NOVEMBER 5
DIFFERENCE
OF MEANS TEST 

Schmidt, Ch.11: "Hypothesis Testing: The Difference between Two Means," pp. 299-319.

The one-sample test has only limited applicability in political research. More interesting research arises for two sample tests. For example, one might test for significant differences between a sample of Democrats and a sample of Republicans, between a sample of developed nations and a sample of underdeveloped nations. We sometimes say that two-tailed tests are used for nondirectional hypotheses, while one-tailed tests are for directional hypotheses. Make sure you grasp this distinction between types of hypotheses.

NOVEMBER 6
THE T-TEST IN POLITICAL RESEARCH

SPSS Users' Guide, Chapter 8, "Using T Tests to Compare Means," pp. 103-119

Assignment: Using the World95 file, prepare a directional hypothesis concerning the differences between "Catholics" and "Muslims" using any of the interval variables in the file as your dependent variable. Because Religion is a polychotomous variable, you will have to choose only two groups--which accounts for the Catholic/Muslim comparison.

Under the Analyze Menu, choose the Compare Means procedures. Select Independent Samples T Test. Use "Religion" as your Grouping Variable and enter "Catholic" for Group 1 and "Muslim" for Group 2. (The Users' Guide shows how on page 111.)

Interpret your results and bring the output to class. You'll see some computer printout from the T-TEST program on the 2/3 exam.

NOVEMBER 7

Optional session: Review for 2/3 examination

NOVEMBER 8
ONE-WAY ANALYSIS OF VARIANCE

Schmidt, Ch.14: "Introduction to the Analysis of Variance," pp. 373-392.

Analysis of variance is suitable for analyzing the effects of a nominal-level independent variable (such as regions of the country) on an interval-level dependent variable (such as percent vote for the Democratic candidate for president). It is especially useful in analyzing the results of psychological experiments, in which subjects are assigned to various "treatment groups" (the independent variable), and the effects of these treatments are assessed on some variable such as memory retention, aggression, problem-solving, and so on. Analysis of variance is less frequently used in social and political research, but the ideas underlying analysis of variance are conceptually important to multiple regression analysis, which is a standard technique in both. In reading this chapter, be sure you understand the within, between, and total sums of squares. You will have to interpret SPSS ANOVA output on the 2/3 examination, so better learn how to read the ANOVA table. Also understand the nature of the F-distribution, which you should be able to distinguish from the t-distribution, to which it is related. (How?)

NOVEMBER 9
ONE-WAY ANALYSIS OF VARIANCE IN RESEARCH

SPSS Users' Guide, Chapter 9, "One-Way Analysis of Variance," pp. 121-133.

Virginia P. Lacy, "Political Knowledge of College Activist Groups: SDS, YAF, and YD," Journal of Politics, 33 (August, 1971), 840-845. (On website)

The Lacy article is an especially clear usage of analysis of variance to analyze differences among activist students during the Vietnam war era.

Assignment: As I mentioned, SPSS 10 believes that it knows what is good for you. SPSS thinks it's OK to use the string variable "Religion" inn the world95 file as a nominal variable for a T-Test. However, SPSS forbids using "Religion" as a nominal variable in ANOVA. So we'll fool SPSS 10 by resorting to an old syntax command. Do the following, using the world95 file:

• Go to the File Menu, choose New and click on Syntax to open a window for a syntax command.

  • In the syntax window, enter this command:

    • means tables=pop_incr by religion/statistics=all.

      • With the cursor on the same line, press control-r.

Using this old command, SPSS will use the string variable "Religion" in an analysis of variance. Study the output in the SPSS viewer and try to interpret it with reference to your readings. It will have one table, "Measures of Association," that I'll need to discuss in class.

Should you want to use SPSS 10 to conduct ANOVA using the string variable "Religion," you'll need to convert it to what SPSS 10 deems is an aceptable nominal variable.

• Go to the Transform Menu and selecting Automatic Recode.

  • Move "Predominant Religion" into the "Variable -> New Name" window.

    • Click on "New Name" and type in "Faith"

      • Click on OK

This procedures creates a new nominal variable, "Faith," and produces the same ANOVA table. However, it does not produce the "Measures of Association" box.

NOVEMBER 12

2/3 EXAMINATION (Go to a review of steps in hypothesis testing)

NOVEMBER 13

Important session
Be There!

Discussion of Research papers

NOVEMBER 14

Optional Session: Discussion of Examination Results