READINGS

GUIDE 1: ISSUES IN MODELING GUIDE 2: TERMINLOGY GUIDE 3: THE LOWLY 2 X 2 TABLE GUIDE 4: BASICS ON FITTING MODELS GUIDE 5: SOME REVIEW, EXTENSIONS, LOGITS GUIDE 6: LOGLINEAR & LOGIT MODELS GUIDE 7: LOG-ODDS AND MEASURES OF FIT GUIDE 8: LOGITS,LAMBDAS & OTHER GENERAL THOUGHTS

OVERVIEW

Due to a variety of technology issues, we have some schedule changes.

Assignment 3 will move to April 2

The paper precis is due March 19.

If you do a rough draft it is due by April 16.

All final drafts are due Tuesday April 28 by 5 PM
 
 

Each student (or possibly student group) will do an analytic paper on a topic of their choice.

Don't worry if you're not familiar with a lot of the technical terms right now, because that's what the course is going to cover! The following is a rough calendar for you to use so that you can approximately plan your semester.

I like to work with a rewrite option. If you can get me a rough draft of your paper by April 16, I will return it to you the 23rd so that you can take feedback into consideration and rewrite it for Tuesday April 28th.  (FIRM DEADLINE)

You may wish to work with other students on a group paper.

General Paper Parameters and what I will need on March 5  19

What I need for March 19:

A brief problem statement of what you are going to do

Your planned data set

Your planned 3-6 variables

What kind of analysis you would like to do

Length: I would like to suggest a paper length equivalent to AT MOST 10 DOUBLE-SPACED PAGES (including diagrams and tables; 15 pages for group papers). Parsimony counts.

You will need to analyze between 3 and 6 variables (total) in your model. Please do NOT use more than 6 variables in this beginning analysis.

You will need a "PROBLEM STATEMENT" at the beginning of your paper. The problem statement must:

• describe what you will study (e.g., adult understanding of science inquiry)

• tell why this is an important topic to study (e.g., we have almost no research on how the typical U.S. adult understands how scientists "know things"; this hampers laypersons as they try to understand new developments in medicine, technology, education and other fields which use scientific methods)

• what your study will add to our current knowledge (e.g., representative samples of adults; examining sample questions about understanding inquiry rather than just memorized facts; examining characteristics of adults that raise or lower inquiry understanding)

• Who or what was studied? (Examples: adults in the United States; members of professional sports teams; Florida corporations with at least 100 employees)

• Whether this is a sample or a population

• How were the data obtained? (Examples: the General Social Survey, 2006; Project TAL, 1997; stock exchange rosters; students in Introductory Psychology courses)

• What is the case base size in your data that you will analyze?

• What constructs do these variables reflect? (e.g., science inquiry measured through a survey question on the experimental method and 4 survey research questions on applied probability)

• How were they measured? If you are doing recoding (e.g., collapsing categories), what do you plan to do?

• How many categories does each variable have? What are the nature of these categories (e.g., nominal, ordinal, numeric?)

• Double check and explain any sizable amounts of missing data (e.g., over 5 percent of the total case base).

What is your planned general method of analysis?

Do you plan to pick a dependent variable?

Are you interested in testing a causal model (if so, what does it look like)?

Do you anticipate any interaction effects (which ones?)

Do you expect any pairwise correlations to be zero? Which ones and why?

AND if you will elect the rewrite option ALSO by April 9   16:

• Include the starting saturated model as well as the model with the best fit. (Everyone)

• The G²s, degrees of freedom, probability levels, R² analogues for the models you discuss (Everyone)

• Evaluate how well your final model fit the data (Everyone)

• SHOW how  to calculate the degrees of freedom for your final model (Everyone)

• Briefly describe any alternative models that also seemed to fit the data (if appropriate) and why you selected the final model you chose

• causal diagrams (as appropriate)

• The partial associations and interaction effects that (a) you hypothesized and (b) that occurred in your results

• the models that you used and tested

What was the final numeric equation for your final model? What kind of coefficients did you use? (e.g.,  or  )

Finally, interpret your model succinctly IN WORDS. How were your variables related? Imagine that you are presenting a conference paper to a NON TECHNICAL audience. Explain the nature of the correlations or interaction effects to your audience.
 

I DO NOT grade on "how well your model works" but rather how well you work (construct and test) your model.

Yes, good spelling, writing and grammar are expected (it doesn't have to be "native English," so don't worry about that one.)

Use articles from some of your favorite journals as a model (although shorter, of course).

You have your choice of datasets, variables and models. For those needing data, I have several sets available including many surveys of Tallahassee between 1974 and 1997 (wow!), the entire General Social Survey (1972-2006), two RDD surveys of Florida, a survey of Warren County, Ohio, the NSF Surveys of Public Understanding of Science and Technology (1979-2006) AND MORE! MORE! All of these are in various SPSS formats (Just ask.)  I encourage you to develop your thesis or dissertation interests if applicable.

If you are interested in any of my datasets, see me by March 19.

WANT TO DO A ROUGH DRAFT? IT IS DUE--ABSOLUTE LATEST--APRIL 16!! HARD COPY.
 
 

OVERVIEW

READINGS

This page created with Netscape Composer
and is best viewed with Netscape Navigator
600 X 800 display resolution.
Susan Carol Losh
January 5 2009
March 17 2009