# Statistics for a Social Scientist

#### wetslope

##### New Member
Hi folks, I am currently working on a PhD in a social science program. Although the statistics training at my department is pretty good for a social science program, they emphasize the applied aspect (very much so) while I would like to have a more thorough theoretical foundation. In this regard, I have decided to take stat courses at the math/stat departments. Before I ask any question let me just give you some background information regarding my previous math/stat training:

-Calculus (3 semesters)
-Linear Algebra
-Differential Equations
-Intro to Proofs
-Prob and Stats(non-calculus)

*The following were done in a single semester long course:

-Basic Probability and Combinatorics
-RVs and Distributions (Normal, t, Binomial, Bernoulli but we did NOT cover: chi-square, poisson, geometric, neg-binomial, etc...)
-Statistics (CLT(did not prove), LoLN,Chebychev's Inequality, CI, Statistical Inference, OLS)

*next semester we will cover "problems" with OLS (heteroskedasticity, autocorrelation, etc), logit/probit regressions, and some of the RV distributions not covered in the first sem. And over the summer I plan to take two semesters of Real Analysis.

1) Given this, do I have to take the undergrad Probability(calc based) and Statistical Inference(calc based) at the math/stat department to take higher level stat courses?

2)Higher level stat coures I "plan" to take (rough list, please advice usefulness to social science research):

-Linear Regression Models
-Time Series Analysis
-Nonparametric Statistics
-Multilevel Models
-Causal Inference

I understand that grad level courses use measure theory (although i have no idea what that is) and my question is does undergrad Real Analysis 2 cover that? Here is a description of the course:

"Equicontinuity. Contraction maps with applications to existence theorems in analysis. Lebesgue measure and integral. Fourier series and Fourier transform"

Are any of those courses not advised for a social scientist? Would you add any other courses to the list?

EDIT: Also let me add that I am not too bad at math, although nowhere close to stellar. (For what its worth GRE-Q was 780)

Last edited:

#### Dragan

##### Super Moderator
Hi folks, I am currently working on a PhD in a social science program.

"Equicontinuity. Contraction maps with applications to existence theorems in analysis. Lebesgue measure and integral. Fourier series and Fourier transform"

Are any of those courses not advised for a social scientist...

What kind of social science program are you in (e.g. Political Science, Psychology, Sociology, Educational Research)?

I sort of doubt that Real Analysis is going to help you....although it might if you were in Economics.

#### wetslope

##### New Member
What kind of social science program are you in (e.g. Political Science, Psychology, Sociology, Educational Research)?

I sort of doubt that Real Analysis is going to help you....although it might if you were in Economics.
I'm taking Real Analysis since graduate level statistics courses seem to require it. How correct is this assessment? I am (or have put myself) in an "awkward" spot: I don't plan to do work on statistical methods themselves in my career but I do want a solid theoretical foundation for my applied work later. I do not want to be one of those people throwing commands into Stata or R not knowing what the hell is actually going on and I want to be able to prove things if need be.
Having said that, I have noticed since my last post that masters level statistics don't really seem to require knowledge of real analysis or measure theory. Clearly, my research has not been exhaustive yet, but what would you recommend?

Last edited: