1. The sudden appearance of Islamist terrorism in Spain

2. Spanish electoral system and voter behaviour

3. Hypotheses on the electoral turnaround

3.1. Latent change hypothesis

3.2. Terror shock hypothesis

3.3. War hypothesis

3.4. Dual news manipulation hypothesis: by the government and against the government

4. Analysis of the electoral impact of the attacks

4.1. Moment of voter decision

4.2. Acknowledged influence of the attacks

4.3. Voter indecision

4.4. Deviation of voter turnout based on applications for postal vote

4.5. Analysis of absentee vote cast before 3/11

4.6. Post-electoral voter assessment of own vote

5. Conclusions

6. Bibliographical references


The impact of the ‘3/11’ terrorist attacks (March 11, 2004) on the Spanish elections held three days later has been the subject of speculation of all kinds. Twelve months later, the time has now come to compare and contrast the hypotheses that have attempted to explain an unexpected electoral result. New data is provided here to help determine the direction and the magnitude of the influence of the attacks on the March 14 elections.

The main explanations for the unexpected electoral turnaround involve four hypotheses: (1) a latent desire for a change of government; (2) the shock caused by the attacks; (3) a desire to punish the government for its position on the war in Iraq; and (4) a dual manipulation of information – by the government and against the government.

After studying these hypotheses and the changes in voting patterns as a result of the attacks, the conclusion reached here is that the electoral turnaround necessarily involved the confluence of each and every one of the three first hypotheses, while the fourth (manipulation of information in both directions) reinforced them. This means that without a latent desire for change, without the intervention in Iraq, and without the attacks, no electoral turnaround would have occurred on March 14, 2004.

By simultaneously using various research tools, we can determine that the effects of the attacks on the elections were: (1) to motivate one million, seven hundred thousand voters who had not been planning to vote; (2) to discourage another three hundred thousand voters from voting – leading to a four-percent increase in voter turnout; and (3) to ‘convert’ one million one hundred thousand voters.

1. The Sudden Appearance of Islamist Terrorism in Spain

‘All terrorists have something in common:

their actions are never fortuitous or lack sense.’

Bruce Hoffman, Inside Terrorism, p. 195.

The sudden outbreak of terrorism during an electoral campaign is nothing new in the history of democracies. In fact, bacteriological terrorism was first used massively against a civilian population in an attempt to change the results of an election[1]. In Spain, the ETA terrorist organization has always tried to be present one way or another to influence voters’ decisions. For example, 14 days before the general elections in 2000, ETA assassinated the Socialist party leader in the Basque Country, Fernando Buesa; a month before the elections to the Basque regional parliament in 1998, the band declared a ceasefire in order to present a scenario more favourable to its demands; and in the electoral campaign leading up to March 14, 2004, ETA once again made its presence known by announcing a ceasefire in Catalonia only, while planning a new attack in Madrid, which was prevented when the Guardia Civil stopped a van loaded with 500 kg of explosives in the very first days of the campaign.

However, the impact of the March 11 attacks on the elections held seventy-two hours later went far beyond anything known before in Spain, given their brutality and their origin[2]. The resulting 192 deaths were equivalent to a fifth of all terrorist attacks in Spain in the previous 30 years[3]. Spanish society had had to deal with the ‘socialization of suffering’ –the strategy developed by radical Basque separatists, but it had never had to bury so many victims on a single day or exchange final campaign rallies for mass funerals[4].

The fact that the attacks were the work of Islamic terrorists was another entirely new thing for Spanish and European society. The electoral implications of this were very different than if ETA had been the author of the attacks, as all international analysts and news media immediately pointed out on the eve of the election. An ETA attack would benefit the governing party, given its clear accomplishments in the fight against terrorism. However, an attack by Islamists would put responsibility for the deaths on the shoulders of the Popular Party, due to its support of the United States in the war in Iraq, which Spanish public opinion clearly opposed.

This situation introduced a third factor which may have determined the final decision of many voters: confusion regarding who had carried out the attacks led to a climate of confrontation between the government, which insisted almost to very end that ‘the main suspect is ETA’, and the most critical of the news media, which suspected from the start that this was the work of Islamists[5]. The news media were right about some things and wrong about others, but with great dedication they took the lead in analyzing and interpreting the information provided by the government, and reached the conclusion that this terrorist attack was of foreign origin (Nacos, 1994 and 2002; Norris et al., 2003; Palmer, 2003).

The brutality of the attacks, their origin, and the confusion surrounding them, made the March 11 elections a sociological/electoral laboratory observed closely by all western democracies. The research presented here sets out to contribute data to help determine the direction and the magnitude of the electoral impact of the attacks. An evaluation is given of the pre-election and post-election polls that have been published; micro-data from the post-election poll by the Sociological Research Centre (CIS) has been analyzed; and the number of applications for postal voting by residents (CER) has been studied as a predictor of voter turnout, as have the results of absentee voting (CERA) and historical voting trends[6].

Studies of political speeches and of the terrorists’ intentions go beyond the scope of this study[7]. The subject of this analysis is Spanish society and its reaction to the attacks.

According to the preface to the 9/11 Report, terrorists set out to ‘rid the world of political and religious pluralism and the right to the plebiscite’. If this was the case in the attacks in Spain, then clearly, the terrorists did not accomplish their goal; quite the contrary: on March 14, Spaniards once again proved Karl Popper’s maxim that ‘the challenges of democracy are overcome with more democracy’. On that day, above and beyond any specific political choice, democracy itself triumphed —democracy and the Socialist Party (PSOE), which obtained the highest level of popular support in history, in absolute terms, with over 11 million votes, representing one in three Spanish adults. In relative terms, this result was second only to Felipe González’s landslide victory in 1982. The fact that this support did not translate into more seats –the 164 obtained by the Socialists left them 12 short of a majority– was due to the strength of the runner-up, the Popular Party, which obtained nearly 10 million votes, giving them 16 seats fewer than the winner.

The following table shows the results of the nine general elections held in Spain since democracy was established.

Table 1. Voting trends (in thousands)

Postal vote by residents23,58426,83626,84728,86029,43830,64932,00633,03933,475
Ballots cast18,27817,93420,95220,08220,35223,40324,80322,81425,483

Source: Ministry of the Interior.

2. Spanish Electoral System and Voter Behaviour

To understand some of the arguments that follow, a quick review of the Spanish electoral system and voter behaviour is in order. First of all, Spain is a parliamentary monarchy in which the Constitution provides control of foreign and defence policy to the president of the government.

Registry in the voter census is automatic and voting is not obligatory[8]. Parliamentary representatives are elected to the lower chamber (Congress) according to a ‘corrected proportional’ system with 52 districts. Population determines how many of the 350 congressional representatives correspond to each district. At the same time, elections to the Senate are held, with each province represented by the four senators who receive the most support: each voter can choose three senators.

This is a ‘strong electoral system’ –one with a great capacity to guide voting trends in a specific direction. For example, since the first democratic elections in 1977, there has been a trend toward an accumulation of votes for the two main parties, which took 82% of all votes in the general elections of 2004.

Although this is a proportional system, the D’Hont Act on the distribution of seats has tended to compensate for this. Furthermore, the large number of electoral districts makes for a disproportionality more typical of a majority system. This means, on one hand, that small variations in voting patterns can significantly alter the distribution of seats and, on the other hand, rather disproportional representation. This systematically leads to poorer results in general elections for smaller nation-wide parties —traditionally the United Left (IU) coalition and the centrist coalitions.

The election campaign officially lasts 14 days —campaign advertising begins earlier but votes are not expressly solicited— and ends one day before the elections, leaving the so-called ‘day of reflection’ on which no political events of any kind are allowed. In accordance with European tradition, paid television advertising is not allowed, but the publicly-owned media give the parties free advertising spaces. Televised debates among candidates are allowed, but it has been a constant that the party with the best chance of winning refuses to participate. The only debates were held before the 1993 elections. In the last general election, it was the Popular Party that refused to participate in debates that most voters did indeed want. Electoral polls can be published up to five days before elections.

There are two kinds of postal vote: one by Spaniards registered as living in Spain (CER) and one for residents living abroad (CERA). A minimal percentage of voters decide to vote by post –about three of every hundred. Specifically, in the March 2004 elections, of a total 26,155,436 voters, 557,533 residents in Spain and 295,934 Spaniards living abroad did so. This means that 3.3% of voters had voted before the 3/11 attacks. The postal vote by residents in Spain (CER) is deposited in the ballot box where the voter is registered. This happens when the electoral colleges close, making it impossible to know its aggregate result, unlike the absentee ballots (CERA), which are counted in each province the week after the elections.

In terms of voter behaviour, it is worth noting that participation in important elections, such as the national ones, is normally high, oscillating in the 70-80% range. The Spanish political scene is very polarised and voting patterns are quite stable, due their ideological component. Progressive cleavage along class and religious lines and the substitution of left-right ideological orientation by nationalist ideology has done little to make voters more independent.

The following figure shows how voters of different political parties place themselves on the left-right continuum (1 being extreme left and 10 extreme right) and on a scale of national identity (1 being ‘I feel Spanish only’ and 5 being ‘I feel Andalusian, Catalan, a citizen of Madrid, etc.’).

Figure 1. Voter self-placement of political ideology and identity, according to declared past vote

n = 10,000.

Source: CIS analysis, September 2002.

Electoral volatility in the general elections is close to 10% when calculated in aggregate terms (Oñate and Ocaña, 1999). Individual volatility is, of course, somewhat greater, but even in times of political change, less than 20% of voters are affected –and less than 15% when political conditions are stable (Gunther, Montero and Botella, 2002). On this occasion, between the 2004 elections and the preceding ones, aggregate changes, including abstention, affected about 11% of voters. When compared to the 1996 elections, this figure drops to 7%. The main electoral changes are the result of differences in abstention —amplified by the system for allocating seats— and not so much the result of electoral volatility. Changing voting patterns are not caused by sudden crises, but rather are the result of gradual pooling of opinion.

Discoveries made in other countries regarding voter decision-making are applicable to Spain: most people have already decided which way they will vote before the election campaign begins; and polls have little impact on them. Polls, like election campaigns, tend not to capture new votes, but rather strengthen voter loyalty, since most people who change their vote have already decided to do so long before the campaign began.

Voters decide on the basis of contrasting opinions. They withdraw their support from the government not just because they are dissatisfied with its accomplishments, but also because they are convinced that the opposition will do better. Citizens are most critical of the government in the areas where the latter is perceived as having the greatest responsibility. Clearly, there is nothing to suggest that Spanish voting trends are not comparable to those seen in other societies.

3. Hypotheses on the Electoral Turnaround

The results of the March 14 elections were not expected by Spanish voters, who only the week before had anticipated a Popular Party victory that never happened.

Table 2. Prediction of PP or PSOE electoral victory

Which party do you think is going to win the elections?
Citigate Sanchís66.712.2
Metras Seis59.06.3
Opina (Cadena SER)70.312.6
Vox Pública67.86.5
Opina (El País)67.611.3

All these polls were carried out before the March 11 attacks. It is therefore impossible to what extent the final result surprised the electorate as a whole. No studies have been published to this respect, but in any case it would not have been possible to precisely measure changes in opinion during those intense 72 hours. Immediately after the elections, the Observatorio Político Autonómico carried out a post-electoral poll that shows that 64% of Spaniards believed that ‘the PSOE would not have won the elections if the events of 3/11 had not occurred’, compared to 23% who believed the Socialists would have won anyway.

Conjecture on the causes of electoral results that neither public opinion nor analysts predicted before 3/11 can be grouped around the following hypotheses:

·        Hypothesis of a desire for a change of government.

·        Terror shock hypothesis.

·        Punishment for the Iraq war hypothesis.

·        Dual news manipulation hypothesis: by the government and against the government.

Poor physical management of the crisis is not considered a possible hypothesis here. The different administrations (municipal, autonomous community, and national) received more praise than criticism for their actions[9].

As the various hypotheses are dealt with here, reference will be made to certain analyses published immediately after the attacks in the Spanish print media. The most sociologically-focused articles have been selected, although the proximity of the events and the format of opinion pieces aimed at the general public make many of them quite political in nature. Also, an attempt has been made to connect these analyses with the hypothesis most prominent in each one, although few of them are limited to the defence of a single hypothesis. It must also be kept in mind that many of the assertions made in these articles cannot be proven in the articles themselves. In any case, they are written by analysts who know their subject well.

3.1. Latent Change Hypothesis

‘The results are not the ones the people expect, but the ones the people want.’

Noelle-Neumann, The Spiral of Silence, 1984.

The latent change hypothesis maintains that the result of the election was the fruit of the desire of the majority to have a change of government. According to this hypothesis, the attacks did not have a significant impact.

Latent Change HypothesisArguing in favour of this hypothesis is Professor Julián Santamaría, former president of the Centro de Investigaciones Sociológicas (CIS) at the time of the NATO referendum. Two days after the elections, he said to United Press International: ‘I do not believe the Al-Qaeda attacks were a deciding factor, but the circumstances surrounding them edged voting towards the opposition.’ The poll by the Noxa Institute, published in La Vanguardia the previous Sunday was the one that came closest to the actual result, giving the PP a scant two-and-a-half point victory. This institute considered this to be a technical dead heat, and according to Santamaría, this was confirmed by polls carried out the day before the attacks[10].Spain’s current ambassador to Cuba, Carlos Alonso Zaldivar, is of the same opinion. In his analysis for the Elcano Royal Institute, ‘Votes and bombs’, he writes: ‘most of the votes that the PP lost, they lost themselves, and it had nothing to do with the massacre in Madrid.’ He writes that ‘the massacre in Madrid did not, therefore, lead to a significant flow of votes from other parties to the PSOE. What it did do was increase voter turnout in the elections.’Likewise, the Mexican consultant María de las Heras maintains in El País that according to analyses done by the Socialist Party, ‘the only thing the terrorists achieved was to bring one million more voters to the ballot box than would have voted for the PSOE under normal circumstances. José Luis Rodríguez Zapatero and his team would have got 90% of their votes with or without the terrorist attack; the other 10% also came from Socialist sympathizers, most of them concentrated in the provinces where the left was the dominant political force’. (‘Terror did not change voter choices on March 14; it simply radicalized them’, El País, 20/3/2004).By contrast, Pilar del Castillo, minister in the last Aznar government (as well as university professor and former president of the CIS), refutes this hypothesis. In a letter to the editor of the daily newspaper ABC (5/4/2004), while not doubting the legitimacy of the election results, she affirms that there are ‘a set of factors that must be considered in order to understand the various reasons for voter behaviour’.PP leader Juan José Lucas says that much the same thing in an article published in La Razón, in which he maintains that the artificial break in the Popular Party’s hold on power will end up lengthening it, as has occurred in other cases in history.

The latent change hypothesis is based on arguments that focus on electoral results before and after March 14, a desire on the part of a majority of voters to have a change of government, and trends in unpublished polls that indicate that the PSOE was catching up with the PP.

Analysis of the results of local elections held on May 25, 2003 and European elections on June 13, 2004, is not sufficient to either prove or refute the latent change hypothesis. The PSOE won the local and autonomous community elections of 2003 (for the first time since 1993) by half a percentage point. However, two factors must be considered along with these figures: on one hand, the lack of regional elections in autonomous communities that achieved their autonomous status through article 151, which disfavours the PSOE, since voter turnout was lower in the two communities where the party obtains the most votes – Andalusia and Catalonia; on the other, and by contrast, the centre-right vote was fragmented by many independent municipal candidatures – something that does not happen in national elections. The proof is that in the 1999 elections, the PP obtained only 38,000 votes more than the PSOE at the municipal level and 933,000 more at the European level, as the following table shows:

Table 3. Votes (in thousands) and percentage of valid vote0}

June 13, 1999 European elections8,41039.7%7,47835.3%
June 13, 1999 municipal elections7,33434.4%7,29634.3%
May 25, 2003 municipal elections7,87634.3%7,99934.8%
June 13, 2004 European elections6,39341.3%6,74143.3%

Note: the percentage for the 2004 European elections is provisional.

Source: Ministry of the Interior.

Neither do the European elections help validate or refute the latent change hypothesis. The Socialist Party once again won, although voter turnout was much lower than on March 14, 2004: only 46% vs. 77%. But the margin of victory over the PP was more modest, both in relative and absolute terms: their five-percent advantage dropped to just over two percent and the PSOE lost 915,000 voters more than the PP.

The Opina Institute estimated that 59% of voters wanted a change of government. Also, there were generally more voters who wanted a Socialist victory than those who wanted the PP to win.

Table 4. Percentage of voters who wanted each party to win

Metras Seis (Colpisa)31.330.5
Opina (Cadena SER)35.741.2

Based on this data, during the campaign the Socialist candidate alluded to the disparity between the predictions and the desire for a victory at the polls, saying that ‘the results are not those that voters expect, but rather those that they desire,’ adopting the thesis of The Spiral of Silence (Noelle-Neumann, 1984). However, although the data is accurate, it is incomplete and offers a distorted picture of voters’ desires.

In fact, when other factors are included in the demoscopic analysis, the latent change hypothesis is weakened:0} 0{among those who wanted a change of government, only half wanted the PSOE to take power.0}:}0{When voter opinion on a change of government is complemented with voter opinion on which party should replace it, we see that one thing is to want the current government to leave power and another thing is to want the main opposition to take power, as the following table shows:

Table 5. Confidence in the government and opinion on the merits of the opposition

(%)Agree strongly/quite stronglyDo not agree at all/much
The PP merits confidence for another four years in office39,653,3
The time has come for the PSOE to take office again35,254,6

Source: Metra Seis for Colpisa, 20/2/2004.

Changes in voter intention do not make a clear case for the latent change hypothesis either, as we can see in direct voter intention, as measured by the CIS during the past two terms of government. On one hand, we can see a clear recovery in intention to vote for the Socialist Party, accompanied by a decline for the PP until the April 2003 poll, carried out immediately after the start of the war in Iraq. On the other, we see a reversal of these trends after that poll.

Figure 2. Trends in direct intention to vote, as measured by the CIS, 1996-20000

Source: CIS Barometer.

If the study is extended to include all the pre-election polls published in the Spanish media during the 30 days preceding the six days before the election (in accordance with Spanish electoral law), we observe the following differences in predicted vote[11]:

Table 6. Comparison of national pre-election predictions

InstituteAverageFieldSample sizePPPSOEIUCIUPNVOthersDiff.PP/PSOE
Metra-SeisColpisaFeb. 6-165,20042.
GallupFeb. 2-202,03643.935.
DemoscopiaABCFeb. 10-2512,76042.
CISJan. 24 – Feb. 1524,10942.
Instituto OpinaEl País Feb. 27 – Mar. 104,00042.
Instituto NoxaLa VanguardiaFeb. 27 – Mar. 202,20041.439.
Vox PúblicaEl PeriódicoMar. 1-302,07142.537.
Celeste-TelLa RazónFeb. 17 – Mar. 502,40442.937.
Sigma-DosEl MundoFeb. 24 – Mar. 2012,50042.
Final result37.642.

Note: all polls were conducted by telephone, except the CIS poll, which is door-to-door.

Between the final results and the ones expected by the polls there was a deviation of 10.4 percentage points in favour of the PSOE. This is the greatest deviation ever recorded in Spain, despite the fact that demoscopic work is not as accurate as in some countries[12], which often leads to post-election debates on the reliability of polls when predictions are not met. This happened in the general elections of 2000, when the average deviation between the six last polls published and the final results obtained by the two main parties was 4.9 percent. In 1996, there was a 6.4 percent deviation, and 5.6 percent in the European elections of June 13, 2004[13].

In addition to these nation-wide polls, many others provincial polls were published[14]. The 16 polls to which we have had access show an average deviation of 11.3 percentage points in favor of the PSOE.

If we focus on the final days before the election, there is no clear evidence that before the attacks there had been a gradual movement toward the opposition party, as discussed above. The Opina Institute offers the following data:

Figure 3. Trends in predicted vote, as measured by the Opina Institute

Source: Opina Institute, for Cadena SER radio.

The contrast between the polls published a week before the elections and those published when the electoral colleges closed enable us to estimate the change in voting trends after the attacks. The interviews for the polls by Demoscopia and the Opina Institute (for Telecinco TV and Cadena SER radio, respectively) were carried out between March 11 and the evening before the elections. These were the first to detect a change in the final result[15], while the interviews carried out at polling stations by Eco Consulting and Sigma 2 (for TVE and Antena 3 television, respectively) did in fact predict the Socialist victory.

Table 7. Election Day polls

Mar. 11-13T5Demoscopia168-17040.6140-14338.396.5
Mar. 12SEROpina Institute154-16039.5151-15940.59-106.1
Mar. 14TVEEco Consulting150-15436.9154-15841.49-116.3
Mar. 14A3Sigma 2153-16138.5152-159416-75.4
Mar. 14Results14837.616442.655.0

Note: only the Eco Consulting and Sigma 2 polls were carried out at polling stations.

To conclude with our analysis of the latent change hypothesis, it is interesting to find out what voters think: the vast majority of respondents, particularly PP voters, believe that the 3/11 attacks ha an impact on the election results:

Table 8. Opinion on the impact of the attacks on the elections

Do you think that the 3/11 terrorist attacks in Madrid had an impact on the results of the general elections
Vox Pública8596.178.278.6

But what kind of impact did they have? According to 43.3% of respondents, the attacks benefited the PSOE, while another 41.4% said they harmed the PP; 10.5% thought they had no impact (Sigma 2 for El Mundo, 26/4/2004). Vox Pública (El Periódico, 25/4/2004) looked deeper into how the attacks motivated voters. In response to an open question with one single recorded response, respondents offered the following answers:

Table 9. How the attacks impacted the elections

How did they have an impact?(%)
The PP was punished for participating in the war on Iraq18.9
The PP was blamed for the attacks; protest vote14.3
They harmed the PP13.6
They favoured the PSOE12.1
The PP was punished for manipulating information8.2
They made people change their vote7.3
They mobilized people to go out and vote6.1
They pushed undecided voters to make up their minds4.2
The ‘final straw’; need for change3.5
People voted out of fear after the attacks3.4
The PSOE manipulated the ‘day of reflection’3.0
People voted in anger, with their hearts, not their minds1.7
Other answers; don’t know; no answer13.1

Source: Vox Pública in El Periódico, 25/4/2004.

Almost certainly, if respondents had been offered multiple choice answers, they would have given more than one answer. In any case, the war in Iraq is most cited cause of voter motivation and most of the other answers given are not incompatible with this.

In conclusion, the latent change hypothesis cannot be accepted as the only valid one to explain the unexpected election results. However, neither can we dismiss the idea that there was greater desire for a change of government than four years before and that the attacks catalyzed this desire in a significant enough minority of voters.

3.2. Terror Shock Hypothesis

‘Each one of these terrorist acts caused anxiety in society which,

in turn, enabled the terrorists to manipulate the public

with new threats to favour their own short- and long-term goals.’

Brigitte Nacos, Terrorism and the media, 1994, p.69.

According to this hypothesis, the shock caused by the terrorist attacks conditioned voting patterns, leading to a different result than would have occurred without them.

Shock HypothesisTo some degree, the shock hypothesis is present in articles in El País by José A. Gómez Yáñez and Josep M. Colomé. Yáñez affirms that ‘we were all upset when we went to the polls, but for these citizens it was a decisive moment that translated into a current of opinion against the PP’ (‘March 14: storm and swell’, 27/3/2004). Colomé, author of Cómo votamos. Los sistemas electorales del mundo, argues in his article that ‘the 3/11 massacre caused the sudden coordination of several million voters in only 48 hours.’ (Shock coordination, El País, 18/3/2004).In an article in ABC, this shock hypothesis is supported by Edurne Uriarte, professor at the University of the Basque Country —who herself suffered an attack by ETA: ‘Some of us thought that we Spanish were much better prepared than others to face terrorism. We already knew what it was like to resist the ongoing threat, the blackmail, and the fear— especially the political leaders of the PP or the PSOE, who with or without an ‘official car and bodyguard’ have been directly in the line of fire, along with policemen, judges, journalists, and so many others. But no — it seems we were prepared only for the selective and discriminate terrorism of ETA. Now that terrorism has become indiscriminate and massive, we must start again from scratch, and deal with the effects of fear and of our disorientation regarding the causes. A great deal of leadership and education on the meaning of terrorism, and specifically of Al-Qaeda’s — a long and complicated task for the new government and for the opposition’ (‘Disorientation and fear’, 6/4/2004).

The crudest version of this hypothesis is sustained by some international observers who have gone so far as to suggest – with a degree of ignorance of Spanish social and political reality – that if the attacks had occurred in another country, voters would have reacted differently. If any society has shown courage in the face of terrorism, it is Spanish society, which has buried more than eight hundred innocent victims in the past three decades.

However, we must not underestimate the lessons to be learned from the shock hypothesis, as if it were a synonym for cowardice or capitulation to terrorism. After the attacks, eight out of ten Spanish citizens were very concerned or quite concerned about security, and felt intense and far-reaching anger, according to the Royal Elcano Institute barometer done two months after the attacks. Conversations among Spaniards in the days after the attacks were full of statements like this one from a mother in Madrid: ‘My daughter slept badly for two weeks after the attacks. She thought she was living in a safe country and suddenly she discovered this wasn’t true.’ And this was the case even though Spain, along with the United Kingdom, was the European country that had most feared an attack – even more than in the United States immediately after 9/11[16]. The fact the terrorists later attempted another attack on the Madrid-Seville high speed railway line and ended up killing themselves when they were located by the police, magnified and prolonged fear after the elections.

For many years, fear of terrorism has been one of the most widespread concerns among Spaniards, though it had been easing slowly since the summer of 2001. Two weeks after the 3/11 attacks, terrorism was once again mentioned as one of the three most important problems felt personally by 37% of respondents – a figure higher than any resulting from ETA attacks over the years.

Figure 4. Trends in the main problems identified by Spaniards (3 open answers – unemployment, terrorism, public safety, immigration, housing, Prestige affair, Iraq war)

Source: the author, with data from CIS barometers.

Concern about terrorism dropped rapidly after the attacks, following the normal process after catastrophes or large-scale tragedies; Figure 4 also reveals this in the case of the sinking of the Prestige and the war in Iraq. The impact on citizens is different in the three cases, affecting 11% in the case of the oil tanker disaster, 19% in the case of the war in Iraq, and 37% after the 3/11 attacks[17].

The days after the Madrid attacks were full of alarm, a perceived lack of public safety, and fear. There was fear not only of new attacks, but of radical reactions on the part of very diverse sectors of society[18]. There were also appeals to this fear, in line with the scheme studied by Pratkanis and Aronson (1996), who propose that recourse to fear as a propaganda tool is most effective when:

(1) There is a serious shock.

(2) A specific recommendation is offered to overcome the threat that caused the fear.

(3) The proposed measures are perceived as effective to deal with the threat.

(4) The person who receives the message believes he can carry out the recommended action.

Many of the proposals suggested from different quarters met the requirements of persuasion through fear.

It would not be fair to say that there was an emotional reaction to the attacks only in a specific sector of voters who changed their vote, or that this sector simply acted emotionally, with no rational component in their decision on voting. In fact, those who acknowledge that the attacks affected their vote most are precisely those with the greatest ability to reflect on their actions. In any case, it is logical to assume that the issue of public safety was very much in voters’ minds when they made their final decision.

‘Personal security has greater weight than collective security, whether national or international, in the opinions and attitudes of each citizen. To the extent that a citizen feels that the state guarantees his own security, he supports the measures it adopts. By contrast, when these measures are perceived as a threat to his own security, the citizen’s reaction is adverse’[19] (Michavila, 2001). These words, written after the 9/11 attacks in the United States, once again proved true in Spanish society.

In democracies, there is a shared assumption that the state has a monopoly on violence in exchange for guaranteeing the security of its citizens. However, when this security is not perceived, social acceptance may break down and activism may develop outside the legal framework, as occurred in Spain before the election (Sørensen, 2004).

In conclusion, although it is not easy to determine to what extent shock in the wake of the attacks was a determining factor in the change of vote, it cannot be dismissed as a plausible hypothesis.

3.3. War Hypothesis

‘The key is not to be found only in the attacks,

but in participation in the war.’

Josefina Elías, director of the Opina Institute

in ¡Pásalo! by Carlos E. Cué, p. 136.

The war hypothesis maintains that voters punished the Partido Popular government for its support of the U.S. military intervention in Iraq, one of the direct results of which was the Islamist attack on Madrid.

War HypothesisProfessor and editorial writer for El Mundo, Jorge de Esteban, expressed this hypothesis clearly in this daily newspaper: ‘one part of the electorate turned on its heels, making Aznar pay for our highly visible involvement in the Iraq war’ (‘Una victoria inesperada’, El Mundo, 17/3/2004).The president of the Spanish Society of Psychology, Andrés Montero Gómez, holds the same opinion. He interprets voter reaction from a psychological perspective, explaining that voters were not thinking about ‘the message they were giving to Islamic terrorism by changing their vote after the attacks’; rather, they changed their vote because they associated the attacks with the war: ‘a certain extreme sector of Spanish politics used all possible means to spread the idea that Aznar was responsible for the attacks by having got us into a war.’ He says that this was accompanied by ‘the sensation that the government was trying to blame ETA and was avoiding mentioning Al-Qaeda for electoral reasons’ (‘Al Qaida sobre una urna’, La Razón, 24/3/2004).Along the same lines, the chief analyst for the Royal Elcano Institute, Javier Noya, maintains that it was the strategy of the PP government ‘toward public opinion during the Iraq war that created a unique and unrepeatable opportunity for the terrorists’ (‘Del 11-M al 14-M: Estrategia Yihadista, Elecciones Generales y Opinión Pública’).

The polls give a number of clues regarding the connection between the attacks and the Spanish government’s position on the intervention in Iraq. According to a poll by the Opina Institute after the elections, more respondents acknowledged that the war in Iraq had influenced their vote more than the 3/11 attacks. There were significant differences according to political party:

Table 10. Acknowledged impact of the attacks and of the war on voter decision

3/11 influenced their vote27.615.633.429.0
Intervention in Iraq influenced their vote41.817.463.554.8

Source: Opina Institute for El País, 30/3/2004.

Respondents acknowledged that the influence of the war was greater than that of the attacks, especially among those who voted for the left; and this influence was greater than that traditionally acknowledged after other events such as election debates or published polls[20].

We could therefore ask: would voters have held the PP responsible for the war in any case? Surely they would have, as they already had done to some extent in the municipal elections of May 25, 2003, but to a much lesser degree. We base this supposition on the following arguments:

(1)   The war was no longer a personal concern for the Spanish. In March 2003, one of every five people polled by the CIS said that the war was among their three main personal concerns, but a year later less than two percent said the same.0} {However, as a result of the attacks, terrorism became a personal problem for 37% of respondents.

(2)   There is a clear relationship between the war and the attacks for a majority of Spanish public opinion, to the point that 64.2% of those interviewed for the Elcano barometer in May 2004 believed that if Spain had not supported the United States in the war in Iraq, the terrorist attacks would not have happened, versus 23.5% who thought they would have happened anyway. The proportion of those who thought that another international terrorist attack would occur in Spain dropped to 44%. Among the causes of this drop was the perception that withdrawing troops from Iraq had reduced the Islamist threat.

(3)   The acknowledged impact of the war was not so much to change voting patterns, but rather to reaffirm them. This is confirmed by matching past votes with intention to vote. Put another way, most of the voters who acknowledged that the war had influenced their decision reaffirmed a past vote against the PP government. This trend is observed even when the question is formulated in terms not of impact, but of change of vote, as was the case of the CIS barometer immediately after the conflict (CIS 2,508), regarding the local elections[21]. Of those who said that the war could change their vote, 26% had voted PP in the previous general elections – similar to the figure for PSOE voters. Most of those who had voted for the PSOE (67%) reaffirmed their intention to vote for the same party, demonstrating much greater loyalty than those who had voted for the PP (27%).

The government was held responsible so quickly (there were only 72 hours between the attacks and the elections) because the ‘terrorism = war’ equation was already established in a critical sector of society, after more than a year of media presence in the military intervention (Iyengar, 1991). The barometer of the Royal Elcano Institute shows that Spaniards’ concern about any kind of terrorism rose directly as the Iraq war approached:

Figure 11. Concern about Islamist terrorism

To what extent are you concerned about the possibility of Islamic terrorist action in Spain?
Nov. 02Feb. 03
Very concerned1930
Quite concerned4651
Not very concerned2213
Not at all concerned105

Source: Barometers of the Royal Elcano Institute.

With so little time between the attacks and the elections, there was no time to prepare analyses of the problem. As is usually the case in such crises, the news media were busy just keeping up with events. The situation was ripe for conclusions that connected the attacks with the war: ‘So, tomorrow we vote, after all, with the campaign cut short by fear. I suppose we will vote against the war. The word terrorism is now a synonym of war,’ wrote Eduardo Haro Tecglen in El País the day before the elections.

Noam Chomsky (2002) lamented that ‘conciseness does not require support or proof’, in reference to how the television media censored him when he tried to express his opposition to the first Gulf War. At the time, he was given the excuse that his explanations were from Neptune and lacked conciseness. This was also the fate of those who tried to blame only the terrorists for the attacks and who tried to separate them from the war in Iraq. Chomsky’s assertion was once again proven: ‘Either you repeat the same conventional doctrines everybody is saying (with no need to give any proof), or else you say something true, and it will sound like it’s from Neptune.’

Some observers feel that this tendency of part of the Spanish electorate to blame the government for the attacks gives the terrorists a reason to keep on committing them. Blaming the government for what is primarily the responsibility of terrorists is quite common (Nacos, 1994). However, this transfer of responsibility for terrorist acts had not occurred in Spain since the 1970s. Since then, Spanish society had more precisely defined who was guilty for terrorist crimes. For example, in July 1997, when ETA kidnapped PP councillor Miguel Ángel Blanco and threatened him with death if his party’s government did not give in to their demand that ETA prisoners be brought to Basque prisons, Spanish society took to the streets to protest against terrorism, not to pressure the government to save the life of the young councilor.

Table 12. Concern about Islamist terrorism

After the attacks, do you think that Spain safer or less safe from international terrorism?0}
{0La misma (no leer)}0{The same 0}18.0
{0Menos seguridad}0{Less safe0}62.4
{0No sabe}0{Don’t know0}8.8
{0No contesta}0{No answer0}1.1
Are you afraid that radical Islamic groups could carry out attacks in Spain?
Don’t know9.8
No answer1.1

Source: Opina Institute for Cadena SER radio, 24/3/2003.

The connection in the Spanish collective mind between the terrorist attacks and the war in Iraq is associated with the phenomenon of polarization observed in public opinion during wartime (Zaller, 1992; Larson, 1996; Sobel, 2001), which in turn is magnified by the news media. Part of public opinion gives great importance to any information that offers proof that the attacks have no connection with the invasion of Iraq, either because they were prepared beforehand, or because the terrorists had other demands such as the withdrawal of troops from Afghanistan, or because countries that have not intervened have also suffered Islamist terrorism. The media followed by this part of public opinion highlight any news to this effect. By contrast, another part of public opinion is convinced that the attacks are the result of the Spanish government’s support for the invasion of Iraq, and the media this part follows hide news to the contrary.

The final question that Aznar received as president directly connected the attacks and the intervention in Iraq: ‘I don’t know if this is the right moment, because it probably is the victims’ moment, but I would very much like to ask you – I don’t know if this will be the last press conference where I’ll be able to ask you this – if you stand by all the foreign policy decisions you have taken in the past two years, knowing that the intentions of the killers could differ, depending on whether they belonged to one group or another’[22].

3.4. Dual News Manipulation Hypothesis: By the Government and Against the Government

‘The press may not be successful much of the time in telling people what to think, but it is stunningly successful in telling its readers what to think about’.

Bernard Cohen, The Press and Foreign Policy, 1963, p. 13.0}

For some, the government manipulated and hid from voters the evidence of an Islamist hand in the attacks, while promoting the idea that ETA was responsible; for others, news media critical of the government insisted from the night of March 11 that Al-Qaeda was responsible, based on inconclusive or fabricated evidence[23].

News Manipulation Hypothesis‘It was not the attack that turned the election around and gave rise to a resounding Socialist victory. Nor was it the evidence that Al-Qaeda was responsible, nor the connection between the attack and the war in Iraq. It was the irresponsible attempt to hide and falsify the truth that brought everything to a boil, including disgust with the war and the lies about Iraq’ (‘Un prólogo y una visión sobre el 11-M y España’).The specialist in electoral psychology Belén Barreiro maintains this hypothesis in ‘14-M: Y hubo sorpresa…’ (El País, 16/3/2004). ‘In general, everything suggests that the Socialist victory was due mainly to a nearly nine-percent increase in voter turnout. It is a well known fact that in Spain abstention is essentially on the left.’ And she considers the causes: ‘But it is not the attack that helps explains the electoral turnaround, but rather the government’s response to this tragic event.’In the same daily newspaper, Josep María Felip says that ‘we must look for the key in the 24 hours before the elections. A perceptive sector of the electorate sensed that erroneous information was coming out of the crisis cabinet formed in the wake of the barbarous 3/11 attacks. It is not a question of whether or not the spokespeople of the crisis cabinet lied, but rather that the impact of the information in the mass media introduced certain doubts about their credibility. (…) The battle was fought in the last 24 hours on the airwaves and on prime time television.’ (17/3/2004). To demonstrate this, he refers to the Demoscopia poll for Telecinco, which was carried out two days before the elections, and even calculates the change in vote caused by the attacks.As for suspected manipulation of information against the government, the chief editor of the EFE news agency during those critical days, Miguel Platón, offers ten conclusions on how news was handled in the wake of 3/11 in his article in Nueva Revista, ‘Entre la información y la manipulación: Un caso periodístico abierto: el 11-M’.For his part, Mario Noya, in Ilustración Liberal, reviews the information offered by Cadena SER radio during those days, in his article, ‘Tres días de marzo en la Cadena Ser’, trying to uncover some of the key ways information was manipulated against the government.

Of the four hypotheses, the two-way manipulation of information is the one that has generated the most journalism, perhaps because most of the books on 3/11 were written by journalists. With the passage of time, some of the accusations made both against the government and against the news media that were critical of the government are slowly being cleared up, in part thanks to the work of the ‘Commission investigation March 11[24].’

It is beyond the scope of this paper to analyze the veracity of the information offered by the various sources. Once again, our interest here is to determine the capacity of this hypothesis to shed light on the events[25]. The post-election study by Observatorio Político Autonómico inquires in various ways about Spaniards’ opinions of the political information given out by the government in those three crucial days: only 36% believed that the government passed on information as it became available; in response to the next question, 62% said they were convinced that the government hid information for electoral reasons[26]. On the other hand, about 38% of those interviewed thought that the information on TVE (the state-operated television network) was good or very good – the same percentage that held the same opinion of Cadena SER, which was more critical of the government. However, the percentage of respondents who felt that the information on TVE was bad or very bad was several times higher (42% vs. 10%) than those who held the same opinion of Cadena SER. The majority opinion, therefore, is that there was manipulation on the part of the government and also, to a lesser extent, on the part of the most critical news media. However, it is difficult to determine to what extent this belief could affect the vote. The figures provided by the Vox Pública poll carried out for El Periódico, discussed above, reveal that far fewer voters ascribed to this hypothesis[27] than to other explanations. However, the available tools make it difficult to determine its true influence.

Strongly reductive psychological mechanisms of cognitive dissonance and selective perception were at work in people’s decisions about who was responsible for the attacks. After the initial moments when hardly anyone doubted that ETA had achieved its dreaded goal of an attack in Madrid, the first people to accept that the authors were Islamic terrorists were those contrary to the intervention in Iraq. This was true even before they had any reliable evidence, and in some cases they relied on evidence that turned out to be false. Meanwhile, those who justified the military intervention continued to give credit to the idea that ETA was behind the attacks. This collective cognitive dissonance between two large segments of the population carried on for considerable time. Both groups believed they had solid arguments to support their own positions, backed by news media which, except on rare occasions, clearly took one side or the other. Two weeks after the tragedy, some people were still convinced that evidence of suicide bombers would surface, as one radio station announced on the night of March 11. Others continued to firmly believe that ETA was somehow involved in the attacks. The former idea was ruled out by the director of the Instituto Anatómico Forense, while experts consider the latter to be very unlikely.

4. Analysis of the Electoral Impact of the Attacks

In order to try to determine the size and type of the electoral impact of the attacks, we have analysed the pre-election and post-election polls that have been published, as well as micro-data from the post-election poll by the CIS, based on 5,377 door-to-door interviews; and the number of applications for postal voting by residents has been studied as a predictor of voter turnout, as have the results of absentee voting and historical voting trends.

The post-election poll by CIS, unlike the one done during the 2000 elections, does not follow up directly on the pre-election poll: if it did, the effects of the attacks could be analyzed with greater precision. Nor does it include past votes from previous elections, making it impossible to determine the transfer of votes. However, it does include some questions that measure the impact of the attacks on voting.

Regarding the publication of the results of the poll by the news media, the following points are worth noting: none of them mentioned that this was the usual post-election poll that CIS carries out after every election; there was systematically incorrect interpretation of filtered questions, inferring results from the general population that in fact related only to one segment of it; and there was a frequent journalistic tendency to treat public opinion as a block, disregarding the capacity of minorities to influence opinion.

4.1. Moment of Voter Decision

When did voters finally decide to vote for a party or coalition? The great majority (83.6%) made their decision ‘quite a long time ago’, in numbers somewhat higher than in previous national elections. Unfortunately, both the answers given and the structure of the questionnaire differ from one poll to another, which enables only a superficial comparison, without the possibility of coming to conclusive results. What can indeed be determined is that, once again, only a minimal percentage of voters come to their decision in the final weeks of the campaign, though they can be decisive in determining the final results.

There are significant differences in the relationship between the time of decision and the choice of party, as Table 13 shows (the table also includes estimates of the number of voters who made their choices at different times).

Table 13. Moment of voter decision according to choice of party

Quite a long time agoDuring the campaign, before the 3/11 attacksAfter the 3/11 attacksVotesQuite a long time agoDuring the campaign, before the 3/11 attacksAfter the 3/11 attacks

Source: author’s own analysis of the post-electoral CIS poll, nº 2,559; n = 5,377.

The table shows the actual percentages and votes, not what would have happened if there had been no attacks. It is logical to assume, for example, that many of the voters who opted for one party or another in the last three days would have done so anyway. Neither does the table include those who did not vote or, more importantly, those who changed party ‘as a result of the 3/11 attacks and their consequences.’ How many of those who had already decided to vote for one party changed to another? If this were known, we could determine to what extent the pre-election polls deviated from the actual trends in voter decision.

When did they decide not to vote? Of the approximately 7.6 million potential voters who finally abstained, 16.9% made their decision not to vote after 3/11. Of these, 24.4% said that if they had voted, they would have voted for the PP, 17.9% for the PSOE, and 6% for IU. This is not a large enough sample to infer the number of voters who abstained from voting as a result of the attacks, but it does suggest that, as with voter turnout, the tendency was to punish the PP more than the PSOE.

4.2. Acknowledged Influence of the Attacks

The questionnaire includes two questions designed to detect the possible influence of the attacks on voting. The first one asks to what extent it influenced voters and, for those influenced, the second asks what effect this had. Seven out of ten (71.3%) say that ‘the 3/11 attacks in Madrid did not in any way affect their personal decision on how to vote’; the rest, about 7.5 million voters, acknowledge that the attacks had a great deal of influence (10.1%), quite a lot (11.4%), or little (7%).

To establish a parallel with Lazarfeld’s classic scheme in The People’s Choice: How the Voter Makes up his Mind in a Presidential Campaign, we could divide the effects of the attacks into: reinforcement, activation, and conversion of voter decision.

The reinforcement of votes was the main effect of the attacks on voter decision (15% of the electorate), once again proving that ‘reinforcement has been declared the dominant effect of political mass communications’ (McCombs and Shaw, 1972).

There was activation of about 1.7 million voters, generally abstentionists less than 40 years old. Among these, the attacks did not ‘form new opinions, but rather raise old opinions above the threshold of consciousness and decision’ (Lazarfeld, 1944).

Conversion, that is, a change in choice of vote, occurred in more than a million voters, most of them middle-aged. In half of the cases, it was the first time they voted for the party they finally chose. Conversion was especially common among those who acknowledged that the attacks had influenced them a great deal or quite a lot.

These three groups acted in accordance with the three hypothesis that explain the role of the attacks in the electoral turnaround – shock, manipulation, and the war; however, we cannot rule out that the latent change hypothesis acted as reinforcement. The influence of the attacks was not the same for all parties, as the following table shows:

Table 14. Influence of the attacks, by votes per party

Did not change my voteEncouraged me to voteChanged my voteDid not change my voteEncouraged me to voteChanged my voteTotal

Note: ‘did not change’ includes those who were not affected and those who reinforced.

Source: author’s own analysis of the post-electoral CIS poll, nº 2,559; n = 5,377.

Post hoc research of this kind enables us to determine how many votes each party gained, but not how many each one lost due to abstention or change of vote. Based on this, what we can affirm based on the data is that the PSOE was the party that received the most support as a result of the attacks, mainly due to the 951,000 people who were going to abstain, but who finally voted. However, voters who changed their votes had greater impact on the elections, since although there were fewer of them (about 700,000), their votes were lost by competing parties.

These figures are consistent with those offered by Sigma-2, which has the advantage of working with larger samples (since voters are questioned as they leave polling stations), enabling us to determine how votes were transferred from one election to the next. According to Sigma-2, the increase in votes for the Socialist Party came from: one and a half million former abstentionists (a number similar to other parties, especially the Popular Party and IU) and more than half a million new voters. However, the Sigma-2 figures do not enable us to determine what proportion of the votes won by the Socialist Party was due to a latent desire for change and how much was due the other hypotheses.

Table 15. Transfers of votes between the elections of 2000 and 2004, according to Sigma-2 (in thousands)


Source: Sigma-2 for El Mundo, 19/3/2004.

Celeste-Tel for La Razón (21/3/2004) offers somewhat different figures, though pointing in the same direction as those from Sigma-2: the winning party, the PSOE, captured two million abstentionists, 700,000 new voters, and 600,000 former PP voters.

Those who changed their vote: Which party would they have voted for? It is not possible to precisely determine which party they would have voted for, since the CIS poll did not inquire to this respect; nor do respondents indicate who they voted for in past elections. However, the political ideology stated by respondents and, in many cases, their demonstrated difficulty in choosing between two options enables us to glean which option they finally rejected.

To begin with, we observe that those mobilized by the attacks – particularly those who changed their vote – take stances that are closer to the political centre than those who traditionally vote for each party. Table 16 shows average political ideology, according to the classic CIS scale that runs from 1 (extreme left) to 10 (extreme right).

Table 16. Political ideology (by party voted for and by influence of the attacks on vote)

VoteDid not influence meReinforced my choiceEncouraged me to voteChanged my voteTotal

Note: 1: extreme left; 10: extreme right.

Source: author’s own analysis of the post-electoral CIS poll, nº 2,559; n = 5,377.

Since, as we have seen, the PSOE was the party that received the most last-minute support, it is interesting to determine the ideological location of these two groups. More than half of Socialist voters whose votes were not affected by 3/11 identified themselves between three and four on the ideological scale. Those who were encouraged to vote PSOE are either most centrist or do not state any political ideology. Only 8% of the votes are from further left than most of the party’s voters, while 56% are from further right.

Another way of determining the possible impact of the attacks on voter decision is by analyzing the question about voter indecision and how the attacks influenced this.

4.3. Voter Indecision

The classic questions asked in the CIS post-election poll regarding voter decisions are, in this case, especially valuable. An analysis of these questions leads to conclusions coherent with those obtained to this point. 78.5% of respondents expressed no indecision regarding whether to vote or their decision about which party to vote for. Among those who finally voted, this figure rises to 86%. However, there are significant differences depending on the party that finally received the votes, as the following table shows:

Table 17. Indecision about which party to vote for, according to party voted for0}

(%)Firm decision onwhich party to vote forConsideredseveral partiesConsidered one partyor abstentionFirm decision onabstention

Source: author’s own analysis of the post-electoral CIS poll, nº 2,559; n = 5,377.

The PNV and the PP are the parties that had the best defined electorate, while CiU, IU, and especially ERC had the most undecided voters among those who eventually voted for them. In the case of the two Catalan parties, it must be kept in mind that in Catalonia there is a broader selection of parties to choose from; and in the case of IU, there is a certain overlap with the PSOE. Due to the small size of the sub-samples, caution is necessary when drawing statistical conclusions. However, we can see where voter indecision lay, and which parties finally took the vote [28].

Voter overlap is greatest among parties that are ideological the most similar. The PSOE won in all areas disputed, the especially important ones being the competition with the other large party (the PP), with abstentionism, and with the IU coalition. The other party that came out a winner is ERC, while the others were only able to ‘beat abstention’, except the PP, which was not even capable of that.

As a group, the Socialist voters that showed the greatest indecision regarding whether to vote PP were the youngest ones, those with the highest levels of education, and those living in cities with 50,000-400,000 inhabitants. For all the parties, these were the voters who were the most indecisive. By contrast, there were no significant differences in terms of gender, religious practice, interest in the campaign, awareness of polls, etc. But it was mainly the voters who said they had been influenced by the 3/11 attacks who showed the greatest indecision; and the more they acknowledged the influence of the attacks, the more undecided they tended to be. As Figure 5 shows, 10% of PSOE voters who acknowledged that 3/11 had had great influence on their decision were undecided about whether to vote for the Socialists or for the PP. This indecision stood at only 2% among those who said the attacks had not influenced them at all.

Figure 5. PSOE voters who considered voting PP, according to impact of 3/11Green: did consider; Red: did not consider; Impact: a great deal, quite a lot, not much, none.

Source: author’s own analysis of the post-electoral CIS poll, nº 2,559; n = 5,377.

Of the Socialist voters who say that 3/11 made them change their vote, 31.4% considered voting for the PP, as did 37.1% of those who decided after the attacks to vote for the PSOE. In conclusion, the relationship between final voter choice and the impact of the attacks is statistically significant[29].

Figure 6. Final decision of undecided voters (PSOE vs. PP), according to influence of 3/11(No influence / Reaffirmed my vote / Encouraged me to vote / Changed my vote) (PP, PSOE, other parties, blank or spoiled ballot, did not answer)

Source: author’s own analysis of the post-electoral CIS poll, nº 2,559; n = 5,377.

Most Socialist voters who changed their vote or were encouraged to vote had already voted for this party on other occasions. These are not, therefore, true conversion, except for one in four voters who opted for this party for the first time. In any case, not everyone who opted for the PSOE for the first time did so because of the attacks. In fact, nearly half (48.7%) of new Socialist voters over the age of 22 did not acknowledge that 3/11 affected their vote, while another 16.2% said it reaffirmed their decision.

One of the effects of the attacks most difficult to measure is their influence on abstentions. Of those who abstained, 16.9% said they made their final decision not to vote after March 11; some did not decide until election day. Of these, 24.4% say they would have voted for the PP, compared to 17.9% who would have voted for the PSOE.

Finally, let us compare the motivations of the voters who acknowledges the influence of the attacks and those of the voters who say the attacks did not influence their vote. We will do this for the three main parties, based on a specific question with a single, free response in each case[30].

PSOE voters not influenced by the attacks expressed ideological motives: ‘It’s the party that best represents the ideas of people like me’ (31.6%); ‘I always vote for this party’ (30.3%); or ‘So that there could be a leftist government’ (8.7%). There is hardly any difference between the motivations of voters who were encouraged to vote by the attacks and those who changed their vote to the PSOE for the same reason: essentially, both were motivated ‘by the 3/11 attacks in Madrid and their consequences’ (48.4%) or ‘to prevent the PP from winning’ (25.2%).

PP voters not influenced by the attacks refer to practical motivations: ‘(The party) has done well in government’ (35.4%) or ‘It is the party best equipped to govern’ (21.1%). We have already determined that only a relatively small proportion of voters who chose this party did so because of the attacks[31]; now, this question shows us that their motivations were similar to those who were not influenced by the attacks: ‘(The party) has done well in government’ (26.8%), or ‘It is the party best equipped to govern’ (12.5%); although 19.6% say they were motivated ‘by the 3/11 attack in Madrid and its results’, and 9% ‘to prevent a coalition government between the PSOE and IU’.

Finally, it is among IU voters that we see the greatest distance between the voters who were influenced by the attacks and those who were not. Among the former[32], a third say they voted ‘because of the 3/11 attack in Madrid and its consequences’, while for another third it was ‘to prevent the PP from winning.’ Those who were not influenced by the attacks voted for IU ‘because it is the party that best represents the ideas of people like me’ (50.6%) or ‘so that there could be a leftist government’ (16%).

4.4. Deviation of Voter Turnout Based on Applications for Postal Vote

The number of applications for postal voting by residents (CER) is a reliable predictor of voter turnout in each province[33]. The 557,533 applications for postal vote represented an increase (based on a deflated census) of 12% over the year 2000, and 5% over the municipal elections of 2003; the figure was down 4% from 1996.

{Mathematically, this figure would imply a considerably higher turnout than four year before, when it was unusually low (70.0% in 2000, 78.1% in 1996, and 76.9% in 1993), but in any case lower than 74.5%, even taking into consideration the atypical behaviour of the Basque Country and Navarre. In the end, 77.2% of voters cast ballots, meaning that the 3/11 attacks raised voter turnout by more than three percent.

In the district of Madrid, where the attacks occurred, a 78% turnout was predicted, based on applications for postal vote (CER) – nearly five percent higher than in 2000, but 2.7% lower than the 80.7% of voters who finally turned out. This was the second highest turnout in Spanish history, second only to 1982.

Table 18. Predicted voter turnout in Madrid, based on applications for postal votes

YearElectionsPostal VoteApplicationsVoter turnoutPredictionError
1991Autonomous communities3,805,48039,09458.7458.790.05
1995Autonomous communities4,082,01556,91670.2070.740.54
2003Autonomous communities4,308,65568,14669.0368.24-0.79

Model: linear regression of voter turnout at 8 am, based on the list of applications for postal vote (CER) in reference to the CER census and the year.

R2 = 0.951 of the general elections model; R2 = 0.929 of the general and autonomous communities model.

4.5. Analysis of Absentee Vote Cast Before 3/11

The voting patterns of the more than one million Spanish residents who were out of the country – referred to in the news media as the ‘emigrant vote’, but more correctly known as the CERA vote (based on census of absentee residents) – cannot be extrapolated to residents in Spain. However, the fact that these votes were cast before March 11 increase their importance as an indicator of trends. Trends in the number of votes for the two main parties is shown in the following figure:

Figure 7. Trends in estimated vote by absentee Spanish residents (CERA)

Source: Ministry of the Interior.

Not only did the PSOE not gain ground on the PP in the Spanish emigrant vote, the gap actually widened from 2.1% in 2000 to 8.8% in 2004. The PP received 44.9% and increased the number of votes it received from emigrants from all autonomous communities, while the PSOE, with 36.1%, held onto its vote in Catalonia, lost votes in Andalusia, Extremadura, and Murcia, and made gains in the rest of the country, especially in Madrid, Galicia, and Asturias. Comparisons of CERA and CER trends in the votes received by the two parties reveal the greatest disparities in Catalonia, Andalusia, and the Basque Country. In these three communities, the PSOE’s results on March 14 were significantly better than the absentee residents’ votes suggested.

There is also certain deviation between the final number of votes and the number predicted by the CERA votes: at the last moment, IU and BNG lost votes, while ERC and PNV made significant gains, as did CiU and CC, though on a smaller scale.

4.6. Post-electoral Voter Assessment of Own Vote

Most people who were encouraged to vote or who changed their vote because of the attacks stood by their decision later. This is not, therefore, a vote they regret, as may be expected in the normal workings of cognitive dissonance reduction mechanisms. Nonetheless, their loyalty is not as great as that of those who felt reaffirmed in their decision or who were not influenced; in fact, one out of four voters who changed their vote because of the attacks would have acted differently if they had known what the final results would be.

Table 19. Voter reaction if they could have known the results

Would have voted for the same partyWould have abstained or deposited a blank ballotWould have voted for a different partyWould have voted
Did not influence me85.510.50.91.1100
Reinforced my choice96.
Encouraged me to vote82.
Changed my vote70.09.715.90.5100

Note: the total accounts for all respondents.

Source: author’s own analysis of the post-electoral CIS poll, nº 2,559; n = 5,377.

As we can see, those who felt that the attacks reaffirmed their vote are those who show less ‘regret’, while three out of ten ‘converts’ would have done something different if they had known what the results were going to be. Regret among converts is lower among those who voted for the two main parties, affecting 25% of those who voted for the Socialists and only 20% of those who voted PP. Those who were encouraged to vote show an intermediate level of regret (14.7%). Once again, regret is lower among Socialist voters (11%) and PP voters (15%).

Table 20. Voter indecision, according to party voted for

VoteFirm decision to vote for a particular partyConsidered several partiesConsidered a particular party and also abstention0}Firm decision to abstain0}NA
Blank/spoiled ballot12.38.213.753.412.3100

Source: author’s own analysis of the post-electoral CIS poll, nº 2,559; n = 5,377.

Voter reaction, if they had known the results beforehand, would have been different depending on the party they voted for. Once again, the PSOE would have lost the most voters, who mostly would have abstained or else voted for the PP or IU. However, it is also the party that would have received the most support if the election results had been known in advance: especially from PP and IU. The PP, and to a lesser extent the PSOE, would also have lost voters to abstention or to the Socialists if they had known the election results in advance. This would have compensated the votes received by the PSOE and by CiU.

5. Conclusions

All research confirms the suspicion of most Spaniards: the attacks in Madrid on March 11, 2004, had a decisive impact on the elections held three days later. This impact, though relatively small, was a determining factor that changed the final result.

We can attribute to the attacks the ‘activation’ of 1.7 million voters who were motivated to vote by the attacks and by the atmosphere surrounding them. This, minus the 300,000 potential voters who abstained for the same reason, meant a four-percent increase in voter turnout. There was relatively little change in voting patters – involving something over a million votes – but the effects of this were great, as voters switched from one party to another.

The concurrence of these phenomena in the changes in voter intention during those 72 hours can be seen in Figure 8, which shows the results of polls by interview date:

Figure 8. Vote estimated by pre-election polls

Note: average of the polls cited in section 3.1 of this paper: before March 11, nine pre-election polls; March 12-13, polls by Demoscopia and Opina for Telecinco and Cadena SER radio outside polling stations, polls by Eco-Consulting and Sigma Dos for TVE and Antena 3.

As for the four hypotheses that explain the unexpected electoral turnaround – existence of a latent desire for a change of government; the shock created by the attacks; punishing the government for its participation in the Iraq war; and manipulation of information by the government and against the government –, we can affirm that these hypotheses are not exclusive, but rather complementary. The first three are necessary preconditions for the electoral turnaround: the shock caused by the attacks activated a backlash against the position taken by the Spanish government in the war in Iraq, and this backlash activated a latent desire for change in a segment of voters who became the deciding factor in the election results. Or to look at it the other way around: without a latent desire for change, without the Spanish government’s support for the war in Iraq, and without the shock caused by the attacks, the change of government would not have occurred. The two-way manipulation of information, by the government and against it, tended to reinforce the process described above.

Narciso Michavila

Professor of Public Opinion and Security at the Instituto Universitario General Gutiérrez Mellado of the UNED

6. Bibliographical References

Barreiro, Belén (2002), ‘La progresiva desmovilización de la izquierda en España: Un análisis de la abstención en las elecciones generales de 1986 a 2000’, Departamento de Ciencia Política y Relaciones Internacionales, Universidad Autónoma de Madrid, WP 5/2002, Madrid.

Barreiro, Belén (2004), ‘14-M: elecciones a la sombra del terrorismo’, Claves de Razón Práctica, nº 141, pp. 14-22.

Bennett, W. Lance, and David L. Paletz (eds.) (1994), Taken by Storm: the Media, Public Opinion and US Foreign Policy in the Gulf War, University of Chicago Press, Chicago.

Boix, Carles, and Clara Riba (2000), ‘Las bases sociales y políticas de la abstención en las elecciones generales españolas: recursos individuales, movilización estratégica e instituciones electorales’, REIS 90, pp. 95-128.

Buzan, Barry (1991), People, State and Fear. An Agenda for International Security Studies in the Post-cold War Era, Longman, Essex.

Chomsky, Noam (2002), Opinión Pública y Propaganda. Conversaciones con David Barsamian, Crítica, Barcelona.

Everts, Philip (2002), Democracy and Military Force, Palgrave, New York.

Everts, Philip, and Pierangelo Isernia (2001), Public Opinion and the International Use of Force, Routledge, Cambridge.

Font, Joan, and Guillem Rico (2003), ‘Spanish Multilevel Turnout: Learning to Vote?’, Departamento de Ciencia Política y Relaciones Internacionales, Universidad Autónoma de Madrid, WP 11/2003, Madrid.

Gunther, Richard, and José Ramón Montero, ‘The Anchors of Partisanship: A Comparative Analysis of Voting Behaviour in Southern European Democracies’, Instituto Juan March, Estudio/Working Paper 150/2000, Madrid.

Gunther, Richard, José Ramón Montero and Jose Ignacio Wert (1999), ‘The Media and Politics in Spain: from Dictatorship to Democracy’, ICSP WP 176/1999.

Hoffman, Bruce (1999), A mano armada. Historia del terrorismo, Espasa, Madrid.

Huddy, Leonie, et al. (2002), ‘Reactions to the Terrorist Attack of September 11, 2001’, Public Opinion Quarterly, vol. 66, pp. 418-450.

INCIPE (2003), La opinión pública española y la política exterior, INCIPE, Madrid.

Instituto Español de Estudios Estratégicos (2003), Seguridad y defensa en los medios de comunicación social Cuadernos de estrategia nº 119, Madrid.

Jerez, Ariel, Víctor Sanpedro and Alejandro Baer (2000), Medios de comunicación, consumo informativo y actitudes políticas en España, Centro de Investigaciones Sociológicas, Opiniones y Actitudes nº 32, Madrid.

Lazarfeld, Paul, et al. (1944), The People’s Choice: How the Voter Makes up his Mind in a Presidential Campaign, Sloan and Pearce, New York.

Lesser, I., et al. (1999), Countering the New Terrorism, RAND, Washington.

Linz, Juan J., and José Ramón Montero (1999), ‘The Party Systems of Spain: Old Cleavages and New Challenges’, Madrid, Instituto Juan March, Estudio/Working Paper 138/1999.

Lippmann, Walter (1997), Public Opinion, Simon and Schuster.

Malesic, Marjan (ed.) (2000), International Security, Mass Media and Public Opinion, University of Ljubljana, Ljublana.

McCombs, M.E., and D.L Shaw (1972), ‘The Agenda-setting Function of Mass Media’, Public Opinion Quarterly, 36, pp. 176-187.

Meulen, Jan van der (2003), ‘Public Opinion and the Role of the Media’, in Jean Callaghan and Franz Kernic (eds.) (2003), Armed Forces and International Security. Global Trends and Issues, Lit, Munster, pp. 299-302.

Michavila, Narciso (2004), ‘Spanish Public Reaction after the Madrid Bombings’, presented at the meeting of the Research Committee on Armed Forces and Conflict Resolution of the International Sociological Association, Ankara, July 7, 2004.

Michavila, Narciso (2002), ‘Opinión pública y conflictos armados. Claves para el análisis’, Revista Sociedad y Utopía nº 19, pp. 163-166.

Michavila, Narciso (2001), ‘Opinión pública mundial tras el 11-S’, presented at the international conference on security studies, Granada, in November.

Miller, J., et al. (1999), Guerra bacteriológica. Armas biológicas y la amenaza terrorista, Ediciones B, Madrid.

Molas, Isidre, and Oriol Bartomeus (2001), ‘Estructura de la competencia política en España (1986-2000)’, Institut de Ciències Politiques i Socials, WP 196/2001, Barcelona.

Montero, Jose Ramón, and Mariano Torcal (1994), ‘Value Change, Generational Replacement and Politics in Spain’, Instituto Juan March, Estudio/Working Paper 56/1994.

Nacos, B.L. (2002), Mass-mediated Terrorism. The Central Role of the Media in Terrorism and Counterterrorism, Rowman and Littlefield, Oxford.

Nacos, B.L., Robert Y. Shapiro and Pierangelo Isernia (2000), Decision Making in a Glass House. Media, Public Opinion and American and European Foreign Policy in the 21st Century, Rowman and Littlefield, Boston.

Noelle-Neumann, Elisabeth (1995), La espiral del silencio. Opinión pública: nuestra piel social, Paidos, Madrid.

Norris, Pippa, Kern Mantague and Marion Just (eds.) (2003), Framing Terrorism: The News Media, the Government and the Public, Routledge.

Oñate, Pablo, and Francisco A. Ocaña (1999), Análisis de datos electorales, Centro de Investigaciones Sociológicas (CIS), Madrid.

Page, Benjamin I., and Robert Y. Shapiro (1992), The Rational Public. Fifty Years of Trends in American’s Policy Preferences, University of Chicago Press, Chicago.

Pratkanis, Anthony, and Elliot Aronson (1994), La era de la propaganda. Uso y abuso de la persuasión, Paidos, Madrid.

Sobel, Richard (2001), The Impact of Public Opinion on US Foreign Policy since Vietnam, Oxford University Press, Oxford.

Sørensen, Henning (2004), ‘Security and Terrorism: A Delicate Balance at the National and Individual Level’, presented at the conference of the RC Armed Forces and Conflict Resolution of the International Sociology Association, July 7-9, 2004, Ankara.

Torcal, Mariano, and Chiber Pradeep (1995), ‘Elites, cleavages y sistema de partidos en una democracia consolidada: España (1986-1992)’, REIS 69, pp. 7-38.

Zaller, John R. (1992), The Nature and Origins of Mass Media Opinion, Cambridge University Press, Cambridge.

[1] In 1984, the Rajneesh sect attempted to take control of Wasco County, Oregon, by poisoning the general population. A total of 751 people suffered salmonella poisoning. The result was the opposite of that sought by the terrorists: ‘On the contrary, the residents of Wasco County, realizing that their town was in danger, registered and voted en masse. The number of voters in the elections of November 1984 was, proportionally, the highest in the history of Oregon. The Rajneeshee candidates were defeated’ (Miller, 1999, p. 39). However, it took more than a year to discover the origin and motivation of this mass poisoning.

[2] During the twelve months after the attacks, about thirty books focusing on this subject were published in Spain. Also, the ‘Commission Investigating March 11, 2004’ offers a large amount of unpublished information on the attacks. The transcript of the hearings is available on the Internet at the website of the Spanish congress:

[3] Information on the victims of terrorism can be found at the website of the Asociación de Víctimas del Terrorismo (AVT),, and at the website of the Basque government (Homeland Office),

[4] Among the dead were 47 foreigners of 14 nationalities: 16 Rumanians, five Ecuadorians, four Bulgarians, four Poles, four Peruvians, three Moroccans, two Hondurans, two Colombians, one Brazilian, one Chilean, one French, one Cuban, and one Guinean.

[5] On April 19, 1995 – the same day that ETA attacked the then leader of the opposition, Jose María Aznar – a bomb killed 169 people in Oklahoma. The first two days after the massacre, Americans, including anti-terrorist experts, thought it was the work of an international Islamist group, as had been the case in other countries. The British newspaper Today even published a cover photo showing a fireman carrying a dead child, with the caption: ‘In the name of Islam’. Two days later – thanks to evidence provided by police – the news was released that the massacre was the work of an American extremist, Timothy McVeigh.

[6] A certain amount of information from the above mentioned sources can be found at the website of the Center for Sociological Research (, the Office of the Electoral Census of the National Statistics Institute (, and the database of electoral results of the Department of the Interior (

[7] This is dealt with rigorously by Javier Jordán in Profetas del miedo, published a month before the attacks, and in a collection of analyses by Antonio Elorza and Fernando Reinares in El nuevo terrorismo islamista. Del 11-S al 11-M.

[8] Cf. Organic Law 5/1985, of June 19, on the General Electoral System.

[9] Some of the figures provided by the Community of Madrid give an idea of the extraordinary response to the attacks. Two hours after the first bomb went off, 291 ambulances were working to take the wounded to 19 hospitals in Madrid. There, they attended to 1,430 wounded in the first nine hours and performed 95 surgical operations. That morning, 162 fire department vehicles were at the four sites of the tragedy. Before the end of the first hour, six mobile blood donor clinics were operating to accept the donations of the thousands of Madrileños who immediately queued up. At the same time, the government of Catalonia sent 500 bags of blood plasma. On the first day, the 112 emergency telephone number answered more than 20,000 calls, with an average response time of 30 seconds. In the first hours, all the examining magistrates in Madrid were put on duty to inspect the bodies of the victims for removal. Eighty-three forensic surgeons from Madrid, plus others who came immediately from Galicia and Catalonia identified 120 bodies in the first 24 hours. The psychological attention unit of the Community of Madrid made more than 3,000 home visits in the first 48 hours, putting 450 professionals on active duty and another 1,200 on call. Another 729 trained volunteers aided the families of the victims. The archbishop made as many priests as necessary available to city officials. After 24 hours, 80% of local transport services were operating again. Taxi drivers, hotel operators, travel agencies, airlines, and department stores immediately offered their help to provide whatever the families of the victims needed.

[10] Of the published polls, Noxa’s was the one that came closest to the final result, deviating only 7.2 points against the PSOE and in favour of the PP. Until now, Noxa’s predictions have always deviated in the other direction –0} 0{overestimating the PSOE’s vote and underestimating the PP’s: by 13.8 points in the Barcelona municipal elections in 2003; 7.5 points in the Catalan regional elections in 2003; and 6.7 points in the European elections in 2004.

[11] Article 69 of Organic Law 5/85 on the Electoral System prohibits the publication or broadcast of electoral polls by any means of communication.

[12] In US presidential elections, the average deviation between predictions and results was 1.1% in November 2004, 2.2% in 2000, 4.1% in 1996, and 2.2% in 1992. (Data from the National Council on Public Polls).

[13] The polls by El Mundo and El País were closer, with deviations of 2.7% and 3.5%, respectively, while the CIS was further off the mark, with a 7.5 percentage deviation.

ANOVAEl Correo GallegoMar. 7La Coruña45284439,112,3
GPSBasque government605Feb. 21Álava3327,12730,89,7
A+M0HeraldoFeb. 15Aragón42403741,36,8
Vox PúblicaLa VozMar. 6Asturias4637,94443,27,2
InvesmarkEl Comercio1.200Asturias48384443,29,1
Metras SeisEl Periódico1.000Mar. 5Castellón5334,74644,616,9
Vox PúblicaDiario de CórdobaFeb. 28Córdoba43443449,814,6
GPSBasque government962Feb. 21Guipúzcoa22241526,39,3
AppendLa Rioja1.060Mar. 7La Rioja}58365044,116,3
KD CreativaCanarias Ahora1.200Mar. 7Las Palmas4332,342341,9
Sigma-20Diario de AvisosMar. 7Las Palmas4825,6423414,2
T sociologíaDiario de NoticiasMar. 7Navarra43273833,612,3
KD CreativaCanarias Ahora1.200Mar. 7Tenerife39262935,819,6
TSALa Opinión1.750Mar. 7Tenerife3529,42935,812,5
Sigma-2Diario de AvisosMar. 7Tenerife3628,32935,814,8
GPSBasque government1.260Feb. 21Vizcaya1923,51926,73,5
Average deviation11,3

[15] This was also picked up by the polls done for the PSOE by Obradoiro de Sociología, according the PSOE campaign manager, Juan Campmany, in his book, El efecto ZP, Barcelona, Planeta, 2005.

[16] The answers to the Gallup International poll in September 2002 to the question, Do you think there could be terrorist attacks in (your country) in the coming weeks? were as follows:

{0España}100{Spain0}{0Canadá}100{Canada0}{0Gran Bretaña}0{Great Britain0}{0Italia}100{Italy0}{0EEUU}100{USA0}
{0Lo creo muy posible}0{I think it’s very likely0}20113212
{0Pudiera ser posible}0{It’s possible 0}426281548
{0No demasiado}0{It’s not likely0}1639394028
{0No lo creo en absoluto}0{I don’t think so at all0}95216389

[17] The pattern of sudden rises and falls in citizen concern is also documented in other countries. In the United States, after the hijacking of the TWA airplane in 1985, 13% of US citizens identified terrorism as the main problem in their country; and it was also the problem most often mentioned. Six months later, this figure dropped to one percent (Hinckley, 1992). During the Iran hostage crisis, five years earlier, the proportion of people who identified terrorism as the main problem was directly related to the volume of news on the crisis in Teheran (Nacos, 1994).

[18] Cf. presentations by the working group on political sociology of the Spanish Congress of Sociology held in Alicante in September 2004.

[19] Cf. Buzan, 1991.

[20] In 1993, 19.8% of respondents (CIS number 2,061) acknowledged that the televised debates had a great deal or quite a lot of influence on their own decision to vote. About 3% of the electorate generally acknowledge that polls influence them (CIS number 1,842 and 2,210).

[21] Responses to the question in the CIS study (2,508, April 2003): ‘And, finally, could the war in Iraq change your vote in the next municipal and regional elections in May?’:

Did not vote1875

[22] The question was put to him by Gabriel Sanz, a journalist working for the Servimedia news agency. It was Aznar’s final news conference after the council of ministers met on March 12, 2004. Aznar answered: ‘You’re right when you started your question by saying that perhaps this is not the right moment. This is the moment for dealing with the things I was talking about before. Thank you very much and good afternoon.’

[23] Both versions of the manipulation hypothesis were dismissed by President José Luis Rodríguez Zapatero, who in his first interview with El País after the elections said: ‘Q.: Did the PSOE demand that the government inform the public about who carried out the attacks? A: That Saturday I asked the Minister of the Interior to make a public statement giving the most possible information both to the PSOE (or to me as president of the government) and to the public (…). I must say that the Minister of the Interior was very reasonable in his direct relations with us, both before and after. Q.: And was it because of that demand that the Minister of the Interior decided to appear publicly to explain the first arrests of suspects? Q.: I think that by then the decision had been made to provide explanations. My assessment is that they couldn’t withhold that information.’

[24] See note 2.

[25] The message that called for the illegal demonstration in front of Partido Popular headquarters the night before the elections is in line with the manipulation argument: Aznar scot free? They call it a day of reflection and Urdaci is working? Today 3/13 at 6 pm, PP HQ, 13 Genova St. No parties. Just the truth. Pass it on!).

[26] Three months later (June 28, 2004), the Opina Institute (for Cadena SER radio), asked questions about manipulation of the news by TVE. These results were obtained: Do you think there is more or less manipulation of information on TVE now? More: 14.2%; same as before: 39.2%; less: 28.2%; DK/NA: 17.4%.

[27] These were the figures given: 8.2% said manipulation by the government was the main cause of the electoral turnaround; another 3% said it was due to manipulation by PSOE on the day of reflection.

[28] The double table shows the number of respondents who say they had trouble deciding between these two parties and the party (or abstention) they finally chose.


Source: author’s own analysis of the post-electoral CIS poll, nº 2,559; n = 5,377.

[29] With 113 available cases and a degree of accuracy higher than 99%, we must dismiss the hypothesis that there was no connection between the vote finally chosen and the impact of the 3/11 attacks. The Pearson chi-square value is 19.804 (p-valor = 0.000).

[30] The question is: What is the main reason that led you to vote for [party or coalition of choice] in these past general elections?

[31] In the sample, n = 56.

[32] In the sample, n = 20.

[33] It is impossible to determine the influence of the attacks on voting patterns by comparing the vote on election day with the postal vote by residents (CER), because these ballots are put in the ballot boxes where each voter is registered and are counted along with the rest of the ballots.