Recoding party identification to Democrats, independents, and Republicans
It often is desirable to simplify the party identification variable so that it has fewer than seven categories. Most often, this is done by recoding the variable to produce three categories: Democrats, independents, and Republicans. There are two possible ways to create these three categories:
We can combine everyone who indicates some attachment to the Democratic Party (strong Democrats, weak Democrats, and independent Democrats) into one category. The same can be done for those who have some attachment to the Republican Party. This will yield three groups: Democrats, independents, and Republicans. The independents in this case will be only the pure independents--those who do not lean toward one of the parties.
An alternative method is to combine strong and weak Democrats into one group, strong and weak Republicans into another group, and all independents (including those who lean toward one of the parties) into a third group. This also yield three groups: Democrats, independents, and Republicans. In this case, more respondents will be included in the independent category and fewer in the Democratic and Republican categories.
We suggest using the first method to recode party identification into three categories, as we feel that those who lean toward one of the parties are often not much different from those who weakly identify with a party. However, some analysts would disagree with this. In the analysis exercises, we have assumed that the first method of recoding party identification is used wherever you are asked to recode the variable. Somewhat different results might be obtained if the second method is used for recoding.