Partisanship

Goal

The goal of this module is to explore basic differences in partisanship by demographic characteristics, including age, race, income, and gender. Crosstabulation and graphs will be used.

Concept

Partisanship is generally defined as an individual's inclination to favor one political party over another. Political scientists do not consider formal enrollment in a political party a necessary part of partisanship. Instead, if a citizen merely feels closer to one party or the other, he or she is considered to be a partisan. Partisanship is often related to an individual's beliefs and attitudes. Knowing someone's partisan identity gives researchers insight into their attitudes toward political issues and preferences for candidates.

Examples of possible research questions about political partisanship:

  • As people age is their partisanship likely to change?
  • Are there differences in partisanship by gender?
  • Does partisanship differ between racial groups?
  • What is the relationship between income and partisanship?
  • How are partisanship and voter turnout linked?

Data for this exercise come from the 2004 American National Election Study, collected by the Center for Political Studies at the University of Michigan. The 2004 ANES consists of a time series study conducted twice, once before and once after the 2004 U.S. presidential election. ANES produces high quality data on voting, public opinion, and political participation to serve the research needs of social scientists, teachers, students, policy makers and journalists who want to better understand the theoretical and empirical foundations of national election outcomes.

Respondents in the 2004 ANES were interviewed before and after the November election. Questions asked of respondents during the survey cover a broad range of topics including attitudes toward candidates and parties, attitudes about different segments of the American public, attitudes on foreign policy matters, specific political behavior, and demographic characteristics.

Data from the 2004 ANES are representative of United States citizens from the 48 contiguous states, who were of voting age on or before the 2004 Election.

This exercise will use the following variables:

  • Party Identification (V043116)
  • Respondent Age (V043249A)
  • Gender (V041109A)
  • Race (V043299A)
  • Household Income (V043293X)

In this exercise, you will examine the relationship between demographic characteristics (age, race, income, and gender) and party identification using crosstabulation and interpreting graphs.

In this dataset, party identification (V043116) is measured on a 7-point scale where "Strong Democrat" is coded as 0, "Independent" is coded as 3, and "Strong Republican" is coded as 6. To simplify the analysis, we created a new party identification variable consisting of three categories: Democrats (coded as 1), Independents (coded as 2), and Republicans (coded as 3). The new variable is called "PARTY.CAT."

Partisanship and Age

For this exercise, the age variable (V043249A) was recoded into six categories: 18-29 (1), 30-39 (2), 40-49 (3), 50-59 (4), 60-69 (5), and 70+ (6). The new variable is called "AGE.CAT."

Run a crosstab of PARTY.CAT and AGE.CAT. Does this sample of survey respondents have more Democrats, Independents, or Republicans? Which party appears to have a larger share of members under 30?

Take a look at the line graph of the same data. You will see that the percentage of each age group by party category is plotted on the y-axis (vertically), and the age categories are plotted on the x-axis (horizontally). Does there appear to be a greater difference in partisanship between those under 40 or those over 60? How can you tell this?

Partisanship and Gender

Research on partisanship concludes that women are more likely than men to identify as Democrats. Is this conclusion supported by the crosstab of PARTY.CAT and gender? Do men and women differ in their attachment to a particular party?

Partisanship and Race

We recoded race (V043299A) into a dichotomous variable called "RACE.BINARY." In the new variable, race is coded as "White" or "Non-white."

Look at the crosstab of partisanship and race. What percentage of whites report being Republican? Are whites or non-whites more likely to be Independent?

Partisanship and Income

In the ANES, the variable measuring household income contains 23 categories. We recoded household income into a new variable with only 4 categories: $0-24,999 (1); $25,000-49,999 (2); $50,000-79,999 (3); and $80,000 or more (4). The new variable is called "HINC4."

Consider the crosstab of partisanship and household income. Compare the highest income level to the lowest income level. Which group is more likely to be Republican? From the group making under $50,000, which party gets the most support?

Think about your answers to the application questions before you click through to the interpretation guide for help in answering them.

Does this sample of survey respondents have more Democrats, Independents, or Republicans? Which party appears to have a larger share of members under 30?

Which party or parties appear to have the greatest membership on the graph? Is there a greater difference in partisanship between those under 40 or for those over 60? How can you tell this?

Are women more likely than men to identify as Democrat? Are men more likely to identify as Republican than Democrat?

What percentage of whites are Republican? Are whites more likely than non-whites to identify as Independent?

Compare the highest income level to the lowest income level. Which group is more likely to be Republican? From the group making under $50,000, which party gets the most support?

Interpretation

Things to think about when interpreting the results:

  • It is always important to consider how missing data might affect the ability to make inferential statements about the population from which the sample was drawn. Large amounts of missing data can limit our ability to generalize sample results. Even small amounts of missing data can affect the results, especially if missing responses are not randomly distributed across categories of the independent (grouping) variable.
  • Weights (mathematical formulas) are often used to adjust the sample proportions, usually by race, sex, or age, to more closely match those of the general population. The analyses used in this guide did not use any weights, which may reduce the generalizability of the findings, but the resulting tables are accurate descriptions of the relationships found between these variables among these respondents.
  • By looking at the row totals, we see that there are 589 Democrats, 115 Independents, and 479 Republicans, therefore there are more Democrats. In the columns associated with the18-29 and 30-39 year-olds, we see that Democrats have 56.5% and 53.5%, respectively and the dark red color in those cells indicates that these percentages are higher than we would expect given the percentage of all respondents who identified as Democrats is 49.8%. Correspondingly, we look at the same categories (18-29 and 30-39) for Independents and Republicans and the percentages are much smaller. Therefore, we conclude that respondents under 40 years of age are more likely to identify as Democrat than either Republican or Independent.
  • When we interpret the line graph, we see that two lines (red and green) are much higher on the graph than the blue line (Independents). The red line (Democrat) is the highest with the green line (Republican) just below, therefore, Democrats and Republicans appear to have the highest membership among the survey respondents. The difference in partisanship between those under 40 is greater than that of those over 60. We can tell this by how wide the distance is between the red and green lines above the age categories in question. The lines are much closer together for the over 60 categories than for those under 40.
  • The crosstab shows that 53.9% of women identify as Democrat, compared to 44.6% of men. This finding adheres with previous research on partisanship and gender. On the other hand, 44.3% of men identify as Republicans compared to 37.3% of women. While it is reasonably clear that women are more likely to identify as Democrat as opposed to Republican, it is not clear that men are more likely to identify as either Democrat or Republican over the other party as the percentages are virtually equal (44.6% Democrat versus 44.3% Republican).
  • 49.2% of white respondents identify as Republican. Non-whites are slightly more likely than whites to identify as Independent in this sample (12.1% compared to 9.0%). However, the relatively medium shading in the cells of the Independent row tell us that we cannot be very confident that this relationship is reflected in the population.
  • When we compare the highest income level, those making $80,000 and over, to the lowest income level, those making up to $24,999, we find that a higher percentage of those in the upper income are more likely to identify as Republican. The majority of the two lowest income categories (57.8% and 52.2%, respectively) identify as Democrat. It does appear that higher income is associated with a lower likelihood of identifying as a Democrat.

Summary

The goal of this exercise was to explore the relationship between demographic characteristics and partisan identification. Taken together, the results of the analyses suggest that there are differences in partisanship by age, income level, race, and gender. However, these analyses only explore bivariate relationships. Multivariate analyses would likely reveal more complicated relationships among the variables used in these analyses. Further, it would also be possible to explore the independent effect each of these variables has on partisanship in the same equation, while controlling for some other potential effects.

CITATION: Inter-university Consortium for Political and Social Research. Partisanship: A Data-Driven Learning Guide. Ann Arbor, MI: Inter-university Consortium for Political and Social Research [distributor], 2009-04-16. Doi: https://doi.org/10.3886/partisanship

Creative Commons License This work is licensed under a Creative Commons Attribution-Noncommercial 3.0 United States License.