March the 17th 2015 was Election Day for the 20th Knesset (parliament) in Israel. With voting being a confidential affair it is impossible to know exactly who voted for each party. Thus it is hard to know how voting is broken down demographically. This is where surveys come in, asking respondents who they intend to vote for (before elections) or alternatively who they voted for (after the elections). However, such surveys are problematic for a number of reasons. Not all respondents are interested in answering, preferring to keep their vote as a secret. Occasionally respondents will name a party other than the one they intend to vote for or voted for out of social desirability or more rarely to mess up the survey results. A key issue in such election surveys is based on who the respondents are. Some sectors of the population may refuse to participate in the survey, being suspicious of authorities. For example, in Israel ultra-orthodox Jews are under-represented in surveys. Other sectors may not be reached depending on how the survey is set up. For instance, older people tend not to feature in on-line surveys, or students often don’t have land lines. So, instead of relying on election surveys, data from the central bureau of statistics can be used to estimate the demographics behind the voters. What can we learn from doing so?
Focusing on the urban populations of cities over 50,000 citizens (30 cities), about 57% of the population in Israel, with around 3,420,000 people having the right to vote. 67.2% of them actually voted. Three of these cities are predominantly of Arab residents (Arab cities from here on; Umm Al-Fahm, Nazareth and Rahat), where 62.3% of those with the right to vote actually voted. In contrast, three other cities are predominantly of ultra-orthodox Jews residents (religious cities from here on; Bnei Brak, Bet Shemesh and Modi’in Illit), where 78.9% of the population voted. In cities where Likud (the major centre-right party in Israel which got the most votes in the elections for the third time in a row) got the most votes, 65.6% of those with the right to vote, voted. Yet in cities where The Zionist Camp was the party most voted for (led by Ha’avoda, the Israeli Labour party, the second most voted party in this election), a higher share of the population voted, 71.5%.
In the nine cities where Labour got most votes, the median population age is 35.6, with 14.5% of the population in those cities aged 65 plus, and 6% of the population aged 20 to 24. In the fifteen cities where Likud got the most votes the median age is 33.56, with 13.6% aged 65 plus and 7% aged 20 to 24. The difference in these populations’ composition may indicate that Likud voters are generally younger whereas voters for Labour slightly older, corresponding to younger populations leaning to right wing parties. This is consistent with research done in Europe that indicates that being male and young raises the probability of voting for the extreme right (see Arzheimer & Carter (2003) for example). Yet in the USA the opposite is observed- with younger voters more democratic and liberal (see Pew Research from 2012 for example).
Strikingly different are the Arab cities where the majority of the population voted for the Joint List (the political alliance of four Arab dominated parties in Israel), where the average median age is 21.3 with 4.3% of the population aged 65 plus and 9% of the population aged 20-24. Also, in the religious cities where the majority voted for Yahadut Hatora (United Torah Judaism, a small ultra-orthodox party), the median age is only 16, with merely 3.7% of the population is aged 65 plus and 7.7% between ages 20 and 24. These sectors are commonly known for having younger populations, having higher fertility rates, and can only be indicative of the majority not voting for Likud or Labour. Indeed, in two of the three Arab cities, the second most voted party was Meretz (a small secular social-democratic party), while in two of the three religious cities the second most voted party was Shas (an ultra-orthodox party).
While looking at the demographics for these cities by majority votes includes non-voters as well, there is some evidence that voting differs according to demographic composition of the urban population. Voting in very young cities tends to be sectoral (Arab or religious), in cities above the average median age voting is more left orientated in older cities. Combining further data from the statistics bureau (such as socio-economic status) with voting patterns may yield more results and potentially serve as an alternative to election surveys. Unfortunately intended or past voting by age or sex, or by place of residence, is often not published in election surveys (what matters most is who is leading in the race to the top!) and therefore it is hard to evaluate how the analysis above compares.