APPLICATIONS FOR MANAGEMENT - 30280 How Changes in Political Trust were affected by Education Levels Post-Economic Recession Grace Glinsman - 3162086 Matilde Bertola - 3142049 Andrea Ballacchino - 3170261 Word Count: 2353 EXECUTIVE SUMMARY: The aim of this research project is to explore the relationship between political trust and education level following the 2007 recession, answering the question ‘Did those with higher education change their political trust more than those with lower education as a result of the economic recession?’ The data being used is from the European Social Survey, specifically Rounds 3 and 5 in the years 2006 and 2010 respectively; focusing on the countries: France, Spain, Bulgaria and Ukraine. Our key variables are the number of years of full-time education and trust level in politicians. In submitting this assignment: 1. We declare that this written assignment is our own work and does not include (i) material from published sources used without proper acknowledgment or (ii) material copied from the work of other students. 2. We have an electronic version of this assignment in our possession. 1. INTRODUCTION In this research project, we aim to answer the question, ‘Did those with higher education change their political trust more than those with lower education as a result of the economic recession?’. The Great Recession of 2007 was a major economic event characterised by declines in national economies globally which lead to catastrophic effects for many countries across Europe, including the fall of many financial institutions, collapse of international relations and political crises within countries. It is important to analyse and keep the discussion about this event going, to avoid a repetition of history by recognising the causes and consequences and noting them for future scenarios. Due to the varying responses of countries’ politicians and governments to the crisis, political trust fluctuated and we want to conclude whether education levels were a factor in this. 2. BACKGROUND The crux of this project is the relationship between education and political trust; what is the nature of their relationship, not only within the scope of the recession, but also throughout time. It is a multifaceted argument with many explanatory factors supporting both a positive and negative linear relationship. On one hand, education levels can raise trust in politics by increasing the stakes and awareness citizens have. Higher educated individuals are more likely to be able to benefit from political processes such as having a higher quality of life and more stable, successful careers. They also have the ability to better understand the political world and take a greater interest in the topic, therewith developing greater trust. Lower education levels can lead to citizens feeling alienated from the political spectrum, and not fully understanding the topics and policies at issue or what they are being told, thereby feeling less trustworthy of politicians. On the contrary, education can diminish trust due to its ability to give citizens the tools to recognise corruption, manipulation and not being able to be tricked by false information. The effects of education on trust levels become more pronounced with steady increases in education, as university level students are the most susceptible to effects of meritocracy and corruption. Individuals can become more critical and dissatisfied with the running of their country, often feeling they could do it better themselves, questioning the traditional processes of democracy and their credibility of the politicians they are being led by, leading to lower trust levels. These arguments can be well concluded by the report entitled ‘How Does Education Affect Political Trust?: An Analysis of Moderating Factors’, Cinar and Kose discover that the effect of education on levels of political trust are very context-dependent, however, one can generally conclude that the two variables are positively correlated in meritocratic countries, and negatively in countries with higher levels of corruption. If we include the recession into our thoughts now, it will have had different impacts on different individuals according to their education levels. The recession was an economic event and had greater life-changing impacts on individuals with lower education levels. They suffered from job loss due to the susceptibility of their jobs to economic downturns, and even if they were fortunate to keep their job, most likely had a reduction in their income level. These both lead to declining quality of life and disability to continue the standard of living they were used to, now potentially not being able to afford the basic necessities. The greater negative impacts on this group may have led to greater anger and dissatisfaction with the politicians running their country and not finding strong enough, quick enough solutions to help the population. This would have led to decreased levels of trust. For individuals with higher education levels, they would have had an increased sense of job security, stable incomes as well as the opportunity for further career advancements, thus leading to not being as impacted by the recession, and so would not be feeling the same levels of fury with the government, therefore their trust levels would not have decreased as drastically, or even at all. 1 These all show that this is a complex research question, which will be interestingly analysed, as a range of conclusions could be made. 3. DATA Data Set: European Social Survey The data we shall be using is a product of the European Social Survey - a multi-country survey given to over 30 countries. It has three aims which are to: monitor and interpret changing public attitudes and values within Europe, advance and consolidate improved methods of cross national survey measurement and finally, develop a series of European social indicators. Hour long face to face interviews on a variety of core topics repeated from previous rounds, also including additional modules, conducted with new samples every round. This report focuses on results of the economic recession and so we decided to analyse the short term impacts as we believe them to be the most powerful and immediate, therewith selecting Rounds 3 and 5 which took place in the years 2006 (right before the recession began) and 2010 (right after the end of the recession) respectively. Countries: Our methodology for choosing which countries to analyse, we considered many variables on which to focus our selection including: geographic location, GDP levels, political leanings of the countries, relative impact size of the regression on the economy as well as picking countries at random using a generator. We opted for the GDP level method as GDP levels give indirect indications of education levels and so we split this variable into four sections: High Income, Upper Middle Income, Lower Middle Income and Low Income levels. To pick a country for each income level, we selected ones we knew were either impacted by or were a part of the cause of the recession as well as those which had available data for rounds 3 and 5. This led us to pick: High Income: France, Upper Middle Income: Spain, Lower Middle Income: Bulgaria, Low Income: Ukraine Key Variables: As we are interested in the concepts of political trust and education levels, there were many variables for us to choose from for our analysis including: trust in politicians, trust in parliaments, trust in political parties, whether the respondent voted in the last elections, highest education level achieved and number of years of education completed. For the political aspect, we chose trust in politicians, as they are the people running political movements and making the decisions on behalf of the citizens, and so when we refer to political trust, we believe this is most referring to the trust we have in the people running our politics. On the education front, we selected years of education completed as it is numerical and easily comparable for all countries, as well as being easily visualised. Trust in politicians (trstplt) is scored from 0 to 10 with 0 representing no trust at all and 10 being complete trust. Education years completed (eduyrs) is answered with full-time equivalents using whole integers. Data Collection: Data for the ESS is collected using one-hour face to face interviews for all participating countries and the following standards are implemented to ensure accuracy and standardisation: (1) Minimum target response rate of 70% (2) Rigorous translation protocols (3) Restricted interviewer workload (48 sample units gross) (4) Quality control back-checks on completed interviews as well as ineligible cases Sampling: ESS follows strict random probability sampling which contain some of the following key rules: (1) All persons must be aged 15 or above living within private households (2) Respondents selected regardless of nationality, citizenship, language, marital or legal status (3) Minimum effective sample size 1,500 or 800 in countries where ESS population is < 2 million 4. MULTIVARIATE ANALYSIS Weights: To begin our analysis, we must generate a new variable overallweight which is equal to the product of design weight (pspwght) and population weight (pweight) to be able to perform our analysis at a 2 European level. The population weight accounts for oversampling of countries, whilst the design weight corrects for their stratification sampling design to be able to make country level conclusions. Outliers: Within our dataset, there are no outliers for trstplt (our dependent variable) due to the nature of the question, respondents are given set responses to select from. Missing Values (Figures 1 and 2) The data was screened for missing values and the number of missing values in our data is 443, which accounts for 2.9% of our dataset, which was then eliminated. Initial Data Observations (Figures 3, 4 and 5) As we can see in the table and histogram below, skewness is 0.59, which is in between the boundaries [-1, 1]. Kurtosis is 2.61, which is in between the boundaries [2,4]. This shows us that the variable does deviate from normality however, not too extremely, and is right-skewed. The average political trust in 2010 was 2.41, lower than 2006 where it was 2.71 as can be seen through the Mean values for both variables, however, for greater accuracy, it would be better to analyse the change in political trust using a linear regression analysis. Linear Regression (Figure 6) The first step of our linear regression is to generate a dummy variable round5 which takes the binary value of 1 for Round 5 and 0 for Round 3. By running a regression of trust in politicians as our dependent variable and round5 as our independent variable, we see an intercept of 2.91 which indicates the average political trust in 2006 (Round 3). The negative coefficient of round5, equal to -0.27, implies the regression is downward sloping; indicating a negative linear regression and that political trust has therefore decreased after the great recession. The two coefficients have a p-value smaller than 5% and are therefore statistically significant. Multiple Regression (Figures 7, 8 and 9) In order to understand whether level of education had an impact on change in political trust, we included in the regression a second independent variable: years of full-time education. The increase in R-squared (from 0.0035 to 0.0043) indicates the introduction of this variable improves model efficiency. The interpretation of this regression leads us to the plotting of two different linear functions: one for the 2006 round and one for the 2010 round. The former has an intercept of 2.74 and a slope of 0.015 (both statistically significant). The latter has an intercept of 2.46, 0.28 lower than the former. This information can be interpreted as the difference between the level of political trust in 2006 and 2010: for every level of education, political trust in 2010 is 0.28 lower than in 2006. Both lines have the same slope, equal to 0.015; the positive sign indicates a positive relationship: a one year increase in education leads on average to a 0.015 increase in political trust. From the last regression, we were not able to understand whether the level of education had an impact on the magnitude of the change in political trust. To do this, we need to introduce a third independent variable, a slope dummy eduyrsXround5, the product between eduyrs and round5. This variable = 0 if the observation was in 2006 and is equal to the number of education years if observation was in 2010. In this case, R-squared is higher and so model efficiency is better. The coefficient of round5 is not significant anymore, meaning that we cannot say the intercepts of the two lines (2006 and 2010) are different. However, the negative value of the slope dummy indicates the magnitude of the difference in political trust before and after the recession differs based on the level of education. The slope of the 2010 line is 0.03 lower than that of the 2006 line: in 2010, an increase in the level of education led to a smaller increase in political trust than in 2006. We can therefore conclude that for people with higher levels of education, the decrease in political trust after the great recession was higher than that of people with low levels of education. 3 We would also like to understand whether there is a difference in the conclusion we reached within different countries. Focusing on France and Bulgaria, introducing a dummy variable (BG = 1 if country is Bulgaria) and a slope dummy (BGXeduyrs = 0 if country is France and to level of education if the country is Bulgaria). BG coefficient is significant and negative, indicating on average political trust in Bulgaria is 1.28 lower than in France. The coefficient of the slope dummy instead is not significant, implying that there isn’t any difference in the effect that education has on the change in political trust. 5. CONCLUSION In short words, following from our analysis we counterintuitively state that a higher level of education led to a stronger decrease in the level of political trust, post-recession. A lower level of education instead did not have as great of an impact on trust directed towards politicians: That said, this is extremely context dependent, as shown by our specific country examples of France and Bulgaria. France, with a higher GDP, is assumed to have higher average education levels, and vice versa of Bulgaria, however, we found that France had higher political trust, reinforcing the idea that political trust cannot be explained by a single factor of education, but also other variables such as potentially age, gender and culture. From this analysis, we highlight the importance of conducting research on the demographics of specific party supporters so that in the event of another recession, they know which groups of individuals tend to lose support for that party and to whom they should direct their campaigns. 6. FIGURES: Figure 1: Missing values for trstplt Figure 3: Histogram for trust in Figure 2: Table showing missing values politicians for eduyrs Figure 5: Summary table for political Figure 6: Linear Regression of Political Figure 4: Summary table for trust in trust Trust against ESS Round politicians Figure 7: Multiple Regression of Figure 8: Multiple Regression of Figure 9: Multiple Regression of Political Trust against Education Level Political Trust against ESS Round and Political Trust against Education Level and Country Education Level and ESS Round 4 7. REFERENCES: European Social Survey European Research Infrastructure (ESS ERIC). (2018). ESS5 - integrated file, edition 3.4 (Austria not included) [Data set]. Sikt - Norwegian Agency for Shared Services in Education and Research. https://doi.org/10.21338/ESS5E03_4 European Social Survey (2021) ‘Round 11 Survey Specification for ESS ERIC Member, Observer and Guest Countries’. European Social Survey. European Social Survey (no date) Data collection, Data Collection | European Social Survey. Available at: https://www.europeansocialsurvey.org/methodology/ess-methodology/data-collection (Accessed: 12 October 2023). European Social Survey European Research Infrastructure (ESS ERIC). (2018). ESS3 - integrated file, edition 3.7 (Latvia and Romania not included) [Data set]. Sikt - Norwegian Agency for Shared Services in Education and Research. https://doi.org/10.21338/ESS3E03_7 Ugur-Cinar, M., Cinar, K. & Kose, T. How Does Education Affect Political Trust?: An Analysis of Moderating Factors. Soc Indic Res 152, 779–808 (2020). https://doi.org/10.1007/s11205-020-02463-z 5