Data & Research Question

I will be analyzing data from the 2018 VDSG “Voter” Dataset to determine whether or not there is a relationship between political party identification, and support for affirmative action. Specifically, I hypothesize that Democrats are more likely to favor affirmative action than Republicans.

Below is a preview of the data:

## # A tibble: 6 x 2
##   PartyIdentification AffirmativeAction
##   <chr>               <chr>            
## 1 Democrat            Favor            
## 2 Republican          Oppose           
## 3 Republican          Favor            
## 4 Republican          Oppose           
## 5 Democrat            Favor            
## 6 Democrat            Favor

Expected vs Observed Values (Null vs Actual)

Expected Values

Below are the expected values, which we would observe if the null hypothesis were true. As a reminder, the null hypothesis is that political party identification and support for affirmative action are independent of one another.

## $expected
##                          voter$AffirmativeAction
## voter$PartyIdentification   Favor Not sure   Oppose
##                Democrat   963.931 692.9668 1327.102
##                Republican 740.069 532.0332 1018.898

Observed Values

Below are the observed values, as summarized from the voter dataset. These values are compared against the expected values to determine whether or not the two variables are independent of one another.

## $observed
##                          voter$AffirmativeAction
## voter$PartyIdentification Favor Not sure Oppose
##                Democrat    1502      850    632
##                Republican   202      375   1714

Upon initial comparison, it seems that there is a substantial difference between the expected values and the observed values. For example:

  • I expect only 963 Democrats to Favor Affirmative Action under the null hypothesis, however I observe 1502 respondents who meet these criteria.
  • I expect only 740 Republicans to Favor Affirmative Action under the null hypothesis, however I observe only 202 respondents who meet these criteria.

Relationship of Interest

I suspect that support for affirmative action (dependent variable), depends on a persons political identity (independent variable). The crosstab below shows how respondents in each political party distribute in their support for affirmative action (row %).

##             
##              Favor Not sure Oppose
##   Democrat    0.50     0.28   0.21
##   Republican  0.09     0.16   0.75

This crosstab lends further general support to my hypothesis.

Chi-Square Test for Independence

## 
##  Pearson's Chi-squared test
## 
## data:  voter$PartyIdentification and voter$AffirmativeAction
## X-squared = 1611.8, df = 2, p-value < 2.2e-16

The results of the chi-square test support my rejection of the null hypothesis. The p-value, indicated in scientific notation as <2.2e-16, is less than the threshold required to claim a statistically significant relationship (p<.05). In other words, there is a statistically signficant relationship between political party identification, and support for affirmative action. Theses variables are not independent of one another.