Together with Daphne Halikiopoulou, we have just published a new LSE blog on why knowing who is likely to vote for a party doesn’t explain a party’s success, summarising our recent article just out in the European Journal of Political Research. In the blog, we argue that improved data and methods have made it easier to understand how people’s attitudes shape their political preferences. Yet, drawing conclusions about national electoral outcomes from individual-level findings risks distorting how we interpret elections.

Certain individual attitudes strongly predict support for certain political parties. For example, attitudes towards immigration are strong predictors of far-right party support.

Yet while this is true, it is often taken to mean that individual attitudes automatically tell us what explains the overall electoral success of a political party. For instance, as attitudes towards immigration are drivers of far-right party support, it is often assumed that immigration explains the success of the far right at the national level.

In a new study, we argue that such interpretations of electoral results are inaccurate. The problem is a methodological one: drawing conclusions about national electoral outcomes from individual-level findings risks committing an “atomistic fallacy”, which can significantly distort how we interpret elections and design policy responses.

The forgotten atomistic fallacy

Over recent decades, improved data and methods have transformed the study of voting behaviour. Indeed, as political scientists, we increasingly rely on individual-level data to explain voting behaviour.

Such data can help us identify which types of individuals are more likely to support particular political parties with greater precision. This has been especially useful in avoiding the ecological fallacy: the error of inferring individual behaviour from aggregate data.

However, this progress has shifted attention away from an equally important issue. When researchers use individual-level relationships to explain national-level outcomes, they risk making the opposite mistake. The “atomistic fallacy” arises when findings about individuals are implicitly treated as explanations for aggregate political phenomena (see Figure 1).

Figure 1: Illustration of the ecological and atomistic fallacies

Source: Vlandas, T. and D. Halikiopoulou (2026).

This distinction matters because relationships observed at one level do not necessarily hold at another. In the context of elections, knowing who is more likely to vote for a political party does not automatically tell us what explains that party’s overall success.

Why individual likelihood does not equal national level importance

A simple example helps illustrate the problem. Suppose a small group of voters has a very high probability of supporting a party, while a much larger group has a lower probability. Even if the first group appears more “important” at the individual level, it may contribute less to the party’s total vote because of its smaller size.

This highlights a key point: as Figure 2 illustrates, the impact of a factor on election outcomes depends not only on how strongly it affects individuals (the magnitude of the effect of an individual level characteristic) but also on how widespread it is in the electorate (the distribution of this characteristic at the national level).

Figure 2: Interplay of distribution and effect size of an individual attribute yielding different far-right party electoral results

Source: Halikiopoulou, D. and T. Vlandas (2025).

As a result, focusing exclusively on individual-level effects can lead to misleading conclusions about what drives electoral success. Variables with strong predictive power at the individual level may have limited aggregate relevance if they apply to only a small segment of voters.

This issue is particularly visible in debates about the rise of far-right parties in Europe. Even if anti-immigration sentiment strongly predicts individual voting behaviour, it does not follow that it is the main driver of aggregate outcomes. What matters is how prevalent these attitudes are across the electorate.

Recent studies suggest that far-right support may be less about a widespread increase in such attitudes and more about their growing salience among a relatively small group. Similarly, other factors, such as economic or status grievances, or their combination into broader voter coalitions, may play a larger role at the national level, even if their individual effects appear weaker.

Three common pitfalls in interpreting election results

Three related but distinct pitfalls associated with the atomistic fallacy can distort our interpretation of election results. First, there is the pitfall of treating the size of an individual-level effect as evidence of its overall importance.

A strong effect does not imply that a factor meaningfully shapes national outcomes if it is not widely shared. Even if strong cultural concerns over immigration make an individual more likely to vote for the far right, it does not follow that cultural concerns over immigration “explain” far-right party success, especially if such concerns are not widespread among the electorate.

Second, comparisons between variables can be misleading. A factor with a larger individual-level coefficient is not necessarily more important than another if it applies to fewer voters. The fact that those with cultural concerns over immigration are more likely to vote for the far right than those with economic concerns over immigration does not mean the former group matters “more” for far-right party success.

Third, changes in individual-level relationships are sometimes used to explain shifts in electoral outcomes. But this can be misleading. For instance, it might be the case that anti-immigration attitudes become a stronger predictor of how a particular individual will vote. Yet it does not follow that this change can therefore explain the overall electoral success of a far-right party. In fact, it is possible that the predictive power of these attitudes may have declined over time.

Taken together, these issues suggest that conclusions about “what explains” election results must be treated with caution when they are based solely on individual-level analysis.

Rethinking how we interpret electoral outcomes

Addressing this atomistic fallacy requires a closer integration of micro- and macro-level analysis. This means paying greater attention to the size and composition of different voter groups, rather than focusing only on individual-level relationships. Researchers can also combine estimates of individual effects with information about how characteristics are distributed across the electorate. This helps clarify how micro-level patterns scale up to national outcomes.

Novel methodological tools could further help disaggregate the different levels of analysis. For example, measuring social structuration allows researchers to assess the extent to which a political party is socially rooted in a particular constituency by estimating the overrepresentation or underrepresentation of a social group in a party relative to its size in the society. In addition, voting bloc decomposition techniques enable researchers to measure the contribution of voting blocs to a candidate’s vote total and to assess how their contributions change from one election to another.

More broadly, our findings highlight the importance of careful interpretation when communicating research beyond academia. Simplified claims about which groups or attitudes “drive” election results can obscure more complex realities and lead to misguided policy recommendations.

For more information, see our new study in the European Journal of Political Research.