The Green Line and the Security Fence:
The Green Line and the Security Fence:
The Green Line and the Security Fence:
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
<strong>The</strong> <strong>Green</strong> <strong>Line</strong> <strong>and</strong> <strong>the</strong> <strong>Security</strong> <strong>Fence</strong>:<br />
Finding a Psychological Factor in Israel’s National Elections<br />
Ari Belenkiy ∗<br />
Ma<strong>the</strong>matics Department, Bar-Ilan University, Ramat Gan 52900, Israel<br />
with<br />
Yosef Grisi<br />
Economics Department, Bar-Ilan University, Ramat Gan 52900 Israel<br />
Abstract. <strong>The</strong> results of <strong>the</strong> 1996 <strong>and</strong> 1999 elections, toge<strong>the</strong>r with data from a 1995 domestic<br />
census, give a clear snapshot of <strong>the</strong> political orientation of <strong>the</strong> Israeli population at <strong>the</strong> time. <strong>The</strong><br />
most important factors of Israeli political preferences: ethnic background, religion, nationality<br />
were first recognized by S. Smooha (1978, 1993). <strong>The</strong>ir relevance was confirmed in twelve<br />
consecutive volumes of <strong>The</strong> Elections in Israel series, edited by A. Arian <strong>and</strong> M. Shamir. <strong>The</strong><br />
recent debate concentrated on <strong>the</strong> question of <strong>the</strong> importance of <strong>the</strong> social class factor.<br />
Several authors since 1996 have pointed out <strong>the</strong> new opportunities for sociological<br />
research presented by <strong>the</strong> 1996 Israeli electoral reform: <strong>the</strong> two-ballot system (separate votes for<br />
<strong>the</strong> prime minister <strong>and</strong> political party) gave voters a chance to express <strong>the</strong>ir ‘global’ (security)<br />
<strong>and</strong> ‘local’ (lifestyle) concerns separately. This paper focuses on <strong>the</strong> ‘global’ issues <strong>and</strong> related<br />
methodological ones.<br />
<strong>The</strong> major result is <strong>the</strong> ma<strong>the</strong>matical description of a fifth factor, which we tentatively<br />
call <strong>the</strong> psychological factor. We measured it by <strong>the</strong> voters’ proximity to <strong>the</strong> <strong>Green</strong> <strong>Line</strong><br />
(hereafter, GL), which separates <strong>the</strong> so-called “territories” (Judea, Samaria, <strong>and</strong> Gaza), populated<br />
by Palestinian Arabs <strong>and</strong> Israeli settlers, from <strong>the</strong> rest of Israel. Applied to <strong>the</strong> aggregate data, <strong>the</strong><br />
results show: <strong>the</strong> closer to <strong>the</strong> GL a Jewish voter lived in 1996 <strong>and</strong> 1999, <strong>the</strong> more biased his vote<br />
was toward <strong>the</strong> Right; roughly, <strong>the</strong> Right lost 3% of <strong>the</strong> vote for each 10 km away from <strong>the</strong> GL.<br />
Though <strong>the</strong> “proximity” or “Distance to <strong>the</strong> GL” (fur<strong>the</strong>r, DGL) parameter alone could not be <strong>the</strong><br />
ultimate measure of <strong>the</strong> psychological factor, it certainly touches on <strong>the</strong> essence of <strong>the</strong> problem as<br />
being directly correlated with <strong>the</strong> time needed for a terrorist to penetrate Israel’s territory <strong>and</strong><br />
commit suicide. We also took care to separate <strong>the</strong> psychological factor from two purely<br />
geographical factors: voters’ proximity to Israel’s external borders <strong>and</strong> <strong>the</strong> sea.
2<br />
Under <strong>the</strong> supervision of <strong>the</strong> first author, a group of students in 2005 conducted a poll of<br />
a group of 1,215 Israeli citizens to see if <strong>the</strong>se issues could be clarified on <strong>the</strong> individual level,<br />
avoiding <strong>the</strong> ecological inferences. <strong>The</strong> poll showed that <strong>the</strong> DGL variable appears to have <strong>the</strong><br />
most Wald-significant coefficient in <strong>the</strong> binary logistic analysis, several times greater than its<br />
geographical analog. It is interesting that it has slightly decayed in magnitude with time, from<br />
early (1996) to middle (1999) to more recent (2001) elections. We associate this decline with a<br />
short period of peace from 1996 to 1999 <strong>and</strong>, after a new break of violence in 2000, an increased<br />
feeling of security among <strong>the</strong> Israeli population since building <strong>the</strong> <strong>Security</strong> <strong>Fence</strong> came on <strong>the</strong><br />
agenda in 2001.<br />
Introduction<br />
In <strong>the</strong> introduction to his book, “Israel: Pluralism <strong>and</strong> Conflict,” Sammy Smooha, a<br />
founding fa<strong>the</strong>r of Israeli political science, wrote:<br />
“<strong>The</strong> 4.6 million residents within ceasefire boundaries of Israel at <strong>the</strong> end of 1975 are<br />
internally separated along five lines resulting in <strong>the</strong> following divisions: Palestinians –<br />
Jews, Israeli Arabs – Jews, Druze – Christian – Moslem Israeli Arabs, religious –<br />
nonreligious Jews, Oriental – Ashkenazi Jews…. Though <strong>the</strong>re is a large <strong>and</strong><br />
exp<strong>and</strong>ing literature on Israel, with one or two minor exceptions none of <strong>the</strong> works has<br />
focused on its multifaceted pluralistic structure. Fur<strong>the</strong>rmore, <strong>the</strong> five pluralistic<br />
divisions are not given equal attention by Israeli sociologists. Palestinian-Jewish<br />
relations, notwithst<strong>and</strong>ing <strong>the</strong> headlines <strong>the</strong>y capture, are rarely studied because of<br />
<strong>the</strong>ir newness, sensitivity <strong>and</strong> perhaps inaccessibility.” (Smooha 1978, 2-3).<br />
In this paper, we attack precisely <strong>the</strong> “inaccessible” issue: Israeli-Palestinian relations.<br />
This issue left its distinct mark on <strong>the</strong> results of all Israeli elections. While it was known<br />
to exist, until now it has eluded quantitative description.<br />
True, <strong>the</strong> territories issue <strong>and</strong> performance of <strong>the</strong> Israeli leaders in <strong>the</strong> area of state<br />
security, as seen by voters, were discussed constantly in <strong>the</strong> literature, <strong>and</strong> a steady
3<br />
increase in <strong>the</strong> importance of <strong>the</strong>se issues in elections from 1984 to 1999 was duly<br />
observed (Shamir <strong>and</strong> Arian 1999: Table 2; <strong>The</strong> Elections in Israel-1999, Table 1.2; <strong>The</strong><br />
Elections in Israel-2001, 16: Fig. 2), but again, <strong>the</strong> quantitative measure of <strong>the</strong> voters’<br />
reactions was missing. It is unclear, for example, how to deduce any meaningful<br />
predictions from <strong>the</strong> fact that 90% of respondents in 1996 <strong>and</strong> 1999 said that <strong>the</strong><br />
territories issue “will influence my vote” (<strong>The</strong> Elections in Israel -1999: 20). Shalev <strong>and</strong><br />
Kis (2002) tried to distinguish a voting pattern in different types of localities, but looked<br />
at <strong>the</strong> latter only from <strong>the</strong> class (income) point of view.<br />
<strong>The</strong> tables <strong>and</strong> graphs that dominate <strong>the</strong> first eight volumes of <strong>the</strong> Elections in Israel<br />
series (1984-1999) show only one parameter at time, although <strong>the</strong>re are at least three<br />
recognized factors of importance mentioned by Smooha: Oriental-Ashkenazi, religiousnonreligious,<br />
<strong>and</strong> Jews-Israeli Arabs issues, which can be alternatively described as<br />
ethnicity, religiosity, <strong>and</strong> nationality. Meantime, a fourth factor, so-called “class<br />
cleavage,” gradually became of importance within Israeli society. To study <strong>the</strong> political<br />
effects of any o<strong>the</strong>r factor, in particular <strong>the</strong> problem of Israeli-Palestinian relations, one<br />
must find a way to control those four parameters. <strong>The</strong> search for a new technique to<br />
examine <strong>the</strong> parameters o<strong>the</strong>r than tables <strong>and</strong> graphs was at a st<strong>and</strong>still when a lucky<br />
event broke <strong>the</strong> impasse.<br />
Data from <strong>the</strong> May 1996 Israeli elections <strong>and</strong> a November 1995 domestic census<br />
provided a unique opportunity to analyze <strong>the</strong> political preferences of <strong>the</strong> different<br />
segments of <strong>the</strong> Israeli population. Never in Israeli history had <strong>the</strong>se two events been so
4<br />
chronologically close. Elections that occurred earlier, in 1988 <strong>and</strong> 1992, <strong>and</strong> those<br />
following in 1999, 2001, <strong>and</strong> 2003, though taking place at regular intervals, are three or<br />
more years away from any census. As a result, only <strong>the</strong> elections in 1996 <strong>and</strong>, to some<br />
extent, 1999, can be paired with reliable data for <strong>the</strong> overall Israeli population, though<br />
<strong>the</strong>y took place during a turbulent period of social <strong>and</strong> economic changes in Israel.<br />
Though one more census, conducted in June 1983, could be similarly paired with <strong>the</strong><br />
1984 elections, a crucial factor was missing: <strong>the</strong> direct election of <strong>the</strong> prime minister,<br />
which took place only in 1996, 1999, <strong>and</strong> 2001.<br />
Shamir <strong>and</strong> Arian (1999) found that ethnicity <strong>and</strong> religiosity have decisive roles in <strong>the</strong><br />
outcome of those three elections; <strong>the</strong> same result for <strong>the</strong> 1999 <strong>and</strong> 2001 elections was<br />
confirmed by Shalev <strong>and</strong> Kis (2002) <strong>and</strong> Andersen <strong>and</strong> Yaish (2003). Let us note, inter<br />
alia, that although <strong>the</strong> latter two papers considered ethnicity <strong>and</strong> religiosity as control<br />
parameters ra<strong>the</strong>r than <strong>the</strong> target of <strong>the</strong>ir research, <strong>the</strong>y oversimplified <strong>the</strong>ir treatment of<br />
<strong>the</strong> ethnicity factor by dividing <strong>the</strong> voters into Ashkenazim (European/American origin)<br />
<strong>and</strong> Sephardim/Oriental Jews (African/Asian origin). 1<br />
First, this is imprecise<br />
etymologically: Bulgarian <strong>and</strong> Romanian Jews are mainly Sephardim <strong>and</strong> vote<br />
haphazardly. Second, it is not clear how Turkish Jews vote. Third, North American Jews,<br />
though Ashkenazim, vote <strong>the</strong> opposite of European Jews: Right, not Left.<br />
An important methodological insight made by <strong>the</strong> authors, who wrote after 1996, was<br />
<strong>the</strong>ir pointing to unique opportunities provided by Israel’s 1996 electoral reform. While<br />
<strong>the</strong> preferences of <strong>the</strong> Israeli voter are complex <strong>and</strong> inseparable (De Marchi <strong>and</strong>
5<br />
Goemans 2001), 2<br />
<strong>the</strong> two-ballot system (separate votes for <strong>the</strong> prime minister <strong>and</strong><br />
political party 3 ) gave <strong>the</strong> voters a chance to express <strong>the</strong>ir ‘global’ (security) <strong>and</strong> ‘local’<br />
(lifestyle) concerns separately (see Israel At <strong>the</strong> Polls: 1996, 90, 127, 258). While Shalev<br />
<strong>and</strong> Kis (2002) <strong>and</strong> Andersen <strong>and</strong> Yaish (2003) concentrated <strong>the</strong>ir analysis mostly around<br />
‘local’ concerns (more precisely, analyzing <strong>the</strong> vote according to class subdivisions<br />
within Israeli society), we shall focus on <strong>the</strong> ‘global’ one, <strong>the</strong> Israeli-Palestinian conflict,<br />
which is basically a personal security or psychological issue.<br />
In fact, <strong>the</strong> attitude toward this single issue gave birth to an inadequate Israeli political<br />
nomenclature: Right <strong>and</strong> Left, where ‘Hawks’ <strong>and</strong> ‘Doves’ would be more accurate. 4 One<br />
can argue that <strong>the</strong>re was a clear division between Right <strong>and</strong> Left, that is, between<br />
‘Hawks’ <strong>and</strong> ‘Doves,’ even prior to 1996 since at least 1993, when <strong>the</strong> Oslo Accords<br />
were signed. However, <strong>the</strong> appearance of such parties as Tsomet in 1992, or <strong>The</strong> Third<br />
Way <strong>and</strong> Israel Ba’Aliya in 1996, with <strong>the</strong>ir mixed platforms on foreign politics, did<br />
muddy <strong>the</strong> waters. 5<br />
<strong>The</strong>refore, Anderson <strong>and</strong> Yaish (2003), in <strong>the</strong>ir study of ‘local<br />
concerns,’ which <strong>the</strong>y identified with ‘party vote,’ offered a ra<strong>the</strong>r complicated (probit 6 )<br />
model wherein several pieces of <strong>the</strong> Israeli political spectrum were included <strong>and</strong> <strong>the</strong><br />
dependent variable was allowed to have a discrete set of values, not just two – for <strong>the</strong> two<br />
major parties. 7<br />
Though different “third parties” offered a different ‘global’ security agenda, a separate<br />
vote for <strong>the</strong> prime minister purified <strong>the</strong> voter’s attitude, separating his local concerns<br />
from <strong>the</strong> security issue. One of <strong>the</strong> Oslo Accord signers, Shimon Peres, in 1996, <strong>and</strong>
6<br />
Ehud Barak in 1999 <strong>and</strong> in 2001 were associated with <strong>the</strong> desire to make concessions to<br />
Palestinians <strong>and</strong> <strong>the</strong>refore with <strong>the</strong> Israeli “Left,” while <strong>the</strong>ir rivals, Binyamin Netanyahu<br />
in 1996 <strong>and</strong> 1999 <strong>and</strong> Ariel Sharon in 2001, opponents of <strong>the</strong> Oslo Accords, were<br />
associated with <strong>the</strong> Israeli “Right.” <strong>The</strong>refore, <strong>the</strong> dependent variable can be treated as<br />
binary, which in turn, allows us to use a traditional, simpler regression model. 8<br />
Looking back, say to 1965 <strong>and</strong> 1969, when “split-ticket voting” was carried out when a<br />
voter chose a party for Knesset <strong>and</strong> local council separately (Arian 1973, 183-6), one can<br />
also discern <strong>the</strong> results we advocate here – but those results went unnoticed. We will<br />
discuss this point later in <strong>the</strong> text.<br />
<strong>The</strong> analysis of <strong>the</strong> Israeli political spectrum that we present in this paper is not unique.<br />
After years of largely descriptive analysis, several authors carried out <strong>the</strong> multilevel<br />
regression analysis. Shamir <strong>and</strong> Arian (1999) performed regression analysis of <strong>the</strong> 1999<br />
elections, using an individual poll of nearly 800 respondents. Debating <strong>the</strong> strong <strong>and</strong><br />
weak aspects of preliminary polls, Shalev <strong>and</strong> Kis (2002) made regression analysis with<br />
aggregate data from <strong>the</strong> Israeli Statistical Bureau, similar to data we used in this paper.<br />
All of <strong>the</strong> above authors avoided treating <strong>the</strong> national (Jewish vs. Israeli Arab) factor<br />
because of some “technical” problems: in one case <strong>the</strong>re were no trustworthy polls<br />
among Israeli Arabs; in ano<strong>the</strong>r, <strong>the</strong>y did not include all o<strong>the</strong>r relevant factors. <strong>The</strong><br />
problems with data <strong>and</strong> <strong>the</strong> way we posed <strong>the</strong> problem (as a “psychological factor” within
7<br />
Jewish Israeli population) forced us to avoid discussing <strong>the</strong> national factor as well. 9 We<br />
attempted to address it in <strong>the</strong> individual poll, though unsuccessfully.<br />
1. Key Question <strong>and</strong> Innovations<br />
<strong>The</strong> key question was to find a ma<strong>the</strong>matical way to describe <strong>the</strong> security factor such that<br />
<strong>the</strong> variable behind it must be independent of <strong>the</strong> major trio, ethnicity, religiosity,<br />
nationality, plus a social factor. Andersen <strong>and</strong> Yaish (2003) in <strong>the</strong>ir study introduced<br />
several control parameters, among <strong>the</strong>m <strong>the</strong> ‘urban locality,’ a binary variable. It must be<br />
recognized immediately as inadequate for our purposes. It has a mixed content of<br />
geography <strong>and</strong> culture <strong>and</strong> <strong>the</strong>refore fails to represent ei<strong>the</strong>r of <strong>the</strong>m. Its inadequacy to<br />
represent “culture” is especially striking: small communities do not lack information<br />
when compared with cities; besides, many kibbutzim can compete culturally with cities. 10<br />
In a more formal objection, one can say that an exact division between urban <strong>and</strong> rural<br />
areas is non-existent in Israel. Introducing <strong>the</strong> urban locality seems a dubious attempt to<br />
represent a st<strong>and</strong>ard ‘center-periphery’ division in a different way. 11<br />
It is not a surprise that urban locality did not feature significantly in Andersen-Yaish’s<br />
regression model (2003). 12 Moreover, <strong>the</strong> sign of <strong>the</strong> coefficient before this variable (see<br />
ibid, 413, Table 3A) suggests that urban people prefer Rightist parties, which might be<br />
true for Jerusalem <strong>and</strong> Sou<strong>the</strong>rn Tel Aviv, but is in direct contradiction to what we know<br />
about communities in Nor<strong>the</strong>rn Tel Aviv <strong>and</strong> “red” (pro-Communist) Haifa. <strong>The</strong> failure<br />
of urban locality forces us to suggest that ano<strong>the</strong>r, more refined set of parameters might<br />
be appropriate to comprehend Israeli reality.
8<br />
In this paper we investigate what Andersen <strong>and</strong> Yaish neglected – <strong>the</strong> problem of a<br />
security or psychological factor – <strong>and</strong> introduce a new parameter, <strong>the</strong> DGL. This is a<br />
quantitative variable measured as <strong>the</strong> shortest distance from any locality to <strong>the</strong> GL. <strong>The</strong><br />
latter refers to <strong>the</strong> 1949 Armstice lines between Israel <strong>and</strong> its neighbors: Egypt, Jordan,<br />
Lebanon, <strong>and</strong> Syria after <strong>the</strong> 1948 Arab-Israeli War, or Israeli War of Independence, <strong>and</strong><br />
encompasses Judea, Samaria, <strong>and</strong> Gaza. This variable was not provided in <strong>the</strong> census<br />
data. We obtained it from <strong>the</strong> Web site of <strong>the</strong> Israeli Ministry of Interior. 13<br />
This variable grasps <strong>the</strong> essence of <strong>the</strong> “psychological” factor (hereafter, P-factor). Why<br />
psychological? Because <strong>the</strong> distance is directly related to <strong>the</strong> time needed for a terrorist<br />
group to infiltrate Israeli territory <strong>and</strong> reach this or that city. Before <strong>the</strong> <strong>Security</strong> <strong>Fence</strong><br />
was raised all along <strong>the</strong> perimeter of <strong>the</strong> <strong>Green</strong> <strong>Line</strong> in 2001-2003, terrorists could<br />
penetrate Israel proper virtually through any place at <strong>the</strong> GL. After an informant's<br />
warning, <strong>the</strong> Israeli police have some time to detect <strong>the</strong> terrorist – <strong>the</strong> time is inversely<br />
proportional 14 to <strong>the</strong> distance from <strong>the</strong> GL to <strong>the</strong> settlement, a possible target of <strong>the</strong><br />
terrorist. When, during 1996-2003, <strong>the</strong> country lived in expectation of new terrorist<br />
attacks daily, <strong>the</strong> Fear on <strong>the</strong> streets was in <strong>the</strong> same (inverse) relation to <strong>the</strong> DGL. <strong>The</strong><br />
most distant places from <strong>the</strong> GL, like Eilat or Golan Heights settlements, did not feel this<br />
Fear at all. True, <strong>the</strong> very first terrorist acts were committed in <strong>the</strong> capitals, Jerusalem <strong>and</strong><br />
Tel-Aviv, but later on, virtually every settlement became a target for terrorists, as follows<br />
from <strong>the</strong> geography of suicide bombings (Haifa, Netania, Beer-Sheba, Hadera, etc.).
9<br />
Since <strong>the</strong> 1993 Oslo Accords, <strong>the</strong> Left has been more inclined to make peaceful gestures<br />
toward Palestinians. This includes not only transferring l<strong>and</strong> <strong>and</strong> dismantling<br />
unauthorized settlements, but also releasing prisoners <strong>and</strong> dismantling some military<br />
checkpoints <strong>and</strong> roadblocks as acts of goodwill. <strong>The</strong>se practices invariably led to easier<br />
infiltration by terrorists into Israel proper <strong>and</strong> to new series of terrorist attacks <strong>and</strong> new<br />
victims – which could not have added to <strong>the</strong> popularity of <strong>the</strong> Left, especially among<br />
those living in proximity to <strong>the</strong> GL. 15 To <strong>the</strong> contrary, <strong>the</strong> Right, enemy of <strong>the</strong> Oslo<br />
Accords, has always been ready for an immediate <strong>and</strong> strong response, such as punishing<br />
any acts of Palestinian violence by permitting new settlement activities, imposing new<br />
checkpoints, <strong>and</strong> setting curfews. <strong>The</strong>se actions could add to <strong>the</strong> feeling of security for<br />
those living closer to <strong>the</strong> GL. In short, this is <strong>the</strong> same “Fear factor.” It is our goal to<br />
show that <strong>the</strong> closer to <strong>the</strong> GL an Israeli Jewish voter lived during <strong>the</strong> late 1990s, <strong>the</strong><br />
more biased his/her vote was toward <strong>the</strong> Right.<br />
One could ask “which came first, <strong>the</strong> chicken or <strong>the</strong> egg?” since <strong>the</strong> reverse chain of<br />
reasoning is also viable – pro-Right Israelis could prefer settling in Jerusalem <strong>and</strong> beyond<br />
<strong>the</strong> GL – in <strong>the</strong> Biblical cities like Hebron, Shechem, Bet El, Efrat, Jericho, <strong>and</strong> Bet<br />
Lehem. This is correct, but <strong>the</strong> only visible reason behind such a preference is religiosity.<br />
Regression analysis would take care of this problem automatically, using religiosity as a<br />
control variable. <strong>The</strong> problem is that, unlike <strong>the</strong> situation with <strong>the</strong> individual poll, we do<br />
NOT have an adequate representation for religiosity in <strong>the</strong> aggregate data. How can we<br />
circumvent it? <strong>The</strong> argument is that for religious people, it makes sense to live IN a<br />
historic Biblical place, not X km away from it! <strong>The</strong>refore, it seems reasonable to exclude
10<br />
<strong>the</strong> “Biblical” areas (<strong>and</strong> only <strong>the</strong>m) from our analysis of aggregate data, which is<br />
tantamount to excluding <strong>the</strong> GL settlements <strong>and</strong> Jerusalem (located at 3-5 km from <strong>the</strong><br />
GL). This solves <strong>the</strong> problem of <strong>the</strong> direction of causation by making <strong>the</strong> reverse<br />
implausible: a move, say, from Haifa (40 km from GL) to Tel-Aviv (20 km to GL) hardly<br />
would be seen as <strong>the</strong> desire to live closer to historic Biblical places!<br />
Still, Jerusalem must be treated separately from <strong>the</strong> GL settlements, since it is already a<br />
modern city, host to large, high-tech industries. A high-tech worker would have no a<br />
priori preference to settle in Jerusalem, ra<strong>the</strong>r than in Tel-Aviv or Haifa. <strong>The</strong>refore, we<br />
decided to keep Jerusalem in <strong>the</strong> aggregate data analysis as well <strong>and</strong> report on differences<br />
between <strong>the</strong> results that include it <strong>and</strong> exclude it.<br />
<strong>The</strong>re might be an additional reason to exclude settlements beyond <strong>the</strong> GL, even in <strong>the</strong><br />
analysis of <strong>the</strong> individual poll where we can control for religiosity. For those who live<br />
beyond <strong>the</strong> GL, <strong>the</strong>re are two reasons to vote Right. One is significantly lower taxes vs.<br />
<strong>the</strong> rest of Israel <strong>and</strong> an opportunity to buy a house several times cheaper than inside <strong>the</strong><br />
GL. 16 Israeli settlements also have enjoyed generous government financing for building<br />
public facilities, special access roads, <strong>and</strong> roads that bypass Palestinian villages, <strong>and</strong><br />
industrial zones, in addition to <strong>the</strong> operation of schools <strong>and</strong> health clinics. This is a socalled<br />
“Quality-of-Life” factor which might be a reason even for non-religious settlers to<br />
vote Right, since <strong>the</strong> Extreme Left (“Peace Now”) since <strong>the</strong> 1990s has raised <strong>the</strong> question<br />
of abolishing tax <strong>and</strong> security advantages for settlers. 17 “Peace Now” is influential within
11<br />
<strong>the</strong> Meretz Party, which was a part of <strong>the</strong> Leftist government coalition in 1992-1996 <strong>and</strong><br />
1999-2001. <strong>The</strong> Rightist government took <strong>the</strong> opposite stance. 18<br />
<strong>The</strong> question is whe<strong>the</strong>r <strong>the</strong> “Quality-of-Life” factor belongs to <strong>the</strong> P-factor. If one wants<br />
to study <strong>the</strong> Fear (or ra<strong>the</strong>r “Fear-of-losing--life”) factor alone, since it is not obvious<br />
how to separate it effectively from <strong>the</strong> former, <strong>the</strong> only solution seems to be to proceed<br />
exactly as we did in <strong>the</strong> previous problem with direction of causation – to exclude <strong>the</strong> GL<br />
settlements from <strong>the</strong> analysis. In case <strong>the</strong> “Quality-of-Life” factor is included in <strong>the</strong> P-<br />
factor, <strong>the</strong> non-religious GL settlements might remain in <strong>the</strong> data. Actually, two sorts of<br />
analysis can emerge – excluding or including GL settlements.<br />
Arguably, <strong>the</strong> P-factor came into <strong>the</strong> limelight due to several terrorist suicide attacks in<br />
Tel-Aviv <strong>and</strong> Jerusalem on <strong>the</strong> eve of <strong>the</strong> 1996 elections, but remarkably, it remained<br />
underestimated until <strong>the</strong> 1999 elections, when <strong>the</strong> Left won by promising quick peace<br />
after a comparatively peaceful lullaby during Netanyahu’s tenure in 1996-99. When this<br />
promise failed, as a result of a new outbreak of violence in September 2000, <strong>the</strong> P-factor<br />
came to <strong>the</strong> fore in <strong>the</strong> 2001 <strong>and</strong> 2003 elections, when <strong>the</strong> Right won a sweeping victory.<br />
To mitigate <strong>the</strong> geographical flavor in <strong>the</strong> P-factor we introduced in parallel two<br />
geographical factors: Location by Sea <strong>and</strong> Distance to <strong>the</strong> Borders of Israel (fur<strong>the</strong>r: DB).<br />
<strong>The</strong> first was measured as a dummy binary variable (1,0) <strong>and</strong> <strong>the</strong> second as <strong>the</strong> shortest<br />
distance to <strong>the</strong> closest border with one of four countries: Egypt, Jordan, Syria, or<br />
Lebanon. Both variables displayed a low significance for election results. This is<br />
consistent with <strong>the</strong> logic of Israeli politics. Both Right <strong>and</strong> Left take serious heed to <strong>the</strong>
12<br />
defense of <strong>the</strong> Israeli borders, <strong>and</strong> residents in <strong>the</strong> vicinity do not feel endangered by <strong>the</strong>ir<br />
proximity. Location by Sea also assumes a “class” factor, since port cities in <strong>the</strong> past<br />
were <strong>the</strong> bulwark of <strong>the</strong> working class. However, it is not a decisive factor, <strong>and</strong>, as we<br />
shall see, two similar voters in Haifa <strong>and</strong> Ashdod have quite different priorities.<br />
Because Israeli Arabs were expected to vote differently than <strong>the</strong> Jews, <strong>and</strong> because <strong>the</strong><br />
1995-1996 suicide explosions created psychological (personal security) problems within<br />
<strong>the</strong> Jewish population alone, we decided not to introduce complicated, difficult to<br />
interpret “mixed variables” but to restrict our analysis to <strong>the</strong> electoral areas with<br />
predominantly (more than 90%) Jewish voters. 19 This cost us about 200 points in <strong>the</strong> data<br />
set (about 9%). Ano<strong>the</strong>r, more technical, problem was to exclude <strong>the</strong> “outliers” (cities too<br />
far away from <strong>the</strong> GL, like Eilat or <strong>the</strong> Golan Heights), <strong>and</strong> we restricted our analysis to<br />
cities within 60 km of <strong>the</strong> GL. This cost us ano<strong>the</strong>r 170 points of data (about 7%).<br />
Both restrictions – 90% <strong>and</strong> 60km – seem arbitrary, but allowed us to focus on <strong>the</strong> major<br />
problem more closely. <strong>The</strong> first parameter (90% Jewish voters) is reasonable in itself<br />
since we cannot completely reject mixed communities with a mixed, but predominantly<br />
Jewish, population, while <strong>the</strong> second (60km) includes <strong>the</strong> major Israeli cities, Haifa in<br />
particular. Changing <strong>the</strong> latter parameter from <strong>the</strong> 60 km to 100 km range shows a slight<br />
decline of <strong>the</strong> P-factor <strong>and</strong> its significance.<br />
2. Difficulties
13<br />
1. <strong>The</strong> first issue is that <strong>the</strong> census provides data about all residents of a particular<br />
electoral area, but not all residents actually vote. Partial justification for our approach is<br />
that <strong>the</strong> percentage of participation in <strong>the</strong> 1996 <strong>and</strong> 1999 elections was almost uniform<br />
within statistical areas, about 80% (Israel At <strong>the</strong> Polls: 1996, 6; <strong>The</strong> Elections in Israel –<br />
2001, 35). 20 <strong>The</strong>refore, we assumed that <strong>the</strong>re was no significant bias in our analysis<br />
under <strong>the</strong> plausible belief that representatives across all <strong>the</strong> groups abstained from voting<br />
r<strong>and</strong>omly.<br />
2. Ano<strong>the</strong>r, more general difficulty comes from <strong>the</strong> very nature of <strong>the</strong> data at h<strong>and</strong>.<br />
Aggregate data (averaged over a statistical area) make any analysis liable to <strong>the</strong> so-called<br />
“ecological fallacy” (see, e.g., Freedman 2001), in which one must restrict conclusions<br />
about correlations of <strong>the</strong> results with <strong>the</strong> size of a particular group, <strong>and</strong> NOT with<br />
individual preferences. A personal poll was necessary to confirm results at <strong>the</strong> individual<br />
level <strong>and</strong> such a poll was conducted by <strong>the</strong> first author with his students in 2005.<br />
Fortunately, this research with aggregate data is partially free from this problem as well.<br />
<strong>The</strong> variable in which we are most interested, <strong>the</strong> DGL, is NOT an average. People vote<br />
at <strong>the</strong> place where <strong>the</strong>y live. Nei<strong>the</strong>r are <strong>the</strong> two geographical variables – DB <strong>and</strong><br />
Location by Sea – averages. This frees our major claim from <strong>the</strong> “ecological fallacy.” 21<br />
On <strong>the</strong> contrary, <strong>the</strong> o<strong>the</strong>r variables, representing controlled parameters (including ethnic<br />
origin, income, religiosity, nationality, age <strong>and</strong> gender), all are represented by averages in<br />
a statistical area. <strong>The</strong>refore, we avoid any claims about <strong>the</strong> vote of an individual carrier of
14<br />
one of <strong>the</strong>se qualities. 22 Still, it is a relief to know that our data both qualitatively (<strong>the</strong><br />
sign of regression coefficients) <strong>and</strong> quantitatively (<strong>the</strong>ir high significance) support<br />
conclusions made by Shamir <strong>and</strong> Arian (1999), Shalev <strong>and</strong> Kis (2002), <strong>and</strong> Andersen <strong>and</strong><br />
Yaish (2003) about <strong>the</strong> major trio (ethnicity, religiosity, nationality) + social factor, while<br />
pointing to several interesting clarifications at <strong>the</strong> same time. This, in turn, justifies our<br />
correct choice of <strong>the</strong> control parameters in our major research directed to ma<strong>the</strong>matically<br />
represent <strong>the</strong> psychological (security) issue. 23<br />
3. Regarding Ethnic Background, we chose <strong>the</strong> smallest possible set of representatives:<br />
ei<strong>the</strong>r entire continents: Asia, Africa, North <strong>and</strong> South America, <strong>and</strong> Europe – or <strong>the</strong><br />
countries that represent <strong>the</strong>ir political preferences most prominently. True, several<br />
countries, like Turkey, Romania, <strong>and</strong> Bulgaria, defy this simplified description, but <strong>the</strong>y<br />
do not influence much <strong>the</strong> results <strong>and</strong> after all we need this variable only as a control.<br />
4. <strong>The</strong> question about religiosity does not exist in <strong>the</strong> census, <strong>and</strong> this is its major<br />
deficiency regarding our analysis. We had to invent a plausible variable to represent it:<br />
large families (7 or more members) could represent <strong>the</strong> religious group while single<br />
people are more likely to be non-religious. This choice is justified a posteriori, since <strong>the</strong><br />
former group votes strongly Right while <strong>the</strong> latter group votes strongly anti-Right.<br />
5. Though all variables, including dependent ones, are qualitative, except Income, after<br />
aggregation <strong>the</strong>y became “quantitative,” which makes <strong>the</strong> technical analysis much easier.<br />
<strong>The</strong>refore, to analyze census aggregate data, we chose linear regression instead of logistic
15<br />
regression, since <strong>the</strong> sign <strong>and</strong> significance of linear regression coefficients allow for<br />
direct interpretation.<br />
6. <strong>The</strong>re is a raging debate on exactly what “class” is. Andersen <strong>and</strong> Yaish (2003) argued<br />
that “class” is defined solely by individual (family) income <strong>and</strong> occupation. Shalev <strong>and</strong><br />
Kis (2002) added to this pair <strong>the</strong> “population density.” We took <strong>the</strong> side of Andersen <strong>and</strong><br />
Yaish so as not to compute an extra parameter. As a proxy for “occupation” we used<br />
“academic degree.” Though usually considered as “education,” an academic degree also<br />
reflects “class” status within Israeli society.<br />
7. In <strong>the</strong> study of P-factor, it would be too costly to discard <strong>the</strong> electoral areas with a<br />
partial Jewish population, say, with 5-10% of Arabs. However, in such areas <strong>the</strong> figures<br />
of <strong>the</strong> vote will not give adequate figures for Jewish voters we are seeking. Indeed, <strong>the</strong><br />
individual poll shows that <strong>the</strong> Arabs vote overwhelmingly Left. Such areas with 5-10%<br />
Arab populations are pretty numerous in Jerusalem <strong>and</strong> Haifa. <strong>The</strong>refore, we introduced a<br />
correction coefficient for <strong>the</strong> dependent variable (vote for <strong>the</strong> Right): K = 100% / (100% -<br />
% of Arab votes) which is adequate for <strong>the</strong> low percentage of Arab voters. Indeed, in <strong>the</strong><br />
area with 10% percent Arabs, <strong>the</strong> result of a 50% vote for <strong>the</strong> Right must be interpreted<br />
as 55% of Jews voted Right – which is adequately represented by K=10/9 = 1.1.<br />
8. To eliminate locations beyond <strong>the</strong> GL appeared to be an inevitable decision, since it is<br />
unclear how to measure <strong>the</strong>m properly. <strong>The</strong> “minus” sign was considered as an option,<br />
but was not a successful one. For example, <strong>the</strong> city of Ariel is located in <strong>the</strong> center of <strong>the</strong>
16<br />
Jewish block in Samaria, 17 km away from <strong>the</strong> GL, but far away from <strong>the</strong> <strong>Security</strong><br />
<strong>Fence</strong>, while several smaller settlements near <strong>the</strong> GL that appeared in <strong>the</strong> poll are much<br />
closer to <strong>the</strong> <strong>Security</strong> fence <strong>and</strong> hence are more insecure. <strong>The</strong>refore, <strong>the</strong> DGL for cities<br />
beyond <strong>the</strong> GL first were set to zero, making <strong>the</strong>m equally insecure.<br />
3. Hypo<strong>the</strong>sis<br />
A zero hypo<strong>the</strong>sis is that <strong>the</strong> major factors, trio (ethnicity, religiosity <strong>and</strong> nationality) +<br />
social, are <strong>the</strong> only significant variables to explain <strong>the</strong> result of <strong>the</strong> Right-Left vote.<br />
An alternative hypo<strong>the</strong>sis is that <strong>the</strong>re is at least one more significant variable, which we<br />
designate as a psychological (security) factor.<br />
4. Data Description<br />
<strong>The</strong> technical part of our research was alleviated due to two fortunate circumstances. <strong>The</strong><br />
first was <strong>the</strong> availability of <strong>the</strong> powerful SPSS 10.0 Data Editor, which can perform<br />
multivariable regression analysis. Ano<strong>the</strong>r piece of luck came from being able to obtain<br />
<strong>the</strong> special arrangement of data necessary for our research. <strong>The</strong> Israeli Central Bureau of<br />
Statistics had prepared a special “integrated” electronic file for <strong>the</strong> 1995 census <strong>and</strong> 1996<br />
elections data for 2,257 statistical areas (Electronic File with Results of Israeli 1996 <strong>and</strong><br />
1999 Elections <strong>and</strong> 1995 Census, 2003, hereinafter, Census data). 24<br />
<strong>The</strong> number of statistical areas came from <strong>the</strong> following arithmetic: <strong>The</strong>re were 1,559<br />
statistical areas comprising cities with population of more than 10,000, <strong>and</strong> 975 areas
17<br />
made up of small communities. In <strong>the</strong> former, a city was subdivided into several<br />
statistical areas to achieve an average number of residents, about 2,000-4,000 in one area.<br />
It appears that both numbers had undergone an additional mixture: some 120 urban <strong>and</strong><br />
120 small statistical areas (each with small numbers of residents or voters) were grouped<br />
toge<strong>the</strong>r with ano<strong>the</strong>r small one, within <strong>the</strong> same geographical territory. <strong>The</strong>refore, this<br />
nuisance factor does not influence our major parameters <strong>and</strong> conclusions. 25<br />
a. Description of variables <strong>and</strong> units of measurements<br />
<strong>The</strong> dependent variables, Netanyahu_96 <strong>and</strong> Netanyahu_99, show <strong>the</strong> percentage of votes<br />
cast for Netanyahu personally in 1996 <strong>and</strong> 1999 respectively.<br />
1. Ethnic Background<br />
This is represented by <strong>the</strong> percentage of those coming from <strong>the</strong> same country of origin<br />
{Turkey, Iraq, Yemen, Iran, Morocco, Libya, Tunisia, Ethiopia, Romania, Bulgaria,<br />
Germany, Pol<strong>and</strong>, former USSR, Latin America, <strong>and</strong> North America} in a given<br />
statistical area. 26<br />
2. Degree of Religiosity<br />
This parameter is not reported in <strong>the</strong> census. We identified it by <strong>the</strong> number of people in<br />
<strong>the</strong> family; variable Person_K shows families with K members. We use only Person_7+<br />
(seven <strong>and</strong> more members) <strong>and</strong> Person_1 (one member), which we assume to represent<br />
<strong>the</strong> two poles – very religious <strong>and</strong> secular.
18<br />
3. Social Class<br />
This parameter is represented jointly by Income <strong>and</strong> Occupation. Income here means<br />
average income within <strong>the</strong> area, in NIS (New Israeli Shekels). We added to this two<br />
variables: Rich <strong>and</strong> Poor – each represents a percentage of those that fall within <strong>the</strong><br />
wealthiest 10% <strong>and</strong> poorest 10%, respectively, of <strong>the</strong> Israeli population. Academic degree<br />
played proxy for Occupation <strong>and</strong> was measured in aggregate data analysis as a<br />
percentage of people with at least a first academic degree. 27<br />
4. Nationality<br />
Variables – Jew, Muslim, Christian, <strong>and</strong> Druze – show <strong>the</strong> percentage of nationalities of<br />
residents in <strong>the</strong> statistical area.<br />
5. Inner <strong>Security</strong> (Psychology)<br />
DGL is given as <strong>the</strong> shortest distance (measured in km) from <strong>the</strong> locality to <strong>the</strong> closest of<br />
three GL segments: Judea, Samaria, or Gaza.<br />
6. Geography<br />
DB is given as <strong>the</strong> distance in km to <strong>the</strong> closest Israeli border.<br />
Location by Sea is a dummy binary variable {1,0} that shows whe<strong>the</strong>r a statistical area is<br />
located by <strong>the</strong> sea (Mediterranean or Red) or not.<br />
7. Age
19<br />
Israeli population was divided into 5 age groups: (18-24), (25-44), (45-64), (64-75),<br />
(75+), each represented by a percentage.<br />
8. Gender<br />
Percentages of Males <strong>and</strong> Females. Census gave <strong>the</strong> number of those older than 15 years.<br />
Let us stress here that we are not overly concerned with <strong>the</strong> exact influence of control<br />
variables (which are aggregated), but only with <strong>the</strong>ir signs <strong>and</strong> <strong>the</strong>ir relevance (supported<br />
by high T-statistics). As for psychological <strong>and</strong> geographical variables, we are concerned<br />
with both <strong>the</strong> value of regression coefficients <strong>and</strong> <strong>the</strong>ir T-statistics.<br />
Correlations<br />
<strong>The</strong> strongest correlation, 0.6, is found between Income <strong>and</strong> Academic Degree. This<br />
means that we have here an example of multicolinearity (Ramanathan, 2002: ch.5), <strong>and</strong><br />
<strong>the</strong>se two variables seem to be interchangeable since both carry <strong>the</strong> same pro-Left bias.<br />
<strong>The</strong> latter also has a strong correlation with different ethnic backgrounds: 0.4 for<br />
Germans <strong>and</strong> North Americans, 0.23 for Polish, 0.2 for South Americans, nearly 0 for<br />
Romanians <strong>and</strong> Bulgarians, <strong>and</strong> -0.2 for Asian <strong>and</strong> African Jews – all at <strong>the</strong> 1% level.<br />
5. Psychological <strong>and</strong> Geographical Factors: simple analysis<br />
It is not a great surprise that <strong>the</strong> DGL must be an important factor. In tiny Israel,<br />
distances are ridiculously small. Jerusalem is located within 3 km of <strong>the</strong> border of Judea;<br />
Beer-Sheba within 17 km of <strong>the</strong> border with Gaza; <strong>and</strong> Tel Aviv <strong>and</strong> Haifa are within 20<br />
<strong>and</strong> 39 km, respectively, of <strong>the</strong> border of Samaria. 28 It is ra<strong>the</strong>r a surprise that <strong>the</strong> P-factor
20<br />
was never considered earlier! To illustrate this point by simple means, we prepared Table<br />
1 that displays <strong>the</strong> pro-Right vote in 15 major Israeli cities, where only electoral areas<br />
with a Jewish population higher than 90% were counted. 29<br />
<strong>The</strong> 15 cities were lined up according to <strong>the</strong>ir distance from <strong>the</strong> GL. All of <strong>the</strong><br />
settlements beyond <strong>the</strong> GL were assigned zero DGL. 30 “# Stat areas” in Table 1 means<br />
<strong>the</strong> number of electoral areas within a city, or, in effect, <strong>the</strong> relative size of <strong>the</strong> city.<br />
City <strong>Green</strong><br />
line<br />
Jeru<br />
salem<br />
Petah<br />
Tikwa<br />
Ashk<br />
elon<br />
Bnei<br />
Brak<br />
Neta<br />
nia<br />
Ramat<br />
Gan<br />
Beer<br />
Sheba<br />
Rishon<br />
Lezion<br />
Reho<br />
vot<br />
Tel<br />
Aviv<br />
Hol<br />
on<br />
Bat<br />
Yam<br />
Ash<br />
dod<br />
Hai<br />
fa<br />
#Stat_areas 82 143 44 20 32 46 41 41 41 26 129 46 39 36 59<br />
DGL 0 3.2 9 10.8 14.3 15 15.6 17 17.3 17.3 20 20 23.5 29.6 39<br />
Loc_by_Sea No No No Yes No Yes No No Yes No Yes No Yes Yes Yes<br />
Dist_Border 32 62 49 65 67 65 49 69 56.7 72 69 74 69 31<br />
1996 87.7 71 57.6 61 88 60.5 48.5 62 51 50 44 51 55.5 66.5 42<br />
1999 79.4 65 50.5 60 87 54 39 55 42 43.4 36 44 48 58 33<br />
Table 1. Pro-Netanyahu Vote vs. Psychological <strong>and</strong> Geographical Factors<br />
in 15 Major Israeli Localities in 1996 Elections<br />
At a glance, one can see extremely strong support for <strong>the</strong> Right near <strong>the</strong> GL <strong>and</strong> that it<br />
gradually decreases going away from it. Of course, for those living in <strong>the</strong> settlements<br />
beyond <strong>the</strong> GL, a vote for <strong>the</strong> Right was <strong>the</strong> obvious choice. Some irregularities versus<br />
linear behavior also can be explained in light of what we learned earlier. <strong>The</strong> extremely<br />
high percentage of pro-Right voters in Bnei Brak (15 km) could be explained by a<br />
disproportional number of haredim (religious Jewish population). Ano<strong>the</strong>r sharp rise in<br />
percent of pro-Right voters in comparatively distant Ashdod (29 km) can be explained by<br />
<strong>the</strong> ethnic component: Russian <strong>and</strong> Moroccan Jews comprise more than two-thirds of its<br />
population. <strong>The</strong> results are shown as a graph (Figure 1):
21<br />
80%<br />
60%<br />
40%<br />
Each city is represented<br />
by a pair of dots of <strong>the</strong><br />
same color: dark for 1996<br />
<strong>and</strong> bright for 1999.<br />
Color signifies location:<br />
Blue: sea<br />
Brown: mountain<br />
<strong>Green</strong>: valley<br />
Yellow: desert<br />
Black: Bnei Brak<br />
Dashed line: regression<br />
1996 with slope –0.3<br />
Dotted line: regression<br />
1999 with slope –0.3<br />
0 20 40<br />
……………..20km………………..…………... 40km<br />
Figure 1. Vote pro-Netanyahu vs. Distance to <strong>the</strong> <strong>Green</strong> <strong>Line</strong> for 15 Major Localities<br />
Figure 1 is very telling. Drawing a straight line through Haifa (DGL = 40km, Right vote<br />
in 1996 = 42%) <strong>and</strong> <strong>the</strong> center point of Greater Tel-Aviv 31 (DGL = 20km, Right vote in<br />
1996 = 48%) one can see <strong>the</strong> downward slope of about 3% per every 10km. A visible<br />
jump upward of Jerusalem <strong>and</strong> <strong>the</strong> settlements near <strong>the</strong> GL are mainly due to <strong>the</strong><br />
religiosity factor, as we explained earlier, <strong>and</strong> we can even measure its intensity in<br />
Jerusalem by comparing a projected 53% vs. 71% de facto.<br />
Figure 1 also shows that in <strong>the</strong> 1999 elections, Netanyahu almost uniformly received 6%<br />
to 8% fewer votes than in 1996, which confirms <strong>the</strong> existence of a firm slope that<br />
depends on <strong>the</strong> DGL. This encourages performing <strong>the</strong> multivariable <strong>Line</strong>ar Regression<br />
analysis.
22<br />
Historical Remark<br />
Historically speaking, <strong>the</strong> same result could be discerned in <strong>the</strong> results of “split-ticket<br />
voting,” when Israeli voters placed <strong>the</strong>ir personal concerns on <strong>the</strong> shoulders of local<br />
municipalities ra<strong>the</strong>r than political parties, while <strong>the</strong> Knesset was likely associated with<br />
<strong>the</strong> “global” security issue. <strong>The</strong> data (Arian, 1983: 186) for <strong>the</strong> 1965 <strong>and</strong> 1969 national<br />
elections clearly point to greater concern for security near <strong>the</strong> GL (Jerusalem) than in a<br />
comparatively secure locality (Ramat Gan). For example, in Jerusalem, <strong>the</strong> hawkish<br />
Gahal party (future Likud party) received 28.3% to Knesset <strong>and</strong> 17.3% to municipality in<br />
1965 <strong>and</strong> 28.3% <strong>and</strong> 16.9%, respectively, in 1969; while in Ramat Gan, <strong>the</strong> numbers<br />
were reversed: 29.2% <strong>and</strong> 39.7% in 1965 <strong>and</strong> 29.7% <strong>and</strong> 39.6% in 1969. As expected, <strong>the</strong><br />
vote for <strong>the</strong> Alignment party (future Labor party) shows just <strong>the</strong> reverse relation between<br />
Knesset <strong>and</strong> municipality votes. After discussing this fact, Arian (1983: 183) explained it<br />
as <strong>the</strong> result of “party organization <strong>and</strong> electoral loyalties toward <strong>the</strong> mayors” 32 <strong>and</strong> did<br />
not proceed to analyze <strong>the</strong> geographical or psychological aspects of <strong>the</strong> facts.<br />
6. Regression Analysis of Aggregate Data<br />
A Five-Level Model<br />
We used <strong>the</strong> linear regression program within <strong>the</strong> SPSS 10.0 for Windows Data Editor.<br />
Regression analysis was performed in five steps. First were considered control variables:<br />
1) demographic, 2) ethnic background, 3) religiosity, 4) social factors <strong>and</strong> only <strong>the</strong>n were<br />
added psychological <strong>and</strong> geographical factors. Were included 1,811 areas with Jews ><br />
90%, <strong>and</strong> 1 km
23<br />
<strong>the</strong> fifth level. After adding social factors, many countries that represent Ethnic<br />
Background became insignificant (North America, Germany, Bulgaria, etc.), so we chose<br />
an alternative variant – to take <strong>the</strong> continent as a whole <strong>and</strong> pick <strong>the</strong> most noticeable<br />
representative from every continent. Only at <strong>the</strong> fifth level did we add three variables<br />
under investigation: DGL, Location by Sea, <strong>and</strong> DB. <strong>The</strong> form of <strong>the</strong> data in <strong>the</strong> Figure 1<br />
resembles logarithmic function <strong>and</strong> suggests trying, not <strong>the</strong> linear distance DGL, but Log<br />
DGL as a variable. All unmarked coefficients are at least *** significant. <strong>The</strong> factors that<br />
did not feature significantly in <strong>the</strong> last two stages, like Gender, Academic Degree, DB<br />
<strong>and</strong> Location by Sea, were entirely removed from <strong>the</strong> model.<br />
1996 Elections<br />
Category Variables I II III IV V<br />
Constant 98.582 97.028 65.805 65.490 71.295<br />
AGE25_44 -.816 -.946 -.372 -.314 -.320<br />
Demography AGE45_64 -1.362 -1.194 -.599 -.404 -.406<br />
AGE65_74 -.176 .736 1.330 .814 .889<br />
YEMEN 1.549 1.313 1.087 1.045<br />
Origin MAROCCO 1.246 .986 .637 .662<br />
POLAND -2.733 -1.712 -1.062 -.973<br />
LATIN_AMER -2.166 -1.569 -1.458 -1.528<br />
Religiosity PERSON1 -.312 -.475 -.513<br />
PERSON7 1.553 1.272 1.201<br />
POOR .108 .110**<br />
RICH -.163 -.160<br />
Income CARS0 .078* .108*<br />
CARS1 .181 .211<br />
CARS2 -.295 -.290<br />
Psychology LOG_DGL -3.048<br />
R-square .103 .432 .576 .635 .644<br />
F-statistics 69.5 196 272 223 216<br />
Dependent Variable: Netanyahu_96<br />
Table 2. Pro-Netanyahu Vote in 1996 Elections – Five-level Model
24<br />
Demographic factors include Age <strong>and</strong> Gender. All <strong>the</strong> displayed Age groups’ votes were<br />
*** (p < 0.001) significant; Age 74+ was a control group, while Age 18-25’ vote came<br />
out insignificant. Gender appeared to be non-significant.<br />
Ethnic background is represented by continents: Yemen for Asia, Morocco for Africa,<br />
Pol<strong>and</strong> for Europe, <strong>and</strong> Latin America for itself. All are **** (p < 0.0001) significant <strong>and</strong><br />
show <strong>the</strong> expected signs: <strong>the</strong> former two are pro Right, <strong>the</strong> latter two, pro-Left.<br />
<strong>The</strong> Religiosity factor is represented by Person_1 <strong>and</strong> Person 7+ variables, which are<br />
**** significant: <strong>the</strong> former has negative sign, <strong>the</strong> latter – positive. One can argue that<br />
this not a surprise, since this factor correlates with <strong>the</strong> previous one, but <strong>the</strong> fact that <strong>the</strong><br />
ethnic background remains **** significant even after adding <strong>the</strong>se two variables<br />
strongly shows that ano<strong>the</strong>r independent parameter exists. <strong>The</strong>ir **** significance<br />
ascertains that <strong>the</strong>y grasp <strong>the</strong> essence of <strong>the</strong> Religiosity factor <strong>the</strong>y represent.<br />
Social factors are represented by Income <strong>and</strong> Academic Degree; <strong>the</strong> latter is insignificant.<br />
<strong>The</strong> former splits into Poor <strong>and</strong> Rich <strong>and</strong> <strong>the</strong> number of cars in <strong>the</strong> family; people with 0<br />
cars vote unclearly; <strong>the</strong> rest has expected signs of <strong>the</strong> coefficients: Poor <strong>and</strong> low middle<br />
class (owners of 1 car) vote pro-Right, Rich <strong>and</strong> high middle class (owners of 2 cars) vote<br />
pro-Left. F-statistics of <strong>the</strong> model drop after adding <strong>the</strong> social factor.
25<br />
At <strong>the</strong> fifth level <strong>the</strong>re were added <strong>the</strong> psychological <strong>and</strong> geographical factors: Log DGL<br />
toge<strong>the</strong>r with DB <strong>and</strong> Location by Sea. <strong>The</strong> first was *** significant while <strong>the</strong> o<strong>the</strong>r two<br />
were insignificant. <strong>The</strong> coefficient -3 before <strong>the</strong> former shows that every 1 km far<strong>the</strong>r<br />
away from <strong>the</strong> GL, <strong>the</strong> Right loses 3/DGL % of votes; i.e., near Jerusalem, about 1%;<br />
near Tel-Aviv, 3/20 = 0.15%; <strong>and</strong> near Haifa, 3/40 = 0.075% vote.<br />
<strong>The</strong> R-square of <strong>the</strong> model reached 0.644 with final F-statistic at 216. Adding all <strong>the</strong><br />
countries for <strong>the</strong> Ethnic Background, not just several, improves R-square insignificantly<br />
to 0.68, while <strong>the</strong> overall F-statistics drop from 216 to 148 (mainly because of <strong>the</strong> large<br />
number of degrees of freedom = 26).<br />
1999 Elections<br />
Running <strong>the</strong> same set of variables against <strong>the</strong> variable Netanyahu_99, we obtained<br />
practically <strong>the</strong> same results (Table 3). Again 1,811 localities were used, where Jews ><br />
90% with 1 km < DGL < 60 km. <strong>The</strong> difference in Table 2 is that Polish background here<br />
featured insignificantly <strong>and</strong> was dropped. We tried continents instead, but again, Europe<br />
featured insignificantly <strong>and</strong> was dropped.
26<br />
Category Variables I II III IV V<br />
Constant 100.52 82.278 51.516 56.857 62.208<br />
AGE25_44 -1.004 -1.079 -.460 -.417 -.421<br />
Demography AGE45_64 -1.554 -1.447 -.754 -.552 -.545<br />
AGE65_74 -.233 .220 1.008 .461 .544<br />
ASIA .479 .364 .302 .287<br />
Origin AFRICA .477 .326 .227 .225<br />
LATIN_AMER -1.308 -.880 -.996 -1.074<br />
Religiosity PERSON1 -.264 -.539 -.568<br />
PERSON7 1.749 1.238 1.179<br />
POOR .117 .118<br />
RICH -.138 -.135<br />
Income CARS0 .174 .198<br />
CARS1 .116 .142**<br />
CARS2 -.369 -.366<br />
Psychology LOG_DGL -2.676<br />
R-square .145 .419 .594 .700 .707<br />
F-statistics 101 217 329 322 309<br />
Dependent Variable: Netanyahu_99<br />
Table 3. Pro-Netanyahu Vote in 1999 Elections – Five-level Model<br />
This time only <strong>the</strong> variable Cars_1 dropped in significance, while <strong>the</strong> overall picture is<br />
practically <strong>the</strong> same: all <strong>the</strong> variables have <strong>the</strong> expected signs <strong>and</strong> are highly significant.<br />
Again, F-statistics dropped at <strong>the</strong> fourth <strong>and</strong> fifth levels. Again, <strong>the</strong> DGL was ***<br />
significant, while <strong>the</strong> DB <strong>and</strong> Location by Sea were insignificant. <strong>The</strong> coefficient -2.6<br />
before <strong>the</strong> DGL shows that Fear was slightly reduced: every 1 km away from <strong>the</strong> GL, <strong>the</strong><br />
Right loses 2.6/DGL % of votes.<br />
<strong>The</strong> R-square of <strong>the</strong> model reached 0.7 with final F-statistic at 309. Adding for Ethnic<br />
Background <strong>the</strong> 16 countries of origin instead of continents does slightly improve R-<br />
square to 0.75, but again causes final F-statistics to drop from 309 to 202.
27<br />
Analysis of <strong>the</strong> Results<br />
<strong>The</strong> fact that DB <strong>and</strong> Location by Sea are not significant might be interpreted in <strong>the</strong><br />
following way. <strong>The</strong> “center-periphery” problem might be considered non-existent in<br />
Israel; this confirms <strong>the</strong> results of Andersen <strong>and</strong> Yaish (2003). <strong>The</strong>re is no “periphery” in<br />
Israel from a political point of view. Indeed, <strong>the</strong> Galilee (mo<strong>the</strong>rl<strong>and</strong> of “periphery”<br />
kibbutzim), “red” Haifa, <strong>and</strong> Nor<strong>the</strong>rn Tel Aviv all voted for <strong>the</strong> Left, suggesting a<br />
conclusive sample from <strong>the</strong> different regions of <strong>the</strong> country.<br />
<strong>The</strong> fact that <strong>the</strong> DB is not significant shows that Israeli borders are not considered<br />
dangerous by residents. This claim is supported by a separate regression analysis for<br />
communities located within 20 km of <strong>the</strong> border, where all three variables feature as<br />
insignificant. While this claim seems obvious with respect to borders with Egypt <strong>and</strong><br />
Jordan, <strong>the</strong> “calm” near <strong>the</strong> two o<strong>the</strong>r borders is remarkable. Though <strong>the</strong>re was no peace<br />
with Syria <strong>and</strong> Lebanon, in <strong>the</strong> eyes of <strong>the</strong> local residents, <strong>the</strong> Israeli occupation of <strong>the</strong><br />
Golan Heights <strong>and</strong> Sou<strong>the</strong>rn Lebanon (until 2000) provided a reliable buffer against a<br />
military threat. We strongly expect to see a rise in <strong>the</strong> significance of DB in <strong>the</strong> 2001<br />
elections, after Israel retreated from Lebanon.<br />
<strong>The</strong> fact that <strong>the</strong> same model works better for <strong>the</strong> 1999 than for <strong>the</strong> 1996 elections comes<br />
as a surprise. Between 1996 <strong>and</strong> 1999, large numbers of new immigrants continued to<br />
arrive in Israel, mainly from <strong>the</strong> former Soviet Union <strong>and</strong> Latin America, transforming<br />
<strong>the</strong> ethnic background of some communities. 33 In addition, this was <strong>the</strong> peak of a shortlived<br />
economic boom, which changed income distribution in Israel in a volatile manner. 34
28<br />
8. Evidence from Individual-Level Data<br />
<strong>The</strong> data from <strong>the</strong> poll conducted by <strong>the</strong> first author with his students in 2005 <strong>and</strong><br />
organized in electronic format as an Excel file by <strong>the</strong> second author was intended to<br />
verify <strong>the</strong> results coming from <strong>the</strong> aggregate data. <strong>The</strong> individual poll has 1,215<br />
respondents, consisting of persons from 127 localities, 36 of which are within <strong>the</strong> GL <strong>and</strong><br />
91 outside, in Israel proper. An advantage was to analyze <strong>the</strong> 2001 vote in addition to <strong>the</strong><br />
1996 <strong>and</strong> 1999 votes. We targeted <strong>the</strong> main socio-demographic characteristics that we<br />
tested earlier: Religiosity, Gender, Age, <strong>and</strong> Academic Degree, toge<strong>the</strong>r with <strong>the</strong><br />
psychological variable, DGL, <strong>and</strong> <strong>the</strong> geographical variable, <strong>the</strong> DB. We dropped <strong>the</strong><br />
Location at Sea since it strongly ** correlated to <strong>the</strong> latter two.<br />
<strong>The</strong> percentages (rounded to <strong>the</strong> closest integer) are:<br />
Number of Cars in 1996/1999/2001: 0 – 41/33/28%; 1 – 40/46/47%; 2 – 17/17/20%; 3 –<br />
2/3/4%; 4 – 0.4/0.6/1.1%.<br />
Number of Children in <strong>the</strong> Family in 1996/1999/2001: 0 – 37/34/29%; 1 – 10/11/12%; 2<br />
– 16/15/16%; 3 – 15/16/17%; 4 – 11/12/13%; 5 – 6/6/6%; 6 – 2/2/2%; 7 – 2/2/2%; 8 or<br />
more - 2/2/2%.<br />
Number of Children in <strong>the</strong> Army in 1996/1999/2001: 0 – 77/73/75%; 1 – 17/19/19%; 2 –<br />
5/7/5%; 3 or 4 – 1/1/1%.<br />
Native-born in Israel - 56%; <strong>the</strong> rest had 32 countries of origin. 80% of our respondents<br />
lived outside <strong>the</strong> GL (Israel proper); 25% lived by <strong>the</strong> sea. Average distances: to <strong>the</strong> GL –<br />
11 km; to <strong>the</strong> Border – 54 km. Male participants amounted to 55% (52% among Jews),
29<br />
while statistical data from <strong>the</strong> Israeli Interior Ministry Web site shows 49.5% males in<br />
Israel (49% among Jews) in 1995-2005. 35<br />
Since it is known (see, e.g., Shalev <strong>and</strong> Kis 2002) that Israelis do not tell <strong>the</strong> truth when<br />
asked about income in a poll, we substituted <strong>the</strong> number of cars in <strong>the</strong> family for <strong>the</strong><br />
Income variable. We thought that <strong>the</strong> number of children in <strong>the</strong> family <strong>and</strong> <strong>the</strong> number of<br />
older children serving in <strong>the</strong> Israeli army might influence <strong>the</strong> individual vote, so we<br />
included both numbers as independent variables.<br />
<strong>The</strong> Religiosity variable acquired <strong>the</strong> meaning it always carries – strictness of<br />
observance. We made it an ordinal with four values: 0 – secular (51%), 1 – traditional<br />
(25%), 2 – orthodox (22%), 3 – ultra-orthodox (2%). <strong>The</strong> Central Bureau of Statistics<br />
gives 45%, 38%, 10%, 7%, respectively, for <strong>the</strong> entire Israeli Jewish population in 2005.<br />
<strong>The</strong> Academic degree variable also became ordinal with four values: 0 – high school<br />
(56%), 1 – first degree (30%), 2 – second degree (11%), 3 – third degree (3%). <strong>The</strong><br />
Central Bureau of Statistics gives 58%, 22%, 20%, respectively, for <strong>the</strong> Israeli Jewish<br />
population above age 15 in 2005; percentages for second <strong>and</strong> third degrees were<br />
combined.<br />
We have 95 Arab votes in <strong>the</strong> poll. All came from three cities, but all voted equivalently:<br />
for <strong>the</strong> Left. Hence <strong>the</strong> variable Nationality (1-Arab, 0-Jew) was not meaningful, <strong>and</strong> as a<br />
result, we excluded Arab voters from <strong>the</strong> regression analysis.
30<br />
<strong>The</strong>re remain 1,120 people between 22 <strong>and</strong> 86 years old who cast <strong>the</strong>ir vote at least once.<br />
Rounded to <strong>the</strong> nearest integer for mean Age <strong>and</strong> median Age were: for all 1,120: 44/46y;<br />
for those 1,082 older than 23y: 45/47y; for those 947 older than 26y: 48/49y. 36 Hence,<br />
after shifting back by 9, 6, <strong>and</strong> 4 years respectively, <strong>the</strong> true mean/median Ages were:<br />
39/40y in 1996, 39/41y in 1999 <strong>and</strong> 40/42y in 2001. Israeli Interior Ministry statistics<br />
show a median 28.5y, but over <strong>the</strong> total population, including <strong>the</strong> youngsters. Those<br />
younger than 19 years old constituted about 37% of <strong>the</strong> total. Recalculating <strong>the</strong> averages<br />
for eligible voters (18 years old <strong>and</strong> older), we found <strong>the</strong> median Age for <strong>the</strong> total falls in<br />
<strong>the</strong> middle of <strong>the</strong> 35-44y old group, a good agreement with <strong>the</strong> data from our poll.<br />
Again, as in <strong>the</strong> aggregate data analysis, <strong>the</strong> DGL variable was measured as <strong>the</strong> distance<br />
to <strong>the</strong> closest point from Judea, Samaria, or Gaza. Since 90% of those participants living<br />
within <strong>the</strong> GL were from <strong>the</strong> city of Ariel, <strong>and</strong> since <strong>the</strong> latter is not considered to be a<br />
religious settlement, we decided to keep all of <strong>the</strong> entries in <strong>the</strong> analysis, setting <strong>the</strong> DGL<br />
for <strong>the</strong>m as zero. <strong>The</strong> negative feature of <strong>the</strong> poll data was that DGL variable correlated<br />
with practically all o<strong>the</strong>r variables: Distance to <strong>the</strong> Border, Location at Sea, Religion,<br />
etc., at <strong>the</strong> ** level. <strong>The</strong>refore, we could not avoid <strong>the</strong> multicolinearity problem. Apart<br />
from that, <strong>the</strong> results in all three elections show strongly <strong>the</strong> importance of <strong>the</strong> P-factor,<br />
<strong>and</strong>, to lesser degree, <strong>the</strong> geographical variable.<br />
Since <strong>the</strong> dependent variables Vote96, Vote99, <strong>and</strong> Vote01 are binary (1-Right, 0-Left),<br />
we ran a binary logistic regression. <strong>The</strong>re are many ways, however, to discuss its
31<br />
goodness of fit – <strong>the</strong> best is to use <strong>the</strong> percentage of predictable outcomes. In all three<br />
cases, elections in 1996, 1999, <strong>and</strong> 2001, <strong>the</strong> percentage of correct predictions was near<br />
68% <strong>and</strong> R-square about 0.215. <strong>The</strong> results for individual variables are displayed in<br />
Tables 4, 5 <strong>and</strong> 6.<br />
Category Variables B S.E. Wald Sig. Exp(B)<br />
Socio-<br />
Demography<br />
Cars_1996<br />
Child_Family<br />
Child_Army<br />
Acad_Degree<br />
Age<br />
Gender<br />
Religiosity<br />
Years_in_Israel<br />
.328<br />
-.043<br />
-.082<br />
-.041<br />
.001<br />
-.202<br />
.349<br />
-.002<br />
.105<br />
.053<br />
.118<br />
.095<br />
.008<br />
.150<br />
.100<br />
.005<br />
9.773<br />
.655<br />
.481<br />
.186<br />
0.20<br />
1.804<br />
12.234<br />
.128<br />
.002<br />
.418<br />
.488<br />
.666<br />
.888<br />
.179<br />
.000<br />
.721<br />
1.388<br />
.958<br />
.921<br />
.960<br />
.999<br />
.817<br />
1.418<br />
.998<br />
Psychology DGL -.088 .009 88.070 .000 .916<br />
Geography DB -.020 .006 10.289 .001 .980<br />
Constant 2.200 .484 20.659 .000 9.021<br />
Table 4. 1996 Elections; Individual Poll; Vote for <strong>the</strong> Right; 994 Respondents.<br />
Socio-demographic variables contributed toge<strong>the</strong>r about 10% to <strong>the</strong> initial 50%, while<br />
Religion was <strong>the</strong> only *** significant variable. <strong>The</strong> cars variable had * significance. <strong>The</strong><br />
Age variable showed * significance only in 2001 elections (young people voted against<br />
<strong>the</strong> Right), but not in 1996 <strong>and</strong> 1999, in agreement with <strong>the</strong> results of <strong>the</strong> 1996 <strong>and</strong> 1999<br />
polls conducted by Shamir <strong>and</strong> Arian (1999: Table 2; <strong>The</strong> Elections in Israel-1999: 18-<br />
19, Table 1.2). <strong>The</strong> variables Gender, Number of Children in <strong>the</strong> Family, Number of<br />
Children in <strong>the</strong> Army, Academic Degree, <strong>and</strong> Years Spent in Israel do not become<br />
significant even at <strong>the</strong> * level.
32<br />
Category Variables B S.E. Wald Sig. Exp(B)<br />
Socio-<br />
Demography<br />
Cars_1999<br />
Child_Family<br />
Child_Army<br />
Acad_Degree<br />
Age<br />
Gender<br />
Religiosity<br />
Years_in_Israel<br />
.290<br />
-.044<br />
-.052<br />
-.006<br />
.007<br />
-.233<br />
.322<br />
.007<br />
.094<br />
.050<br />
.113<br />
.091<br />
.007<br />
.140<br />
.093<br />
.005<br />
9.626<br />
.801<br />
.214<br />
.004<br />
1.023<br />
2.783<br />
12.136<br />
1.909<br />
.002<br />
.371<br />
.644<br />
.949<br />
.312<br />
.095<br />
.000<br />
.167<br />
1.337<br />
.957<br />
.949<br />
.994<br />
1.007<br />
.792<br />
1.380<br />
1.007<br />
Psychology DGL -.083 .009 89.267 .000 .920<br />
Geography DB -.020 .006 12.151 .001 .980<br />
Constant 1.392 .422 10.876 .001 4.023<br />
Table 5. 1999 Elections; Individual Poll; Vote for <strong>the</strong> Right; 1,081 Respondents<br />
Adding <strong>the</strong> DGL variable in all three cases increased <strong>the</strong> result by about 7-10% – from<br />
60% to 67-70%. Each time, this variable became significant at <strong>the</strong> *** level. <strong>The</strong> "odds<br />
ratio" (<strong>the</strong> Exp(B) in <strong>the</strong> tables) equal to (for example) 0.92 case in 1999, suggests<br />
diminution by 2% of <strong>the</strong> voters for Right for each km from <strong>the</strong> GL (since 0.92 = 0.48 /<br />
0.52) which is remarkably consistent with <strong>the</strong> results of aggregate data analysis. Fur<strong>the</strong>r,<br />
adding <strong>the</strong> Distance to <strong>the</strong> Border variable in all three elections increased <strong>the</strong> result by<br />
only a tiny 0.5-1%. <strong>The</strong> latter variable was significant at <strong>the</strong> ** or *** levels. Its<br />
coefficient, equal to -0.02 in all three elections, was 4.4 times smaller in absolute value<br />
than <strong>the</strong> one before <strong>the</strong> DGL in 1996, 4.1 times smaller in 1999, but only 3 times smaller<br />
in <strong>the</strong> 2001 elections. This means that <strong>the</strong> P-factor gradually fell from 1996 to 2001, in<br />
contrast to “security identified as a major issue before elections” (<strong>The</strong> Elections in Israel<br />
– 2001, 16, fig. 2), which shows a significant drop of security concerns in 1999 that rose<br />
anew in <strong>the</strong> 2001 elections.
33<br />
Category Variables B S.E. Wald Sig. Exp(B)<br />
Socio-<br />
Demography<br />
Cars_2001<br />
Child_Family<br />
Child_Army<br />
Acad_Degree<br />
Age<br />
Gender<br />
Religiosity<br />
Years_in_Israel<br />
.208<br />
-.038<br />
-.134<br />
-.050<br />
.013<br />
-.242<br />
.380<br />
.001<br />
.083<br />
.049<br />
.107<br />
.088<br />
.006<br />
.134<br />
.092<br />
.005<br />
6.268<br />
.614<br />
1.568<br />
.316<br />
4.859<br />
3.248<br />
17.281<br />
.070<br />
.012<br />
.433<br />
.211<br />
.574<br />
.028<br />
.072<br />
.000<br />
.791<br />
1.231<br />
.963<br />
.875<br />
.952<br />
1.014<br />
.785<br />
1.463<br />
1.001<br />
Psychology DGL -.065 .008 70.228 .000 .937<br />
Geography DB -.022 .005 16.077 .001 .978<br />
Constant 1.313 .397 10.916 .001 3.718<br />
Table 6. 2001 Elections; Individual Poll; Vote for <strong>the</strong> Right; 1,120 Respondents<br />
Unfortunately, <strong>the</strong> data for <strong>the</strong> Ethnic variable (2-Sephardi, 1-Ashkenazi, 0-Arab) was<br />
incomplete. Some 350 Jewish voters did not identify <strong>the</strong>mselves properly: many may<br />
have considered <strong>the</strong>mselves as “mixed ethnic.” Hence, we dropped this variable from <strong>the</strong><br />
regression analysis. <strong>The</strong> straightforward bivariate correlation for 770 valid cases shows a<br />
0.1** coefficient between <strong>the</strong> Ethnic variable <strong>and</strong> each Vote variable. This, of course,<br />
means that Sephardim are more likely to vote Right.<br />
<strong>The</strong> addition of <strong>the</strong> Ethnic Background variable did not improve <strong>the</strong> final score<br />
significantly, only by about 1% each time. This seems strange at first glance; however,<br />
<strong>the</strong> same effect for individual polls was observed by Shamir <strong>and</strong> Arian (1999: Table 2) in<br />
<strong>the</strong>ir research on 1996 elections, where Ethnicity, which featured *** in <strong>the</strong> short (sociodemographic)<br />
list of variables, lost one * after adding so-called “issues” <strong>and</strong> lost ano<strong>the</strong>r<br />
* after adding so-called “performances.” An even more striking effect was observed in<br />
<strong>the</strong> 1999 elections (<strong>The</strong> Elections in Israel-2001, Table 1.2), where in <strong>the</strong> most general
34<br />
model, <strong>the</strong> Ethnicity factor became only ** significant, but lost one * after adding issues,<br />
<strong>and</strong> lost <strong>the</strong> last * after adding “performances.”<br />
<strong>The</strong> correlation of individual votes in <strong>the</strong> 1996 <strong>and</strong> 1999 elections (those older than 26y,<br />
including Arab votes) in our poll is 95.6%, which means that 4.4% of voters switched<br />
sides, while in reality 6.6% voters changed <strong>the</strong>ir affinity to Netanyahu in <strong>the</strong> 1999<br />
elections (from 50.5% in 1996 to 43.9% in 1999, see <strong>The</strong> Elections in Israel – 2001, 8).<br />
<strong>The</strong> difference between 4.4% <strong>and</strong> 6.6% is not * significant. 37<br />
Fur<strong>the</strong>r, our poll shows a 93.3% correlation of <strong>the</strong> individual vote in <strong>the</strong> 1999 <strong>and</strong> 2001<br />
elections (those older than 23y, including Arab votes), which means that 6.7% of voters<br />
changed sides. In <strong>the</strong> 2001 elections, Barak got 18.5% fewer votes, compared with <strong>the</strong><br />
1999 elections. However, in 2001 <strong>the</strong>re was <strong>the</strong> lowest turnout for <strong>the</strong> elections ever<br />
known: 2.7 million vs. 3.2 million in 1999 (<strong>The</strong> Elections in Israel-2001, 8). It is known<br />
that many of Barak’s 1999 supporters, Arabs included, were reluctant to vote in 2001,<br />
while our poll included only people who did vote in 1999 <strong>and</strong> in 2001. Despite this subtle<br />
point, overall, we believe our poll represents fairly <strong>the</strong> entire Israeli population in its<br />
diversity.<br />
9. <strong>The</strong> Individual Poll vs. Aggregate Data<br />
<strong>The</strong> poll highlighted <strong>the</strong> problems known as “ecological inference” regarding <strong>the</strong> Income<br />
<strong>and</strong> Ethnic background variables.
35<br />
1. <strong>The</strong> aggregate data analysis supports conclusions, derived by Shalev <strong>and</strong> Kis (2002,<br />
Fig. 4.1) in a similar analysis, that <strong>the</strong> poor vote for <strong>the</strong> Right. <strong>The</strong> individual poll cannot<br />
address this problem meaningfully since people do not report <strong>the</strong>ir income honestly. <strong>The</strong><br />
substitute for income – <strong>the</strong> Number of cars a family owns – in <strong>the</strong> poll shows a positive<br />
coefficient in favor of <strong>the</strong> Right, but it is more likely that <strong>the</strong> Number of cars alone is<br />
NOT an adequate measurement of Income.<br />
2. According to <strong>the</strong> aggregate data analysis, <strong>the</strong> major role in <strong>the</strong> political choice of <strong>the</strong><br />
Israeli voter was played by his Ethnic background, or equivalently, by <strong>the</strong> way his/her<br />
parents voted. This might imply that Israeli society is still a “ga<strong>the</strong>ring of exiles,” ra<strong>the</strong>r<br />
than a cohesive new cultural entity. However, <strong>the</strong> individual-level poll did not support<br />
this conclusion. <strong>The</strong> answer to this dichotomy (a new example of “ecological inference”)<br />
comes from a well-known fact (Freedman 2001): people prefer to live near <strong>the</strong>ir social peers<br />
ra<strong>the</strong>r than ethnic/cultural peers.<br />
However, <strong>the</strong> aggregate data analysis highlighted a new fact: a simple division of <strong>the</strong><br />
Israeli population into Ashkenazim <strong>and</strong> Sephardim/Oriental is inadequate to discover<br />
people’s political preferences. Turkish <strong>and</strong> South American Jews gravitate toward<br />
European views, while North American Jews vote as do Oriental Jews. Religious fervor<br />
is likely <strong>the</strong> most defining factor behind this division. When <strong>the</strong> poll was able to address<br />
<strong>the</strong> latter factor comparatively adequately, <strong>the</strong> role of <strong>the</strong> Ethnic Background was<br />
completely obliterated.
36<br />
10. Major Conclusions: <strong>the</strong> Role of <strong>the</strong> P-factor<br />
Summarizing, we addressed <strong>and</strong> partially solved several interesting methodological<br />
problems, which for many years have been beyond <strong>the</strong> grasp of practitioners in <strong>the</strong> field.<br />
<strong>The</strong> variable we introduced, <strong>the</strong> DGL, appears to capture <strong>the</strong> bias toward <strong>the</strong> Right of a<br />
Jewish Israeli voter who resides near <strong>the</strong> GL, due to <strong>the</strong> obvious threat coming from<br />
behind <strong>the</strong> GL. Among Palestinians, <strong>the</strong> level of (hypo<strong>the</strong>tical) support for different types of<br />
armed attacks remains high: 92% for attacks against soldiers, 92% for attacks against settlers, <strong>and</strong><br />
58% for attacks against civilians inside Israel. 38<br />
1. According to <strong>the</strong> aggregate data analysis for <strong>the</strong> 1996 <strong>and</strong> 1999 elections, <strong>the</strong><br />
psychological factor, as measured by <strong>the</strong> DGL, played a significant role in <strong>the</strong> outcome of<br />
<strong>the</strong> 1996 <strong>and</strong> 1999 Israeli elections, though less significant compared with ethnic,<br />
religious, <strong>and</strong> social factors. <strong>The</strong> quantitative claim is that for every 1 km away from <strong>the</strong><br />
GL inside Israeli territory, <strong>the</strong> Right loses 3/DGL% of <strong>the</strong> vote. A more refined<br />
measurement should be found for cities inside <strong>the</strong> GL; for example, to <strong>the</strong> closest<br />
dangerous place, or perhaps to <strong>the</strong> <strong>Security</strong> <strong>Fence</strong>.<br />
According to <strong>the</strong> individual poll, <strong>the</strong> P-factor was *** significant in all three elections.<br />
<strong>The</strong> "odds ratio" speaks of Right losing 2% of <strong>the</strong> vote every additional kilometer far<strong>the</strong>r<br />
from <strong>the</strong> GL. It was <strong>the</strong> most Wald-significant factor, several times more Waldsignificant<br />
than <strong>the</strong> geographical factor, Distance to <strong>the</strong> Border, <strong>and</strong> even more Waldsignificant<br />
than ano<strong>the</strong>r well-known *** factor, Religiosity, though it is lower than <strong>the</strong><br />
latter in intensity. A scientist with <strong>the</strong> same income <strong>and</strong> color of yarmulke is more likely
37<br />
to vote Right in Tel Aviv (20 km) than in Haifa (39 km). A non-religious dockworker is<br />
more likely to vote Right in Ashkelon (11 km) than he would if he lived in Ashdod (29<br />
km).<br />
2. <strong>The</strong> analysis of <strong>the</strong> aggregate data involving two geographical variables (Distance to<br />
<strong>the</strong> Border <strong>and</strong> Location by Sea), indicates that traditionally assumed center-periphery<br />
bias is nonexistent in Israel, at least since 1996.<br />
3. <strong>The</strong> direction of causation problem, which appeared along <strong>the</strong> way <strong>and</strong> is not<br />
automatically resolved by doing regression analysis, was partially resolved by placing<br />
certain restrictions on <strong>the</strong> DGL – excluding data for historic Biblical places within <strong>the</strong> GL<br />
from <strong>the</strong> analysis.<br />
11. Final Remarks <strong>and</strong> Open Problems<br />
<strong>The</strong> P-factor came to <strong>the</strong> forefront in 1996 due to <strong>the</strong> suicide bombings in Jerusalem <strong>and</strong><br />
Tel Aviv in 1995-96, though likely it was implicit in all previous elections, as well. This<br />
conclusion, expressing <strong>the</strong> importance of <strong>the</strong> P-factor, could be drawn from <strong>the</strong> results of<br />
“split-ticket voting” of <strong>the</strong> past when Israeli voters placed <strong>the</strong>ir personal concerns on <strong>the</strong><br />
shoulders of <strong>the</strong> local municipalities, ra<strong>the</strong>r than political parties, <strong>and</strong> addressed <strong>the</strong>ir<br />
psychological (security) concerns to <strong>the</strong> Knesset at large. <strong>The</strong> data presented in (Arian<br />
1973) for 1965 <strong>and</strong> 1969 national elections clearly point to greater security concerns near<br />
<strong>the</strong> GL (Jerusalem) than <strong>the</strong>y do for a comparatively secure locality (Ramat Gan).
38<br />
In September 2000, a new round of Palestinian terrorist suicide attacks interrupted <strong>the</strong> lull<br />
of <strong>the</strong> late 1990s. 39 <strong>The</strong> idea of constructing <strong>the</strong> <strong>Security</strong> <strong>Fence</strong> finally appeared on <strong>the</strong><br />
agenda in 2001. This is signaled by <strong>the</strong> reduction of <strong>the</strong> coefficient before <strong>the</strong> DGL, but<br />
Ariel Sharon, <strong>the</strong>n representing <strong>the</strong> Right, still got twice as many votes as Barak, on <strong>the</strong><br />
Left. Only <strong>the</strong> blind could miss <strong>the</strong> P-factor, but it would be important to check <strong>the</strong><br />
consistency of <strong>the</strong> coefficient before <strong>the</strong> DGL: (-3/DGL % per 1 km) with a reliable set of<br />
control parameters.<br />
Though by <strong>the</strong> next elections on January 28, 2003, <strong>the</strong> <strong>Security</strong> <strong>Fence</strong> had not yet been<br />
completed, its construction was underway, supported by <strong>the</strong> entire Israeli political<br />
spectrum. 40<br />
A feeling of increased security guaranteed ano<strong>the</strong>r sweeping victory for<br />
Sharon. Again, <strong>the</strong> importance of <strong>the</strong> P-factor would be of interest. We predict <strong>the</strong><br />
discovery of a drop of significance of <strong>the</strong> P-factor in <strong>the</strong> 2003 elections, <strong>and</strong> its possible<br />
disappearance in <strong>the</strong> latest, March 2006 elections True, a more refined analysis might be<br />
needed here, since <strong>the</strong> 2003 <strong>and</strong> 2006 elections lacked a major ingredient of our analysis,<br />
<strong>the</strong> two-ballot system. 41<br />
<strong>The</strong> relevance of <strong>the</strong> “distance” parameter we have introduced here for <strong>the</strong> general<br />
political or sociological discourse can be tested in <strong>the</strong> elections of countries where two<br />
rival populations live in clearly defined areas, such as <strong>the</strong> French-speaking vs. <strong>the</strong><br />
English-speaking in Canada, or <strong>the</strong> Flemish <strong>and</strong> Walloons in Belgium.<br />
Notes 42<br />
∗ Corresponding author. E-mail belenka@mail.biu.ac.il
39<br />
1 This is <strong>the</strong> old approach in Israeli sociology; see Shamir <strong>and</strong> Arian 1983. According to Shamir<br />
<strong>and</strong> Arian 1999: “Ethnicity effect is prominent as of 1977, but <strong>the</strong>re is no clear trend.”<br />
2 See also discussion about “cross cutting cleavages” in Israel At <strong>the</strong> Polls: 1996, 255.<br />
3 Until 1996, <strong>the</strong> Israeli system was a one-ballot system with proportional party list representation<br />
in <strong>the</strong> Knesset, a one-house parliament with 120 seats.<br />
4 This is a well-known Israeli political misnomer; see, for example, Shamir <strong>and</strong> Arian 1999;<br />
Andersen <strong>and</strong> Yaish 2003.<br />
5 <strong>The</strong> Third Way was to give up <strong>the</strong> “territories,” although holding <strong>the</strong> Golan Heights, while<br />
Israel Ba’Aliya did not clearly identify itself in <strong>the</strong> political spectrum (see Israel At <strong>the</strong> Polls:<br />
1996, 136).<br />
6 See, for example, its description in Ramanathan 2002 (ch. 12.2).<br />
7 Andersen <strong>and</strong> Yaish (2003) included in <strong>the</strong> analysis <strong>the</strong> marginal parties, extreme Left <strong>and</strong><br />
extreme Right.<br />
8 Like <strong>the</strong> one used by Shamir <strong>and</strong> Arian 1999, see also <strong>The</strong> Elections in Israel – 2001, 48-49.<br />
9 Those who need to underst<strong>and</strong> <strong>the</strong> vote in Arab sector may consult, for example, Israel At <strong>the</strong><br />
Polls: 1996, 103; <strong>The</strong> Elections in Israel – 2001, 55-103.<br />
10 Until recently, every kibbutz had its own writer(s).<br />
11 Even taking this as a hypo<strong>the</strong>sis to verify <strong>the</strong> number of residents, a good quantitative variable<br />
is a much better parameter.<br />
12 In Model A, it is significant at 21% level; in Model B at 8.4% level.<br />
13 This data seems unavailable by now; <strong>the</strong>refore, we had to measure it “by h<strong>and</strong>” for <strong>the</strong> poll by<br />
measuring <strong>the</strong> distances with help of <strong>the</strong> Google Earth program. This caused some differences in<br />
distances between <strong>the</strong> aggregate data <strong>and</strong> <strong>the</strong> poll data.<br />
14 Ei<strong>the</strong>r in arithmetic (-x) or multiplicative (1/x) sense. Here we discuss <strong>the</strong> former case only.<br />
15 See some supporting evidence in <strong>the</strong> Report of <strong>the</strong> Special Committee to Investigate Israeli<br />
Practices Affecting <strong>the</strong> Human Rights of <strong>the</strong> Palestinian People <strong>and</strong> O<strong>the</strong>r Arabs of <strong>the</strong> Occupied<br />
Territories (A/55/373/Add.1) (9 October 2000) at<br />
http://domino.un.org/UNISPAL.NSF/5ba47a5c6cef541b802563e000493b8c/d6f6fd922cd766168<br />
5256989005a3a80!OpenDocument<br />
16 See, e.g., Internet website http://www.ynet.co.il/articles/0,7340,L-2227415,00.html of Israeli<br />
newspaper Yediyot Achronot: settlers enjoyed 130 million NIS (c. $30 mln) in tax reductions in<br />
2001; more particular: an average eligible tax payer settler in Judea <strong>and</strong> Samaria (total 34, 430
40<br />
men) paid 6,456 NIS taxes less, in Gaza (total 1,890 men) – 8,934 NIS less than he would pay in<br />
Israel proper.<br />
17 “Peace Now” claims that 77% of settlers are those of “quality-of-life.” (see report at<br />
http://www.peacenow.org/policy.asp?rid=&cid=3377). <strong>The</strong> figures in <strong>the</strong> report are not supported<br />
by <strong>the</strong> table <strong>and</strong> clear-cut statistical analysis.<br />
18 Foundation for <strong>the</strong> Middle East Peace reports, for example, that on December 13, 1996 (see<br />
Web site at http://www.fmep.org/settlement_info/1996_settlement_timeline.html), <strong>the</strong> Netanyahu<br />
cabinet approved <strong>the</strong> restoration of unspecified levels of benefits <strong>and</strong> subsidies to settlers <strong>and</strong> to<br />
manufacturing, industrial, <strong>and</strong> commercial enterprises locating in settlements, which will now<br />
enjoy “A”-level national priority area status. <strong>The</strong> Rabin government had earlier canceled or<br />
reduced some of <strong>the</strong>se incentives.<br />
19 All of our predecessors excluded Arab voters from analysis for various reasons.<br />
20 Participation in <strong>the</strong> 2001 Elections dropped to 62.3%.<br />
21 We are grateful to Professor Gary King of Harvard for a discussion on this point.<br />
22 True, if we knew <strong>the</strong> variances for every statistical area, it would be possible to infer some<br />
statistical statements about individuals as well. Unfortunately, variances are not reported by<br />
statistics, only averages. Still, though not reported, <strong>the</strong> variances might not be too large. People<br />
prefer to live near <strong>the</strong>ir social peers. <strong>The</strong>re are a few “mixed” communities, such as Ramat Gan,<br />
but even in such a community within electoral areas <strong>the</strong> variance is likely to be small.<br />
23 <strong>The</strong> first trio is now a st<strong>and</strong>ard reference point in every study; see, e.g., Friedl<strong>and</strong>er et al 2002.<br />
24 <strong>The</strong> Electronic File with <strong>the</strong> 1996 <strong>and</strong> 1999 election results was created by <strong>the</strong> group Mikum<br />
<strong>and</strong> later sold to <strong>the</strong> Israeli Statistical Bureau. <strong>The</strong> number of electoral areas was higher than <strong>the</strong><br />
number of statistical areas, about 6,500 in number.<br />
25 <strong>The</strong>re were several more comparatively minor twists in <strong>the</strong> data, like special balloting urns for<br />
soldiers <strong>and</strong> diplomats. Due to <strong>the</strong> small number of eligible voters among <strong>the</strong> Arabs in East<br />
Jerusalem <strong>and</strong> <strong>the</strong> Druze villages on <strong>the</strong> Golan Heights, all of <strong>the</strong>ir electoral <strong>and</strong> statistical areas<br />
were united in one. This does not affect our research since we are concerned with predominantly<br />
Jewish areas.<br />
26 <strong>The</strong> country of origin was assigned to <strong>the</strong> second generation according to how <strong>the</strong>y identify<br />
<strong>the</strong>mselves, even though <strong>the</strong>y were Israeli-born.<br />
27 Israeli statisticians chose to record <strong>the</strong> number of years spent in study: 0-4, 5-8, 9-12, 13-15,<br />
16+, which in fact prevents us from identifying <strong>the</strong> actual number of people holding BA/BS or
41<br />
Masters degrees. We refer to <strong>the</strong> last group as holding BS degrees, though we certainly missed<br />
some of <strong>the</strong> previous group, those listed with 15 years of study.<br />
28 For Jerusalem, it was measured between <strong>the</strong> center of Jerusalem <strong>and</strong> <strong>the</strong> center of East<br />
Jerusalem, whose status is uncertain.<br />
29 <strong>The</strong>refore, Jerusalem additionally lost 5 statistical areas out of 148; Tel-Aviv lost 27 (all of<br />
Yaffo) out 156; while Haifa lost 26 out of 87.<br />
30 This decision obliterates a more complex picture for communities beyond <strong>the</strong> <strong>Green</strong> line.<br />
Jordan valley communities have always been seen, within <strong>the</strong> settlement movement, far<strong>the</strong>r Left<br />
than <strong>the</strong> communities closer to <strong>the</strong> <strong>Line</strong>, like Ariel <strong>and</strong> Ma'aleh Adumim.<br />
31 This, in addition to Tel-Aviv proper, includes Ramat Gan, Holon, Bat Yam, Rishon Lezion <strong>and</strong><br />
probably Petach Tikwa <strong>and</strong> Rehovot.<br />
32 Teddy Kolleck in Jerusalem <strong>and</strong> Abraham Krinitzi in Ramat Gan.<br />
33 Governmental policy was to disperse new immigrants uniformly across <strong>the</strong> country.<br />
34 In those years, <strong>the</strong> Israeli economy grew about 2% annually. Though <strong>the</strong> boom improved <strong>the</strong><br />
living st<strong>and</strong>ards of <strong>the</strong> entire population, redistributing <strong>the</strong> wealth through governmental<br />
channels, <strong>the</strong> poor received <strong>the</strong>ir increase by artificial means, such as transfer payments to<br />
religious institutions <strong>and</strong> <strong>the</strong> creation of government jobs, ra<strong>the</strong>r than making <strong>the</strong>ir fortunes by<br />
starting new businesses. After <strong>the</strong> “bubble” collapsed, <strong>the</strong>y became as poor as before.<br />
35 See Web site for <strong>the</strong> Israel’s Ministry of Interior (Hebrew text): http://www.moin.gov.il<br />
36 <strong>The</strong> st<strong>and</strong>ard deviation was 13-14 years.<br />
37 <strong>The</strong> aggregate data show a ** 95.9%.correlation between votes in 1996 <strong>and</strong> 1999 elections.<br />
38 Opinion poll # 3, conducted by <strong>the</strong> Palestinian Center for Policy & Survey Research, between<br />
19-24 December 2001. <strong>The</strong> total sample size of this poll is 1357 Palestinians 18 years <strong>and</strong> older.<br />
<strong>The</strong> margin of error is 3% <strong>and</strong> <strong>the</strong> non-response rate is 3%. See http://www.pcpsr.org/index.html.<br />
39 <strong>The</strong> so-called “second Intifada.”<br />
40 Despite strenuous <strong>and</strong> vocal resistance from <strong>the</strong> world community <strong>and</strong> <strong>the</strong> International Court<br />
of Justice at <strong>The</strong> Hague, in particular. See, e.g., <br />
41 <strong>The</strong> number of small splinter parties on <strong>the</strong> margin became so large in 1999 that in 2001, <strong>the</strong><br />
Knesset decided to return to <strong>the</strong> original one-ballot system.
42<br />
42 References<br />
1. Andersen R. <strong>and</strong> Yaish M. 2003. “Social cleavages, electoral reform <strong>and</strong> party choice: Israel’s<br />
‘natural’ experiment.” Electoral Studies 22:399-423.<br />
2. Arian A. 1973. <strong>The</strong> Choosing People. <strong>The</strong> Press of Case Western Reserve Univ. Clevel<strong>and</strong><br />
<strong>and</strong> London.<br />
3. Electronic File with Results of Israeli 1996 <strong>and</strong> 1999 Elections <strong>and</strong> 1995 Census. 2003. State<br />
of Israel, Central Bureau of Statistics, Jerusalem.<br />
4. Freedman D.A. 2001. “Ecological inference <strong>and</strong> <strong>the</strong> ecological fallacy.” International<br />
Encyclopedia of <strong>the</strong> Social <strong>and</strong> Behavioral Sciences. Oxford: Elsevier Science.<br />
5. Friedl<strong>and</strong>er D., Eisenbach Z., Ben-Moshe E., Lion-Elmakias L., Hleihel A., Lunievski S., <strong>and</strong><br />
Ben-Hur, D. 2002. Changes in Educational Attainments in Israel since <strong>the</strong> 1950s: <strong>The</strong> Effects<br />
of Religion, Ethnicity <strong>and</strong> Family Characteristics. Hebrew Univ. of Jerusalem <strong>and</strong> Central<br />
Bureau of Statistics.<br />
6. Israel At <strong>the</strong> Polls: 1996. 1998. Edited by Elazar D.J. <strong>and</strong> S<strong>and</strong>ler S. London, Portl<strong>and</strong>, OR:<br />
Frank Cass Publishers.<br />
7. De Marchi S., <strong>and</strong> Goemans H. 2001. “Bargaining <strong>and</strong> complex preferences: examining <strong>the</strong><br />
case of <strong>the</strong> Israeli electorate.” Durham, N.C.: Duke Univ. Paper presented at <strong>the</strong> Annual<br />
Meeting of <strong>the</strong> American Political Science Association, San Francisco, 2001. Preprint.<br />
8. Ramanathan R. 2002. Introductory Econometrics with Applications, 5 th ed., Mason, Ohio:<br />
South-Western, Thompson Learning.<br />
9. Shamir M <strong>and</strong> Arian A. 1983. “<strong>The</strong> ethnic vote in Israel’s 1981 elections.” Electoral Studies 1:<br />
315-331.<br />
10. Shamir M. <strong>and</strong> Arian A. 1999. “Collective identity <strong>and</strong> electoral competition in Israel.”<br />
American Political Science Review 93: 265-277.<br />
11. Shalev M. with Sigal Kis. 2002. “Social Cleavages among non-Arab voters. A new analysis.”<br />
In: <strong>The</strong> Elections in Israel – 1999. Arian A. <strong>and</strong> Shamir M., eds. Tel-Aviv: <strong>The</strong> Israel<br />
Democracy Institute, pp. 67-96.<br />
12. Smooha S. 1978. Israel: Pluralism <strong>and</strong> Conflict. London: Rutledge & Kegan Paul Ltd.;<br />
Berkeley <strong>and</strong> Los Angeles: Univ. California Press.<br />
13. <strong>The</strong> Elections in Israel – 2001. 2002. Arian A. <strong>and</strong> Shamir M., eds. Tel-Aviv: <strong>The</strong> Israel<br />
Democracy Institute.