good. This relates nicely to literature in Aldrich, Shepsle and Bonchek.
You might look at educational or income differences on your independent variable, to see if different kinds of people think differently about the election.
Christa,
You seem to be in great shape. Do you need help specifying the independent variable? Calculating the absolute differences in candidate distances is the way to go. You may want to use "euclidean" distances (see Hinich and Munger or the more complex parts of the Shepsle and Bonchek chapter on spatial modelling). Essentially, you square the difference rather than absolute difference, implying that greater distances have an even greater (squared) turnout effect.
Brian,
This seems fine but keep in mind that the ultimate dependent variable is turnout, not patriotism.
So you might find patriotism --> turnout, then explore who is mroe or less patriotic.
The nice thing about regression is that it controls for correlation between independent variables, so even though income and patriotism may be correlated, you can use both of them in your regression.
Yes use other controls. Don't just use partisanship, use the "folded" partisanship measure.
Post a Comment
<< Home