Respondent 148711 'took' a whole mandate from a party — is it true? The truth behind the 'noise' in the polls | Shmuel Rosner

Only this was really missing for us. As if we didn't know that we are surrounded by cruel enemies. As if we didn't remember — it is still very fresh — the surprise attack of October 7. As if we didn't enter a withdrawn and suspicious mood in the face of a hostile and fickle international reality. As if all these were not enough, "sources" came, who cannot be exposed for censorship reasons, and revealed a conversation between "countries", whose names also cannot be exposed for censorship reasons, who plotted, or at least one of them plotted, to promote peace agreements with Israel in order to lull it to sleep ahead of a war at a more convenient time.

Only this was really missing for us. Another reminder that the world is a cold and cruel place. Another reminder that Israel, even when it received some validation from its neighbors thanks to agreements, understandings, interests, common enemies, common friends, never approached a status of full, permanent legitimacy, which there is no longer a need or reason to challenge.

Yes, it signed several important peace agreements. No, it cannot close its eyes. Yes, it manages complex relationships, some of them hidden, with very many countries in the environment, including those that prefer to hide the connection. No, it cannot rely on these relationships, unless they are backed by great Israeli power. Yes, it can continue to talk about peace as an ideal, and it would be good if it continues to talk about it as an ideal. No, it cannot assume that the ideal will be realized in the dreamy way that suits anthems, or speeches, or emotional ceremonies.

Only this was really missing for us. Another reminder. As if we don't know.

Respondent 148711

Respondent number 148711 is a man close to 50 years old. His education is not academic. His income is a bit below average. He does not believe that the Prime Minister really intends to draft Haredim who do not study. He is against the inclusion of Ra'am in the coalition, even if Yoav Segalovitz joins the party. The answers he gave to our poll are not always consistent. And yet, he taught us an important thing. He taught us to better formulate what we know and what we do not know.

I had to pay attention to respondent number 148711 because of my negligence. In the middle of the day, in the middle of data collection, I became curious. So I downloaded the data already collected to see partial results. I didn't notice that I was downloading the results exactly when respondent 148711 was still answering the questionnaire. That is, I calculated a result when his questionnaire was only half full (and in fact, closer to a quarter). A glitch.

After an hour, when I downloaded the data again, and calculated again, it turned out that this is not just a glitch — this is a glitch worth a mandate. Respondent 148711 with a quarter of a questionnaire took a mandate from Likud and added a mandate to Religious Zionism. Respondent 148711 with a full questionnaire added a mandate to Likud and took a mandate from Religious Zionism.

And let it be said in advance: respondent 148711 is not a Likud or Religious Zionism voter. In fact, he is not voting for any party at the moment. He does not know who he will vote for. But the algorithm of mandate distribution, which throws every respondent to every poll to the place suitable for them in the overall calculation, threw him to one place when the questionnaire was partially full, and to another place when the questionnaire was full, and the mandate calculation changed.

Mandate polls are an event that the public does not always find easy to understand. Why should a party go down by a whole mandate this week compared to last week, when nothing happened that relates to it? And why should it go up by two? Here, sometimes all it takes is a respondent like 148711 to change his mind (or a pollster who doesn't notice that he only filled out a quarter of a questionnaire). Sometimes one respondent is what makes the difference between headlines like "upward trend for Likud" or "dive in the polls for Lieberman" — bombastic headlines behind which lies a daily, trivial reality.

In the case of respondent 148711, the difference was one more mandate or one less mandate for two parties, which were anyway above the electoral threshold. But what if 148711 had given his vote to a party that is a bit below the electoral threshold? Every week we turn to different people. This week we encountered the specific preference of this voter. Maybe a week ago we turned to a person similar to him — who preferred the Jewish Home or Religious Zionism. Such changes are an inevitable part of a poll. And this means that for every mandate result, one must take into account a kind of statistical plus-minus.

Let's check what would happen if the same respondent, 148711, had a slightly different preference. Let's assume he had given his vote to the Jewish Home, which this week, in a Channel 13 poll, did not pass the electoral threshold, what would happen then? I checked. The answer is that for all sorts of reasons related to the type of respondent and the weighting, he would have transferred the Jewish Home over the electoral threshold. Or then other parties would be affected, because if the Jewish Home jumps from zero to 4, one needs to take these four mandates from other parties. "Yashar!" would go down by a mandate. Also Yisrael Beiteinu would go down by a mandate, and so on. The full result is not so important, because this is not the poll and this is not the point. What is important is the principle: one respondent, for one poll, would have changed the picture.

Such an influence can of course be measured and neutralized in all sorts of ways. We turn to several hundred people every week — a sample size sufficient to ensure that pluses and minuses will not be too large. A large sample neutralizes the noise to a large extent, but does not eliminate it. There still remains statistical noise.

And more: pollsters can identify individually an event like that of respondent 148711 and try to ensure that this one respondent will not shuffle the cards for all the parties (and so that there will be no doubt: the decision not to transfer the Jewish Home over the electoral threshold this week was made after the case of 148711 was already identified). And yet, it is worth putting a finger on this event to clarify to everyone who follows the polls what we repeat almost every week: a very large part of the fluctuations you see from week to week, or from one pollster's poll to another's, are statistical noise and nothing more.

Statistical signal

Now let's make it a bit more complicated and even more interesting. A full mandate forecast provides one number for each party (Shas 7 mandates, "Beyachad" 13 mandates). But such a forecast relies on answers of hundreds of people and on a distribution method that translates percentages of answers into mandates, and every pollster does this distribution a bit differently (because one needs to estimate things like participation percentages of different sectors and the influence of surplus agreements and more).

Following the semi-amusing event with 148711, we became curious and took one poll, from recent times, and in relation to it we asked a simple question: what would happen if every respondent separately gave a different answer to the voting question? Let's assume, if from all the respondents to the poll, everyone remains exactly the same, except for Moshe from Bnei Brak who says Shas instead of Otzma Yehudit. Or everyone remains exactly the same, except for Miriam from Haifa, who says "Yashar!" instead of the Democrats.

Pay attention to the result of this test: almost 20 thousand scenarios of voting change were examined, each time of one respondent. In 3,415 of the cases the mandate forecast changed. That is, these are cases where a change of answer of one person moved at least one mandate. More than that: in the specific poll we checked, 740 out of 780 respondents — almost 95% of the respondents — had a theoretical ability to move a whole mandate by changing one answer. Also Moshe from Bnei Brak, also Miriam from Haifa, also Dorit from Efrat, also Uri from Kiryat Ono.

Shall we say it again? Almost every respondent to the poll had the ability to change a vote so that they would move a whole mandate in the final result (the respondent of course does not know that they can do this, and they also do not know to which party they need to change their vote in order to move a mandate). You will say: aha, here is the proof that polls are worth nothing. Here is the proof that if Hila from Jerusalem got up this morning on the wrong side, she will change the whole mandate picture.

Wait with this conclusion. Because after the test that determines what happens to the mandates, comes also the test of what happens to the blocs. And here is revealed the ability to ignore the noise and identify the sound. We said that in 3,415 scenarios of a single position change, the mandate forecast changed. Now we will ask: and how did this affect the general picture? If one ignores the noise and looks for the signal, one discovers that it did not affect much. And this is exactly the point.

In 2,745 of these scenarios mandates went up and down, and the size of the coalition bloc remained exactly as it was (in this poll, 50 mandates). That is, in 80% of the cases where a position change of one respondent also changed the mandate picture, the bloc picture remained identical. While in 16% of the cases the coalition received another mandate, in 3% of them it lost a mandate, and in 1% of the cases it lost two.

And in other words: after all the simulations, tests, rises and falls of parties because of small changes in answers to the poll, we remained with a stable picture in total: the coalition has 50 mandates, or a bit more, or a bit less. If one ignores the noise, and small movements here or there of this or that party, it turns out that this is — at least for now — a quite stable situation.

Ardan's potential

Everything you will read here deals with a party that does not exist. That is — from last week it exists, but in the polls. Until then it was examined only as a party that does not exist. A party on paper. A speculation party. A party that one needs to imagine before answering a question that concerns it, and not all of us have imagination.

This means that a large part of what you will read here needs to be taken with a grain of salt. Maybe it will turn out that in the transition from a party that does not exist to a party that exists, from an idea to execution, many changes occurred. Maybe it will turn out that Israelis who didn't think at all about Gilad Ardan as a leadership option, suddenly flock to him. Maybe it will turn out that he has charisma that we didn't know before. Maybe disappointment in other parties will drive voters towards him. Maybe. Meanwhile, what we have is findings about a party that does not exist — which we publish when it is already almost existing. With all the necessary reservations.

Gilad Ardan and Yuli Edelstein do not have a detailed platform, but they have a record. They do not have an old party, but they have rich experience. No one has yet been impressed by their party, no one has yet been disappointed by it. It did not vote for a law that angered you, it did not sit in a government that disappointed you, it did not pass a reform that pleased you, it did not fight for a cause that is important to you. Those who consider voting for it are required to invent for themselves a thing that is supposed to be. Whether it will be or will not be, it depends on the candidates, and also on their voters, if such are found.

In several polls since it officially arose, the party did not pass the electoral threshold. But, as mentioned, until a moment ago, it was not yet a party. Its real test will begin in a week or two, after it exhausts its effort to soar. What is true, in polls we conducted in recent weeks, and some of which examined among other things which voters would consider voting for which party, we found about 10% of voters who say they would consider voting for Ardan's party. This is a lot of people, but a bit less than the rate of those who consider Religious Zionism, and quite similar to the number of those who consider Blue and White. These considerers — the potential audience of the party — can be measured to try to understand who they are.

This is not exactly the audience of the small center parties (which this week, in most polls, none of them crossed the electoral threshold). This is also not the audience of the right that sits today in the government. This is a third thing, which can be identified in the answers to questions on various topics.

Let's start with the question of self-definition, because it is relatively simple. 51% of those considering Ardan's list define themselves as "moderate right" or "right close to the center" — compared to 34% in the entire sample. And pay attention to the opposite figure: "deep right" people are only 4% of this audience. Among those considering Likud, the rate of "deep right" people is 18%, five times more.

From here to content, and it is appropriate to start with the judicial reform. Why start specifically with it? Because many tests we conducted identify this topic as especially significant among coalition supporters. Of course, security is also important to them, and economy, and other things. But what defines them differently from rival camps is the topic of the judicial reform (or revolution, each according to their taste).

We also asked potential voters of Ardan and Edelstein about the judicial reform, and there are three possible positions that were presented to them: in favor of reform in the direction the coalition wanted, even without broad agreement; in favor of significant reform, but only in case there is agreement on it; and a third position — against significant reform. Among Likud voters, the dominant position is in favor of reform even without agreement. Only 17% of them say "reform only with agreement".

Among "Yashar!" voters, from the other side of the bloc barricade, the dominant position is with agreement, but many of "Yashar!" voters say "there is no need for significant reform". The potential voters of Ardan and Edelstein are in the middle: 21% for the first option, 66% for the second, and only 7% for the third.

66% in favor of reform with broad agreement. This is the highest rate measured by us in any group of potential voters on this question. And what is interesting is not only this figure, but also what is around it: there are almost no potential Ardan voters who want Yariv Levin's reform even without agreement, and there are almost no potential Ardan voters who oppose deep reform.

In fact, the two groups that fought in the streets for two years are almost not represented in the audience of potential Ardan and Edelstein voters. If you are looking for a sociological justification for such a list, it is found in this line. Not in Ardan's biography, not in Edelstein's seniority. In this line. Because this is the line that a very significant part of Israel's citizens agree with.

From here to the Haredim. This is the topic where Edelstein's presence is dominant, because of it he reached a break with Likud. 59% of those considering the list support the obligation of all or most Haredim to be drafted and the imposition of sanctions on those who do not draft. Among those considering Likud, the parallel figure is 28%, and this is probably the sharpest line that separates this audience from Likud. And if it seems to you that it is only about an argument over clauses of a draft law, here is a figure that teaches that it is about a more significant gap: 40% of potential Ardan and Edelstein voters want a broad government with parties from both blocs, without Haredim and without Arabs. Only 5% of them want a right-wing coalition similar to today's.

Up to here everything looks consistent. Moderate right, statesmanlike, wants reform with agreement, without partnership with Haredim, in favor of a broad government. And then comes the question of Judea and Samaria. We asked what Israel should do in this territory. 42% of potential Ardan and Edelstein voters say to strengthen control over the Palestinians and expand the settlements. 41% of them say to strive for full separation. A difference of less than a percent between two positions that are currently the dominant positions — seemingly opposite — of the public in both blocs.

To understand how unusual the division on the question of Judea and Samaria is, here is the same question as it is answered in several other groups of those considering voting for other parties. Likud considerers: 72% for strengthening control versus 17% for separation. Religious Zionism considerers: 89% versus 6%. "Yashar!" considerers: 13% versus 54%. Almost every group we checked is decided to one side. The group of Ardan-Edelstein considerers is the only one where there is an audience almost equal for both positions. On the question of the territories, Ardan-Edelstein's audience does not behave like a party audience. It behaves like a coalition of parties.

The question of the territories — and this is a thing that is very important to pay attention to — is also the main question that distinguishes Ardan and Edelstein's audience from Tropper and Hendel's audience. Among those considering the Jewish Home, like in the audience of most parties, there is a clear decision: towards separation (55%). Say from now on: Tropper and Hendel's party is more left-wing than Ardan and Edelstein's party. One can assume that this will be used by Ardan to differentiate from his main competitor in the area between the blocs. But pay attention: if he chooses a very hawkish line, he will immediately lose about half of the audience that considers him and prefers separation over deepening control and annexation.

That is, one can mobilize this audience with a message about the reform. One can mobilize it with a message about the Haredim. One can lose 40% of it in one sentence about the future of the territories. In any direction. This week we used information and data of the "Madad" site, including election polls for Channel 13 and polls for other parties, including JPPI. We also used reports of "Kan Hadashot" and N12, data of the Central Bureau of Statistics and polls of all media outlets. Prof. David Steinberg assisted in writing the article about respondent 148711.tguvot@maariv.co.il