Assignment 8: Statistical Control

Assignment 8: Statistical Control

Date Assigned: January 31/February 2

Title of Assignment: Statistical Control

Due Date: February 7/9

Marks: 15 {see numbers in brackets for amount for each question or task}

Up to this point in the course, we have tended to isolate two variables from the complex interplay of political life in order to consider the relationship between a dependent and one or more independent variables. Now we look at relationships in the context of other variables. We recommend that you approach this exercise as a part of your ongoing research as many of you will be able to re-examine relationships you studied for the bivariate assignment (among others).

Use crosstabulations (also known as contingency tables) to investigate the interrelations among three variables: an independent variable, a dependent variable, and a test or control variable. If you are still unclear about these terms, please consult the Meier and Brudney chapter or the Korey text. Each describes some of the relationships that may be found when a simple two-variable relationship (independent and dependent variables) is tested by introducing a third variable.

Follow these steps to complete this assignment:

1) Select an independent variable and a dependent variable

IV: _______________________________________________ ---->

DV: _______________________________________________

2) What is your hypothesis? In a sentence or two, justify your hypothesis {1}.

________________________________________________________________________

3) Crosstabulate these variables.

4) Analyze the contingency table. What is the measure of association? Is this relationship statistically significant {2}?

________________________________________________________________________

5) Select a control variable that you think affects this relationship.

_________________________________________________________

6) What effect do you hypothesize it will have on your original relationship? Why? {1+}

________________________________________________________________________

7) Recode, if needed, your control variable into two or three categories. Re-run the original crosstabulation while controlling for this third variable.

8) Present the new tables clearly and aesthetically. Analyze EACH partial relationship in a few sentences. {4}

9) In a simple table, show the measures of association of your original crosstabulation, as well as for each of your control tables like this:

	Tau-b	Significance
Original	0.21	0.002
Controlled for low education	0.10	0.082
Controlled for moderate education	0.19	0.023
Controlled for high education	0.33	0.001

Are the differences in the measures of association in your control tables different enough from the original table (i.e. using the 1/3 rule or by looking at the standard error) so as to suggest the control variable has an effect on the relationship? Is this difference substantively significant? {2+}

10) What pattern of elaboration have you discovered? Do the results illustrate “replication,” “specification,” a spurious relationship or what? Explain. Was it consistent with your hypothesis? Explain why this is so. Conclude using your own words on what effect the control variable has on your original relationship? Speculate on the causal order of the relationship, especially if you think the relationship is spurious. {3}

11) Hand in all of your output including syntax.

Be thoughtful and creative in identifying relationships to test. We will reward efforts to test plausible or likely controls. So, make sure that your hypothesis about how the control variable affects your original relationship makes “sense.” You may investigate relationships with a classmate, but the written analysis you hand in must be your own solo efforts.