A Jeanne Site
Statistics and Methods
California State University, Dominguez Hills
University of Wisconsin, Parkside
Latest update: July 1, 2000
Faculty on the Site.
- Introduction to the Statistics and Methods Resources Page
A sociologist's understanding of and need for math? No formulas, honest?
- Reading Tables
What is all this gibberish?
- Reading Graphs
Is this modern art?
- Using Dummy Tables
So who's calling whom a dummy?
- To Organize Complex Text
You mean you can draw a picture to make sense of this mess?
- To Plan Your Own Research
How could you forget to ask whether they're male or female?
- Reading Frequencies and Crosstabs
80% of them "hated school?" And these are the "good kids?"
- Using Frequencies and Crosstabs
Don't you think we ought to see if females like school more than males do? Wouldn't we have to look at both variables, attitude towards school AND gender?
- Programs That Do It All For You
The Statistical Package for the Social Sciences. So this is the software that does it all for you? Oh, It's not free??
Survey Documentation and Analysis. Oh, THIS is the software that's free, and that you an use from home??? Wow, school gets better all the time!
Geographic Information Systems. Not free, but CSUDH has a site license. Why would I care? Oh, there's free access to GRASSlinks at Berkeley? And this is an important area if I'm interested in ecology and environmental issues?
- Terms of Art
You mean the way that people know I'm smart is if I can throw around some of the terms, and I really do know what they mean? Is that like when they taught us vocabulary building, ten words a week???
- Descriptive Statistics
So my personal experience is only good for understanding people that have experiences like mine? You mean I shouldn't presume to think that someone in Tanzania feels just like I do, or sees the world the same way???
You mean these programs will give me the percentages, so I don't have to calculate them? Yippee!
I get it. Frequencies are just group snapshots. They show me how many of the people I studied were in each of the categories that I created. Is THAT what a grade curve is? But how did they study us to get the grade curve???
- Bar charts
You mean I can draw pictures of how many of the females like school and how many of the males like school, and these programs will draw the pictures for me, and I can print them up, and won't that make my term paper long and impressive????
- Tests of Significance
You mean that all these tests tell me is that there's a pretty good chance that there is something out there and that these numbers didn't just come up by chance? And I'm only 95% sure of THAT???
So, if I want to know how strong a relationship there is, I have to square r? Does the program do that for me???
So this test just tells me that I can be 95% sure that the average scores for the two groups are really different, and I didn't get this result by chance? You want me to learn all this to be 95% sure there is "something" our there other than random chance???
I see, this is the same foolishness, except with nominal data. Is this much ado about nothing, or what???
- Measures of Association
Well, if this is the real thing, why didn't we start with it? You mean these tell me how much better I can predict one variable if I know the other variable? Like, how much better can I predict whether you like school, if I know you are male or female?
So this one is for nominal data? So I could use it to tell how much better I could predict your political party identification if I knew your religious affiliation? Cool!
And this one is for ordinal data? So I could use it to tell how much better I could predict your rank on liberalism if I knew your rank on education level?? I got it!
This one's a little more complex? Can I leave that for an advanced course?
- Inferential Statistics
So this is where the problem with personal experience comes in? You mean that in order to draw conclusions about large groups on the basis of what I find in my personal experience or in my study, I have to make sure that the people I apply my results to are really like the people I studied. So how do I decide what "like" means? Oh, we call that the representativeness of the sample? I'll bet that leads right into the modernist/postmodernist issue!
- Population and Sample
So there are simple guidelines I can follow that will be acceptable for deciding if my sample is "representative?"
Is that what probability has to do with all this? When I just look at a small sample, and then infer or generalize what I find to a whole population, I have to worry about the probability that the sample gives me similar results to what I would have gotten if I had studied the whole population? So that's what all that 95% stuff is about? We're 95% sure that the sample we studied is "representative" or "like" the population.
- Representativeness of Sample
That's easy. I just have to be sure I don't study laboratory rats, and then extend my conclusions from their behavior to that of college sophomores. Right?
- Mathematical assumptions which can't be ignored
Now how on earth would I know about mathematical assumptions? I don't do math, remember? Oh, you mean assumptions like that the variable of intelligence is "normally distributed," like in the bell curve? Don't they make that assumption when they set up the grade curves, with very few As, mostly Cs, and few Fs? Oh, you mean I have to worry about whether grades would be a variable distributed the same way over all students as intelligence is distributed over the whole population? After all, students in college classrooms are not likely, if admissions standards work, to be deadbeats. So why should there be as many Fs as As? Maybe I do need to look at some of these assumptions.