Almost a million voters have not decided who to vote for: the figure missing from the polls that could change the picture

For almost four years, news editions have been publishing election polls at least once a week — and as the target date, October 27, approaches, the pace of publications only increases, until it becomes almost daily. But among all the numbers presented on the screen or in the pages of newspapers, there is one figure that almost never reaches the viewer: how many of those surveyed in the sample do not know who to vote for at all.

This is not a marginal figure. A poll presenting "Party A - 22 mandates, Party B - 18 mandates" creates an impression of high accuracy, but behind these polished numbers hides a whole layer of respondents who answered "don't know". In most cases, they are simply not included in the final result: pollsters redistribute the votes of those who have decided, and present a picture that looks complete - even if in practice it is based only on part of the sample.

Will those undecided become the factor that decides the upcoming elections, without the pollsters seeing it coming?

The assumptions behind the method

First, let's explain how the method works: when a sample of 500 respondents is surveyed, not everyone answers clearly who they intend to vote for - some answer "don't know" or "haven't decided yet". In the election polls presented in the media, pollsters calculate the percentages of support for parties only from those who gave a clear answer. Thus, in practice, if out of 500 respondents 75 answered that they are undecided, the result presented is based only on the remaining 425 answers.

This method is not arbitrary, but relies on a certain methodological assumption: that the distribution of opinions among the undecided is similar to the distribution among those who have already decided. That is, pollsters assume that the balance of power between the parties, as measured among those who answered clearly, will also reflect to a reasonable extent what will happen among those who have not yet decided.

This assumption is valid only under one condition: that the undecided are truly distributed randomly among the parties, just like the general public. But to check if this assumption is indeed met, experts explain to Globes, pollsters use several complementary tools. Already in the questionnaire itself, it is customary to ask the undecided between which specific parties they are undecided - for example, between two parties within the same bloc, compared to parties on both sides of the political map - and thus it is possible to identify whether his indecision is concentrated within one bloc, or whether it is distributed between the blocs.

4 characteristics of the typical Israeli undecided voter

1. Basic demographic profile: mainly a secular-traditional woman, a resident of the north of the country, of Mizrahi origin, with post-secondary non-academic education.

2. Geographic-economic location: income below the general average. Among residents of the periphery who were severely affected in the war, indecision is higher.

3. Political identity: towards the current elections, indecision is concentrated mainly among voters identified with right-wing parties, and in particular in the Religious Zionism sector.

4. Information consumption and political knowledge: consume less news (both in traditional media and on networks), possess lower political knowledge.

Source: Reichman University and expert estimates.

---

The gap between the poll and the ballot box

Alongside this, it is still an assumption that does not necessarily reflect reality, and behind the way data is presented in polls there are two main problems: the first is that the viewer receives the processed product without knowing how many people hesitated at all, and without knowing that every percentage he sees has already undergone statistical adjustments that are not revealed to him. And this is not a theoretical matter: these numbers influence strategic decisions within parties, sometimes even candidates' considerations of whether to stay in the race - and even the decisions of the voting public itself.

The second problem, and perhaps the more significant of the two, is that a significant segment of the public is not presented in the poll, while in practice it just hasn't decided yet, so the forecast presented may be far from the actual result. Those hesitating - or by their professional name "floating votes" - are exactly those who may ultimately decide the elections.

Therefore, often a gap is created between the picture presented to the public for months and the result at the ballot box - a gap that can be expressed in a turnaround that no one saw coming, simply because the polls did not really measure those who had not yet formed their opinion.

"We are three months before the elections, and the rate of undecided is currently about 15%. This is between 18 and 20 mandates - about 930 thousand eligible voters," explains Meir Ben Haim, an expert in research methods and an external lecturer at Tel Aviv University. "Although this is not an unusual figure at this stage, this time there is a multitude of small parties hovering on the electoral threshold, so the undecided may ultimately decide the elections."

Reuven Harari, co-CEO of the Gal Hadash research institute, presents a similar estimate: "This is a number significant enough to influence the results, and it may accompany us until election day. Usually, between 10% and 15% of the public does not decide until the last moment, sometimes right at the ballot box." He adds that in the past there were even campaigns that went out just before the closing of the polls and managed to attract the undecided, and therefore even the last polls are not accurate. "Anything can happen," he concludes.

Not only in Israel: more and more voters in the world decide at the last moment

The phenomenon of the undecided in elections is not local, according to a new study published this month in the British journal of Oxford University, Parliamentary Affairs. The phenomenon is prevalent around the world, and researchers even note that the rate of undecided voters is growing over the years, in parallel with a general increase in voter volatility.

What is behind the change? According to the researchers, the main reason is that the traditional political divisions that defined Western politics throughout most of the 20th century, such as economic class or religious affiliation, have been replaced by new and more complex divisions: generation, culture, education, and identity. The result is that a person can identify with one group on an economic issue, and with a completely different group on a cultural or value issue - a situation that researchers call "cross-pressures".

Bottom line, all these data add up to one picture: this is not a marginal technical flaw in the polls, but a built-in gap that the media chooses not to present. Almost a million eligible voters whose political distribution is unknown will continue to be absent from the mandates presented on the screen - even when the probability increases that they will tip the scales and not distribute randomly as the methodology assumes.