The international community took on following the break-up of the
Document Sample
DRAFT
Charting a Challenged Election:
Turnout Distribution Analysis as a Tool for Quick Election Assessment
John Brady Kiesling
Abstract: Election observers come under enormous pressure to announce immediately
whether a given election was free and fair enough to confer legitimacy on the victor. Vote
fraud of various kinds is endemic , but no mechanism exists to assess the scale and
probable impact of that fraud.
In the 1996 and 1998 Armenian presidential elections, vote fraud affected many voting
precincts but not all or even most. The commonest forms of vote fraud had the effect of
raising reported voter turnout. By aggregating detailed precinct polling data by voter
turnout percentage and graphing the number of votes cast for each candidate in each
turnout band, differences in voting patterns at lower and higher turnouts become
immediately evident. It is possible to show in crude terms, by the divergence from a
standard distribution curve, the extent to which fraud or pressure distorted the results.
The method warrants further study and could be refined through the application of
sophisticated statistical methods. Every democratic society, including the U.S., should
insist that detailed precinct-by-precinct election results, comparable from one election to
the next, be made available in electronic form as a defense against electoral manipulation.
An election should endow the winners with the democratic legitimacy necessary to carry
out the function of their elective office. The mere fact of holding an election is not
enough to generate useful legitimacy. Habits of electoral fraud and profound voter
skepticism persist in many countries on the road to democracy. Therefore the
international community monitors elections in emerging democracies to deter fraud and
bolster public confidence that the electoral process conveys the will of the electorate.
This research is drawn from the international monitoring of the 1998 and 2003 Armenian
Presidential elections, plus equivalent data from the 2002 Minnesota and Florida general
elections. I took part as an election observer under the auspices of the Office for
Democratic Institutions and Human Rights (ODIHR), a component agency of the
Organization for Security and Cooperation in Europe (OSCE). 1 In Armenia, in 1998,
ODIHR assembled more than 180 foreign observers, supported by local interpreters and
drivers. There were long-term teams in each province (marz), deploying the short-term
observers that fanned out around the countryside on Election Day. Each four-person
team visited as many polling places as possible, and stayed for the all- night vote count at
one suspect precinct. 2
The election observer teams assembled by ODIHR have developed a reasonably powerful
set of tools for judging the quality of election law and procedures. 3 Their tough
assessments are made public very soon after each monitored election. Though ODIHR is
willing to state clearly that a given election did not meet international standards, it has
resisted pronouncing on whether the official results of a given election accurately
reflected the will of the voters.
2
This hesitation is natural. Even the largest and most efficient monitoring mission cannot
observe every activity of every voting place. There is no scientific method to justify a
claim to godlike knowledge of what the outcome would have been in a perfect system.
The 2000 U.S. Presidential elections brought to public attention the unwelcome
knowledge that even elections in established democracies are a far less clear and
objective process than we would like to believe.
The failure of the international community to challenge a fraudulently elected ruler has
costs that probably outweigh short-term political expediency. The 1996 reelection of
Levon Ter-Petrosyan as President of Armenia supports such a conclusion. Ter-
Petrosyan's challenger, Vazgen Manukyan, claimed that his own victory had been
overturned by late- night fraud in the vote tallying, a conclusion some international
observers supported. The U.S. government and other OSCE players satisfied themselves
with token criticism. 4 Ter-Petrosyan took office in a sea of opposition protests. Even
without international criticism, Ter-Petrosyan found himself a tainted and unpopular
president, increasingly dependent on the powerful figures in his government whose
intercession had provided the vital margin of victory. Finally, Ter-Petrosyan was forced
to resign, after a brief, unproductive tenure.
The 1998 election that replaced Ter-Petrosyan with Robert Kocharyan distorted the
popular will as well. Detailed examination of both rounds of that election led to the
conclusion that well over 100,000 more votes had been cast in favor of the ultimate
winner than there were voters to cast them. Once again the international community was
put in the awkward position of having to decide whether to offer or withhold legitimacy.
The ODIHR mission groped for a methodology to quantify the extent of the fraud, but
ultimately offered no pronouncement on whether Robert Kocharian, the winner by an
ostensible 59-41% landslide, received a genuine popular mandate.
The Fraud Problem
Unfree, unfair elections tend to follow consistent patterns that leave traces in the election
results. 5 A prudent central government would prefer a narrow but p lausible victory to a
landslide. 6 Some local officials and clan/tribal/economic leaders, however, justify and
preserve their privileged local status by delivering an impressive share of the vote to
whomever they judge likely to win. There is still in many formerly totalitarian countries
an instinct to seek a turnout as close to 100% as possible, with the highest possible
percentage for the expected winner. However, such a high turnout cannot legitimately be
achieved in Armenia, where electoral registers still contain hundreds of thousands of
Armenians who emigrated for economic reasons and did not return for elections.
In multi-party elections conducted under international supervision, most precincts have at
least one committed opposition member in the electoral commission, or there are outside
observers present. Under such circumstances, a plurality of precincts will have
reasonably fair balloting and accurate counting and reporting, and a further large group of
precincts will have only minor irregularities -- typically the use of state and parastatal
mechanisms to assure that a maximum number of voters are brought to the polls, fortified
with the promise of cash or other benefits that will accrue from an appropriate outcome.
Kiesling, Elections, 9/9/2010
3
Where local power structures offer incentives to voters, they take steps to ensure that
bribed or coerced voters actually vote as promised. In Armenia, this was typically
achieved though "carousel voting," a method that did not require active collusion by
election officials. Each voter would be given outside the polling station a pre- marked
ballot to conceal in a pocket. At the polling station, the voter would receive a normal,
unmarked ballot. Entering the polling booth, the voter would cast the pre- marked ballot
and conceal the unmarked one. He would then return to the outside election supervisor
with the unmarked ballot, to be marked with the correct vote and given to the next
customer. Though such fraud is in itself neutral regarding the number of people who
vote, the presence of an organization involved in carousel voting is usually associated
with an aggressive effort to mobilize voters and thus a higher turnout.
More aggressive fraud, such as large-scale ballot box stuffing, miscounting or
invalidation of opposition ballots, or falsified protocols, requires active complicity from
members of the election commissions. In the 1998 Armenian elections, observers found
enough ballot boxes containing stacks of identically marked ballots to document the
common practice of spending the final minutes of election day forging the signatures of
absent voters and casting ballots on their behalf. 7
Turning away opposition voters triggers court challenges and publicity. The safest forms
of electoral fraud increase the reported turnout non- linearly over the course of election
day. ODIHR employs a useful method of having its observer teams record the time of the
visit and the number of votes cast up to that time at each polling station visited. By
graphing turnout over time with a large enough sample of precincts, it is possible to
detect the distortion in voting patterns resulting from last-minute ballot-box stuffing.
ODIHR monitors the integrity of the precinct vote tally reporting by obtaining copies of
as many precinct protocols as possible. These protocols are checked against the official
published results. Finally, ODIHR's election analyst compares the outcome in precincts
in which ODIHR observers have monitored and verified the count against precincts
where the counting took place without foreign witnesses. If the outcome diverges
substantially in monitored versus unmonitored precincts, ODIHR can draw various
conclusions about the results.
ODIHR's microscopic examination of specific precincts provides crucial data on the
mechanics of electoral fraud. When electoral misconduct reached the point of perverting
the electoral outcome, statistical methods can be found to quantify it. When a significant
number of precincts share a common phenomenon of turnout being inflated by additional
votes for only one candidate, a graph of turnout using precinct-by-precinct results will
produce a distribution curve of voter behavior that diverges substantially from a normal
bell curve.
Methodology
The 1998 Armenian presidential elections were the starting point for this research, which
led me back to a study of the 1996 election results as well. In both elections there was a
Kiesling, Elections, 9/9/2010
4
strong empirical correlation between observed election misconduct and precinct voting
results characterized by extremely high voter turnout and massive support for the winning
candidate.
The International Foundation for Electoral Systems (IFES), a U.S.-funded NGO, had
wisely insisted as a condition for assistance that the Armenian Central Election
Commission make available to the public and the international community almost
immediately after the elections Microsoft Excel spreadsheet files giving the detailed
precinct-by-precinct results of the elections for the whole country. 8
Using a spreadsheet of the detailed precinct results it is a simple matter to sort precincts
by voter turnout (total votes cast/number of registered voters) and percentage of the vote
for each candidate (votes/total valid votes). Sorting the list by turnout will highlight at top
and bottom the non-negligible number of precincts for which some error or serious
omission has been made in recording the results. The 1998 Armenian election CEC
spreadsheet I received contained dozens of substantial errors, either missing data or
obvious transpositions from column to column. Less obvious transpositions, such as the
number of votes for each candidate, are not detectable. Sobering, however, is that even
U.S. election data shows similar obvious errors, such as precincts with purported voter
turnouts of as high as 400 percent. Once obvious transpositions are corrected, and
uncorrectable precincts purged, the spreadsheet, sorted by turnout, is ready for use.
A column is created using the MROUND function (from the Excel add- in Analysis Tool-
Pak) to round each precinct turnout percentage to a whole number. With 1000 or more
precincts, rounding turnout to a multiple of 2% or 2.5% achieves a convenient level of
detail. Using the Data:Subtotal menu, segregate the data by rounded turnout, summing
the columns for registered voters, total votes, and votes for each candidate. Compress the
display to show only the subtotal amounts. It is then a simple matter to chart on a line
graph the total number of votes per candidate on the y-axis versus turnout percent band
on the x-axis.
Given enough data points (at least several hundred precincts and several hundred
thousand voters), one would expect a fairly symmetrical bell curve of vote distribution by
turnout band, with the peak located near the overall turnout percentage countrywide. To
test this analytical method, I applied it to precinct voting data from the 2000 U.S.
presidential elections in Minnesota and Florida. 9 Figure 1 gives the outcome of my
turnout distribution analysis for the precinct results in Minnesota, some 4076 precincts
and a total turnout of 2.45 million voters, 75.2% of registered (including late-registering)
voters. Gore won this election by a 2% margin, 1.17 million to 1.10 million.
As one would expect in Minnesota, the results are a clean distribution curve with a
maximum at 76%, consistent with the overall reported voter turnout. Results for Bush
and Gore track closely, albeit with a shift to favor Bush as turnout increases. I graph on a
secondary y-axis (right side scale) the difference in vote percent between Gore and Bush
at each level of voter turnout. A linear regression trend line suggests a slight correlation
between increasing turnout and increasing support for Bush.
Kiesling, Elections, 9/9/2010
5
Figure 1
Minnesota 2000 Presidential Elections
450000 150.0%
400000
100.0%
350000
y = -0.0208x + 0.4819
R2 = 0.2904
300000 50.0%
Total votes
250000 Gore
Voters
Bush
0.0%
Nader
200000 gore-bushpct
Linear (gore-bushpct)
150000 -50.0%
100000
-100.0%
50000
0 -150.0%
3 4 4 5 5 6 7 7 8 8 9
4 0 6 2 8 4 0 6 2 8 4
% % % % % % % % % % %
Precinct Turnout
In Florida (Figure 2) the results are similar. With some 5900 precincts included, and 5.8
million voters, the bell curve is relatively smooth. The voting peak, at 70% turnout, is
slightly higher than the calculated overall turnout of 66.7%. Comparing the jagged peak
in the distribution curve to a smooth curve would allow one to estimate the net effect of
more successful Republican voter mobilization (or, for a conspiracy theorist, voting
machine tampering) on turnout.
Figure 2
Florida 2000 Presidential Elections
700000 80.0%
60.0%
600000
y = -0.0145x + 0.3789
40.0%
R2 = 0.3477
500000
20.0%
BUSH
400000 GORE
Votes Cast
0.0%
NADER
TURNOUT
-20.0% Gore-Bush%
300000
Linear (Gore-Bush%)
-40.0%
200000
-60.0%
100000
-80.0%
0 -100.0%
24 tal
28 tal
32 tal
36 tal
40 tal
44 tal
48 tal
52 tal
56 tal
60 tal
64 tal
68 tal
72 tal
76 tal
80 tal
84 tal
88 tal
92 tal
96 tal
l
ta
To
To
To
To
To
To
To
To
To
To
To
To
To
To
To
To
To
To
To
To
20
Precinct Turnout
Kiesling, Elections, 9/9/2010
6
The Arme nian Presidential Elections
If Minnesota and perhaps Florida show the predictable pattern for a democratic election,
Figure 3 shows a very different curve, the results of the highly controversial 1996
Armenian presidential elections in which incumbent Levon Ter-Petrosyan defeated
Vazgen Manukyan and others by a reported 51.9% to 41.1%. Election night was marked
by a mysterious interruption in the counting process, and thousands of citizens later took
to the streets to protest massive manipulation of the results.
The official curiosity of OSCE observers focused on a missing 22,000 ballot coupons in
Yerevan. These coupons were an additional security measure, and their number was of
interest because the discrepancy they signaled was just enough to put Ter-Petrosyan over
the 50% threshold for a first-round victory. It was assumed, however, that Ter-Petrosyan
had in any case a wide lead. The question of the coupons was ultimately dropped.
After Ter-Petrosyan's ouster, however, various Armenian personalities felt freer to
disclose their discomfort in having played unethical roles in his reelection. It became
clear, in retrospect, that there were two stages to the election fraud. The first had been
widespread fraud in many precincts on election day. This had been enough to give Ter-
Petrosyan a sizeable lead over Manukyan, but not enough to cross the 50% threshold.
Late at night, as this short- fall became apparent, senior officials in Yerevan scrambled to
produce extra votes. In their haste to correct the results, they had little time to spread the
fraudulent votes unobtrusively across hundreds of polling stations or cope with the
accounting problem of 22,000 more ballots than coupons.
Figure 3
1996 Armenian Presidential Election
160000 80.0%
140000
60.0%
120000
40.0%
y = 0.0323x - 0.2905
R2 = 0.687
% Margin for winner
100000
Number of Votes
Valid votes
20.0%
Ter-Petrosyan
80000 Manukian
TP-VM%
0.0%
Linear (TP-VM%)
60000
-20.0%
40000
-40.0%
20000
0 -60.0%
3 3 4 4 5 5 6 6 7 7 8 8 9 9 1
0 5 0 5 0 5 0 5 0 5 0 5 0 5 0
% % % % % % % % % % % % % % 0
%
Percent turnout
Kiesling, Elections, 9/9/2010
7
As Figure 3 shows, however, the result of graphing 1218 precincts or multiple-precinct
communities and 1.25 million valid votes by turnout (grouped in 2.5% bands), the
election produced a curve that diverges dramatically from a normal distribution. The
peak of the distribution curve is at 52.5 %, well below the official turnout of 58.8%.
There are numerous spikes. As is readily apparent, voting behavior at the peak of the
distribution curve shows a consistent, clear advantage to opposition challenger
Manukyan. Only in the areas of elevated turnout does the pattern shift, with a close
correlation between increased turnout and increased margin of victory for the incumbent.
There are possible explanations for substantial differences in voting behavior and turnout.
Segregating Armenian results by urban/rural and small/large precinct, however, showed
no significant difference in voter behavior. Nor could the urban/rural distinction explain
the massive oversupply of votes at unusually high turnouts. Given the huge percentage of
the Armenian population that had migrated to Russia or elsewhere for economic reasons,
and the primitive state of most voter registries, turnouts much beyond 70% would have
been physically impossible in most areas of the country.
Armed with Figure 3 in 1996 in addition to ODIHR's report of widespread fraud, it would
have been easier for the U.S. Government to have agreed with opposition complaints that
Ter-Petrosyan had stolen the election. The 1998 Armenian elections, however, were in
significantly less clear. Figure 4 shows the distribution of votes in the first round, in
which Robert Kocharyan won a plurality against multiple rivals, though his rival Karen
Demirchyan enjoyed a small but consistent lead in lower turnout bands.
Figure 4
Armenia: Votes Against Turnout, March 16, 1998
Regions by community, Yerevan by precinct
180000
160000
140000
120000
Kocharyan
100000 Demirchyan
Votes
Badalian
80000 Manukyan
Total Turnout
60000
40000
20000
0
3 3 4 4 5 5 6 6 7 7 8 8 9 9
3 8 3 8 3 8 3 8 3 8 3 8 3 8
% % % % % % % % % % % % % %
Kiesling, Elections, 9/9/2010
8
That the election was problematic can be seen in Table One.
Table One
Election Registered Turnout turnout Chart
turnout
Date Voters % change maximum
21/9/96 2210189 1333204 60.3% 52.5%
16/3/98 2277195 1458135 64.0% 124931 57.5%
30/3/98 2300816 1567702 68.1% 109567 57.5%
The overall statistics showed a major, unexplained increase in turnout since the 1996
presidential elections, an increase not registered by observers on the ground. This was
followed by an even more striking increase in turnout –109,000 additional voters -- from
the 1998 first round to the second round run-off with Karen Demirchyan, though election
observers detected no increase or if anything a slight decrease in voter numbers from two
weeks before. The turnout distribution chart shows a large discrepancy between
officially reported turnout of 68% and the peak of the turnout distribution curve at 57.5%.
Study of Figure 5 (note shift of x-axis direction from previous graphs) shows that the
credibility of Kocharyan's crushing victory over Demirchyan in the run-off was
undermined by the near-draw at the more reliable part of the distribution curve.
Figure 5
Armenia: Votes against turnout, March 30, 1998
by region communities and Yerevan precincts
100000
90000
80000
70000
60000
Votes
50000
40000
30000
20000
10000
0
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
0%
.5
.0
.5
.0
.5
.0
.5
.0
.5
.0
.5
.0
.5
.0
.5
.0
.5
.0
.5
.0
.5
.0
.5
.0
.5
0.
97
95
92
90
87
85
82
80
77
75
72
70
67
65
62
60
57
55
52
50
47
45
42
40
37
10
RK 2nd Rnd KD 2nd Rnd
Kiesling, Elections, 9/9/2010
9
The final chart, Figure 6, shows the utility of consolidating first and second round
precinct voting data in a single spreadsheet. Doing so makes it effortless to aggregate
precincts based on the number of additional voters they reported in the second round.
(On a secondary axis is plotted the absolute number of precincts reporting a given change
in voter turnout, with the maximum, predictably, at zero change in number of voters).
Statistically speaking, every new voter (a vote cast in the second-round but not the first)
correlated to an additional vote for Kocharyan. In the polling stations in which turnout
officially decreased, Kocharyan and Demirchyan ran a statistical dead heat.
Figure 6
Armenian Elections 98: Added Votes Against Precinct Turnout Change
Kocharian Gains from Inflated Turnouts
60000 700
642
50000
600
40000
500
Number of Precincts
30000
400
Total Votes
20000
300
10000
248
200
0
+
0
950
900
850
800
750
700
650
600
550
500
450
400
350
300
250
200
150
100
-100
-150
-200
-250
-300
-350
-400
-650
50
-50
142
1000
126
100
-10000 76
64
33 38 34
24 28
16 19 17
4 3 3 2 5 9 6 9 8 3 5 8
-20000 0 0 1 1 0
Precinct turnout change (50-vote bands)
RKVotes KDVotes precincts
Conclusion
Turnout distribution analysis offers a useful snapshot of the extent to which outside
forces have intervened, legally or not, to distort the normal distribution curve of voter
behavior. In the Armenian case, the graphical picture coincides with a reasonably sober
assessment of what took place on the ground. Tested on additional elections and
improved by more sophisticated statistical methods, it could be a useful complement to
the tools already available to election observers.
Any quantitative method can be outwitted by systematic geographic distribution of
fraudulent votes, but this more systematic fraud is very difficult to manage, given the
varying strength of opposition parties and the self-serving behavior of local political
bosses. In any case, fraud can be detected by the rigorous scrutiny of an organized team
of experienced election observers. It is crucial to witness all phases of the elections,
including the counting process, to secure control copies of precinct returns for
Kiesling, Elections, 9/9/2010
10
comparison with the official data placed on the spreadsheet, and to make a statistical
comparison between observed and unobserved precincts.
A vital component of any truly democratic electoral process is making the detailed voting
results available to anyone who asks for them. In some respects the U.S. is still an
emerging democracy as well. Each U.S. county has its own philosophy about how to
make results available. Few report precinct-level data on-line, fewer still list the number
of registered voters in conjunction with that data, and almost none put that data in an
easily usable electronic format. The next federal reform of U.S. electoral processes
should require the Secretary of State or senior election official of each state to publish
downloadable electronic spreadsheet with the compiled precinct-by-precinct election
results of each county, including number of registered voters and method of voting used.
To the extent possible, precinct boundaries should remain the same from one election to
the next, allowing fast, powerful statistical analysis to restore voter confidence that their
will is faithfully reflected.
John Brady Kiesling
westtothesea@hotmail.com
1
Since March 2003 I am no longer an emp loyee of the U.S. Govern ment. The data used in this research is
for the most part publicly available, thanks to the work of IFES in Armen ia. My v iews regarding the
conduct of the Armenian elections in no way imply o fficial concurrence by the U.S. govern ment, OSCE or
ODIHR. I owe an enormous debt of gratitude to U.S. Embassy Political Assistant Alla Bakunts, my
interpreter and guide, and to former U.S. Embassy Economic Assistant Haik Gugarats, who inspired key
statistical insights. My respect for the work o f ODIHR and its dedicated specialists is immense. I would
like to single out Dr. Soren Theisen, the Danish Caucasus historian, for h is wisdom and hospitality in th e
wilds of Syunik Marz and Copenhagen. Finally, my thanks to the Hellenic Studies Program and the
Woodrow Wilson School of Princeton University for their support and insights.
2
Office for Democratic Institutions and Human Rights, REPUBLIC OF A RM ENIA PRESIDENTIA L
ELECTION, MARCH 16 AND 30, 1998, FINA L REPORT at :
http://www.osce.org/documents/odihr/1998/04/1215_en.pdf
3
Election observer missions made up of polit icians, such as those of the Parliamentary Assembly of the
Council of Europe and the Co mmonwealth of Independent States, tend to take a more h ierat ic and less
rigorous view of their role. Many Armenians, at least, had the impression that their purpose was to witness
and bless the recurring miracle of democracy and catch the first flight ho me after the obligatory banquet.
4
State Depart ment Press Spokesman Burns endorsed Ter-Petrosyan's victory on October 22, 1996, one
month after the elect ion, after painful discussions within the U.S. gov ernment. See
http://dosfan.lib.u ic.edu/ERC/briefing/daily_briefings/1996/9610/961022db.html
ARMENIA: Observers: `Serious Viol ati ons' Fail To Affect Vote Result
Yerevan SNARK in English, 1700 GMT 24 Sep 96 Tuesday, September 24, 1996
(FBIS Transcribed Text) YEREVA N, SEPTEM BER 24.
(SNA RK). Simon Osborn, Coordinator of the international observers group, announced today the
preliminary conclusion of observers fro m OSCE and the Bu reau of Democratic Institutions for Hu man
Rights. According to Osborn, serious violations of the law were recorded during the presidential election,
Kiesling, Elections, 9/9/2010
11
however they did not affect the result at the current stage. Preliminary results show that Levon Ter-
Petrossian garnered 51.94% and Vazgen Manukian won 41.15% o f votes.
Regarding the violat ions Osborn said they were recorded during the presidential campaign and in the
election day. The state-run TV granted to Levon Ter-Petrossian more time than to his contenders.
Moreover, Vazgen Manukian and Ashot Manucharian were refused to appear on TV paid channel a week
before the elections. The stealing of ballots and the ballot box fro m a precinct in Yerevan coincided with
energy cut there Simon Osborn considered to be the most serious violation in the election d ay. Observers
also noted that officers violated secret ballot and made servicemen vote in favor of a certain candidate.
Osborn expressed concern over the fact that representative of the Internal Affairs and Defense Ministries
were presented at precincts in Yerevan, Kotaik, Ararat and Armavir in the election day. Nevertheless, the
international observers believe the presidential elections were held legally.
5
See the report by the Office for Democrat ic Institutions and Human Rights, REPUBLIC OF A RM ENIA
PRESIDENTIA L ELECTION, MA RCH 16 A ND 30, 1998, FINA L REPORT at
http://www.osce.org/documents/odihr/1998/04/1215_en.pdf. There is a melancholy consistency of
electoral misconduct around the globe. The ODIHR report of the 2003 election shows similar flaws. It is
sobering how much fraud can take p lace undetected by even trained observers.
6
A Kocharyan advisor told the author in a private conversation that Kocharyan's election team had been
aiming for 55% of the vote, a clear v ictory but by a credible margin.
7
Ibid, p. 12-13.
8
Access to this detailed election data is usable electronic form should be a basic minimu m demand of any
international organization or citizen watchdog group anywhere in the world. The file should include, at a
minimu m, region, mun icipality, and polling precinct identifiers, nu mber of registered voters, number of
valid ballots, number of spoiled/invalid ballots, and number of votes for each candidate. Since election -to-
election co mparison is a powerfu l tool for detecting vote fraud, the same precinct names and boundaries
should apply fro m one elect ion to the next, to the extent possible.
9
Fro m the Federal Elect ions Project web site of the American University School of Public Affairs,
http://www.american.edu/academic.depts/spa/ccps/elections/states.html Dav id Lublin and D. Stephen Voss.
2001. "Federal Elections Project." A merican University, Washington, DC and the University of Kentucky,
Lexington, KY. A few dozen precincts with null data or turnouts of less than 20% or mo re than 100% were
excluded on grounds of probable recording error. Absentee ballots were perforce ignored, since they were
not tabulated in the precincts in which the absentee voter was registered. A fuller study would take
absentee ballots into account.
Kiesling, Elections, 9/9/2010
Shared by: Jun Wang
Jun Wang
Prof.
Dr
China Agricultural ...
About
Some of Those documents come from internet for research purpose,if you have the copyrights of one of them,tell me by mail vixychina@gmail.com.Thank you!
Related docs
Other docs by hcj
