4

Unit 2: Adding control variables

Examining a bivariate relationship between dichotomous variables, with a dichotomous control variable.

o  Run cross tabulation on bivariate relationship

§  Analyze-Descriptives-Crosstabs;

§  insert I.V. & D.V.

§  request cells: observed counts, column and total percentages;

§  request statistics: Chi square & Cramer’s V

o  Request a control variable

§  Return to crosstab command

§  insert C.V. into box under “layer 1 of 1”

Sample results statement for crosstabulation of nominal level dichotomous variables with a dichotomous control variable.

While a majority (76.6%) of respondents in the 1996 GSS sample support capital punishment (DPS), there is a moderately strong (V= .129), statistically significant difference (p.<.000) in DPS by sex, with 82.6% of men favoring capital punishment, in contrast to 71.6% of women.

When political orientation is held constant, support for capital punishment is greater for both conservative men (90.0%) and women (79.9%) than for non-conservatives (78.2% and 67.2% respectively). Additionally, the percentage difference between men and women’s DPS increased for both conservatives (10.1% difference) and non-conservatives (11.0% difference), resulting in a moderately strong, statistically significant difference in DPS by sex for both conservatives (p.<.002; V=.140) and non-conservatives (p.<.000; V=.121).

The number of tables needed for control variable problems =

The number of attributes of the CV + 1

REMEMBER: ALWAYS EXAMINE THE BASIC IV/DV ASSOCIATION FIRST!!!!

1)  Examining a bivariate association and adding a control.

Q.1: Is there a statistically significant difference in support for marijuana legalization (GRASS) between male and female respondents in the GSS96TAB sample and is this more likely to hold true for white than black respondents?

( alternative wording: and do sex difference in support for marijuana legalization differ among white and black respondents OR and do sex difference in support for marijuana legalization remain when race is held constant?)

Independent variable (IV): SEX Dependent variable (DV): GRASS

Control variable (CV): RACE

Step 1: Examine the basic IV/DV association: Is there a statistically significant difference in support for marijuana legalization (GRASS) between male and female respondents in the GSS96TAB sample

Table 1: Sex differences in support
for marijuana legalization
SEX
GRASS / male / female / total
legal
Not legal
Total
Pearson X2:p.
/ Cramer’s V=
Source: 1996 GSS tab subset

Step 2: Examine the basic IV/DV association with the control condition: …and is this more likely to hold true for white than black respondents?

Table 1B: Sex differences in support
for marijuana legalization among
black respondents
SEX
GRASS / male / female / total
legal
Not legal
Total
Pearson X2:p.
/ Cramer’s V=
Source: 1996 GSS tab subset
Table 1A: Sex differences in support
for marijuana legalization among
white respondents
SEX
GRASS / male / female / total
legal
Not legal
Total
Pearson X2:p.
/ Cramer’s V=
Source: 1996 GSS tab subset

Switching IV and CV

Q.2: Is there a statistically significant difference in support for marijuana legalization (GRASS) between white and black respondents in the GSS96TAB sample and is this more likely to hold true for male than for female respondents?

Independent variable (IV): RACE Dependent variable (DV): GRASS

Control variable (CV): SEX

Step 1: Examine the basic IV/DV association: Is there a statistically significant difference in support for marijuana legalization (GRASS) between white and black respondents in the GSS96TAB sample

Table 1: Race differences in support
for marijuana legalization
RACE
GRASS / White / black / total
legal
Not legal
Total
Pearson X2:p.
/ Cramer’s V=
Source: 1996 GSS tab subset

Step 2: Examine the basic IV/DV association with the control condition: …and is this more likely to hold true for male than female respondents?

Table 1B: Race differences in support
for marijuana legalization among
female respondents
RACE
GRASS / White / Black / total
legal
Not legal
Total
Pearson X2:p.
/ Cramer’s V=
Source: 1996 GSS tab subset
Table 1A: Race differences in support
for marijuana legalization among
male respondents
RACE
GRASS / White / Black / total
legal
Not legal
Total
Pearson X2:p.
/ Cramer’s V=
Source: 1996 GSS tab subset

3.) Another wording: Do gender differences in support for capital punishment vary by race in the 1996 GSS sample?

Independent variable (IV): Dependent variable (DV):

Control variable (CV):

Step 1: Examine the basic IV/DV association:

Table 1:
CAPPUN / total
favor
oppose
Total
Pearson X2:p.
/ Cramer’s V=
Source: 1996 GSS tab subset

Step 2: Examine the basic IV/DV association with the control condition

Table 1B:
CAPPUN / total
favor
oppose
Total
Pearson X2:p.
/ Cramer’s V=
Source: 1996 GSS tab subset
Table 1A:
CAPPUN / total
favor
oppose
Total
Pearson X2:p.
/ Cramer’s V=
Source: 1996 GSS tab subset