Election polls and mathematical illiteracy

I keep running into people who say that Zehut hasn’t passed the threshold since the deadline for submitting lists.

It’s enough to make someone who understands math despair.

Here’s a link to the Central Elections Commission’s polling page.  On this page, you can find links to the actual poll reports that the polling companies submit to the CEC. Let’s take a look at the most recent one on there. It’s a Midgam poll from 20 August.

On the first page, it says that the poll was commissioned by Walla News, and carried out by Midgam Consulting and Research under the leadership of Meno Geva. It claims to reflect the entire population of the State of Israel, ages 18 and up, and says that the sample they used was 510 respondents (about the number of people employed at a medium sized mall). And that the margin of error is plus or minus 4.4% (more than the threshold). It also says that the sampling they used was “stratified sampling.”

I won’t go into the fact that stratified sampling only works for parties that have distinct population concentrations that support them. Sectors, as we call them in Israel. And that not all parties are sectoral parties. I’ve already written a piece on that.

What I want to talk about is the margin of error. What does “margin of error” mean? It means that since they’re using such a small sample to try and reflect an entire country, they recognize that they can’t be more accurate than a range of about 8.8% (about 11 seats in Knesset). That’s enormous. But ultimately, a margin of error that size doesn’t make much difference for large parties like the Likud and Blue & White, and those are the important ones, right?

In this poll, for example, the Likud received 23% of the responses. That means that 117 of the 510 people who responded to the poll said they plan to vote for the Likud. But here’s the thing. It doesn’t mean that the Likud is actually projected, based on this poll, to win 23% of the vote on election day. It means that the Likud is projected to win anywhere between 18.6% and 27.4% on election day. That means anywhere from 22 to 33 seats. But the number of seats they reported for the Likud is 31. Not even the 27-28 that 23% of 120 would have given, but 31, because they distributed the “lost votes” from Zehut and Otzma Yehudit themselves, in order to have everything add up to 120.

They don’t tell you that more people answered “don’t know” than either Shas or the Democratic Camp. And in a footnote to their report, they tell the reader (though this isn’t reported by the media) that the data does not take into account those who are torn between parties and the degree of certainty they have in their intention to vote for a certain party. This means that 10 people who say, “I don’t really know… I guess maybe the Likud” are given the same amount of weight as 10 people who say, “Zehut. No question about it.”

If they were to ask “On a scale of 1 to 10, how sure are you that you intend to vote for the party you selected,” and factor that number into the poll, the results would look very different. But then again, if they were willing to put more than the bare minimum of work into the poll in the first place, they’d use a larger sample than 510 people.

Graphic by Lisa Liel based on Midgam report to CEC for poll of 20 Aug 2019

Here is a table that shows the results of this Midgam poll. I’ve shaded the information that gets reported by the media. That’s all you see. But it’s not the real story. For example:

  • The poll shows that up to 19 seats, almost 1 in 6, won’t commit to voting for any of the parties listed. Those 19 seats can go anywhere, to anyone, and this poll gives no indication whatsoever of where they’re likely to go.
  • The poll shows that Zehut could get anywhere between 0 and 8 seats, and that Otzma could get anywhere between 0 and 7 seats. Both of these give a decent chance of passing the threshold, but by only reporting the percentage of people who explicitly responded Zehut or Otzma, they give the public the impression that neither party is passing the threshold. Which is not supported by the data.
  • In fact, with a 4.4% margin of error in each direction, there’s no statistical difference between any of the parties other than the Likud, Blue & White, the Joint List, and Yemina. All of the rest can pass the threshold or not pass the threshold, and the poll results don’t say much about them beyond that.
  • And, again, by omitting the degree of certainty the respondent has for supporting the party they selected, the results become even less reliable.

It’s a truism that the average person doesn’t really care enough to think about issues, let alone math. They want sound-bites flowing out of the TV and radio and into their heads – pre-chewed, pre-digested bits that don’t require anything of them.

But we can’t afford that any more. The State of Israel stands at a critical moment. Trump’s “Deal of the Century” (or perhaps Jared Kushner’s) is headed our way, and none of the parties on the right or the left will stand up to Trump and say “no”. Trump may have nothing but good feelings for Israel, but not intending harm doesn’t mean he won’t cause harm.

We desperately need a strong right wing government that will put enough pressure on Netanyahu to prevent him from buckling under American pressure. If you believe in the principles that Zehut represents, if you want to be able to buy good avocados that don’t taste like water and aren’t rotten, if you want to be able to buy decent pineapples at a price that isn’t 10 times that of other countries, if you want to be able to afford an apartment for yourselves and for your children, if you want your children to remain here, in our land, and not run off to make their fortune in countries that prize entrepreneurship, rather than trample it, if you believe that this is our land and that refusing to defeat the enemy is suicidal, you must vote for Zehut on 17 September. The poll results are not an excuse for you to evade this responsibility. They don’t say what the media is telling you.

About the Author
Lisa Liel lives in Karmiel with her family. She is a member of the Zehut party, works as a programmer/developer, reads a lot, watches too much TV, does research in Bronze/Iron Age archaeology of the Middle East, and argues a lot on Facebook.
Related Topics
Related Posts