Monday, November 10, 2008

Independent, dependent, and confounding variables

My little sister is taking a poli sci class and needed help with her project so I took a chance and said I'd work with her. This was a *slight* mistake on my part; it'd been so long since I'd had to worry about interpreting graphs and data and correlations since my epidemiology class. But I still walked her through it and explained the importance of the independent variables, and the dependent variables. I wanted her to test a null hypothesis of the political ideologies between those who felt strongly about religion and the general population. For some strange reason she wasn't requested to account for confounding variables (that might come later, though) with regards to who followed what religion and their political ideology. Can you see the problem with this?
I think age might be a confounding variable, but I'm not sure (it's hard to definitively identify confounders). Is it really age or religion that decides political ideology? I think being old, and religious makes you more likely to follow a conservative ideology, and being young makes you more likely to follow a liberal ideology regardless of religion. Yet, your religion might have an influence on your ideology. So, the only way to tell if age is confounding or not would be to have the religious groups stratified (this way you could see if younger members of differing religions follow liberal ideologies and older members of differing religions follow conservative ideologies) or to have a case-control type study (limit the people who answered the questions to being a certain age group). Can you think of anymore sources of error?

2 comments:

Lara Newell said...

No. Your post was confusing. Too bad you will be gone next year when I take biostatistics.

Unknown said...

I'm not sure I understand what was confusing, knothead. Being older might have an impact on your political ideology, but it's hard to determine to what extent without trying to eliminate this factor. So, you'd have people all of the same age and compare their religious and political affiliations/ideologies, or you'd have people of varying ages and you'd compare their religion and ideologies. If the null hypothesis were true, that there is no difference between religious people's ideologies and the general population's ideologies, then you could conclude that your sample size was a good match to the general population. But the null hypothesis is usually the opposite of what you'd believe, since you'd assume religion affects ideology. The null hypothesis is just a way of determining the difference between groups or data and if it's significant enough to use an alternate hypothesis or if your results were caused by chance.

 
Creative Commons License
This work is licensed under a Creative Commons Attribution 3.0 Unported License.