Ecanal Special Report No. 1, July 2006 



REVIEW OF INCONSISTENCIES IN MEXICO'S PRESIDENTIAL 
ELECTION OF JULY 2nd 2006 


Contents 
Summary and Conclusions 
Page 
INTRODUCTION 5 
I. MACRO PROBLEMS 5 
II MICRO PROBLEMS 13 
III. SIGNIFICANCE OF INCONSISTENCIES 14 
IV. ENVIRONMENTAL PROBLEMS 14 
V. THE DEMAND FOR A BALLOT-BY-BALLOT RECOUNT 16 

REVIEW OF INCONSISTENCIES IN MEXICO'S PRESIDENTIAL ELECTION
OF JULY 2nd 2006 


SUMMARY AND CONCLUSIONS 

Mexico's presidential election of July 2nd 2006 was apparently a
close race between the two leading candidates: Andrés Manuel López
Obrador (AMLO) and Felipe Calderón (FC). Nonetheless, many question
marks have been raised about the process of vote counting and the
environment that prevailed before the day of election. These are the
grounds on which large groups of people find the results unacceptable
or demand further certainty that irregularities did not represent a
biased pattern across the process against AMLO. If such question marks
are ignored, major and possibly increasing protests are likely to take
place. A president elect who lacks high credibility would find it very
difficult to lead government, especially when he should face very
difficult decisions. To name a few:

(1) The new president will find that public spending was growing at
very high rates since the end of 2005 and may have gone out of control
at this stage ^Ö arresting it will be essential to make enough room to
implement new programs. Yet this calls for a degree of authority that
cannot be challenged.

(2) Since 2005 it has become clear that maintaining the volume of oil
production is all the time more difficult. To decide the focus of
priorities and implement the necessary discipline in Pemex will
require great capacity to negotiate and freedom from vested interests.

(3) Tax collection has been falling as a result of reductions in
income tax rates which have not been offset by new sources of tax
revenue The difficulties of finding the right mix of taxes between
income, VAT, and social security contributions are compounded by high
levels of tax evasion and by the promise of a flat rate for income tax
made, perhaps naively, by FC during the campaign. This flat rate would
cause in our estimation a further reduction in tax revenue of 2
percentage points of GDP, given the high concentration of collection,
if the rate is set at a significantly lower level than the present top
rate.

 (4) Public sector pension systems record increasing deficits which
are until now covered by budgetary transfers from the federation, but
this strategy is not feasible over the medium to long term and,
moreover, it is unwise. Solving this problem will require not only a
financial strategy (on which the current administration has a
proposal, yet exceedingly costly), but mainly a political negotiation
with unions, which must be led by the President.

(5) Bank credit for the private sector remains highly concentrated on
consumer loans and is unlikely to be diverted to business loans
without the Government creating the right incentives.  Banks will
naturally prefer to continue lending for consumption, where margins
are very high and risk is low, but this not desirable or healthy for
global resource allocation.

(6) Some major effects from NAFTA-liberalization of corn have already
been felt and thus caused emigration to the northern border, major
cities and the US. The impact of the final stage of tariff reduction
scheduled for January 2008 will be felt early next year, however, as
the tariff will fall to a level sufficiently low to displace Mexican
producers, while US agriculture continues to be
subsidized. Negotiations on this sector ought to be starting now; for
any government to lack a clear course on this major problem will be
recipe for further social tension.  These are major reasons why Mexico
needs a presidency strong enough to emerge from the election, if not
in terms of the share of votes, at least in not being challenged on
moral grounds. It will be impossible for a government to succeed in
all political negotiations and policy implementation without an aura
of legitimacy.  Thus, the treatment of claims over all aspects of the
election by the Federal Electoral Tribunal will be essential for
medium-term confidence and macro stability. The claims we discuss have
been coming from different parties, including citizens, and need
further analysis before conclusions can be reached on their impact on
votes. Information is still flowing at the closing of this report,
such that new data will have to be included for a fair assessment. At
this stage, however, there are sufficient grounds to allow people to
shape an opinion or at least to know the main objections to the
election. The complexity of the electoral process in some instances
suggests that undertaking a thorough and independent examination of
all substantial claims may not be feasible now, when the impartiality
of the entity in charge of the vote count (IFE ^Ö Mexican Electoral
Institute) is in doubt. This gives rise to the growing demand for a
ballot-by-ballot recount, which will probably continue to grow in
coming weeks.

Since the evening of July 2nd, evidence which casts doubts on the
certainty of electoral results has included many points. Below is a
summary.

(1) IFE failed to name a likely winner on the night of the election,
even though it had Felipe Calderón (FC) ahead of Andrés Manuel López
Obrador (AMLO) by 1% of the vote, a respectable figure of 400,000
votes.

(2) There were 283,448 fewer votes for President than for senators. In
many ^Ñadjacent' voting stations (^Ócontiguas^Ô in IFE parlance),
there were no votes cast for President.

(3) In states where AMLO was leading, the difference between votes for
senator and votes for president was the largest (in favour of senator
votes). In states where FC was leading, the difference between votes
for president and those for senator was the largest (in favour of
president votes). The extreme cases were Tabasco (96,450 votes less
for President) and Nuevo León (41,290 votes more for President). In
six states were AMLO led, votes for President were less than those for
senator by 313,882. In states where FC led, votes for President were
more than those for senator by 111,178.

(4) On July 2nd-3rd, IFE had counted tally sheets of 117,000 voting
stations, leaving aside 13,000 voting stations because they had
anomalies that called for further examination. Although the result in
favour of FC with the count of 117,000 stations was 1%, the result of
the 13,000 stations that were counted later 4% in favour of
AMLO. Thus, the question arises as to how such a large sub-set of
voting stations had a different structure of votes than the 117,000
voting stations. Furthermore, IFE never explained what the
inconsistencies of the 13,000 stations were and how it could resolve
them within hours on July 3rd. When the PREP count was concluded with
a difference in favour of FC of 0.6%, its reliability had been put
under serious doubt.

(5) Following PREP, the next stage of the electoral count was the
count by district, which involves a count of tally sheets in all 300
districts. During this count district councils are authorized to open
ballot boxes for a manual recount of votes. AMLO's coalition demanded
52,000 boxes to be opened on the grounds of irregularities that were
visible in varying degrees. District councils only authorized opening
2,873 boxes. IFE did not explain what had Ecanal Special Report No. 1,
July 2006 been the outcome of such recounts, except that in 95% of
cases the number of ballots differed from the data in tally
sheets. Absence of a detailed report lent itself for each party
showing counts from individual districts. According to Ricardo Monreal
of AMLO's coalition, the recount of these boxes gave AMLO an
additional 102,000 votes. Only 15 days later did IFE issue aggregate
figures of the recount, indicating that opening 2,830 boxes had
resulted in lower figures of votes for all candidates. It did not
explain, although this may be understood, that these reductions
applied to the basis of PREP figures.  The reported reduction in votes
runs counter to the figures on upward revisions on votes for all
parties resulting from district count with respect to the PREP. The
total votes that were recorded in district count in addition to those
in PREP were 3,164,684 votes, of which AMLO received 1,131,844.

(6) The count by districts resulted in the amendment of many
discrepancies between tally sheets and the PREP, some of which had
been identified on July 2nd-3rd. It also included those voting
stations that had been left out of the PREP. As was mentioned, votes
for all three candidates were found to be greater than in PREP,
including for Roberto Madrazo (RM). But AMLO's additional votes were
16.3% greater than the additional votes for FC and 15.8% greater than
the additional votes for RM, both significant differences.

(7) With the results of district count, the small margin of error that
IFE had indicated the PREP count would have (0.03%) was significantly
exceeded: the correction over the first PREP result was 8.7 percent.

(8) Intervention of President Fox in favour of FC and against AMLO in
the form of statements to the press and commentaries to business
groups has been documented extensively and now rests with the
Electoral Tribunal. AMLO's coalition considers that this intervention
made the election unfair.

(9) Intervention of Secretary of Communications, Mr. Pedro Cerisola,
in favour of PAN is also documented and rests with the Electoral
Tribunal.

(10) Claims that the electronic counting of votes was biased and
therefore showed paths not typical of stochastic processes have been
submitted by professional analysts and mathematicians and may justify
further enquiries

(11) An on-going analysis of null votes (Walter Mebane and Yuriko
Takahashi, of Cornell University --see http://polmeth.wustl.eduy),
which is part of their analysis of elections in the US and now in
Mexico, has shown an abnormally high number of null votes in
Guanajuato and Nuevo Leon. In these states AMLO's coalition had a low
level of representation at the stations (33.9% in Guanajuato). This
analysis uncovers a positive correlation between high numbers of null
votes and abnormally low votes for AMLO.

This report has been based on official information from the IFE and
analysis offered by analysts who examined the vote counting process
throughout July 2nd-6th. It also draws on part of the information that
AMLO's coalition has made available.

INTRODUCTION 

This report discusses why Mexico's presidential election has raised
many doubts on its rectitude and aroused protests among voters who
believe that results of such election should be disputed.

Doubts arise from three different sets of problems: 

(1) Macro problems regarding aggregate vote figures released during
the vote counting period and also regarding the electronic system
employed by the Federal Electoral Institute (IFE);

(2) Micro problems, arising from irregularities in 52,000 of the
130,477 voting stations in the country's 300 electoral districts; and
(3) Environmental problems surrounding the campaigns, especially
forces external to the political parties that may have hurt the
election impartiality, objectivity, legality, and certainty that the
law requires of any election

I. - MACRO PROBLEMS 

Macro problems became evident as early as 8 pm on July 2nd, when the
pollsters, encouraged by the IFE did not ^Ö as was scheduled - release
results from their exit polls.

Rapid count 

During the election on July 2, the PAN's exit polls had established a
lead for FC over AMLO. This created incentive for PAN vote promoters
to call for votes from PRI supporters in districts where votes were
still being cast, under the argument that ^Óthe PRI had collapsed and
the race remained only for AMLO or FC to win^Ô. It also allowed vote
promoters of PAN to seek votes in poor districts, sometimes in
exchange for money. Some cases of vote buying have been documented,
yet no attempt to measure their aggregate impact has been made.

One example was the teachers' leader Elba Esther Gordillo, who called
PRI governors in northern states, urging them to induce votes in
favour of PAN, as the PRI was apparently falling behind. Ms. Gordillo
is on record calling the governor of Tamaulipas, Eugenio Hernández and
asking him to help votes for FC and to ^Ósell^Ô this as a favour to
FC. Documentation of this call now rests with the Federal Electoral
Tribunal in the file submitted by AMLO's coalition calling for
ballot-by-ballot recount.

At 11 pm, the IFE explained that it would not release rapid count
figures because the difference in voter preference between PRD
candidate Andres Manuel Lopez Obrador (AMLO) and PAN candidate Felipe
Calderon (FC) was too close and both candidates had at various times
alternated in the lead.

The absence of rapid count figures and of exit polls alarmed many who
had constantly raised questions about the independence of the IFE's
council from government institutions or parties close to the
government. One case is that of the leader of the teachers' union,
Elba Esther Gordillo, who was instrumental in the appointment of the
council in 2003.

In the absence of rapid count reports, observers' only references
were the most recent polls before the election and exit polls from
firms retained by various parties. In 10 polls previous to the
election, shown in Annex 1 on page 7, six showed AMLO in the lead; one
showed him tied with FC; and three showed FC in the lead. In the
evening of July 2nd, the pollster retained by AMLO (Covarrubias)
obtained the result that AMLO was ahead of FC by 1.2% of the vote; in
the end, this pollster gave him 2.5 percentage points above FC. Two
other pollsters retained by PAN (Arcop and Gea-Isa) showed FC ahead by
2 and 3.2 percentage points respectively.

Both candidates claimed to have won the election based on these polls
as soon as IFE announced that it could not name a winner.

 ANNEX 1LAST POLLS BEFORE ELECTION AND RAPID COUNT STATISTICS 

Percentages of votes 

POLLSTER CALDERON AMLO 
Last Polls: 
Zogby 35 31 
Parametria 33 37 
Indemerc 32 33 
Gea-Isa 41 36 
Marketing Politico 37 34 
El Universal 34 36 
Mitofsky 33 36 
BGC 34 34 
Reforma 34 36 
Demotecnia 31 35 
Rapid Counts: 
AMLO Pollster Rapid Covarrubias) 37.1 34.6 
FC Pollster Rapid (Arcop) 37 35 
FC Pollster Rapid (Gea) 33.4 36.6 

Source: http://www.opinamexico.org/encuestas-electorales.phtml 

The announcement of the IFE at 11 pm was ambiguous. It mentioned that
although FC was ahead of AMLO by 1.0% of the vote in the as yet
unfinished PREP count, it could not take this for granted and that
reliable results would only come from the district count starting on
July 5th. The PREP system was counting all tally sheets as they were
coming from the 300 electoral districts, a process that continued
through July 3rd.

Program of Preliminary Electoral Results (PREP)

The PREP system aroused additional questions and suspicions owing to
three problems:

(1) Early advantage for Felipe Calderón granted by order of vote count
The IFE caused, by the order of its vote counting, an early and high
advantage for FC over AMLO.  This was by counting first votes from
districts in which FC was expected to lead, even though these
districts are in the north. There has not been an answer to why the
IFE did not start counting and reporting the results from Mexico City,
where the early advantage would have been for AMLO.  Questions remain
regarding a possible attempt to create a bias that would benefit FC
once a difference in his favour was established and remained for the
entire count.

(2) Regular difference in percentages of votes for FC steady at 1%
Once an early lead was established for FC in the count of PREP, the
votes accumulated through subsequent vote counting exhibited a regular
pattern of slow convergence: FC percentage was falling, while AMLO's
was rising. After 1 am of July 3rd, the accumulated votes showed a
permanent lead of 1% of the vote in favour of FC, insensitive to the
entry of new districts in which either AMLO or FC were significantly
ahead of each other. This is unlikely to happen in stochastic
processes, such as a close election, where voting reports kept flowing
from districts. These anomalies of the PREP are described separately,
in Annex.

(3) Larger number of votes for Senator than President An interesting
finding is that the number of votes for senator exceeded those for
President by 283,448. This is significant for various reasons:

(a) More people could vote for President than for Senators, as many
special voting stations were placed to allow travelers to cast a vote
only for President at a district other than their home.

(b) There were 40,000 votes for President cast abroad and processed by
IFE.

(c) The absence of any vote for President was more frequent in voting
stations ^Óadjacent^Ô to the main voting station, which suggests vote
shaving in only some stations. Alternatively, the explanation would be
that in many adjacent stations voters rejected all presidential
candidates when they had not rejected them in the main voting station.

(d) In states such as Tabasco, where AMLO was leading, there were
96,450 fewer for President than for Senator. Similarly, senatorial
votes far exceeded the vote for president in other states where AMLO
led, such as Michoacán, Veracruz, Oaxaca, Chiapas, and DF (in total
313,882 less votes in the six states). In contrast, in states where FC
was leading, there were more votes for President than for
Senator. This is the case in Nuevo León (41,290 more votes for
President), Sinaloa, Baja California, Morelos, Sonora, and Chihuahua
(111,178 more votes for president in all seven states).  

(e) In states such as Tabasco and Michoacan, the difference between
votes for Senator and those for President meant that in each voting
station many voters did not cast a ballot for President when they had
done so for senator. In Tabasco this number would have been 41 on
average per voting station. In Michoacan it would have been 11.

(f) Voters in many stations had to stay in line for half an hour, one
hour or even longer before they could cast their votes, suggesting a
motivation to vote. The intense presidential campaign had made the
names of presidential candidates and their personalities well known to
the voters, which was not the case with candidates for senate seats,
whose names are often ignored by the public. The table showing
differences in such votes by state is in Annex 3 below.  Annex 3 Votes
for President and votes for senators in election of July 2nd 2006

Per Sheet Per Sheet 
Votes Votes Difference Sheets Sheets Difference Votes Votes Difference 
State President Senator President Senator President Senator 
Tabasco 718,416 814,866 -96,450 2,374 2,368 6 303 344 -41 
Michoacán 1,313,117 1,368,485 -55,368 5,347 5,329 18 246 257 -11 
Veracruz 2,666,041 2,714,639 -48,598 9,140 9,121 19 292 298 -6 
Oaxaca 1,153,159 1,201,205 -48,046 4,445 4,437 8 259 271 -12 
Distrito Federal 4,690,671 4,723,960 -33,289 12,134 12,100 34 387 390 -3 
Chiapas 1,055,039 1,087,170 -32,131 4,745 4,735 10 222 230 -8 
Guanajuato 1,845,784 1,866,223 -20,439 6,070 6,057 13 304 308 -4 
Puebla 1,888,115 1,902,631 -14,516 5,989 5,975 14 315 318 -3 
San Luis Potosí 848,240 862,698 -14,458 3,071 3,067 4 276 281 -5 
Tamaulipas 1,117,642 1,131,324 -13,682 3,909 3,898 11 286 290 -4 
Yucatán 684,968 698,111 -13,143 2,111 2,108 3 324 331 -7 
México 5,482,214 5,490,474 -8,260 15,304 15,260 44 358 360 -2 
Nayarit 321,583 324,785 -3,202 1,391 1,386 5 231 234 -3 
Quintana Roo 351,542 354,670 -3,128 1,130 1,130 0 311 314 -3 
Campeche 271,606 273,791 -2,185 893 897 -4 304 305 -1 
Zacatecas 486,638 488,741 -2,103 2,382 2,379 3 204 205 -1 
Coahuila 881,567 883,082 -1,515 3,060 3,049 11 288 290 -2 
Baja California Sur 163,554 164,811 -1,257 650 648 2 252 254 -2 
Durango 491,269 490,987 282 2,153 2,140 13 228 229 -1 
Hidalgo 898,121 896,671 1,450 3,035 3,027 8 296 296 0 
Guerrero 824,863 823,025 1,838 4,383 4,372 11 188 188 0 
Tlaxcala 401,489 399,388 2,101 1,252 1,247 5 321 320 1 
Jalisco 2,769,407 2,767,274 2,133 8,029 8,006 23 345 346 -1 
Querétaro 625,686 623,244 2,442 1,790 1,781 9 350 350 0 
Aguascalientes 403,094 400,343 2,751 1,224 1,221 3 329 328 1 
Colima 243,602 239,355 4,247 759 757 2 321 316 5 
Chihuahua 1,019,833 1,014,653 5,180 4,258 4,239 19 240 239 1 
Sonora 837,721 831,501 6,220 2,974 2,965 9 282 280 1 
Morelos 689,629 681,531 8,098 2,049 2,037 12 337 335 2 
Baja California 902,433 890,087 12,346 3,472 3,462 10 260 257 3 
Sinaloa 798,587 760,643 37,944 4,247 4,231 16 188 180 8 
Nuevo León 1,703,721 1,662,431 41,290 5,001 4,985 16 341 333 8 
SUM 38,549,351 38,832,799 -283,448 128,771 128,414 11 287 290 -3 

Source: Elaborated with data from PREP system through 6 pm of July 3rd 2006 

This analysis contradicts some simplistic explanations that have been
offered regarding this difference, which they argue is caused by
voters not liking any of the presidential candidates, but still
showing a preference for a senator.

With regards to the operation of adjacent stations, voter
disenfranchisement, or removal from the voter registration list,
occurs frequently; in some cases by error and in others by design. The
disenfranchised voter, when he attempts to vote at the assigned
polling place, will find that he is not on the list. He is then
directed to so-called ^Óadjacent stations^Ô, where he is allowed to
vote if his voter registration card is deemed to be valid.

There are thousand of incidents reported in the ^Óadjacent^Ô voting
stations. In many stations, large numbers of PAN voters were found to
vote early at adjacent stations, which consumed the supply of
available ballots (750) that were allocated to the voting station. In
many cases, these voters were on the registration list, but merely
went to the adjacent station and used up the ballots.  Subsequent
voters who wanted to cast a ballot were unable to do so for lack of
ballots. These would be those voters who were removed from the voter
registration list, who were predominantly voters for AMLO's
coalition. This has been documented as a result of many notes sent by
frustrated voters.  See (http://www.juliohernandez.com.mx).

Accumulation of doubts on the reliability of the PREP system grew
during the night of July 2nd and early morning of the 3rd, as was
attested by many members of the press and analysts who followed the
process on the web. AMLO indicated shortly after 11 pm that he was
claiming victory based on his own pollster's exit polls. FC claimed
victory based on the still incomplete reading of PREP, which gave him
a 400,000 vote lead. Roberto Madrazo (RM) completely disqualified the
PREP as a reliable count and indicated he should wait for the
following stage, i.e. the recount of tally sheets by electoral
district councils. Developments are described in Annex 4.

District count 

After PREP was concluded, the second stage of the vote count and the
definitive one in the process is the count by the 300 district
councils of all tally sheets. This process started on July 5th, being
the only opportunity for each party's representative to express
doubts on the correct data in tally sheets and consequently to demand
that boxes from certain stations are opened and votes recounted.

District councils rejected most of the demands from AMLO's coalition
to open up boxes.  AMLO's coalition demanded to open 52,000 boxes,
but councils only agreed, frequently with great reluctance, to open
only around 2,873 boxes. AMLO was credited with additional votes for
President in the following cases, although other cases have been
mentioned by commentators:

In Tabasco, 20,000 votes more for AMLO over votes for FC 
In Zacatecas, 6,000 votes more for AMLO over votes for FC 
In Nuevo León, 9,000 votes more for AMLO over votes for FC 
In Durango, 1,200 votes more for AMLO over votes for FC 
In Coahuila, 20,000 votes more for AMLO over votes for FC 

In addition, in Aguascalientes (a state governed by PAN) there were
missing ballots for President in the boxes of 10 voting stations which
the councils examined partially. Differences between votes for
President and those for Senators varied between 94 and 193 votes in
each of the ten voting stations.

District councils, usually under the lead of the representatives of
IFE or PAN refused to allow for more boxes to be opened. Perhaps for
that reason, the recount of tally sheets followed the same pattern as
the PREP. This does not save the PREP, however, from serious
criticisms. For one thing, the PREP had not included 3 million votes
in the count of July 2nd-3rd, which it explained had to be excluded in
order to avoid delays as part of an agreement with all political
parties dating last February. Nevertheless, the IFE had omitted to
mention that these votes were excluded in all its communications on
vote count through July 3rd.

For another, the IFE had stated that the PREP system had a degree of
error of 0.03 percent.  The actual error of the PREP in the case of
votes for AMLO can be established as follows:

When the PREP's count established a lead of FC at 1%: Votes for AMLO
were 13,575,091; When the PREP counted some of the 3 million votes
originally set aside: Votes were 13,624,506 -- (error of 0.4% in
preliminary count); With district count after opening 2,873 ballot
boxes: Votes 14,756,350 -- (error of 8.7% in preliminary count and of
8.3% in final PREP.)


It must be noted that the district count resulted in upward
adjustments in votes for president for all candidates, including
RM. Nevertheless, the adjustment obtained for AMLO was 16.3% higher
than the adjustment for FC and 15.2% than the adjustment for RM,
differences that are significant.

In addition, the percentages of votes reported at the end of district
count (35.89% for FC and 35.31% for AMLO) were calculated over the
number of total votes including annulled votes. If their votes had
been weighted against the total of valid votes only, the shares of
both candidates would have exceeded 36%. If the officially reported
percentages were to be maintained at 35.89% and 35.31%, however, it
might mean that there are still 1.168 million votes still unaccounted
for. IFE has not explained this anomaly or corrected the percentages
accordingly.

Thus, considering many technical allegations on the improbability of
such a regular pattern to occur in such an election and discrepancies
found between tally sheets and ballot boxes whenever these were
opened, the demand to count vote by vote introduced by AMLO has gained
broad support from many parties, including members of other
coalitions. One key issue is that the public has become increasingly
aware of the findings of additional votes that reduce the margin of
difference in favour of FC as any new iteration in the counting of
votes has been done.

While this demand has gained strength in the public's opinion, IFE
representatives had access to ballot boxes sitting in the 300
districts over the past two weeks, boxes which had been sealed after
the district count. IFE representatives opened boxes without legal
authority, without informing public opinion or the political
parties. When they were discovered by observers and AMLO denounced
them on Wednesday July 12th, IFE explained that it wanted to confirm
votes in some cases and to satisfy request of information from the
Electoral Tribunal.

The electoral law prohibits any box being opened in the absence of all
political parties' representatives. AMLO's coalition has confirmed
that the Electoral Tribunal made no such request to the IFE.

Thus, a tentative conclusion is that the district count and the IFE
lack the confidence of at least one major party and of a large number
of voters. On Tuesday July 11th PAN also submitted claims to examine
and eventually annul a large number of ballots from voting
stations. If PAN articulates a case claiming vote or tally sheet
irregularities, then the credibility of the district count will be
further eroded and hence the credibility of the IFE.

II. - MICRO PROBLEMS 

These are numerous problems such as: 

(1) Double counting of votes for PAN in the PREP; i.e. votes in tally
sheets entered twice into PREP.

(2) Differences in votes between tally sheets and PREP figures,
usually against AMLO and in favour of FC.

(3) Identical number of votes for FC in ^Óbasic^Ô voting stations as
in ^Óadjacent^Ô voting stations, especially in districts where FC was
ahead of AMLO.

(3) Large number of annulled votes in districts in favour of AMLO.

(4) Insufficient ballots for voters in districts favoring AMLO.

(5) An inexplicably larger number of votes for President than for
Senators, as was described above.  This seems to happen only in some
states and in some voting stations, but further analysis is needed to
reach conclusions on the nature of this anomaly.  For a sample of
these problems, see Annex 6. It is impossible to determine the bias
that might be attributed to these problems. It is clear from claims
laid by AMLO's coalition that in the majority of cases of ballot boxes
opened, the number of votes is different from than reflected in tally
sheets. This gives grounds to the demand for opening at least 52,000
boxes where irregularities were noted. As this represents 39.8% of all
voting stations, the demand is now simply to open all ballot boxes if
doubts on this election are going to be solved satisfactorily.

One example of how haphazard the decision to open up boxes was during
district count can be seen in the case of a district in
Querétaro. Here the representative from the IFE was against opening
the box, but exceptionally the representative of PAN supported the
petition to open. In this case the tally sheet had listed 200 votes
for FC more than votes were in the box. The tally sheet was corrected
accordingly.

III. - SIGNIFICANCE OF INCONSISTENCIES 

As much information in the media continues to flow and while AMLO has
submitted his case to the Electoral Tribunal, citizens continue to
scrutinize information available on the election and to bring their
findings to interested parties or to the press.

One claim that has been discussed at length already is the suspicious
trends in both the PREP count and the district count, as was described
in annexes 3 and 5. These trends have called the attention of
pollsters and mathematicians, who have suggested that only an
administration of data processing with a particular end in mind would
be capable of yielding the regular difference observed between the
percentage of votes for AMLO and that for FC during the two counts,
i.e. the PREP and the district count.

>From the many anomalies that remain unexplained, the errors exhibited
in data sheet and the likely bias against AMLO found in many vote
recounts, the neutrality and impartiality of the IFE is at a stake.

Not resolving the doubts hanging over the vote counting, the election,
arguably the most hotly-contested in Mexico's history, will lack
credibility and the president elect will find it very hard to govern.

IV. - ENVIRONMENTAL PROBLEMS 

This election was unusually contaminated by the intervention of
external forces which, apart from the political parties, were making
an effort to shape voters' opinions and influence the thinking of
groups, such as business, investors, owners of new houses, and
beneficiaries of social programs in a way favourable to one candidate
(generally FC) and unfavourable to another (generally AMLO).

President Fox's attempts to harm AMLO's public image by discrediting
the merits of his proposed economic strategy and warning against the
danger of changing such a strategy has been established beyond any
doubt. The instances in which Fox referred to AMLO or to his proposals
in a negative way in press conferences or in comments to reporters
have been documented and now rest with the Electoral Tribunal.

In addition, perhaps as a result of President Fox's campaign and
perhaps also because of uncertainty regarding AMLO's economic program
on taxes and tax privileges, the Consejo Coordinador Empresarial (CCE)
ran TV spots that ought to have caused anxiety among the public, as
they referred to the likelihood of currency devaluations, loss of
personal savings, or loss of property for people if the current
economic strategy were to change.

On top of this, PAN conducted the most significant part of its
campaign describing AMLO as a ^Ódanger to Mexico^Ô, comparing him with
Hugo Chávez, President of Venezuela, and warning voters that he would
increase the public debt and cause an economic crisis.

AMLO's coalition asked IFE, who has sufficient authority, to order a
cancellation of these TV campaigns. IFE's council turned down
unanimously AMLO's claim that the ^Ódanger to Mexico^Ô spot was
defamatory and ought to be taken off the air. The spot continued to
run on TV until the Electoral Tribunal determined that it was against
the law. By that time, the negative campaign had already damaged
AMLO's image in the public's eye.

Similarly, IFE reacted tardily to the CCE spot on imminent economic
crisis and requested its removal. The CCE refused until IFE threatened
with a court order, but by the time it was removed, it had caused
damage to AMLO.

IFE on many occasions asked President Fox to refrain from intervening
in the campaigns and from making comments that could be interpreted as
directed against one candidate. Fox continued with his own campaign
and only at times adjusted his behavior temporarily, without changing
his stance or his readiness to comment at any time against AMLO.

On July 3rd the Minister of Communication, Mr. Pedro Cerisola, is on
record for having called the governor of Tamaulipas, Eugenio
Hernández, thanking him for the votes he induced for PAN. Mr. Cerisola
advised the governor to speak with PAN's President, Mr. Manuel Espino,
and hear about another way in which the PAN needed some "help".

On the basis of time consumed on radio and TV, the spending of PAN,
CCE and other business organizations in the negative campaign against
AMLO amounted to $100 million only for a week prior to the
election. This represents a considerable burden against the fairness
of the election, even more so because total spending of PAN exceeded
the limits on campaign spending authorized by IFE. Although in the end
these figures will have to be examined in light of evidence by the
Electoral Tribunal, the impression in many of AMLO's followers is that
external forces should not have been allowed by IFE to damage this
campaign.


V. - THE DEMAND FOR A BALLOT-BY-BALLOT RECOUNT 

The difference of votes counted by IFE and by the district councils in
favour of FC over AMLO is only 243,934. While so many claims exist of
unfairness in this election owing to the intervention of forces
external to the political parties and of those authorities entitled to
take part in it, and while so many doubts arise regarding the
rectitude of the vote counting, it has become necessary to obtain
total certainty regarding the rightful winner.

It is more likely that discrepancies between ballots and tally sheets
exist in poor districts, where voters have low levels of education and
might be unable to follow all instructions to cast a vote properly. As
AMLO support is strongest in these districts, null votes may be
biased. Thus, the demand to recount votes is reasonable.

It is possible that criteria in the voting stations to annul votes
were not applied in a homogenous way across the country. Although the
law indicates that votes must be annulled when voters indicate more
than one preference, many of the annulled votes that AMLO's coalition
is submitting to the Tribunal do not suffer from this anomaly. Thus,
it is possible that citizens who made decisions regarding the votes in
the voting stations were not sufficiently trained to make the right
choices and decisions. Again, this risk is higher in poorer
districts. The high number of annulled votes (2.2% of the total)
deserves further examination in light of the high rate of
participation of people and of the abnormally high share of these
votes in Guanajuato and Nuevo León. Furthermore, the association is
close between a low figure of votes for AMLO and abnormally high
figures of null votes, at least in some states.

A large difference in the amount of resources which the two main
political forces had for this campaign allowed PAN to have
representatives at each of the 130, 477 voting stations. While
responsibility for having assigned representatives falls entirely on
the organization of AMLO's coalition, this does not justify that any
of the other parties including IFE, overlooked irregularities at the
voting station. Given the claim of major irregularities, it is thus
appropriate to ensure that improper practices were not being
replicated on a major scale.

21 July 2006 

List of Annexes 
Annex 1 Last poll before the election and rapid count statistics 
Annex 2 Anomalies in steady difference of votes between FC and AMLO from 1 am of 3 July 
Annex 3 Votes for President and for senators and differences by state 
Annex 4 Chronology of the vote count 
Annex 5 Paradoxes of district count of tally sheets. 
Annex 6 Sample of micro problems affecting many vote districts