Why do voters have stable attachments to political parties?

Tutor’s comments

None yet.

Essay

Some voters have stable attachments to political parties due to a combination of sociological and cognitive influences, both of which can be explained through a bounded rationality framework. Historically, social cleavages were the strongest contributor to voters’ stable attachments, alongside generational inheritance of partisanship independently of shared social characteristics. However, the ability of these sociological factors to explain the empirical findings has declined significantly over time, and they are not capable of fully explaining observed data from the 1970s onwards. In part, this is because the question’s assumption that voters do indeed have stable attachments to political parties, whilst still true for a sizeable proportion of citizens, is no longer sound across the board. This partisan dealignment provides evidence for the rational choice account over purely sociological explanations. In this essay, I will first clarify what is meant by “stable attachment” and present the sociological “Michigan model” along with historic arguments in its favour. I then show that the Michigan model’s predictions cannot be squared with recent empirical data, and outline the more explanatorily powerful rational choice account. Finally, I demonstrate how the waning strength of party identification across Western democracies and persistent instability of partisan attachments in developing democracies provide further evidence for the valenced partisanship account, and conclude that a rational agent approach gives the best explanation of why certain voters currently have stable attachments.

The term “stable attachment” was popularised in the literature by the Michigan model of voter choice, which held sociological factors like party identification to be the most important in determining who citizens would support in elections (Holmberg 2009, 557). When introducing the concept, Campbell et al. (1960, 121) stipulate that partisan attachment is not merely given by an individual’s voting history or even party membership status, but rather their “affective orientation” towards the relevant political group. To avoid begging the question and presuming an absence of deliberative rational components in the formation of such attachments, though, it may be more helpful to think of a stable attachment as when a voter remains instinctively supportive of a political party over an extended period of time. This conceptualisation still captures the idea that stable attachment involves feelings of loyalty and personal identification but avoids giving the impression of trivially dismissing the hypothesis that these affects are created by utility-maximising rational processes.

According to the Michigan model, sociological factors, specifically early socialisation and membership of social cleavages, play the central role in the development of voters’ stable attachments to political parties. Data from polling conducted in America in 1958 show that roughly 80% of individuals who had two Democrat-supporting parents maintained that partisan orientation, with the figure at around 70% for inheritance of Republican attachment (Campbell et al. 1960, Table 7.1). Further data are marshalled to demonstrate the durability of these attachments, with 93% and 80% of strong Democrats and Republicans respectively never having changed parties (Ibid, Table 7.2). This connection between personal background and party identification was not unique to America. In Britain, the dominant social cleavage until the mid-20th century was class, which had an extremely strong correlation with political attachment whether measured by vote share, party membership, or surveys asking directly about party identification (Clarke et al. 2004, 40-41). The strength of these partisan attachments also appeared to grow with age, something attributed to the fact that initial predispositions towards one particular party would reinforce themselves by influencing the way in which individuals interpreted political developments (Campbell et al. 1960, 133). Taken in the round, these observations provided strong evidence to support the hypothesis that sociological factors were the primary influencers of party identification.

However, developments in partisanship since the 1960s undermine this account. First, causality between voting behaviour and party identification appears to flow in the opposite direction to how the Michigan model would predict. Whilst the account does not imply that stable attachments are translated one-for-one into votes at election time, since not all partisan citizens will turn out to cast their ballot, it does view party identification as the “unmoved mover” located at the start of the causal chain determining voting behaviour in elections. Yet analysis of data in the Netherlands over a 25-year period shows that not only is party identification more volatile than voting behaviour, the latter appears to causally precede the former, with voters seeming to change their identification to match how they actually voted ( Thomassen & Rosema 2009 See also Rethinking partisanship: Some thoughts on a unified theory in the same book. ).

There are further problems for the sociological approach when analysing stable attachments as the dependent rather than independent variable. Since an individual’s early-life experiences and social attributes generally remain fixed over their lifetime, and the underlying social structure of a nation changes only very gradually with the effects of generational turnover, one would expect party identifications to remain relatively constant at both a micro and macro level, if these attachments actually are the result of sociological factors. This has not consistently been the case within Western democracies since the mid-20th century, though.

Even at the time the Michigan model was formulated, voters exhibited a considerable amount of flexibility in party identification. In the four-year period from 1956-1960, over 40% of respondents to the American National Election Studies survey did not maintain a stable partisan attachment Using so-called “turnover tables”. (Clarke & McCutcheon 2009). The substantial number of voters whose identifications changed in this short timeframe is not something the sociological account is well-placed to explain. At the population level, the proportion of British citizens without strong partisan attachments has increased steadily since the 1970s (De Vries & Hobolt 2020). The trend is mirrored across almost all The USA is a notable exception, where strong party ID has been increasing over time. This is, according to Holmberg, something that “many political pundits in the US seem unaware of”. One hypothesis about why we have this difference is that it’s because American politics is more candidate-centred than the European party-centred style of campaigning. Not sure of the details of this theory though. But America is a counterexample to cognitive mobilisation/modernisation theory, unless we’re prepared to say that it has de-modernised. Western countries: in a sample of nineteen advanced industrial democracies, all but two saw the level of political identification percentage fall over the course of the late 20th century (Dalton 2002, 26). This widespread and rapid phenomenon of dealignment is not compatible with the view that sociological factors are responsible for political identification, and indeed challenges the titular question’s assumption that voters do in fact have stable attachments to political parties.

Explaining voters’ stable attachments to political parties through the framework of bounded rationality provides a richer account of voting behaviour which fits much better with the past 50 years’ data. Individual voters can be modelled as rational agents seeking to maximise their utility subject to external constraints, including limits on their cognitive capacity (Simon 1955; Downs 1957, 207). According to Downs’s original rational choice model, voters would support and delegate their decision-making power to the party whose policies were closest to their own position in policy space, but analysis of British opinion polling suggests that this is not quite the case (Sanders et al. 2011). Rather, because most parties profess the same terminal goals (for instance, growing the economy or improving healthcare), voters select parties on the basis of which seems to be the most competent, honest, or strong – what Stokes (1992, 146) termed “valence issues”. The fundamental similarity between the positional and valence competition accounts, though, is that they treat voters as computationally constrained utility-maximising agents who employ heuristics to assist with their deliberations over select which party to support.

It is this use of shortcuts which provides a justification for why voters may have stable attachments to political parties. Citizens begin their adult lives with some pre-formed prior beliefs about the relative merits of each party, but gradually adjust these views over time with retrospective appraisals of how each party performed in government, with the effect of each update attenuating over time because of the mathematics of Bayesian updating. A voter’s party identification at a given time is simply the accumulation of these judgements There’s evidence that voters form their opinions based on the party leader as an individual too. This is a further cognitive shortcut – it’s easier to decide what you think about the party figurehead than to form a view about the entire organisation. This is particularly useful if valence competition is happening. , which have a degree of stability to them (Fiorina 1981, 94). When tested against polling on British voters’ party identification in three periods from 1966 to 1998, this conception of party identification as a running tally of backward-looking evaluations fits the observed data well, and significantly better than the Michigan model would (Clarke et al. 2004, 200). The rational choice model is therefore able to accommodate the fact that some voters frequently move parties, because significant new information has caused them to revise their overall judgements.

Moreover, the bounded rational agent approach correctly identifies which voters are likely to have the most stable attachments. Cognitive constraints vary between individuals, and one would expect more politically sophisticated There’s a definition of political sophistication in Kroh & Selb (2009): it “encompasses the cognitive skills to process information and a willingness to collect political information (Converse 1975).” individuals to obtain and assimilate information regarding the performance of political parties more frequently, and thereby have relatively changeable partisan attachments. As a result, the cognitive mobilisation hypothesis concludes that better-educated voters should have a greater level of political participation but tend to involve themselves in more active ways Dalton has an early paper on the consequences of cognitive mobilisation. than party membership, such as via protests and grassroots campaigning. These predictions are borne out by the data. In six out of seven European countries investigated, a higher level of education was associated with weaker party identification (Dalton 2002, 33), and detailed analysis of partisanship in Germany found that the least sophisticated voters simply held on to the party identification they inherited from their parents (Kroh & Selb 2009). Modernisation in Western democracies means that younger voters are increasingly well-educated, and citizens now have abundant access to information on which to base their retrospective judgements. We should therefore expect a weakening Though there’s a dissenting note in this paper which finds that cognitive mobilisation increases party attachment. of partisan attachment over time – exactly as has been the case (Dalton 2002, 34). Therefore, rational agent account not only successfully predicts which voters are likely to have more stable attachments, but also provides reasons behind the medium-term pattern of dealignment

Finally, the rational choice account can explain the lack of strong party identification in new democracies, where surveys have found that partisan attachments are generally weak or entirely absent (Holmberg 2009). While the sociological model would struggle to articulate why this would be the case, given that emerging democracies have social cleavages just as established ones do, the valenced partisanship approach gives a clear answer: people’s evaluations of the young political parties change substantially over short time periods, because the parties’ minimal existing reputations Except this paper says that, in Spain at least, new parties develop loyal backers quite easily. make voters’ accumulated judgements of them very sensitive to new information There’s another theory, found in Ejik, that party attachment in developing countries isn’t useful as a policy shortcut because parties keep changing their ideological positions. Stokes argues that much political competition, particularly in clientelistic developing democracies, is of the valence nature. But these two accounts seem to contradict each other? If people aren’t voting based on policy issues, then rapid changes of parties’ position in the policy space shouldn’t prevent people from developing party identifications if they’re useful for other things (e.g. as proxies for competence). .

So, to conclude, voters have stable attachments to political parties which function as useful shortcuts when deliberating over how to vote. Differences in cognitive constraints between voters explain why some have more durable political identifications than others. The partisan attachments of less sophisticated voters will tend to have a closer relationship with parental beliefs and sociological attributes, because these are the simplest possible proxies for which party is most likely to maximise an individual’s interests. More sophisticated voters use retrospective evaluations to continually assess which party appears the most competent, the mathematics of which mean that this Bayesian updating-based heuristic can manifest itself as an attachment which becomes stronger over time. A purely sociological account is incapable of explaining the volatility of party identification at the individual level, the medium-term reduction in the number of voters with strong party loyalties in Western countries, or the lack of stable partisan attachment seen in emerging democracies. The rational agent framework coupled with cognitive mobilisation theory fits with observed trends much better, and is able to provide a convincing account of which voters will have stable attachments, why, and what their partisanship will be determined by.

Bibliography

Campbell, A., Converse, P., Miller, W., & Stokes, D. E. (1960). The American Voter. John Wiley & Sons.

Clarke, H. D., Sanders, D., Stewart, M. C., & Whiteley, P. (2004). Political Choice in Britain. Oxford University Press.

———, & McCutcheon, A. L. (2009). The dynamics of party identification reconsidered. Public Opinion Quarterly, 73(4), 704–728. https://doi.org/10.1093/poq/nfp051

Dalton, R. J. (2002). The decline of party identifications. In Parties without Partisans (pp. 19–36). https://doi.org/10.1093/0199253099.003.0002

De Vries, C. E., & Hobolt, S. B. (2020). Voter loyalty. In Political Entrepreneurs: The Rise of Challenger Parties in Europe (pp. 63–85). https://doi.org/10.23943/princeton/9780691194752.003.0004

Downs, A. (1957). An Economic Theory of Democracy. New York : Harper.

Fiorina, M. P. (1981). Retrospective voting in American national elections.

Kroh, M., & Selb, P. (2009). Individual and contextual origins of durable partisanship. In Political Parties and Partisanship. Routledge.

Holmberg, S. (2009). Partisanship reconsidered. In Oxford Handbook of Political Behaviour. (pp. 557–570). https://doi.org/10.1093/oxfordhb/9780199270125.003.0029

Sanders, D., Clarke, H. D., Stewart, M. C., & Whiteley, P. (2011). Downs, Stokes and the Dynamics of Electoral Choice. British Journal of Political Science, 41(2), 287–314. https://doi.org/10.1017/s0007123410000505

Simon, H. A. (1955). A behavioral model of rational choice. The Quarterly Journal of Economics, 69(1), 99. https://doi.org/10.2307/1884852

Stokes, D. E. (1992). Valence Politics. In Electoral Politics. Oxford University Press

Thomassen, J., & Rosema, M. (2009). Party identification revisited. In Political Parties and Partisanship. Routledge.