The recent Spanish general election produced a very difficult research scenario, driven by volatility and new emerging political trends, plus economic constraints and limited methodological innovation
The emergence of two “new” political parties, Podemos (on the left) and Ciudadanos (centre right) in the 2015 municipal elections, dramatically changed our political party system. The 2015 general elections, where the Partido Popular (PP) party with 28.7% of the votes, did not obtain the support of the chamber to form a government, led to political instability. Six months later, we were called to a second election. After June 2016, PP obtained 33% of the valid votes, and could form a government in a second round.
In an atmosphere of strong political and social tension between Catalonia with a separatist agenda, and the central government, the Supreme Court sentence of PP officials for corruption in an irregular financing network, led to a motion of censure on 1 June 2018 and as a result, Pedro Sanchez, leader of the socialist party (PSOE), became the new president of Spain, with the support of Podemos and the nationalist/independentist groups. The difficulties of approving a new budget led to fresh elections ten months later.
A dynamic situation
In the latest Spanish general elections held at the end of April, we could observe these dynamics: the consolidation of the Spanish multi-party system, the reinforcement of the autonomist candidacies, and the emergence of an extreme right party.
The multi-party consolidated due to a clear victory of the socialist party, an historical debacle of the PP, which becomes irrelevant in Catalonia and the Basque Country, a fast growth of Ciudadanos which disputes the hegemony of the right, and relatively poor results for Unidas Podemos, who, however, could govern together with the PSOE.
Nationalists, regionalists, and independentists grew strongly in the parliament, from the most obvious ERC (Esquerra Republicana de Catalunya), increasing from 9 to 15 seats, to the most discrete CC (Coalicion Canaria), doubling the number of seats, from 1 to 2.
Finally, the emergence of rightest VOX with 10% of votes, entering the national parliament for the first time with 24 seats in line with recent outcomes in other countries in Europe, where extreme right parties obtained 13% in Germany, 21% in France, 27% in Austria and 46% in Poland.
How the polls performed
For the first time in recent Spanish history, the right was fragmented in three parties, which, in addition to the left-right and nationalist-non-nationalist voting axis, made the electoral results extremely difficult to analyse. Despite this, certain predictions were close to the outcome, such as the tendency for the governmental party to concentrate the vote, and for the Spanish left to win in scenarios of high turnout (75.75%).
However, other possible factors such as the hidden vote, or the over-declared vote, were much harder to identify. During the time of ETA terrorism in the Basque country, we established a long-term pattern to estimate the percentage of people who avoided the interviewers, refused to respond, or simply lied about their vote to right parties. We could also estimate how proud some voters felt about voting for certain left parties’ and how much they wanted to let others know about it. These learnings were stable and allowed corrections to improved predictions.
In a two parties’ scenario the traditional weighting by vote recall helps to compensate for the logical biases of working with a sample. It also seems to be very valuable in the case of historic multiple-party scenarios, although with lower efficiency (Cabrera-Álvarez, Pablo and Escobar, Modesto, 2019 Reis.Rev.Española de Investigaciones Sociológicas. Nº 165. Enero- Marzo, 2019. Pp45-64). However, the Spanish political scenario today is rapidly changing from one election to another.
The high percentage of people delaying their voting decision to the last minute, limits the efficiency of forecasting, regardless of the sample size. Traditionally figures of undecided voters were around 20%, but March’s pre-electoral survey by CIS (Center for Sociological Research), showed 42% of voters were undecided. A poll conducted by IPSOS the day before the election on 26/27 April, still reflected that 15% of respondents were uncertain.
The important role that electoral debates play and their contribution to the decision of their “followers” (thinking of Lazarsferld’s maximum of 8% of swingers voters from right to left or vice versa) is very well known. In the past Spanish general elections, two electoral debates were held with audience figures of 49% and 44% respectively. These debates played a very important role in voting decisions. According to the same survey carried out by IPSOS, 40% of interviewees declared that TV debates helped them to decide on their vote.
Polling in general
21% of the total turnover of the research sector in Spain in 2017, was online research vs. 16% CATI, and 13% CAPI (source ANEIMO). In total more than 240 political polling surveys were carried in Spain between January 2018 and the general elections [1]. Despite this we did not see new methodological approaches in political polling. What we did see was the consolidation of the personal face-to-face methodological approach from the work done by CIS, as a reference of quality. It is nevertheless fair to mention that the new geographical approach of some researchers, or the estimations done by assembling multiple secondary sources [2] (El País, April 2019) worked well on predicting final outcomes.
In terms of types of political studies, the election will probably be remembered in Spain, as the last in which an exit poll was carried out. The continuous improvement of the vote count in Spain(end results available within 3 hours for well above 90% of votes counted) among the fastest in the world, has made these polls become a media element more than a statistical tool and the media impact of exit polls today hardly exceeds the first hour after the closure of polling stations. Being an expensive tool, which implies a high reputational risk for the research company, this type of study is very likely to disappear in a scenario of permanent cost reduction.
[1] https://es.wikipedia.org/wiki/Anexo:Sondeos_de_intenci%C3%B3n_de_voto_para_las_elecciones_generales_de_Espa%C3%B1a_de_2019
[2] https://elpais.com/politica/2019/04/03/actualidad/1554295033_268698.html
