Keywords: Political behavior, participation and culture

In political communication research, partisan labels limited to Democrat or Republican may prevent a deeper understanding of American voters and the political dynamics involved in election campaigns. If the 2016 election has made one thing clear, it is that neither Democrats nor Republicans are homogeneous groups. The commonly used labels mask important factions among voters within parties that have critical implications regarding media choices, world values, and voting behavior. Categorizing individuals that support diverse values into one of two groups glosses over such important distinctions.

Additionally, most research on political party identification has focused on partisan voters and the consequences of political polarization. Though the portion of independent voters has steadily grown over the past decades, far fewer studies have examined them. Too often researchers treat independents as a monolithic entity, using a single category to identify independents. Labeling individuals who do not identify with one of the two major parties as “Independent,” essentially groups together apples and oranges, obfuscating meaningful within-group differences.

In this context, the typical labels may not provide enough precision to understand and predict issue attitudes and voting behavior. Deriving from a strong theoretical basis, we build a more nuanced ideological typology that can match the current political environment and news media landscape in its fractionalization.
To formulate a contemporary typology, this study explores distinct clusters of voters that emerged in the context of the 2016 Presidential election. Using national survey data we collected shortly before the November elections (N = 2,582), we conducted a cluster analysis to classify individuals into subgroups that share similar profiles of opinions concerning 17 different personal values and worldviews.

We implemented two complementary analyses, using K-means clustering. In the first stage, we conducted a cluster analysis on the entire sample, identifying nine clusters of the voting electorate defined by common outlooks on political and social values. Responses were standardized twice, once by an individual's responses, and once across sample, in order to account for variation in response tendencies among our sample. Following the statistical analysis, we validated the results by examining the clusters and differences in their voting behaviors, political-social issue priorities, as well as their media diets. This qualitative validation was also conducted as part of the process to name the nine clusters, a task which we found as daunting as defining the appropriate number of clusters. In the second act, we repeated the analysis procedure on sub samples, based on the self-stated ideology in order to compare them to the clusters that emerged in the initial analysis.

Our results suggest that voters’ political attitudes are not organized in a single-ideology dimension. The political ideology is rather a multidimensional trait that should be measured in a more elaborated way so that can properly predict people’s political behavior including their voting choices. Furthermore, there were no "pure" clusters, meaning that all nine groups were comprised of self-claimed Republicans, Democrats, and Independents, further testifying the need for more meaningful ways to classify citizens.