As the campaign reaches the halfway point, how ‘predictive’ are the polls right now?
In the past 3 elections, the polling 2 months out was remarkably close to the eventual results, certainly regarding the blocs, suggesting accuracy and pretty settled public opinion
This week marks two months since the 2022 election was called, and the halfway point of the mammoth four-month campaign. For those of us following closely, it feels like a lot longer.
We shouldn’t really be surprised. The fifth campaign in three years, this was always going to be a long slog.
And with a weary public, a long, hot summer, and the Jewish festivals just around the corner, the parties sensibly have decided to operate on a low flame in the first phase of the campaign, saving their budgets and any tricks they may have for the final stretch.
It has always been felt that once the summer holidays come to an end, the public will begin to tune in more closely to political developments. So with that in mind, we have decided to use this week’s column to answer two questions:
What, if anything, has changed in the polling over the first half of the campaign? And most importantly, how much trust should we put in polling as a predictive tool at this stage?
1. What has happened so far?
Looking at changes since the campaign began in late June, we can identify two “winners” and two main “losers” of the first part of the campaign.
The big winners so far have been Religious Zionism and Otzma Yehudit (who – having flirted with the idea of running separately – on Friday announced that they will indeed run under a joint list. For more on the dynamics behind their decision, see our article on the subject from earlier this month). Combined, the two parties are now polling at 11.7 seats, an increase of 2.6 since the campaign began (and virtually double the six seats they won in last year’s election).
Prime Minister Yair Lapid’s Yesh Atid has also had a solid campaign so far, increasing by 2.3 seats to 23, and it will hope to keep this momentum going as it looks to get into the mid, and even high, twenties by Election Day.
A third party – Defense Minister Benny Gantz’s Blue and White, now known as the National Unity Party – has increased by 4.8 seats, from 8 to 12.8, in the past two months. However, when you take into account the fact that it merged with Gideon Sa’ar’s New Hope, which began the campaign averaging 4 seats before effectively folding into Blue and White, the gain is a much more modest 0.8.
The main loser so far is the Zionist Spirit party, which began the campaign as Yamina, polling at 4.4 seats. Today it is at 0 in our averaged poll of polls, and has not passed the threshold in any of the last 10 polls. If this continues, pressure will no doubt grow on Ayelet Shaked to drop out of the race ahead of the deadline for list submission in two weeks.
The other “loser” so far is actually Likud, down 2.6 seats in our average, from 35.1 to 32.5. When the campaign began, it was the biggest party by almost 15 seats; now the gap is “just” 9.5.
However, Benjamin Netanyahu is unlikely to be too concerned by this decrease on its own. Ultimately, Israeli elections are a bloc game: losing seats to satellite parties generally means very little in the overall context of the election.
And in terms of the blocs, the overall trend so far is – just about – positive for Netanyahu. The Netanyahu core bloc, made up of Likud, the Haredi parties and Religious Zionism/Otzma Yehudit, began the campaign at 58.8 seats in our average, 2.2 short of a majority. In the past two months it has gained half a seat, and now stands at 59.3. Conversely, the current government has lost half a seat, and now stands at 55.2 seats.
Simply put, the Netanyahu bloc has gained very slightly in the first half of the campaign, but is still short of 61 seats in our average (and in most polls).
So the simple answer to our first question “what has changed?”: Not much.
2. How predictive are polls two months before an election?
In answering this second question, we must begin with a cautionary note about the predictive qualities of polls in general. As we always tell our clients, polling should be seen as a snapshot in time, rather than a predictor of the future.
In-depth public opinion polling can tell us a lot about a society and dominant trends, and certainly can play a crucial role in strategy and message development. However, public opinion polls have not always been reliable predictors of the future – even less so when all we have access to are the topline numbers.
This does not of course mean that they have no role to play in helping us to understand what is likely to happen, or that all polls are equally limited. A good poll taken a few days before an election will usually give a pretty good indicator of the likely results (within a margin of error); a low-quality poll taken months ahead of time, less so.
The question here, then, is what can polls tell us two months out? To attempt to answer it, we will look at the past three election cycles: In 2021, 2020 and September 2019.
The graph below shows the difference between the polling average on January 22, 2021, two months before the 2021 election, and the eventual results. Blue means the party exceeded its polling, red means it undershot the polls. Most significantly, the bigger the number, the further the polls were out, while the closer the number is to zero, the more “predictive” the polling was.
As you can see, when it comes to parties, a huge amount changed in the final two months of the campaign. Labor was averaging zero seats two months out, yet ended up with seven seats. Religious Zionism similarly jumped four seats, from two to six, over the final two months. In contrast, Gideon Sa’ar’s New Hope was flying high as the second-biggest party, with 16.3 seats, before ultimately winning just six. Naftali Bennett’s Yamina had a similar collapse.
But when it comes to the blocs, the polling was spot on. The parties that pledged to recommend Netanyahu and join his government (which included Yamina at that stage) averaged 58.6 seats two months out, and ended up at 59.
So while the intra-bloc situation changed hugely, the early public opinion polls gave an extremely accurate picture of the overall political context, strongly indicating that Netanyahu would get very close to forming a government, but ultimately fall short.
If we go back to 2020, the situation was partially reversed. While acknowledging that there were various mergers on the right and left in the final weeks – so we have had to categorize the parties in a slightly imprecise way – we see that the polling for parties on January 2 was fairly consistent with the overall result. In fact, all but one party ended up within two and a half seats of its polling two months out.
When it comes to the blocs though, the early polling was slightly less instructive, and we saw a two-seat shift from the anti-Bibi bloc to the pro-Bibi bloc in the final two months. This is not a major shift in the general scheme of things, but it probably was enough to lay the groundwork for the formation of a Netanyahu-led unity government between Likud and Blue and White after that election.
Going back to the September 2019 election, the polling average on July 17, 2019, gave a pretty good indication of the ultimate result two months later. Here, each party ended up within 3.5 seats of its early campaign polling, with many of the parties hitting the bullseye with almost one hundred percent accuracy. Most importantly, when it comes to the blocs, there was almost no shift from the polling two months out to the eventual results.
The main takeaway here – and the answer to our second question – is that over the past three elections, the polling two months out has been remarkably – and surprisingly – close to the ultimate results, certainly when it comes to the all-important blocs. In none of the three elections was there more than a two-seat shift between the blocs in the final two months, and in two there was almost no change at all.
Of course, this could mean one of two things. Either the polling was good, or public opinion was so settled that very little substantive changed over the course of the campaign. The answer is probably a bit of both. But either way, looking at this data, it is hard to escape the conclusion that we should be giving more weight to the current polling than many currently do.
What does this mean looking ahead?
Things will definitely heat up in the next two months, and significant changes in the structure of the race are therefore very possible.
But based on the polling in the first half of the campaign, as well as the data from the previous three cycles, it certainly looks like we are heading for another extremely close election, in which the pro-Netanyahu bloc will get very close to the 61 seats it needs to form a government.
Will he finally get there, or again fall just short? At the moment, it is just too close to call.
