Spain’s in the Mood or is it? Public Mood on Immigration
Virginia Ros, University of Manchester
The Public Mood is a measure of overall global attitudes towards government policies, a reflection of whether the population is in favour or against the policies that a government is implementing or planning. It has been used to explain the dynamics of the opinion—policy link (Stimson 1991; Franklin and Wlezien 1997; Soroka and Wlezien 2010; Wlezien 1995). While traditionally, and for reasons of comparability, wither single survey question is used as a time series (Lancee and Pardos-Prado 2013) or a latent factors are created as non-dynamic concept (Tabachnick and Fidell 2007), mood proposes to take several different questions and use them to create a latent variable that includes all the possible facets of the issue. It then tracks changes in preferences towards policy issues by collecting a variety of different survey questions on the particular issue as captured in different survey projects to create a time-series of sentiment towards the issue. It thus overcomes two problems:
- Collecting comparable indicators covering a long enough time series to explore changes in public opinion (Page, Shapiro, and Dempsey 1987)
- Including all the dimensions of the particular issue (Ben-Nun Bloom, Arikan, and Lahav 2015; Card, Dustmann, and Preston 2005)
Mood has one core advantage to specific measures, as it includes as many items as possible, it circumvents problems of measurement error with regard to specific items (Bound, Brown, and Mathiowetz 2001).
It is yet possible to develop mood measures that are more specific than overall global attitudes towards government policies if mood is disaggregated by policy issues (Stimson 1991): To measure public opinion towards a particular issue, a variety of questions on the specific topic from various survey projects can be used. To calculate mood the following information need to be recorded:
- Unique variable name to calculate the differences in percentages for each repeat item
- Date of data collection to be able to account for time
- The exact number of respondents in each survey to weight the percentages
On the basis of this information, so-called survey marginals, i.e. the percentage of responses that are given to a specific category or a sum of categories in a survey question, can be calculated. This is the proportion of respondents reporting a negative preference to the respective issue item. To be included in the immigration mood measure, the survey questions need to comply with two criteria: (1) they must be directive, i.e., force respondents to reveal a positive or negative preference towards immigration, and (2) they must at least be asked twice. This is crucial because the public mood pairs the survey marginals creating dyadic ratios of change in public preferences towards immigration. This raises the question about the match of the survey questions. In other words, how far do the survey questions have to match to be counted as a repeat item? If the question includes changes in the exact question wording or response categories vary, some information might be lost over time.
Two alternative approaches for inclusion have been proposed. (1) Starting with the maximum number of items and, subsequently, subtracting the least relevant ones in the creation of the time series in a stepwise building process (Stimson 2012) or (2) beginning with a smaller number of items that are thought to measure public opinion and subsequently adding more items if they make a meaningful contribution to the time series (Soroka 2003).
Analogous to the Principal Component Analysis (Stimson, Thiébaut, and Tiberj 2012), the mood algorithm then pools all correlated questions in a time series calculating the average variation of the dyadic ratios (the difference between the survey marginals over time) and weighting each of this ratios by the number of responses in each survey. Finally, based on the date included in the database, the algorithm creates a time series extrapolating the data points that are not present from the variation calculated.
One the one hand, this approach overcomes the problem of measuring change over time, because it uses the change in perceptions itself to measure mood. On the other hand, it is ensured that the mood incorporates all aspects and the multidimensionality of public opinion by aggregating as many survey items as possible. In comparison to other measures of public preferences, such as issue saliency using most important problems (Hobolt and Klemmensen 2008), mood reflects positive or negative positions of public opinion towards a particular policy issue of immigration. It thus gives a sense of the preferred direction in which the public wants public policy to shift.
However, one caveat of compiling different questions from different surveys is that they all measures are different regarding the question wording and how the data have been collected. Hence, changes in public opinion may occur as a consequence of these differences instead of a true shift in public perceptions. Mood overcomes this issue, by giving more weight to surveys with a higher number of respondents.
In the following I present first results of a measurement of immigration mood providing a directive measure of public opinion towards immigration in Spain (2000 to 2014). Even though this is not the first time that the public mood has been applied to a single issue (see for example the Policy Agenda Project, 2014), it is one of the first times a disaggregated mood has been measured in a context outside the United States.
I began by creating a database with survey marginals looking at survey questions towards immigration. Following Soroka (2003), I began by including a smaller, more focused number of survey questions from international surveys that included Spain, such as European Social Survey (ESS), the Eurobarometers (EB), the International Social Survey Programme (ISSP), and the World Value Survey (WVS). In addition, I also included measures from specific surveys about immigration conducted by the Spanish public polling institutes, such as the Centre for Sociological Research (CIS) and by a private polling institute called Anàlisis Sociòlógicos, Económicos y Politicos (ASEP).
For example, questions included in the immigration mood are:
- Do you think that the presence of immigrant workers coming from less developed countries have contributed to increase unemployment among Spaniards or, on the contrary, that it hasn´t had a significant effect? Measured on a 3-point directive scale
- I’m going to read out some statements. For each, please tell me whether you tend to agree with it or tend to disagree? People from these minority groups are enriching the cultural life of Spain. Measured on a 3-point directive scale
- In your opinion, do immigrants receive much more than they give, more than they give, as much as they give, less than they give or much less than they give from the State. Measured on a 5-point directive scale.
In a trade-off between quantity and quality of the survey items, only those that had a high loading in the model were finally included. The threshold of ±0.3 in the loading score – below/above which the item was dropped – proved to be efficient and when low score variables are dropped the percentage of variance explained by the model increases. The result is a variable that measures the percentage of negatives views about immigration and includes the views of Spaniards on immigration in relation to the job market, the welfare state and the concerns that immigration arises for the culture of the native population. The result is a variable that captures the percentage of negative views towards the issue of immigration in Spain over time.
Figure 1 Evolution of immigration public mood (2000-2014)
Source: Data from CIS, EB, ESS, EVS, ISSP, ASEP
Figure 1 shows the evolution of the percentage of responses showing a negative perception of immigration. Overall, Spain displayed a negative trend in their perception of immigration (2000-2014). Negative perceptions of immigration have steadily increased from 2000 to 2011. Before this year, immigration was a not much of a concern in Spain (only 3% of the population was foreign born according to the Institute of National Statistics) and immigration as a policy issue was not salient at all. Before 2000, the percentage of negative views oscillated around 30% in decline. However, 2000 marks a turning point. The percentage of people with negative views towards immigration begins to increase. This coincides with a steady rise in the foreign population in Spain, on average, by 2% a year. In 2003 and 2007 there seems to be a change in the trend but it is just a small decrease in negative public opinion that does not last more than a year in any case. This stable increment in negative perceptions of immigration means that, in the years 2000 to 2012 the percentage of the Spanish population showing negative opinions towards immigration increase from 28% in the first semester of 2000 to 41% in the last semester of 2011. This trend matches the figures showing the evolution of the foreign population in Spain that can be seen in Figure 2. However, the steeper increase from 2008 to 2011 might also be due to the economic crisis. This relation with the economic crisis might explain also why, after 2012, once the crisis seems to have been accepted by the Spanish population and the shock is gone, the trend in negative perceptions of immigration changed declining slightly and, by 2014, the percentage of the Spanish population showing negative views towards immigration is again below 40%.
Figure 2: Evolution of the percentage of immigrant population in Spain 2000-2011
Source: Spanish National Statistics Office (INE)
To conclude, this is the first time an issue specific measure of mood has been implemented outside the US context. The mood measure overcomes two problems that single items could not solve: (1) It covers a complete time series within a particular context and issue area providing two data points per year increasing the amount of data to run further analyses and (2) it gathers all the dimensions studied by the literature (McLaren and Johnson 2007) by including several items that are clearly related to different aspects of the concerns that immigration might arise. Sadly, in the case of Spain we observe increased scepticism towards immigration. Spain – just like in many other European countries – seems to be in the mood with immigration.
 The algorithm used to do this process is available through a software program, called WCalc, or as a function for R or as a C++ source code from James Stimson’s web page. The R code to run the algorithm is available in James Stimson’s webpage (https://dl.dropboxusercontent.com/u/14523416/Extract.r) and can be changed to adapt it to specific needs.
 See Stimson (1991) for a thorough technical introduction on how to build the public mood.
Ben-Nun Bloom, Pazit, Gizem Arikan, and Gallya Lahav. 2015. ‘The Effect of Perceived Cultural and Material Threats on Ethnic Preferences in Immigration Attitudes’. Ethnic and Racial Studies 38(10):1760–78.
Bound, John, Charles Brown, and Nancy Mathiowetz. 2001. ‘Measurement Error in Survey Data’. Handbook of Econometrics 5:3705–3843.
Card, David, Christian Dustmann, and Ian Preston. 2005. Understanding Attitudes to Immigration: The Migration and Minority Module of the First European Social Survey.
Franklin, Mark N. and Christopher Wlezien. 1997. ‘The Responsive Public’. Journal of Theoretical Politics 9(3):347–63.
Hobolt, Sara Binzer and Robert Klemmensen. 2008. ‘The Dynamics of Issue Diversity in Party Rhetoric’.
Lancee, Bram and Sergi Pardos-Prado. 2013. ‘Group Conflict Theory in a Longitudinal Perspective: Analyzing the Dynamic Side of Ethnic Competition’. International Migration Review 47(1):106–31.
McLaren, Lauren and Mark Johnson. 2007. ‘Resources, Group Conflict and Symbols: Explaining Anti-Immigration Hostility in Britain’. Political Studies 55(4):709–32.
Page, Benjamin I., Robert Y. Shapiro, and Glenn R. Dempsey. 1987. ‘What Moves Public Opinion?’ The American Political Science Review 81(1):23–43.
Soroka, Stuart. 2003. ‘Media, Public Opinion, and Foreign Policy’. The International Journal of Press/Politics 8(1):27–48.
Soroka, Stuart and Christopher Wlezien. 2010. Degrees of Democracy: Politics, Public Opinion, and Policy. Cambridge University Press.
Stimson, James A. 1991. Public Opinion in America: Moods, Cycles, and Swings. Westview Press Boulder, CO.
Stimson, James A. 2012. ‘On the Meaning & Measurement of Mood’. Daedalus 141(4):23–34.
Stimson, James A., Cyrille Thiébaut, and Vincent Tiberj. 2012. ‘The Evolution of Policy Attitudes in France’. European Union Politics 13(2):293–316.
Tabachnick, Barbara G. and Linda S. Fidell. 2007. Using Multivariate Statistics. 5th ed. Boston: Pearson/Allyn & Bacon.
Virginia Ros is PhD Graduate in Politics from the School of Social Science at the University of Manchester. Her expertise is on Spanish politics of immigration. This blog post is based on her PhD dissertation.