EDA Study Shows 2008 CA Prop 8 Results Appear to Have Been Corrupted

Click Here to download the Introduction as a pdf [5]


The following study of suspect Proposition 8 election results in Los Angeles County, CA, is drawn from data gathered in EDA's Election Verification Exit Poll [6] (EVEP) analysis of the 2008 Presidential election, which reports similarly questionable election results in several states.

Although this exit poll analysis cannot provide conclusive proof of election fraud (because such proof would require access to memory cards and computer code accorded proprietary exemption from public examination) it does provide the strongest indirect proof available that election results have almost certainly been altered by manipulation of the computerized voting systems.

Deviations between exit polls and official results far outside margins of error, cannot be explained away by demographics or polling factors. The facts established in these reports cannot responsibly be dismissed or evaded.

Election Defense Alliance calls on legislators, secretaries of state, attorneys general, the voting public, and especially candidates in upcoming elections, to read these reports and seriously confront their implications.

  ___________________________________________________________________

 An EDA Investigative Report

'Exhaustive analysis of exit polls conducted in Los Angeles County has led to the conclusion that the vote count for Proposition 8 (the ban on same-sex marriage) appears to have been corrupted.

There were not enough Republican voters to account for the disparity between the exit poll and official results
even if every Republican non-responder voted for Proposition 8.

The Edison-Mitofsky exit poll showed a similar disparity statewide,
indicating that altered vote counts may not be limited to Los Angeles County.'

For the Introduction and Executive Summary:
http://electiondefensealliance.org/files/intro&exsum.pdf [7]

CITIZEN EXIT POLLS IN LOS ANGELES COUNTY: AN IN-DEPTH ANALYSIS 

Richard Hayes Phillips, Ph.D.

 

Download the PDF [8]

Appendices added

Related report: Introduction and Executive Summary [9]
_________________________________________

Abstract

 Exhaustive analysis of exit polls conducted in Los Angeles County has led to the inescapable conclusion that the vote count for Proposition 8 (the ban on same-sex marriage) was corrupted. The data were drawn from questionnaires filled out by 6326 voters at ten polling places scattered across Los Angeles County, and were properly adjusted to match the gender, age, race, and party affiliation of the electorate.

For Proposition 4 (which would have required parental notification and a waiting period for minors seeking abortions), the official results differ from the adjusted exit poll data by only 0.64%. But for Proposition 8, the disparity between the official results and the adjusted exit poll data is 5.74%, enough to affect the margin by 11.48%. Because Los Angeles County comprised 24.23% of the statewide electorate, an error of that magnitude would have affected the statewide margin by 2.78%, accounting for most of the official 4.48% statewide margin of victory. There were not enough Republican voters to account for the disparity between the exit poll and the official results even if every Republican non-responder voted for Proposition 8. The Edison-Mitofsky exit poll showed a similar disparity statewide, indicating that altered vote counts may not be limited to Los Angeles County. 

Click Here to download the Introduction as a pdf [5]


The following study of suspect Proposition 8 election results in Los Angeles County, CA, is drawn from data gathered in EDA's Election Verification Exit Poll [6] (EVEP) analysis of the 2008 Presidential election, which reports similarly questionable election results in several states.

Although this exit poll analysis cannot provide conclusive proof of election fraud (because such proof would require access to memory cards and computer code accorded proprietary exemption from public examination) it does provide the strongest indirect proof available that election results have almost certainly been altered by manipulation of the computerized voting systems.

Deviations between exit polls and official results far outside margins of error, cannot be explained away by demographics or polling factors. The facts established in these reports cannot responsibly be dismissed or evaded.

Election Defense Alliance calls on legislators, secretaries of state, attorneys general, the voting public, and especially candidates in upcoming elections, to read these reports and seriously confront their implications.

  ___________________________________________________________________

 An EDA Investigative Report

'Exhaustive analysis of exit polls conducted in Los Angeles County has led to the conclusion that the vote count for Proposition 8 (the ban on same-sex marriage) appears to have been corrupted.

There were not enough Republican voters to account for the disparity between the exit poll and official results
even if every Republican non-responder voted for Proposition 8.

The Edison-Mitofsky exit poll showed a similar disparity statewide,
indicating that altered vote counts may not be limited to Los Angeles County.'

For the Introduction and Executive Summary:
http://electiondefensealliance.org/files/intro&exsum.pdf [7]

CITIZEN EXIT POLLS IN LOS ANGELES COUNTY: AN IN-DEPTH ANALYSIS 

Richard Hayes Phillips, Ph.D.

 

Download the PDF [8]

Appendices added

Related report: Introduction and Executive Summary [9]
_________________________________________

Abstract

 Exhaustive analysis of exit polls conducted in Los Angeles County has led to the inescapable conclusion that the vote count for Proposition 8 (the ban on same-sex marriage) was corrupted. The data were drawn from questionnaires filled out by 6326 voters at ten polling places scattered across Los Angeles County, and were properly adjusted to match the gender, age, race, and party affiliation of the electorate.

For Proposition 4 (which would have required parental notification and a waiting period for minors seeking abortions), the official results differ from the adjusted exit poll data by only 0.64%. But for Proposition 8, the disparity between the official results and the adjusted exit poll data is 5.74%, enough to affect the margin by 11.48%. Because Los Angeles County comprised 24.23% of the statewide electorate, an error of that magnitude would have affected the statewide margin by 2.78%, accounting for most of the official 4.48% statewide margin of victory. There were not enough Republican voters to account for the disparity between the exit poll and the official results even if every Republican non-responder voted for Proposition 8. The Edison-Mitofsky exit poll showed a similar disparity statewide, indicating that altered vote counts may not be limited to Los Angeles County. 

Click Here to download the Introduction as a pdf [10]


The following study of suspect Proposition 8 election results in Los Angeles County, CA, is drawn from data gathered in EDA's Election Verification Exit Poll [6] (EVEP) analysis of the 2008 Presidential election, which reports similarly questionable election results in several states.

Although this exit poll analysis cannot provide conclusive proof of election fraud (because such proof would require access to memory cards and computer code accorded proprietary exemption from public examination) it does provide the strongest indirect proof available that election results have almost certainly been altered by manipulation of the computerized voting systems.

Deviations between exit polls and official results far outside margins of error, cannot be explained away by demographics or polling factors. The facts established in these reports cannot responsibly be dismissed or evaded.

Election Defense Alliance calls on legislators, secretaries of state, attorneys general, the voting public, and especially candidates in upcoming elections, to read these reports and seriously confront their implications.

  ___________________________________________________________________

 An EDA Investigative Report

'Exhaustive analysis of exit polls conducted in Los Angeles County has led to the conclusion that the vote count for Proposition 8 (the ban on same-sex marriage) appears to have been corrupted.

There were not enough Republican voters to account for the disparity between the exit poll and official results
even if every Republican non-responder voted for Proposition 8.

The Edison-Mitofsky exit poll showed a similar disparity statewide,
indicating that altered vote counts may not be limited to Los Angeles County.'

CITIZEN EXIT POLLS IN LOS ANGELES COUNTY: AN IN-DEPTH ANALYSIS 

Richard Hayes Phillips, Ph.D.

 

Download the PDF [8]

Appendices added

Related report: Introduction and Executive Summary [9]
_________________________________________

Abstract

 Exhaustive analysis of exit polls conducted in Los Angeles County has led to the inescapable conclusion that the vote count for Proposition 8 (the ban on same-sex marriage) was corrupted. The data were drawn from questionnaires filled out by 6326 voters at ten polling places scattered across Los Angeles County, and were properly adjusted to match the gender, age, race, and party affiliation of the electorate.

For Proposition 4 (which would have required parental notification and a waiting period for minors seeking abortions), the official results differ from the adjusted exit poll data by only 0.64%. But for Proposition 8, the disparity between the official results and the adjusted exit poll data is 5.74%, enough to affect the margin by 11.48%. Because Los Angeles County comprised 24.23% of the statewide electorate, an error of that magnitude would have affected the statewide margin by 2.78%, accounting for most of the official 4.48% statewide margin of victory. There were not enough Republican voters to account for the disparity between the exit poll and the official results even if every Republican non-responder voted for Proposition 8. The Edison-Mitofsky exit poll showed a similar disparity statewide, indicating that altered vote counts may not be limited to Los Angeles County. 

Click Here to download the Introduction as a pdf [11]


The following study of suspect Proposition 8 election results in Los Angeles County, CA, is drawn from data gathered in EDA's Election Verification Exit Poll [6] (EVEP) analysis of the 2008 Presidential election, which reports similarly questionable election results in several states.

Although this exit poll analysis cannot provide conclusive proof of election fraud (because such proof would require access to memory cards and computer code accorded proprietary exemption from public examination) it does provide the strongest indirect proof available that election results have almost certainly been altered by manipulation of the computerized voting systems.

Deviations between exit polls and official results far outside margins of error, cannot be explained away by demographics or polling factors. The facts established in these reports cannot responsibly be dismissed or evaded.

Election Defense Alliance calls on legislators, secretaries of state, attorneys general, the voting public, and especially candidates in upcoming elections, to read these reports and seriously confront their implications.

  ___________________________________________________________________

 An EDA Investigative Report

'Exhaustive analysis of exit polls conducted in Los Angeles County has led to the conclusion that the vote count for Proposition 8 (the ban on same-sex marriage) appears to have been corrupted.

There were not enough Republican voters to account for the disparity between the exit poll and official results
even if every Republican non-responder voted for Proposition 8.

The Edison-Mitofsky exit poll showed a similar disparity statewide,
indicating that altered vote counts may not be limited to Los Angeles County.'

CITIZEN EXIT POLLS IN LOS ANGELES COUNTY: AN IN-DEPTH ANALYSIS 

Richard Hayes Phillips, Ph.D.

 

Download the PDF [8]

Appendices added

Related report: Introduction and Executive Summary [9]
_________________________________________

Abstract

 Exhaustive analysis of exit polls conducted in Los Angeles County has led to the inescapable conclusion that the vote count for Proposition 8 (the ban on same-sex marriage) was corrupted. The data were drawn from questionnaires filled out by 6326 voters at ten polling places scattered across Los Angeles County, and were properly adjusted to match the gender, age, race, and party affiliation of the electorate.

For Proposition 4 (which would have required parental notification and a waiting period for minors seeking abortions), the official results differ from the adjusted exit poll data by only 0.64%. But for Proposition 8, the disparity between the official results and the adjusted exit poll data is 5.74%, enough to affect the margin by 11.48%. Because Los Angeles County comprised 24.23% of the statewide electorate, an error of that magnitude would have affected the statewide margin by 2.78%, accounting for most of the official 4.48% statewide margin of victory. There were not enough Republican voters to account for the disparity between the exit poll and the official results even if every Republican non-responder voted for Proposition 8. The Edison-Mitofsky exit poll showed a similar disparity statewide, indicating that altered vote counts may not be limited to Los Angeles County. 

The Ballot Propositions

 

Citizen exit polls were conducted by trained volunteers on behalf of Election Defense Alliance (EDA) on November 4, 2008 at 37 sites in eight states, including ten polling places in Los Angeles County. The purpose was not only to collect demographic data (gender, age, race, and party affiliation) for election analysis, but also to reach a large enough sample of voters at the polls to verify (or question) the official results. In Los Angeles, four ballot propositions (as well as the presidential election and some local contests) were included on the questionnaire handed to voters. Two of these propositions are of limited use as analytical tools for election verification:

 

Proposition 1A was a bond issue for a high-speed passenger train. It passed with 52.7% of the vote, according to the official results. Support for Proposition 1A was geographically based. Among the 25 most populated counties in California (those with more than 100,000 ballots cast), Proposition 1A received more than 60% of the vote in seven counties on or near San Francisco Bay (San Francisco 78.4%, Marin 65.2%, Sonoma 63.9%, Alameda 62.8%, San Mateo 61.1%, Santa Clara 60.4%, and Santa Cruz 60.2%). Nowhere else was support for Proposition 1A that high, although it did receive 59.1% of the vote in nearby Monterey County. Its next best showing among the 25 most populated counties was 55.6% in Los Angeles County. These would be the counties with the most to gain from the passage of Proposition 1A.

 

Proposition 11 was a Constitutional amendment involving redistricting of legislative districts. It was defeated in 2005, but passed in 2008 with 50.9% of the vote, according to the official results. It was defeated soundly in San Francisco County, receiving only 36.9% of the vote. But in 21 of the 25 most populated counties, the vote was very close; support for Proposition 11 ranged only from 46.8% to 55.8% (the other exceptions being 59.1% in Placer County, 57.5% in Marin County, and 44.8% in Alameda County), with little or no apparent correlation with party affiliation. In no county other than San Francisco did Proposition 11 receive more than 60% or less than 40% of the vote.

 

The other two ballot propositions are well suited for direct comparison. Proposition 4 was a Constitutional Amendment requiring a waiting period and parental notification before termination of a minor’s pregnancy. It had been defeated twice before, in 2005 and in 2006, and was defeated for the third time in 2008, winning only 48.0% of the vote statewide, according to the official results. Proposition 8 was a Constitutional Amendment eliminating the right of same-sex couples to marry. According to the Edison-Mitofsky statewide exit poll, Proposition 8 was headed for defeat by a margin of 52% to 48%. Instead it passed statewide with 52.2% of the vote, according to the official results. Because of the unexpected passage of Proposition 8, election integrity advocates have wondered aloud if the official results were legitimate.

 

TABLE 1: OFFICIAL RESULTS IN THE TWENTY-FIVE MOST POPULATED COUNTIES

 

	Yes 1A	No 1A	Yes 4	No 4	Yes 8	No 8	Yes 11	No 11
Alameda	62.8%	37.2%	35.7%	64.3%	38.0%	62.0%	44.8%	55.2%
Contra Costa	54.4%	45.6%	41.7%	58.3%	44.6%	55.4%	51.8%	48.2%
Fresno	55.5%	44.5%	60.0%	40.0%	68.7%	31.3%	52.4%	47.6%
Kern	50.8%	49.2%	65.5%	34.5%	75.3%	24.7%	47.6%	52.4%
Los Angeles	55.6%	44.4%	46.2%	53.8%	50.0%	50.0%	47.6%	52.4%
Marin	65.2%	34.8%	25.6%	74.4%	24.9%	75.1%	57.5%	42.5%
Monterey	59.1%	40.9%	43.8%	56.2%	48.4%	51.6%	46.8%	53.2%
Orange	43.6%	56.4%	54.4%	45.6%	57.7%	42.3%	55.0%	45.0%
Placer	39.6%	60.4%	50.3%	49.7%	59.8%	40.2%	59.1%	40.9%
Riverside	49.0%	51.0%	60.1%	39.9%	64.7%	35.3%	53.3%	46.7%
Sacramento	47.7%	52.3%	46.2%	53.8%	53.9%	46.1%	55.8%	44.2%
San Bernardino	46.2%	53.8%	59.2%	40.8%	66.8%	33.2%	51.7%	48.3%
San Diego	48.5%	51.5%	52.8%	47.2%	53.8%	46.2%	54.4%	45.6%
San Francisco	78.4%	21.6%	24.6%	75.4%	24.8%	75.2%	36.9%	63.1%
San Joaquin	53.3%	46.7%	55.6%	44.4%	65.5%	34.5%	50.3%	49.7%
San Luis Obispo	46.3%	53.7%	44.6%	55.4%	51.1%	48.9%	50.6%	49.4%
San Mateo	61.1%	38.9%	37.4%	62.6%	38.2%	61.8%	51.0%	49.0%
Santa Barbara	54.2%	45.8%	42.5%	57.5%	46.4%	53.6%	53.1%	46.9%
Santa Clara	60.4%	39.6%	43.1%	56.9%	44.2%	55.8%	53.2%	46.8%
Santa Cruz	60.2%	39.8%	27.8%	72.2%	28.7%	71.3%	50.3%	49.7%
Solano	53.9%	46.1%	50.2%	49.8%	55.9%	44.1%	52.3%	47.7%
Sonoma	63.9%	36.1%	33.2%	66.8%	33.5%	66.5%	53.0%	47.0%
Stanislaus	49.0%	51.0%	59.8%	40.2%	67.9%	32.1%	52.0%	48.0%
Tulare	44.9%	55.1%	65.8%	34.2%	75.1%	24.9%	49.8%	50.2%
Ventura	48.5%	51.5%	49.5%	50.5%	52.9%	47.1%	51.7%	48.3%
State Totals	52.7%	47.3%	48.0%	52.0%	52.2%	47.8%	50.9%	49.1%

 

NOTE: These were the twenty-five counties with 100,000 or more ballots cast. The thirteen counties with more than 250,000 ballots cast are shown in blue. The percentages do not include “undervotes” – those who did not vote on the proposition. Only the “yes” and “no” votes are included in the calculations.

 

Proposition 4 is clearly the most reasonable benchmark with which to compare Proposition 8, because both were hot-button social issues with overlapping support among the electorate. Exit poll data bear this out. In the ten polling places combined, 66.63% voted in favor of both propositions, or against both propositions; only 23.08% voted for one proposition and against the other (6.67% voted on one or the other, but not both; and 3.62% voted on neither proposition). (See Table 2)

 

TABLE 2: EXIT POLL DATA FOR ALL TEN POLLING PLACES COMBINED,

SHOWING OVERLAPPING ELECTORATE ON PROPOSITIONS 4 AND 8

 

 

Statewide, as stated above, Proposition 8 received 52.2% of the vote and Proposition 4 received only 48.0% of the vote, a differential of 4.2%, according to the official results. This pattern, with Proposition 8 running ahead of Proposition 4, appeared in 24 of the 25 most populated counties, the differential ranging from 0.2% in San Francisco County to 9.9% in San Joaquin County. The only exception was Marin County, where Proposition 4, with 25.6% of the vote, ran 0.7% ahead of Proposition 8, with 24.9% of the vote.

 

If the official results are true and correct, one would expect those results to be closely matched by exit polls. In the ten polling places in Los Angeles County where citizen exit polls were conducted, Proposition 8 ran 3.68% ahead of Proposition 4, according to the official results. But according to the exit polls, the reverse was true – Proposition 4 ran 2.01% ahead of Proposition 8 (see Table 3), which amounts to a disparity of 5.69%. Looked at another way, both propositions ran better in the official results than in the exit polls -- Proposition 4 by 2.06%, and Proposition 8 by 7.75%, which amounts to the same differential of 5.69%. Again, these percentages do not include “undervotes.”

 

TABLE 3: COMPARISON OF EXIT POLL DATA AND OFFICIAL RESULTS FOR

PROPOSITIONS 4 AND 8, ALL TEN POLLING PLACES COMBINED

 

 

NOTE: In California, absentee voters can hand deliver their ballots at the polls. In this report, absentee voters are not included in the exit poll data or in the official results, so the data are directly comparable.

 

This disparity appears in all ten polling places, and always in the same direction. (See Table 4) Proposition 8 always fares better relative to Proposition 4 in the official results than in the exit polls; the disparities range from 2.2% at Topanga to 10.5% at Long Beach. Considered alone, Proposition 8 fares better in the official results than in the exit polls in 9 of 10 polling places, by 7.75% overall; the differentials reach 11.3% at Eagle Rock, and 17.7% at Glendale, and the disparities between the margins of victory or defeat (that is, the point spread) would be twice as much. For example, at Glendale, Proposition 8 lost by 29.0% in the exit poll, but passed by 6.4% in the official results; the disparity in the point spread is 35.4%.

 

When comparing exit poll data with official results, it is common practice to compare the difference in the point spread. Think of it this way. If we are dividing a bushel of apples, and I have three more than half, and you have three less than half, I have six more than you do. This method works even if there are more than two choices on the ballot, in which case the difference between the exit poll data and the official results may be greater for one candidate than another.

 

Proposition 4, considered alone, fares better in the official results than in the exit polls by only 2.06%, which suggests that the sample of voters responding to the exit poll was quite representative of the electorate. Moreover, Proposition 4 fares better in the official results than in the exit polls in only 5 of 10 polling places, which is exactly what one would expect from a reliable exit poll. The disparities should balance out – some in one direction, and some in the other. The very fact that this was the case in regard to one proposition but not the other suggests the official results for Proposition 8 are wrong.

 

There are four possible reasons for a large disparity between exit polls and official results: (1) a basic flaw in the exit poll methodology; (2) many voters lying on the questionnaire; (3) a non-representative sample of voters responding; or (4) the official results being erroneous or fraudulent. Let us consider the first three possibilities one at a time.

 

It is very hard to argue that some flaw in the exit poll methodology would be responsible for a glaring disparity concerning Proposition 8 and not with Proposition 4. While I shall leave it to those who conducted the exit polls to explain their methodology, I must note that the exit polls in Los Angeles County were organized by Judy Alter, who had already done this five times previously. The number of voters responding – 6326 in Los Angeles County alone – was larger than the statewide sample relied upon by Edison-Mitofsky, and amounted to 54.28% of the total ballots cast in these ten polling places.

 

It is difficult to believe that many voters lied on the questionnaires, as this explanation would require that voters in all ten polling places lied about how they voted on Proposition 8, but told the truth about Proposition 4. Note that in 8 of 10 polling places, the disparity between the exit poll percentages and the official results is greater for Proposition 8 than for Proposition 4.

 

And it seems unlikely that in ten different polling places, scattered all over Los Angeles County, in a variety of neighborhoods with many different ethnic groups, the voters responding to the exit poll comprised a more representative sample for Proposition 4 than for Proposition 8. These were the same voters, in the same polling places, on the same day. Both Proposition 4 and Proposition 8 were hot-button social issues. More often than not, those who supported one proposition support the other, and those who opposed one proposition would oppose the other.

 And yet, in all 10 polling places, Proposition 8 fares better relative to Proposition 4 in the official results than in the exit polls, by 5.69% overall. And in 9 of 10 polling places, Proposition 8 fares better in the official results than in the exit polls, by 7.75% overall. Such glaring disparities are a red flag. A serious investigation is warranted.

 

TABLE 4: COMPARISON OF EXIT POLLS AND OFFICIAL RESULTS FOR

PROPOSITIONS 4 AND 8, ALL TEN POLLING PLACES INDIVIDUALLY

 

 

 

 

 

 

 

 

 

 

TABLE 4: COMPARISON OF EXIT POLLS AND OFFICIAL RESULTS FOR

PROPOSITIONS 4 AND 8, ALL TEN POLLING PLACES INDIVIDUALLY (continued)

 

 

 

 

 

 

 

 

 

 

According to the official results, Proposition 8 was defeated by 599,602 votes statewide. Los Angeles County comprised 24.23% of the electorate for Proposition 8, with 3,246,959 persons voting on it (see Table 5). If 5.69 to 7.75% of the votes on Proposition 8 in Los Angeles County were shifted from one column to the other, from “no” to “yes,” as suggested by the exit polls in ten different polling places, this would affect the margin (the point spread) by twice those percentages, by some 370,000 to 500,000 votes. And if the vote count was altered elsewhere in the state, the will of the voters may have been reversed.

 

TABLE 5: OFFICIAL RESULTS IN LOS ANGELES COUNTY AND STATEWIDE

 

 

A working hypothesis must be that the official results are true and correct, and that the disparities between the exit polls and the official results are due to non-representative samples of voters responding to the exit poll. In order to test this hypothesis, we must compare the demographics (gender, age, race), and the party affiliations, of the voters who participated in the exit polls (the responders) to the voters not responding (the refusals). This underscores the importance of collecting “refusal data,” as was done in this poll. The exit pollsters noted the gender, race, and estimated age of each voter who was approached but declined to respond. These data can be compared to the responses on the questionnaires filled out by the participating voters. Based upon this information, the raw data for the exit poll can be adjusted accordingly, to better reflect the demographic makeup of the electorate.

Gender Bias

When the demographic data from the exit polls are examined, a gender bias is immediately apparent. Among the 5451 responders who identified their gender, 3220 (59.07%) were women, and 2231 (40.93%) were men. This imbalance is found at all ten polling places (see Table 6).

 

TABLE 6: EXIT POLL RESPONDERS, BY GENDER

 

 

This gender imbalance would obviously have had some effect on the outcome of the exit polls, because the data show a “gender gap” in the vote count. In the ten polling places combined, Proposition 4 gained the support of 41.88% of the men and 39.62% of the women, a differential of 2.26%. Proposition 8 gained the support of 40.48% of the men and 35.84% of the women, a differential of 4.64% (see Table 7).

 

TABLE 7: EXIT POLL VOTE TOTALS FOR BALLOT PROPOSITIONS, BY GENDER

 

 

This “gender gap” appears in 9 of 10 polling places, the lone exception being Long Beach, where women were more likely than men to support both ballot propositions. However, there were some polling places where the differential was insignificant – for example, Proposition 4 at Locke High School and in Lynwood, and Proposition 8 in Santa Monica and Topanga. A detailed breakdown of the vote on the ballot propositions, by gender, age, race and party affiliation, for each polling place, is presented in the Appendix.

Ideally, one would find out how many men and women voted at the polls and adjust the exit poll data accordingly. However, in Los Angeles County, poll books are not available for public inspection, and the Registrar of Voters wants $250 for two countywide files from which it would take weeks to extract the data manually and match up the voter histories with the gender. Under the circumstances, the best approach is to make a conservative assumption, and a 50-50 split meets this requirement because women generally do outnumber men at the polls. Moreover, support for both Proposition 4 and Proposition 8 was greater among men than among women, so adjusting to a 50-50 split will overstate support for both propositions.

To adjust for the “gender gap” is a straightforward mathematical process. Let us begin with a simple hypothetical example not dissimilar to the exit poll data. Suppose there are 25 responders in the exit poll. Only 20 identify their gender, and of these, 12 (60.0%) are women, and 8 (40.0%) are men. The vote count is 17 (68%) to 8 (32%). The breakdown is 9 to 3 among women, 5 to 3 among men, and 3 to 2 among those who did not identify their gender.

To adjust the vote count to a 50%-50% balance by gender, we multiply the numbers for women by 50/60 (or divide by 1.2), and we multiply the numbers for men by 50/40 (or divide by 0.8). I prefer to divide, because the adjustment factors are easy to determine – 1.2 for 60%, 0.8 for 40%, and so on. The adjusted count becomes 7.5 to 2.5 among women, and 6.25 to 3.75 among men, and remains 3 to 2 among those who did not identify their gender. The adjusted total count is now 16.75 (67%) to 8.25 (33%) – not much of a difference (see Table 8).

 

TABLE 8: HYPOTHETICAL EXIT POLL ADJUSTMENTS

 

 

Let us suppose, in another hypothetical example, that all women voted one way and all men voted the other way. If the breakdown is 12 to 0 among women, 0 to 8 among men, and 3 to 2 among those who did not identify their gender, for a total vote count of 15 (60%) to 10 (40%), the adjusted count becomes 10 to 0 among women, 0 to 10 among men, and 3 to 2 among those who did not identify their gender, for an adjusted total count of 13 (52%) to 12 (48%) – only an 8% difference (see Table 12). This begins to explain what pollsters mean when they talk about the “margin of error.” There are very real limits as to how inaccurate an exit poll can be (see Table 9).

 

TABLE 9: HYPOTHETICAL EXIT POLL ADJUSTMENTS

 

 

When this methodology is applied to the exit poll results from Los Angeles County, in order to adjust for the obvious gender imbalance among the responders, we find that it makes very little difference at all (see Table 10). Proposition 8 still fares 7.29% better (instead of 7.75%) in the official results than in the exit polls. Proposition 4 still fares 1.84% better (instead of 2.06%) in the official results than in the exit polls. And the disparities, relative to each other, are still 5.45% (instead of 5.79%).

 

TABLE 10: COMPARISON OF EXIT POLL DATA, ADJUSTED FOR GENDER,

AND OFFICIAL RESULTS FOR PROPOSITIONS 4 AND 8

 

 

These adjusted numbers are derived from the sum totals of the adjustments calculated for all ten polling places individually. The calculations are set forth in full detail in the Appendix, and the adjusted numbers for each polling place for Propositions 4 and 8 are presented here (see Table 11).

Even with the percentages adjusted to account for gender imbalance, Proposition 8 still runs better in the official results than in the exit polls in 9 of 10 polling places, by as much as 17.3% in Glendale, 11.1% in Eagle Rock, 7.1% in Topanga, 6.3% in Santa Monica, and 6.3% at Lockhurst. Proposition 8 still runs better relative to Proposition 4 in the official results than in the exit polls in all ten polling places, by as much as 10.2% in Long Beach, 8.9% in Glendale, 7.7% in Lynwood, 6.7% in Berendo, and 6.5% in Eagle Rock (see Table 11).

Thus it is shown that the glaring disparities between the exit polls and the official results for Proposition 8 are simply not attributable to “gender bias” in the exit polls.

 

TABLE 11: COMPARISON OF ADJUSTED EXIT POLL DATA, ADJUSTED FOR GENDER, AND OFFICIAL RESULTS FOR PROPOSITIONS 4 AND 8

 

 

 

 

 

 

 

 

 

 

TABLE 11: COMPARISON OF ADJUSTED EXIT POLL DATA, ADJUSTED FOR GENDER, AND OFFICIAL RESULTS FOR PROPOSITIONS 4 AND 8 (continued)

 

 

 

 

 

 

 

 

 

 

Age and Race Bias

 

To examine the possibility that the disparities between the exit polls and the official results are due to non-representative samples with respect to age or race, it is useful to examine the voting patterns of these demographic groups as set forth in detail in the Appendix. Presented here are summary tables for all 10 polling places combined. Bear in mind that the characteristics of the electorate will vary among the different polling places.

 

With respect to Proposition 4, (which would have required parental notification and a waiting period before termination of a minor’s pregnancy), the age of the voter made almost no difference. In the exit poll, Proposition 4 was supported by 40.1% of voters under 30, 40.1% of voters between 30 and 59, and 41.8% of voters aged 60 or older. But support for Proposition 8 (that is, opposition to same-sex marriage), was clearly correlated with age. Proposition 8 was supported by 31.7% of voters under 30, 38.8% of voters between 30 and 59, and 48.5% of voters aged 60 or older (see Table 12). Thus, an oversampling of voters under 30, or an undersampling of voters over 60, or both, would cause the exit poll to understate the support for Proposition 8.

 

TABLE 12: EXIT POLL RESULTS FOR PROPOSITIONS 4 AND 8,

BY AGE GROUP, ALL TEN POLLING PLACES COMBINED

 

 

With respect to Propositions 4 and 8, the race of the voter did make a difference, most obviously with white voters, among whom the vote was 81.2% against Proposition 4 and 82.8% against Proposition 8 (see Table 13). Clearly, this is not a representative sample of white voters in Los Angeles County as a whole, although it could be representative of these 10 polling places, which include some of the most liberal communities in the county (e.g. Topanga and Santa Monica). It is an unfortunate shortcoming of the exit poll that no predominantly Republican areas were covered. But this does not invalidate the results. It only means that in these 10 polling places, an undersampling of white voters would overstate the support for Propositions 4 and 8, and that an oversampling of white voters would overstate the opposition.

Among the other races identified in both the exit poll questionnaires and the refusal data, support for Proposition 4 was strongest among Latinos, from whom it received 56.9% of the vote, compared to 51.3% among blacks and 50.1% among Asians (see Table 13). Thus, an undersampling of Latino voters could have understated the support for Proposition 4, but probably not by enough to make much of a difference.

Support for Proposition 8 was strongest among black voters, from whom it received 60.5% of the vote, compared to 48.8% among Latinos and 45.7% among Asians (see Table 13). Blacks
were the only racial group among whom the support for Proposition 8 was stronger than for Proposition 4.

As stated above, 66.53% of the electorate voted the same way on both propositions. Those who voted for Proposition 4 and against Proposition 8 amounted to 17.7% of Latinos, 13.5% of blacks, and 13.4% of Asians. But 21.4% (208 of 970) black voters made the opposite choices, supporting Proposition 8 while opposing Proposition 4, whereas only 10.6% of Latinos (193 of 1828) and 9.6% of Asians (55 of 571) did so (these data are set forth in detail in the Appendices).

Thus, an undersampling of black voters would understate support for Proposition 8, whereas an undersampling of Latinos relative to Asians, or vice versa, would have made little difference.

 

TABLE 13: EXIT POLL RESULTS FOR PROPOSITIONS 4 AND 8,

BY RACE, ALL TEN POLLING PLACES COMBINED

 

 

Demographic information on the voters not responding to the exit poll, known as “refusal data,” was collected at 6 of 10 polling places (Taft, Santa Monica, Topanga, Lockhurst, Eagle Rock, and Lynwood). These include 4 of the 5 polling places with the greatest disparities between the exit polls and the official results for Proposition 8.

 
TAFT 

At Taft High School there was a 6.0% disparity between the margins of defeat (the point spreads) for Proposition 8 in the exit poll and in the official results. Proposition 8 was defeated by 194 (66.0%) to 100 (34.0%) in the exit poll, and by 351 (63.0%) to 206 (37.0%) in the official results (see Table 4). Among non-responders, Proposition 8 was defeated much more narrowly, by 157 (59.7%) to 106 (40.3%), or else the official results are not true and correct.

 

The refusal data for Taft High School show that the exit poll sample was quite representative with respect to age group (see Table 14). However, white voters were severely underrepresented (by 19.2%), while Latinos, blacks, and Asians were all overrepresented (by 2.2%, 4.9%, and 12.0%, respectively). This was the most extreme imbalance reflected in the refusal data for any of the six polling places. Altogether, 212 white voters refused to participate in the exit poll; these represented nearly two-thirds of the white voters (22 of 321, or 66.0%), and nearly all of the refusals (212 of 226, or 93.8%). As it happens, 120 of the 212 (57.5%) were men, whereas 57 of 95 (60.0%) white responders were women (among those who revealed both race and gender).

 

TABLE 14: EXIT POLL RESPONDERS AND REFUSAL DATA, TAFT

 

 

The “gender bias” at all 10 polling places has already been addressed. At Taft, adjustment of the sample to a 50%-50% gender balance showed a 0.6% rise in support for Proposition 8 and a 0.2% rise for Proposition 4.

However, adjustment of the exit poll sample according to race has a greater effect, and in the opposite direction. Of the 47 exit poll responders who identified themselves as Asian, 21 (48%) voted for Proposition 8, and 23 (52%) voted against it (three made no choice). Of the 19 exit poll responders who identified themselves as black, 8 (42%) voted for Proposition 8, and 11 (58%) voted against it. Of the 10 exit poll responders who identified themselves as Latino, 4 (44%) voted for Proposition 8, and 5 (56%) voted against it (one made no choice).

These percentages being nearly equal, and the numbers being small, an oversampling or undersampling of Asians, blacks or Latinos, relative to each other, would have little effect upon the results.

But an undersampling of white voters would have understated the opposition to Proposition 8 because, of the 109 exit poll responders who identified themselves as white, only 35 (33%) voted for Proposition 8, and 71 (67%) voted against it (three made no choice).

If we adjust the exit poll results by race, using the same methodology shown above for adjustment by gender, the end result is 32.6% for Proposition 8, and 67.4% against it. This represents a 1.4% drop in support for Proposition 8, and increases the disparity between the exit poll and the official results from 6.0% to 8.8% (the mathematics are shown in detail in the Appendix).

SANTA MONICA

 

At Santa Monica there was a 12.8% disparity between the margins of defeat (the point spreads) for Proposition 8 in the exit poll and in the official results. Proposition 8 was defeated by 452 (88.6%) to 58 (11.4%) in the exit poll, and by 610 (82.2%) to 132 (17.8%) in the official results (see Table 4). Among non-responders, Proposition 8 was defeated much more narrowly than among responders, by 158 (68.1%) to 74 (31.9%), or else the official results are not true and correct.

According to the official results there were 762 ballots cast at the polls. There were 534 responders to the exit poll and, according to the refusal data, there were 237 refusals, for a total of 771 – a discrepancy of nine voters (there may have been only 228 refusals). But though the refusal data may not be precisely correct, all but 23 (9.7%) of those refusing were described as white (11 were Latino, 11 were Asian, and one was black). Nearly half of white voters, 214 of 433 (49.4%) refused to participate; 107 were men and 107 were women. Based upon the refusal data, whites were underrepresented by 8.6% in the exit poll. Latinos, blacks and Asians were overrepresented by 1.0%, 1.3%, and 6.3%, respectively. (See Table 15)

 

TABLE 15: EXIT POLL RESPONDERS AND REFUSAL DATA, SANTA MONICA

 

 

Regarding Proposition 8, the demographic data do not explain how 31.9% of all the non-responders at Santa Monica could have voted for the ban on same-sex marriage. In the exit poll, only 11.4% voted in favor. There was no demographic group within which support for Proposition 8 even approached 31.9%, the strongest support, 17 of 77 (22.1%), coming from voters aged 60 or older. Among ethnic groups, the strongest support, 10 of 60 (16.7%), came from Asians.

 

TOPANGA

At Topanga there was a 14.4% disparity between the margins of defeat (the point spreads) for Proposition 8 in the exit poll and in the official results. Proposition 8 was defeated by 528 (93.8%) to 35 (6.2%) in the exit poll, and by 917 (86.6%) to 142 (13.4%) in the official results (see Table 4). Among non-responders, Proposition 8 was defeated much more narrowly than among responders, by 389 (78.4%) to 107 (21.5%), or else the official results are not true and correct.

The refusal data for Topanga seem entirely trustworthy. Officially there were 1078 ballots cast at the polls (not including 99 absentee ballots). Of these, 822 voters were approached by the exit pollsters; there were 584 responders, and 238 refusals. According to the refusal data, voters under 30 were underrepresented by 3.5% in the exit poll, and voters between 30 and 59 were overrepresented by 4.1%. The refusal data show that the sample was quite representative with respect to race (see Table 16).

 

TABLE 16: EXIT POLL RESPONDERS AND REFUSAL DATA, TOPANGA

 

 

Regarding Proposition 8, the demographic data do not explain how 21.5% of all the non-responders at Topanga could have voted for the ban on same-sex marriage. In the exit poll, only 6.2% voted in favor. The strongest support, 17 of 107 (15.9%), was among voters aged 60 or older. Among ethnic groups, the strongest support, 3 of 16 (18.8%), came from Asians.

 

LOCKHURST

 

At Lockhurst Elementary School there was a 13.4% disparity between the margins of defeat (the point spreads) for Proposition 8 in the exit poll and in the official results. Proposition 8 was defeated by 372 (64.7%) to 203 (35.3%) in the exit poll, and by 672 (58.0%) to 487 (42.0%) in the official results (see Table 4). Among non-responders, Proposition 8 was defeated much more narrowly than among responders, by 300 (51.4%) to 284 (48.6%), or else the official results are not true and correct.

 

The refusal data for Lockhurst seem entirely trustworthy. Officially there were 1186 ballots cast at the polls (not including 180 absentee ballots). Of these, 831 voters were approached by the exit pollsters; there were 596 responders, and 235 refusals. According to the refusal data, voters under 30, and voters over 60, were underrepresented in the exit poll (by 2.0% and 3.3%, respectively), and voters between 30 and 59 were overrepresented (by 5.3%). Latino voters were underrepresented by 2.4% and black voters were overrepresented by 2.0% (see Table 17).

 

TABLE 17: EXIT POLL RESPONDERS AND REFUSAL DATA, LOCKHURST

 

 

Support for Proposition 8 was strongest among black voters, of whom 12 of 26 (46.2%) voted for the ban on same-sex marriage. Support was 46 of 157 (29.3%) among whites, 24 of 66 (36.3%) among Latinos, and 27 of 79 (34.2%) among Asians. Thus, an oversampling of black voters would overstate the support for Proposition 8. When the numbers are adjusted with respect to race, the result is 35.2% Yes and 64.8% No, a change of only 0.1%. When we adjust the numbers for Proposition 4, the result is 34.9% Yes and 65.1% No, which is no change at all (the mathematics are shown in detail in the Appendix).
 

Proposition 8 was supported by 27 of 100 (27.0%) of voters under 30, 111 of 344 (32.3%) of voters between 30 and 59, and 43 of 111 (38.7%) of voters aged 60 or older. Thus, the undersampling of voters with respect to age group involved both extremes (27.0% and 38.7%). When we adjust the numbers accordingly, the result is 35.4% Yes and 64.6% No, a change of only 0.1%. Similarly, when we adjust the numbers for Proposition 4, the result is 34.7% Yes and 65.3% No, a change of only 0.2% (the mathematics are shown in detail in the Appendix).

EAGLE ROCK

At Eagle Rock there was a 22.6% disparity between the margins of defeat (the point spreads) for Proposition 8 in the exit poll and in the official results. Proposition 8 was defeated by 521 (70.9%) to 214 (29.1%) in the exit poll, and by 709 (59.6%) to 480 (40.4%) in the official results (see Table 4). Among all non-responders, Proposition 8 must have passed overwhelmingly, by 266 (58.6%) to 188 (41.4%), or else the official results are not true and correct. How this could have happened cannot be explained by the demographic data. In the exit poll, the strongest support for Proposition 8 was 67 of 128 (52.3%) among Asians, 51 of 127 (40.2%) among voters aged 60 and older, and 53 of 161 (32.9%) among Hispanics.

The refusal data for Eagle Rock seem entirely trustworthy. Officially there were 1209 ballots cast at the polls (not including 111 absentee ballots). Of these, 1090 voters were approached by the exit pollsters; there were 757 responders, and 333 refusals. According to the refusal data, voters between 30 and 59 were overrepresented by 2.5% in the exit poll, and voters aged 60 and older were underrepresented by 3.1%. Latino voters were overrepresented by 3.1% and white voters were underrepresented by 4.8% (see Table 18).

 

TABLE 18: EXIT POLL RESPONDERS AND REFUSAL DATA, EAGLE ROCK

 

 

An oversampling of Latinos (and a slight oversampling of Asians) would have overstated the support for Proposition 8. In the exit poll, support for Proposition 8 was 53 of 161 (32.9%) among Latinos, compared to 24 of 184 (13.0%) among whites, 3 of 15 (20.0%) among blacks, and 67 of 128 (52.3%) among Asians.

When we adjust the numbers by race, the result is 28.3% Yes and 71.7% No, a change of 0.8% in both columns, and the disparity between the margins of defeat for Proposition 8 in the exit poll and the official results increases to 24.2%. When we adjust the numbers for Proposition 4, the result is 31.8% Yes and 68.2% No, a change of 0.9% in both columns, and the disparity between the margins of defeat for Proposition 4 in the exit poll and the official results increases to 11.6% (the mathematics are shown in detail in the Appendix).

In the exit poll, support for Proposition 8 was 20 of 122 (16.4%) among voters under 30, 118 of 435 (27.1%) of voters between 30 and 59, and 51 of 127 (40.2%) among voters aged 60 and older. Thus, if voters aged 60 or older were underrepresented in the exit poll it would have understated the support for Proposition 9.

When we adjust the numbers according to age, the result is 29.5% Yes and 70.5% No, a change of 0.4% in both columns, and the disparity between the margins of defeat for Proposition 4 in the exit poll and the official results decreases to 21.8%. When we adjust the numbers for Proposition 4, the result is 33.0% Yes and 67.0% No, a change of 0.3% in both columns, and the disparity between the margins of defeat for Proposition 4 in the exit poll and the official results decreases to 9.2% (the mathematics are shown in detail in the Appendix).

Thus, adjusting the exit poll data according to race changes the results in one direction, and adjusting the exit poll data according to age changes the results in the other direction. But none of the changes come close to accounting for the disparities between the exit polls and the official results.

LYNWOOD

 

At Lynwood there was a 13.8% disparity between the margins of victory (the point spreads) for Proposition 8 in the exit poll and in the official results. Proposition 8 was approved by 712 (54.9%) to 584 (45.1%) in the exit poll, and by 1621 (61.8%) to 1004 (38.2%). (See Table 4) Among all non-responders, Proposition 8 must have passed overwhelmingly, by 909 (68.4%) to 420 (31.6%), or else the official results are not true and correct.

The refusal data for Lynwood seem entirely trustworthy. Officially there were 2736 ballots cast at the polls (not including 176 absentee ballots). Of these, 1932 voters were approached by the exit pollsters; there were 1396 responders, and 536 refusals (see Table 19).

 

TABLE 19: EXIT POLL RESPONDERS AND REFUSAL DATA, LYNWOOD

 

 

In the exit poll, support for Proposition 8 was 341 to 205 (62.5%) among voters between 30 and 59, and 83 to 31 (72.8%) among voters aged 60 or older. Among voters under 30, Proposition 8 was defeated, losing by 195 to 287 (40.4%). The refusal data show that voters under 30 were overrepresented by 6.5%, and that voters between 30 and 59 and voters aged 60 or older were underrepresented by 3.5% and 2.9%, respectively (see Table 26). This would have substantially understated the support for Proposition 8. If we adjust the numbers according to age, the result is 56.4% Yes and 43.6% No, a change of 1.5% in both columns, which reduces the disparity to 10.8% (the mathematics are shown in detail in the Appendix).

In the exit poll, support for Proposition 8 was 412 to 372 (52.6%) among Latinos, 190 to 125 (60.3%) among blacks, 7 to 6 (54%) among whites, and 6 to 1 (86%) among Asians. The refusal data show that blacks were proportionately represented in the exit poll (see Table 26). But the refusal data do show that 50 of 59 Asians who were approached refused to participate, so that Asians were underrepresented by 2.8% in the exit poll, while only 266 of 1070 Latinos who were approached refused to participate, so that Latinos were overrepresented by 3.2% in the exit poll.

It is possible to adjust the numbers according to race, in an attempt to rectify this imbalance, although the result, being based upon a sample of only nine Asian voters (and only 14 white voters), is inherently imprecise. The result is 55.6% Yes and 44.4% No, a change of 0.7% in both columns, which reduces the disparity to 12.4% (the mathematics are shown in detail in the Appendix).

The exit poll adjustments at Lynwood amounted to 1.0% for gender, and 1.5% for age, and 0.7% for race. We have no direct way to adjust the exit poll data according to age, race, and gender combined. But even if these adjustments are compounded, the result is a change of 3.2% in both columns, which still leaves a disparity of 7.4% between the margins of victory for Proposition 8 in the exit poll and in the official results.

Thus it is shown through exhaustive analysis that the disparities between the exit polls and the official results, which turn up consistently in the vote count for both the presidential election and for Proposition 8, cannot be attributed to gender bias, age bias, or racial bias in the samples of voters polled. There seem to be only two possible explanations remaining. Either the samples were not representative with respect to party affiliation, or the official results are not true and corre
 

Party Affiliation

For those refusing to participate in exit polls, gender can be determined, race observed, and age approximated, by the pollsters themselves. But this is not true of party affiliation. One cannot tell from appearance if the person refusing is a Republican or a Democrat. Sometimes the only way to determine the party affiliation of the refusals is to examine the poll books, tally up the numbers of voters from each party who voted at the polls, calculate their percentages, and compare them with the data from the exit poll. If party affiliation is not listed in the poll books they must be compared with the voter rolls, making the process even more time-consuming. In this way it can be determined if the sample of voters polled was representative with respect to party affiliation. Failure to do this, for whatever reason, is a serious deficiency in the art of exit polling.

Poll books are simply not available for inspection on Election Night, which is why exit poll results are not adjusted in a timely manner with regard to party affiliation. Tallies of the party affiliations of those actually voting at the polls can be collected on Election Day by having poll watchers, with “walking lists” showing party affiliation, stationed inside the polling place, checking off the names of voters who sign in at the polls, while the exit pollsters are outside the polling place interviewing voters. This could have the undesired effect of inhibiting voters from participating in the exit polls, out of concern that the questionnaires they fill out will not remain anonymous.

In some jurisdictions, the party affiliations of the actual voters are compiled and provided by the election officials themselves. Also there are private companies who maintain voter data bases from which such data can be gleaned.

We knew that the disparities observed between the exit polls and the official results at the polling places in Los Angeles County might not be due to an erroneous or fraudulent official vote count. They might just as easily have been caused by disproportionate numbers of Republican voters refusing to participate in the exit polls. This would have caused support for John McCain and for both ballot propositions to be understated in the exit poll results.

One of the questions addressed in this study, and one of the very purposes of exit polling, is to determine if the official results are true and correct. Therefore, even though McCain got only 13.48% of the vote in the exit polls, compared to 18.72% in the official results for the same 10 polling places, we could not assume that McCain voters were undersampled in the exit polls, and that the official results were true and correct. But neither could we assume that the exit polls were accurate, and that the official results were fraudulent. All conclusions derived from unwarranted assumptions are likewise unwarranted. This made it all the more important to find out if registered Republicans were undersampled in the exit poll, which is quite possible to do, because party registration is a matter of record.

In the exit polls, at the 10 polling places combined, there were 6326 responders. Of these, 4174 (65.98%) were Democrats, and only 646 (10.21%) were Republicans; 331 (5.23%) identified another party (Green, Libertarian, Peace and Freedom, or American Independent), and 1175 (18.57%) checked none or declined to state. The percentage of Republicans among the responders ranged from 2.2% at Locke High School to 25.5% at Taft High School (see Table 20).

It was necessary to compare these rather low percentages of Republican responders to the true percentages of registered Republicans who voted at the polls, and to adjust the exit poll data accordingly.

 

TABLE 20: PARTY AFFILIATION OF EXIT POLL RESPONDERS

 

 

In Los Angeles County the election results, precinct by precinct, are updated numerous times during the weeks following an election. This was especially necessary in 2008 because, according to the Office of Elections, there were so many newly registered voters that not all the information could be entered into the voter database in a timely manner. Many of these voters had to cast provisional ballots, and until the official voter database was completed, those ballots could not be counted because their validity could not be determined. For this reason, the numbers in the voter databases of the private companies may not precisely match the final official results, but they are close enough because we are dealing with party affiliation as percentages, not as raw data.

The proper comparison is with voters at the polls, not with total ballots cast, because absentee voters are not included in the exit poll data. The breakdown of voters at the polls, by party affiliation, is given below (see Table 21).

 

TABLE 21: PARTY AFFILIATION OF VOTERS AT THE POLLS

 

By comparing the tables above, it is seen that Republican voters were undersampled in all 10 polling places, by ratios ranging from 1.16 to 1 at Taft High School to 2.5 to 1 at Topanga. In the ten polling places combined, the ratio is more than 3 to 2. This matters very much, because support for Propositions 4 and 8 was clearly correlated with party affiliation, and Proposition 8 was the more polarizing of the two.

A majority of Republicans supported both measures, but support was stronger for Proposition 8 (70.1%) than for Proposition 4 (63.1%). Majorities of Democrats, third-party voters, and unaffiliated voters opposed both propositions, and in all three cases, the opposition to Proposition 8 was stronger than to Proposition 4 (see Table 22).

 

TABLE 22: VOTE ON PROPOSITIONS 4 AND 8 BY PARTY AFFILIATION

 

 

In the10 polling places combined, the response rate among Democrats was 60.2%, ranging from 44.2% at Glendale to 84.8% at Santa Monica, whereas the response rate among Republicans was only 36.7%, ranging from 21.6% at Topanga to 53.3% at Santa Monica (see Table 23). This underscores the importance of ascertaining the party affiliations of those who voted at the exit poll sites, in order to adjust the exit poll data accordingly; and it raises the challenge of how to encourage Republicans to participate in the exit polls.

 

TABLE 23: RESPONSE AND REFUSAL RATIOS BY PARTY AFFILIATION

 

 

But when the exit poll data are adjusted according to party affiliation, only some, not all, of the disparity between the exit polls and the official results is accounted for. The result is an increase of 1.42% in the support for Proposition 4 and an increase of 2.01% in the support for Proposition 8. These adjustments in the exit poll data only make the official results more suspect, because the numbers for Proposition 4 are explained and the numbers for Proposition 8 are not. The disparity between the exit polls and the official results for Proposition 4 is reduced from 2.06% to 0.64%, well within the margin of error, whereas the disparity for Proposition 8, although reduced from 7.75% to 5.74%, is still enough to affect the margin by 11.48% (see Table 24).

 

TABLE 24: COMPARISON OF EXIT POLL DATA, ADJUSTED FOR PARTY AFFILIATION, AND OFFICIAL RESULTS FOR PROPOSITIONS 4 AND 8

 

 

When the exit poll data are adjusted according to party affiliation for each of the ten polling places individually, Proposition 4 runs more strongly in the official results than in the adjusted exit polls in 5 of 10 polling places, exactly half, which is what one would expect from a reliable exit poll, whereas Proposition 8 runs more strongly in the official results than in the adjusted exit polls in 9 of 10 polling places, almost all, which is what one would expect in a rigged election.

The disparities for Proposition 8 run as high as 3.8% at Lockhurst, 4.3% at Santa Monica, 4.7% at Topanga, 4.9% at Berendo, 6.1% at Lynwood, 6.3% at Eagle Rock, and 14.2% at Glendale. If election fraud is the reason for the disparities, one out of seven voters at Glendale had their votes on Proposition 8 reversed. Bear in mind that these disparities affect the margin of defeat or victory (the point spread) by twice these amounts. At Glendale, Proposition 8 was defeated in the adjusted exit poll by 22.0%, but carried the official count by 6.4% -- an astonishing difference of 28.4%. The calculations are set forth in full detail in the Appendix, and the adjusted numbers for each polling place for Propositions 4 and 8 are presented here (see Table 24).

Looked at another way, Proposition 8 was defeated in the Glendale exit poll by 392 (64.5%) to 216 (35.5%), but passed in the official results by 872 (53.2%) to 766 (46.8%). Among all non-responders, Proposition 8 must have passed overwhelmingly, by 656 (63.7%) to 374 (36.3%), or else the official results are not true and correct.

Among the exit poll responders, there were no demographic groups that supported Proposition 8 by such an overwhelming margin. Only among voters aged 60 or older did Proposition 8 receive even a majority – 26 (53%) to 23 (47%). Proposition 8 did enjoy overwhelming support among registered Republicans in Glendale, passing by 57 (68.7%) to 26 (31.3%), but at that ratio, it would take about 700 Republicans to bring the vote count for Proposition 8 up to 53.2%.

There appears to be no valid explanation for the official vote count for Proposition 8. We have properly adjusted the exit poll data to account for gender, age, race, and party affiliation, and a huge discrepancy remains. In the adjusted exit poll data, the opposition to Proposition 8 is still stronger than the opposition to Proposition 4, while the reverse relationship appears in the official results (see Table 25). I can only conclude that the official vote count for Proposition 8 may be fraudulent.

 

TABLE 25: COMPARISON OF ADJUSTED EXIT POLL DATA, ADJUSTED FOR PARTY AFFILIATION, AND OFFICIAL RESULTS FOR PROPOSITIONS 4 AND 8

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

  

Conclusion

What we are seeing in the numbers from Los Angeles County represents a profound threat to self-governance and to personal autonomy. The most basic and fundamental right of the people in a democracy is the right to have our votes counted as cast. The most personal and private decision of one’s life is whether and whom to marry. In Los Angeles County, it appears that the one was violated in order to deny the other.

What this tells us yet again is that electronic vote counting cannot be trusted. Votes can be altered without leaving a trace, the citizens are none the wiser, and not even the election officials are certain of the count. 

As long as electronic voting is used in our electoral process, it underscores the importance of exit polls -- not only as an instrument to analyze the voting patterns of the electorate, but more importantly, as a means to verify or challenge the accuracy of the official vote counts. It confirms the importance of careful analysis of exit poll results, with proper adjustments to account for sample bias with respect to gender, age, race, and party affiliation, showing all calculations as in a mathematics exam, because the raw exit poll data is never the whole story. And it heralds the monumental contribution of the 140 citizen volunteers who devoted their time, effort and energy to election protection in Los Angeles County. These people volunteered out of a love for their country and a distrust of electronic vote counting. Because we are unable to verify the security, accuracy and reliability of the computers themselves, exit polls may be our best available means to check the vote count.

The likelihood that the vote count in Los Angeles County was corrupted makes it incumbent upon Edison-Mitofsky to release the raw data for their exit polls. Their unadjusted data appears to have shown a statewide disparity of 4.2% for Proposition 8, strikingly similar to the citizen exit polls in Los Angeles County. Even if the raw, unadjusted data is normally considered proprietary, Edison-Mitofsky has a civic duty and a moral responsibility to release it, so that we can subject their data to the same scrutiny as our own data has undergone. We, the people, need to know if our votes have been counted accurately, or if the altered vote counts extend beyond Los Angeles County.

 

Sensitivity Analysis

Some have suggested that the reason for the disparity between the exit polls and the official results for Proposition 8 could be a sample bias among Republican voters. This theory assumes that the non-Republicans who responded to the exit poll were representative of the non-Republicans in the electorate, but that the Republican responders were more likely to have voted against Proposition 8 (that is, for gay marriage) than were the Republican refusals.

This theory derives from two established facts: in the 10 polling places combined, only 15.6% of the voters at the polls were registered Republicans and only 36.7% of them participated in the exit poll. It is entirely plausible that conservative Republicans would be more reluctant than moderate Republicans to participate in an exit poll being conducted in heavily Democratic precincts, in what Marj Creech and I have long called the “hostile territory scenario.”

Given the available data, the only way to test this scenario is to assume that non-Republican responders were representative of non-Republicans in the electorate and to stick to the data adjusted according to party affiliation; and to assume that all Republican non-responders, each and every one of them, voted for Proposition 8. This should reveal, once and for all, whether there were enough Republican non-responders to explain the official results, or whether it is mathematically impossible to do so.

In the 10 polling places combined, there were approximately 1172 Republican non-responders, accounting for 10.1% of the electorate. Among the Republican responders who indicated a choice on Proposition 8, 70.1% voted for it, and this was the percentage used when adjusting the exit poll data to account for party affiliation.

Thus, even if each and every Republican non-responder voted for Proposition 8, it could only account for 3% of the electorate (30% of 10%), and cannot explain the disparity between the adjusted exit poll data and the official results.

Looked at another way, 35.9% of non-Republican exit poll responders voted for Proposition 8; at that rate, about 1492 non-Republican non-responders would have voted for Proposition 8 (in addition to 2360 exit poll responders, including Republicans, who voted for it). This still leaves us 1473 votes short of the official count for Proposition 8 (5325 minus 3852), and there were only about 1172 Republican non-responders.

Thus, even if each and every Republican non-responder voted for Proposition 8, it is not enough to explain the official results (see Table 26).

The best example is at Glendale, where 24.9% of non-responders, about 260 of 1045, were Republicans. Even if they all voted “Yes” on Proposition 8, the official results cannot be explained unless 396 of 785 other non-responders, more than half, also voted “Yes” on Proposition 8. But among exit poll responders at Glendale, Proposition 8 was supported by only 26.0% of Democrats, 26.8% of third-party voters, and 42.0% of unaffiliated voters, as shown in the Appendix. Even among Republicans, only 68.6% supported Proposition 8 – a solid majority, but not 100%. Thus, the official results are mathematically impossible to explain unless one assumes that all samples of voters, of all parties, were non-representative.

When one analyzes Proposition 4 in the same manner, a very different story emerges. In the 10 polling places combined, 38.9% of non-Republican exit poll responders voted for Proposition 4; at that rate, about 1617 non-Republican non-responders would have voted for Proposition 4 (in addition to the 2402 exit poll responders, including Republicans, who voted for it). This leaves us only 667 votes short of the official count for Proposition 4 (4686 minus 4019), and there were about 1172 Republican non-responders; if only 56.9% of these voted for Proposition 4, the official results can be explained. (See Table 27

The interpretation that the disparity between the exit poll data and the official results for Proposition 8 is due to a corrupted vote count is bolstered by the fact that the official results for Proposition 4 are so easily explained by data from the very same exit poll – the same voters, the same day. The official results for Proposition 8 may very well be fraudulent.

The interpretation that the official results for Proposition 8 are true and correct not only requires that Republican exit poll responders were very different from Republican non-responders, but that Democratic, third-party, and unaffiliated exit poll responders were likewise non-representative of the electorate.

This argument would render useless and invalid any exit poll conducted anywhere – in Ukraine, Georgia, Iran, Iraq, Mexico, Ohio or California – because it would allow the results of any exit poll to be dismissed on the unwarranted assumptions that the official results are true and correct, that the exit poll responders must not be representative of the electorate, and that the exit poll results must therefore be wrong. But exit polls have a long history of being remarkably accurate reflections of the electorate. In Germany and elsewhere, exit polls are relied upon to forecast the winners of the elections, and citizens are content to wait for days to hear the official count.

 

TABLE 26: DATA FOR REPUBLICAN AND NON-REPUBLICAN

RESPONDERS AND NON-RESPONDERS, PROPOSITION 8

 

 

 

TABLE 27: DATA FOR REPUBLICAN AND NON-REPUBLICAN

RESPONDERS AND NON-RESPONDERS, PROPOSITION 4

 

 

____________________________________________________

 

Appendices

Attachment	Size
CA-Prop8-Exit-Polls-LA-County-RHP.pdf [8]	419.22 KB