Summary: This short essay explores four models for understanding political parties and ideologies:

1. Each party has an ideology that represents positions that fall somewhere on the left-right spectrum;
2. Each party represents a temperament or underlying principle, such as traditionalism or progress;
3. Each party represents an interest-group coalition, such as the workers or business;
4. Each ideology represents whatever its major associated political party stands for at the moment. In turn, per Cohen, March and Olsen (1972), any political party is a “collection of choices looking for problems, issues and feelings looking for decision situations in which they might be aired, solutions looking for issues to which they might be the answer, and decision makers looking for work.”

I argue that the first three models don’t fit US politics by themselves, and the last one (a “garbage-can” model) has some validity.

We are familiar with a model in which each political party promotes an ideology, and we can place the various parties’ ideologies on a spectrum to tell how far apart they are, where the median lies, and whether the right or left is more influential. When this model is applied to US politics empirically, the typical finding is that our parties have moved apart or “polarized.”

Verlan Lewis (2021) has argued that most empirical measures of polarization do not inquire into the content of the left or right positions. They identify statistical clusters that they label as ideologies, but they do not tell us what the ideologies stand for. Closer inspection reveals that the meaning of the ideological labels has changed drastically over time.

As Lewis notes, “in the 1960s, liberal MCs [Members of Congress] tended to vote against tax increases and in favor of tax cuts, while conservative MCs tended to vote just the opposite.” This statistical relationship was very strong. The words “liberal” and “conservative” later changed their meanings so that conservatives are now the tax-cutters.

Lewis also illustrates his critique of the standard “static” model with the examples of three 20th-century senators: “‘Cotton’ Ed Smith (D-SC, 1909–1944), Henry ‘Scoop’ Jackson (D-WA, 1953–1983), and Ron Wyden (D-OR, 1996–present).” All three have the same ideological score–left of the median–on the influential DW-NOMINATE scale, yet “Smith was a racist demagogue who opposed the New Deal, Jackson was a ‘neoconservative’ who supported both the Great Society and the Vietnam War, and Wyden is a ‘progressive liberal’ who opposes racism, has sought to reform entitlement spending, and opposes militarism.”

Lewis concludes, “As we can see, what it meant to be a ‘liberal’ MC in the 1930s was very different from what it meant to be a ‘liberal’ MC in the 1970s, and both are very different from what it means to be a ‘liberal’ MC today.” 

We might try to detect some underlying values or dispositions that define ideologies over time. One candidate: conservatives want to preserve something or return to the past, whereas progressives want to move forward.

I think that American progressives from 1932 until 1970 were, indeed, temperamentally oriented to change, while conservatives during that period wanted to hold onto traditions. Since then, however, I observe that progressives often want to preserve and conserve institutions that have become traditional (neighborhood public schools. welfare programs, unions) whereas conservatives from Reagan and Gingrich to G.W. Bush (not to mention Trump) embrace radical change. The temperamental orientation of the ideologies has switched.

A third possible model assumes that parties change their positions–and even their temperaments–but they retain the same core interest groups over time. We might expect a given country to have a party for the workers and one for the bourgeoisie, with potentially a third for the peasants. Perhaps the US has only bourgeois parties, but Republicans rely on business-owners and professionals from suburbs and small towns in the North, while Democrats depend on farmers plus urban industrial workers.

The problem with this third model is that the parties prove surprisingly likely to change their interest groups. Indeed, upscale professionals in northern suburbs are now at the heart of the Democratic coalition, while rural people in the South are core Republicans; and Northern industrial workers tilt to the GOP. Each of these groups has switched sides.

Nor is this pattern unique to the USA. The UK Labour Party, formed to represent industrial workers, drew 38 percent of the most advantaged social stratum in the 2024 General Election, compared to the Tory’s 18 percent. Labour performed a little worse among semi-skilled and skilled laborers than among managerial and professional employees. In France, the supposedly left-wing New Popular Front performed worst among workers (ouvriers), and was the top choice of the managerial class (cadres). The German Social Democratic Party, formed in 1875 to represent workers, now performs better among white-collar workers with high education.

if these models based on issues, temperaments, or interest groups fail, what model could work? I’d turn to Cohen, March and Olsen (1972), who posited that any “organization is a collection of choices looking for problems, issues and feelings looking for decision situations in which they might be aired, solutions looking for issues to which they might be the answer, and decision makers looking for work. … To understand processes within organizations, one can view a choice opportunity as a garbage can into which various kinds of problems and solutions are dumped by participants as they are generated” (Cohen, March & Olsen 1972).

If this model applies to politics, then a given party is not a manifestation of any specific principles, nor an agent for a given demographic coalition. It is a space within which various actors can participate, yielding various outcomes over time. In turn, an ideology–at least in a regime like the USA–is mainly the name for that set of views that is currently held by one of the parties.

In that case, it is not illogical if the word “liberal” comes to mean entirely different policy positions over time; that is the outcome of people dumping “various kinds of problems and solutions” into the garbage can of the Democratic Party, which then represents “liberalism.” (And the same for the GOP and conservatism.)

In fact, I don’t think the garbage-can model quite works for US parties. They do retain some philosophical premises and portions of their coalitions over substantial periods, and to some extent, their changes in positions reflect changes in the external world. For example, the parties may have switched their positions on isolationism versus interventionism because the main perceived adversary was communism for 45 years–but not before or after that.

Still, the first three models don’t fit by themselves, and the garbage-can model has some validity.

---

Sources: Lewis, V. (2021). The problem of Donald Trump and the Static Spectrum Fallacy. Party Politics, 27(4), 605-618; Cohen, M. D., March, J. G., & Olsen, J. P. (1972). A garbage can model of organizational choice. Administrative science quarterly, 1-25. See also: Mapping Ideologies as Networks of Ideas; in defense of institutions as “garbage cans”; what if political parties structure our thinking for us?; UK election results by social class; social class inversion in the 2022 US elections;  class inversion as an alternative to the polarization thesis; class inversion in France and what does the European Green surge mean?