Twitter as tool for political analysis

Twitter analysis is becoming a source of knowledge in many areas. Right now many scientists and professionals are using it for better understanding and measuring corporate and institutional reputation. There are also attempts to monitor election processes through Twitter analysis.

This kind of studies are perfectly in line with all our empirical studies based in news measurement and content analysis, applied to the field of sport brands (Economics, Sport and Intangibles research group, ESI), and more widely, to companies and politics (Media, Reputation and Intangibles center, MRI).

A tweet, and tweet content analysis has its own rules and social community rules, but it share some points in common with news and news content analysis. Both news and tweets are a defined and closed unit of information. Both come from free and open sources and (normally) can be reached by all (specially if we refer to the online version of newspapers news).  A tweet has its own specificities, mainly derived from the restriction to a message of 140 characters max.

We like Twitter as source of reputation analysis. We like it a lot. We consider it a relevant tool.

As for political analysis, the utility of Twitter has to be demonstrated. In previous analysis run by our MRI center, we find that news were a good predictor of elections outcome in Spanish General Elections, 2008. It was also a correct tool to predict voting preferences in United States 2008 Presidential Elections.

As mentioned, Twitter is being used by scientists and practitioners in political analysis. For instance, Spanish newspaper Publico showed the live evolution of presence in Twitter of all candidates.

Twitter and PSOE Primary Elections

We present in this post our results concerning Twitter analysis of an ongoing electoral process. It refers to the election in Spain of the new leader of the main party in the opposition, the Spanish Socialist Party (PSOE). There are right now two official pre-candidates counting with a sufficient number of party elector endorsement. This is Alfredo P. Rubalcaba (61 years old) and Ms Carme Chacón (41 years old).

We feel that Twitter is well adapted to provide relevant information in a primary election context, as the problem of ideological bias inside the Twitter community tend to vanish, as both candidates come from the same party and share basically similar political principles. People in Twitter interested in PSOE electoral process (PSOE voters or supporters as well as people disliking PSOE) show their preferences by referring to the candidates. The same can be said about the current US republican primary elections.

Of course, number of tweets where a politician is mentioned is not a direct measure of political power and voting intention. Many people, specially inside the Twitter community, post messages for attacking or mocking a political rival rather than for supporting it. But, whatever the content of a tweet is positive or negative, a tweet becomes a measure of political influence. If you are not politically relevant, you will no receive attention from Twitter users, outside the set of your direct followers. Quantity of tweets shows influence, quality of tweets shows reputation. This is why number of tweets can provide information about political power and eventually about voting expectations.

It is true that Twitter political barometer cannot be translated into election winning odds, at least in the PSOE primaries, as voting is not made available to all Socialist Party voters, but only to a small number of electors, who can choose the candidate freely and not based in public opinion preferences.

Twitter Political Barometer: Daily Political Trend

In any case, we show in this post the evolution of the presence of both Socialist candidates since the end of December 2011, when the electoral process started.

The figures that we provide contain a basic but crucial correction of data that it is not usually performed by other analysts (like the measure provided by newspaper Publico). We exclude from our analysis all tweets mentioning both candidates, and we retain only those talking exclusively about Rubalcaba or Chacón alone.

We provided a first measurement two weeks ago, published in MRI Universidad de Navarra site. There we commented the results for a Spanish audience. As we present here the results to international readers (only 5% of this blog readers come from Spain), we do not comment the political reasons of the ups and downs of each candidates. We simply point out here that the Twitter momentum of each candidate can be explained by specific events.

The way we present the results are as follow: a value 50 means that each candidate is receiving during the last 24 hours the same amount of tweets. Values above 50 means that Rubalcaba is more present than Chacón in Twitter. Reaching a value of 100 means that all tweets referred to Rubalcaba in the previous 24 hours, while a value 0 means that all messages went to Chacón.

Our analysis shows that Carme Chacon is leading the race since mid January. The only exception is last week-end tweeting coverage to the formal support given by former Prime Minister Felipe Gonzalez to Alfredo Rubalcaba. This prevalence faded away quickly, and right now Chacon is again the main reference in Twitter.

As many Twitter analysis say, tweets are nourished mainly by information created by traditional news provider, and not by content genuinely emerging from Twitter community. One singular exception are trending topics (TT) not related with news. In this sense, even the Twitter CEO, Dick Costolo, stated recently that Twitter is not a media company:

‘We don’t create our own content; we’re a distributor of content and traffic. We’re one of the largest drivers of traffic to other media properties, [namely] to other online web properties, even to films’

Mashable Social Media, 31 January 2012

But the evolution of this time series provides also information produced by Twitter. It shows how Twitter community reacts to different news and events related with the candidates and the primary campaign. Peaks, their value and comparison with other peaks is telling us how relevant the Twitter community considers that any single event is, and how it influences the election race.

As the present case refers to a local political event we do not explain each single event behind the news and thereafter its impact on tweets for each candidate. But readers familiar with the PSOE electoral campaign can appreciate and evaluate the impact of the different political movements made by each camp, Rubalcaba and Chacón.

Twitter Political Barometer: Political Intensity Index

We provide a second measure. This new one refers to the degree of interest that Twitter community shows concerning the PSOE electoral process.

The second figure shows a composite measure of the number of tweets about Alfredo Rubalcaba or Carme Chacón, in a 24h basis.

This is of course a relevant complementary measure to the first one that presented the political candidate leading the race in terms of share of news by day. It is important to know the intensity of the debate and its evolution in time for assessing the impact of the campaign for the political interests of each candidate.

A value 1.0 corresponds to the average degree of tweets per day published during the time period. values bigger than 1 mean that in that moment Twitter users are talking more about one or both candidates than in the average trend during the campaign.

According to our results, top Twitter momentum of the political campaign up to now was reached by 8 January 2012, when one of the contenders, Carme Chacón, officially launched her bid as candidate for leading the Spanish Socialist party. Twitter messaging was at that moment four times higher than average. When Alfredo P. Rubalcaba presented his candidacy, Twitter Political Intensity Index reached a value of 1.2, ten days before his rival.

We are observing an increase of interest during this last week. By the end of January the Index took value 1.7. Nevertheless, we are not observing yet an explosion of interest inside Twitter community, not that the campaign is reaching its final stages before voting takes place.

Twitter Political Barometer: Overall Impact Index

The last measure we want to present in this post is somehow a combination of two previous measures.

We present an Overall Impact Index of each candidate in Twitter. The basis of this measure is the sum of all tweets received by each candidate since the start of the political fight, day by day. This time, we exclude again the tweets where both candidates were mentioned. We calculate the percentage of all tweets going to Rubalcaba and Chacon. Of course, sum of both gives value 1. Also, as we follow just two candidates, we have a symmetric evolution.

We have introduced a correction in the data, giving a smaller weight to older tweets. Our aim is to show the present perception of who is commanding now the race inside the Twitter community. Recent tweets have now a bigger impact that tweets published one week or one month ago.

We find that right now Chacon is leading the race, with 55% of all tweets since the start of the campaign. She became the main reference in Twitter since she announced officially her program to become the new leader of the Socialist Party, by 8 January 2012.

Alfredo P. Rubalcaba was the leading reference in Twitter between December 29 2011 and January 7 2012. This corresponds to the interval where he announced officially his candidacy, and when he questioned the adjustment program proposed by Mariano Rajoy’s conservative government.

During the last week covered in this graph, Carme Chacón maintains her advance in a rather steady path. There was an upward trend favorable to Alfredo Rubalcaba around January 25, but this trend has reversed again in favor of Chacón.

Twitter Political Barometer: How Powerful Is PSOE Third Way?

Like in other primary voting process, there are at this PSOE primaries some party members who present their candidacy to become the party leader, ahead from mainstream candidates.

Mainstream candidates are Alfredo P. Rubalcaba and Carme Chacón. Alfredo P. Rubalcaba was Vice-president in the previous socialist government with José Luis Rodríguez Zapatero. He fighted against Mariano Rajoy as Prime Minister candidates in the 20 November 2011 elections that his party finally lost. Carme Chacón was Defense Minister in Zapatero’s government. She renounced to assume the leadership in past elections. Both are the current reference in the Socialist party.

Besides these two natural leaders, the outliers that emerged and announced their intention to become a candidate were, among others, Emiliano García-Page and Antonio Quero.

Emiliano García-Page is the major of Toledo, young and promising politician. After launching several proposals and contacts, he recently renounce to present as candidate and endorsed Afredo P. Rubalcaba project a couple of days ago.

Antonio Quero is a civil servant in Brussels European institutions. He presents himself as a third way, rooted in a more participative basis. He has received support from movement ‘Bases en Red’. He affirms that he counts with some 10% of electors endorsements. The minimum required to formally become a candidate is 20%.

As we argued above that presence in Twitter can be seen as a measure of political influence, we propose to use this tool in order to show how relevant are García-Page and Antonio Quero among Twitter users.

We have been monitoring their presence in Twitter altogether with main candidates Rubalcaba and Chacón.

We present below our results.

We compare the presence of ‘third way’ candidates to the sum of tweet for mainstream Rubalcaba and Chacón. Our results are undisputable: both Quero and García-Page have played a marginal role in the primaries process inside Twitter community. Antonio Quero captured 1.4% tweets about the primaries, Emiliano García-Page another 1.0%. The overwhelming remaining 97.6% went to both Rubalcaba and Chacón.

If these results mean something, we expect that Quero will obtain far less than the 10% of total endorsements that he is claiming to have.

We close the analysis of this issue by focusing the results in the last and decisive week. Social media impact and visibility during the week preceding the nomination congress provides information about the trend and political chances of an emerging unexpected surprise.

Again our data shows that Antonio Quero is not perceived inside Twitter community as a relevant alternative, as his weights decreases to a mere 1.0% of total weekly tweets, even lower than the attention gathered during the whole electoral period.

It is true that, as political precedent, José Luis Rodríguez Zapatero became the PSOE leader as a surprise, as he was not mainstream and popular candidate before the congress. But we feel that this time Twitter says us that such an unexpected last minute reversion of favoritism is not plausible.

PS: Antonio Quero announced in the first day of nomination Congress, by Friday night, his decision to abandon the race as candidate, before the deadline for presenting the required endorsements was closed.

Addendum 1: The impact of ‘Chacón es Zapatero con faldas’

Today, Friday 3 February the PSOE Congress opens. During this week-end the electors will choose their new leader.

Yesterday, one of the Socialist Party barons, Juan Carlos Rodríguez Ibarra, former President of the Extremadura region pointed out in an radio interview at RNE that contender Carme Chacón was someone like Zapatero, but with a skirt. (‘Carme Chacón es como Zapatero pero con faldas’). This was probably meant to show that Chacón will follow the same social and economic model that the one proposed by former Spanish Prime Minister José Luis Rodríguez Zapatero. Elements of this model is currently highly criticised by many, even inside the Socialist Party, and is considered the source of the unprecedented electoral failure suffered by PSOE in 20 November 2011 general elections.

Whatever the intentions pursued by R. Ibarra, many people have considered this statement as unlucky, deplorable (Marcelino Iglesias) or even outrageous because its sexist profile. General perception by Twitter is also negative.

We propose a measurement of the impact of this statement in Twitter. This event can be considered as a political communication mistake, as the ‘attack’ to Chacón was launched by a supporter of Alfredo P. Rubalcaba (male), but its general negative reaction provides new support to Carme Chacón as women leader.

We want to assess the impact of Ibarra remarks in Carme Chacón visibility in Twitter in these crucial hours just before the final voting, which is announced to be very tight.

The figure shows the share of all tweets about Rubalcaba or Chacón going to Carme Chacón. This is an 8 hours average value. A value above 50% shows that Chacón is winning in number of tweets.

Ibarra remark came by 2 February in the afternoon.

Our results show that just before that moment, Rubalcaba was gaining momentum, and Chacón counted with 60% of all tweets. Since the remarks spread in Twitter, Chacón increased her presence up to almost 80-20 share. This morning, its presence moves between 60 and 75% of tweets.

If we exclude references to Ibarra and the skirts, Rubalcaba was managing to become the reference  as Chacón values decreased even to 45%.

Addendum 2: Rubalcaba and Chacón in the Twitter just before nomination

I am publishing this new section by Saturday 4 February 2012, 11h30 am (CET), just a couple of hours before voting and the nomination of the new leader (Secretario General) of PSOE takes place. Apparently, nobody can right now predict who will win, as number of endorsements are similar. Both camps claim right now that they count with the majority of votes.

We present here the track of the presence of both candidates in Twitter in the crucial stage of the nomination process. The PSOE Congress started yesterday, Friday 3 February. In the following figure we present the share of tweets reached by each candidate, day by day since 1 February, up to today 4 February, at 10 am.

Our results show that Carme Chacón has increased substantially her advance in the crucial last days of the electoral process. Remember that the distribution during the whole electoral process was Chacón 55% and Rubalcaba 45%. In the final four days the ratio in Twitter turns practically to a 4:1 for Chacón against Rubalcaba. 75.1% of all tweets refer to Carme Chacón, while 24.9% talk about Alfredo P. Rubalcaba.

In next figure we present the share of tweets for each candidate, day by day since February 1st. We show also the evolution of the intensity of the debate (by the size of the bars, as value 100 means the average number of daily Tweets during these last four days). In all cases, Carme Chacón completely dominates the Twitter debate, as she moves between 73% and 79% of all daily tweets.

Addedum 3: Epilogue (up to now)

PSOE delegates elected new party leader (Secretario General) last Saturday February 4. Voting took place by 2pm, after both Carme Chacón and Alfredo P. Rubalcaba delivered their programmatic speeches. Result was expected to be announced by 4ph, but it was delayed for 90 minutes. Apparently, the result was so tight, that al least one recount of votes was needed.

Alfredo P. Rubalcaba was elected with 487 votes. His rival, Carme Chacón, received 465 votes. There were other 3 votes. This means that Rubalcaba got 51% of support. Even if this is a joke (but a true no fake photo), the following photo is an explanation by Rubalcaba to parliamentary Rosa Díez (Feb 6) about how he did to win the elections. It reflects how dramatic was the election process and the importance of last minute negotiations to capture those decisive 11 votes.

We close our analysis of this post by presenting the evolution of Twitter presence of both candidates during the 38 PSOE Congress.

First figure refers to daily timeline during the week of the Congress. We take value 1 for the average daily presence of both candidates during the five days previous to the Congress. It started by Friday 3 February.

As already identified in previous figures in this post, Carme Chacón become the reference in Twitter and had a comfortable advance as the Congress commenced. As expected, social media attention exploded in the elction day (Saturday 4 February), as the index took value 11. Both candidates are followed with the same intensity.

Once Rubalcaba becomes the new leader, Twitter attention turns as expected to Rubalcaba. Sunday 5 February was the Twitter aftermath of the election, and was also the day that the Congress devoted to define the new political guidelines. Twitter media attention decreased to 3.4 points for Rubacaba and 1.5 points for Chacón.

We observe that Twitter relevance of Carme Chacón is vanishing rapidly since her defeat.

Election day was the crucial day and the one that captured by far the most social media attention. We present in the following figure a focus of Twitter activity during Saturday 4 February 2012.

We present hourly based measures. Now, value 1 is the average number of news received by both candidates in average during the five days before the elective Congress. They refer to tweets about one of the candidates not mentioning the other one.

Rubalcaba delivered first his programmatic speech. Of course, during his speech (by 12h) he became the reference in Twitter above Chacón. Then came Carme Chacón speech. It created almost three times more tweets than during his rival speech, reaching an index value of 1200 points. In terms of social media impact, Carme Chacón benefited from presenting in second place. Our results show that the presenting order is a very relevant issue for social media impact. Nevertheless, in this case this impact was not determinant, was voting was closed to party electors.

Communication of the winner of the election came by 17h25. Of course, this was translated into an explosion of tweets, that propelled Rubalcaba presence in Twitter to a daily maximum of 1470 points. Beaten rival received 470 points of attention in Twitter. Twitter messaging dropped quickly, in a matter of less than 2 hours.