People like to talk - also about politics. They talk about ‘how things are going’, which typically comes down to government performance: everyone has an opinion about whether the society is moving toward something better or worse and the government is always responsible.

These speculations materialize through democratic elections, where people pass judgment on the past behavior of the government. Instead of just talking cheap, people take concrete action by voting. To reach a decision, they ask themselves ‘am I doing better now than I did before the last election?’ If the balance of all considerations is positive, the voter will reward the government by voting for it. If negative, the voter will vote for the opposition.

For political scientists this simple model of voter choice is a dear child with many names: it is called retrospective voting, economic voting, or performance voting model. Many ordinary citizens, (non-political scientists), will undoubtedly recognize the above reasoning as something they typically do before deciding how to vote. Electoral volatility is on the rise, which suggests that people are becoming more likely to evaluate government performance after each election, instead of simply voting for the party they (and their forefathers) always have voted for.

In all its simplicity and intuitiveness, the model is very appealing for political scientists. Unsurprisingly, it has been hugely popular among scholars of voting behavior. There seems to be good reason for this. An impressive number of well-executed studies have shown that economic evaluations of government performance are an important driver of vote choice in Western democracies.

Despite its usefulness, some premises of the model have nevertheless not been properly scrutinized. Among other things, the model assumes that voters are aware of at least 1) how the economy of the relevant unit (e.g. municipality, country) is doing compared to the previous mandate period, 2) who the incumbents are and 3) who the opposition is.

It cannot be taken for granted that the typical voter knows these things. The very first survey results from the US in the 1940’s showed what political theorists had for so long suspected. Ordinary citizens are poorly informed even about the most basic facts pertaining to politics. Considering that this was well-known to the scholars who wrote the pioneering works about retrospective voting during the 1960’s and early 1970’s, it seems odd that citizens’ information levels have not been taken better into account within this field of research.

So, it then becomes an empirical question to be investigated whether and to what extent people actually are aware of those facts that the retrospective voting theory expects of them. I am aware of two studies that address this question - a number that is quite unproportioned in comparison to studies that explain vote choice through retrospective assessment. Both studies only look at knowledge of government composition, totally ignoring knowledge of the economy, which is difficult to measure. In their 2013 study in Electoral Studies, David Fortunato and Randolph Stevenson found that about one-third of voters in majoritarian systems are unable to identify the incumbent government. While also including some other European countries, their evidence came mostly from the UK.

Based on the study by Fortunato and Stevenson, the glass could arguably be half full or half empty. Given the rather pessimistic tone of the political knowledge literature, their findings could be seen as somewhat optimistic if two-thirds of voters pass this crucial test. Unlike much of the literature that has branded ordinary citizens as hopelessly ignorant, it seems that the majority of citizens are in fact capable of retrospective voting. But, a majoritarian system with only a very few parties seriously competing for power is as easy as it gets for the voters. Knowing whether it’s the Conservatives or Labour who are calling the shots at the moment is hardly very demanding.

In a proportional electoral system, coupled with a higher number of parties and numerous possible government coalitions, naming the government correctly is a much more difficult task. As Sara Hobolt and Jeffrey Karp have shown, coalition governments are much more common than single-party governments, and coalitions of several parties are by no means exceptional. The second study, which has examined the knowledge premises of retrospective theory, is by the author of this blog post. Published recently in The International Political Science Review, my analysis found that ignorance about government composition is much more widespread in Finland, which is a pretty standard case of a country with multi-party coalition governments that typically align along the left-right dimension. Although generally considered a nation with an extremely well-educated electorate, only 38 percent of the Finnish survey respondents were able to correctly name the government. At the time of the survey there were three major parties, who set the tone in Finnish politics. About 67 percent could correctly place these parties either in government or opposition, which is arguably a clearly better achievement. Placing the smaller government parties posed problems for most respondents. Interestingly, people who had voted for an opposition party were more likely to know who was in the government. Knowing whom to punish seemed to be a better motivator than being able to reward – at least in Finland.

So what should we learn from this? In my opinion, two lessons seem most essential. Firstly, it is obvious that a large portion of democratic electorates do not know who they should reward or punish in elections. If we would also measure correct perceptions of economic performance of governments, the share would be further greatly reduced. Particularly in the more complex multi-party systems people find it quite difficult to say who is in the government. Secondly, the findings suggest that the theoretical foundations of retrospective voting should acknowledge the assumptions they make about citizens’ political awareness. Although the evidence is restricted to a few studies it is hardly very surprising that many people simply cannot be expected to vote according to political scientists’ expectations because they just are not as sophisticated as our models hope they would be.

Voters may judge retrospectively, but they may not be able to cast a vote accordingly. It is necessary for anyone who builds their understanding of voting on any theory to understand the limits in factual knowledge that people typically have about politics. Some people are better than others at connecting the dots between what they want from politics and the candidate that shares the same values.