Does Education Change Political Attitudes? Evidence from a Kenyan School ExperimentEdward Miguel, UC Berkeley and NBERMichael Kremer, Harvard and NBERRebecca Thornton, University of MichiganWGAPE 12/2008 Education and Politics 2Education, development, and democracy• What impact does education have on politics?-- Classic modernization theory (Lipset 1959) posits that wealth, education, democratization emerge together-- Education broadens horizons beyond family / tribe, provides access to media and information, empowers people to act collectively to make demands on the state-- Was Africa’s 1990s democratization wave driven, in part, by schooling investments made a decade earlier?• The cross-country patterns underlying modernization theory have come under attack (Acemoglu et al. 2007)WGAPE 12/2008 Education and Politics 3Education, development, and democracy• At the micro-level of analysis, there is strong evidence that educated citizens are more likely to care about politics and to vote (Verba and Nie 1972)-- Logan and Bratton (2006): large positive correlations between education, political participation in 15 Sub-Saharan African countries -- However, with few exceptions, omitted variable bias issues are not adequately dealt with. More / less educated people could differ in many unobservable ways (e.g., personality, family background, ability)• Milligan et al (2004) find strong positive effects of education on voting in the US but not the UK. Positive impacts on political knowledge in both countriesWGAPE 12/2008 Education and Politics 4An education experiment in rural Kenya• This paper exploits a randomized experiment carried out in 2001-2002 in rural western Kenya (Busia district)-- Grade 6 girls in 34 primary schools were offered a merit scholarship if they scored well on academic exams. 35 schools were controls. (N=3200 girls)• In the short-run, the program produced large test score gains, of 0.25 s.d. -- Student and teacher attendance improved• What is the impact of this increase in human capital on political attitudes and knowledge 4-5 years later?WGAPE 12/2008 Education and Politics 5Talk outline• Introduction• The Girls Scholarship Program in Kenya• 2001-2002 Short-run impacts (summary)• 2005-2007 Long-run impacts on:-- Education-- Political attitudes, knowledge-- Gender, ethnic and religious identity, social capital• Discussion / conclusionWGAPE 12/2008 Education and Politics 6The Girls Scholarship Program (GSP) in Kenya• The top 15% (district-wide) of grade 6 girls in program schools won:– 500 KSh (US$6.40) for school fees, for two years– 1000 KSh (US$12.80) cash, for two years– Public recognition, certificate at an award ceremony• The total cash component was US$38, in a setting where annual per capita income is close to US$400-- Two cohorts competed (2001, 2002)• Recent school incentive program evaluations in rich countries have produced mixed results (Angrist and Lavy 2002, Leuven et al 2003)WGAPE 12/2008 Education and Politics 7Map of AfricaKenyaWGAPE 12/2008 Education and Politics 8BusiaWGAPE 12/2008 Education and Politics 9WGAPE 12/2008 Education and Politics 10Sample attrition in the initial study• Initially in two neighboring districts, Busia and Teso-- Teso district has long been a difficult setting for NGO work, due to the opposition of some local leaders-- High sample attrition rates in Teso district complicate estimation of effects there (wide treatment bounds)-- Sample follow-up in 2001 in Busia was high at 79% (program schools) versus 78% (control), while in Teso it was low at 53% (program schools) versus 65% (control)• The follow-up 2005-2007 survey was restricted to Busia-- Not a problem for the current study since our focus is the impact of education, not of merit awards per seWGAPE 12/2008 Education and Politics 11Summary of earlier findings (Kremer et al 2008)• The randomization “worked” at producing comparable program and control groups (Table 1)WGAPE 12/2008 Education and Politics 12WGAPE 12/2008 Education and Politics 13Summary of earlier findings (Kremer et al 2008)• The randomization “worked” at producing comparable program and control groups (Table 1)• Large test score gains for both cohorts (2001, 2002) of 0.25 standard deviations (Table 2, Figure 1)-- We cannot reject equally large gains for girls throughout the baseline (2000) test score distribution, even at the bottom-- Gains in teacher attendance (reducing absenteeism by one third), and also student attendance• Test score gains persisted through 2003WGAPE 12/2008 Education and Politics 14WGAPE 12/2008 Education and Politics 15WGAPE 12/2008 Education and Politics 16The 2005-2007 Tracking Survey• A panel dataset of rural Kenyan young adults (2001-07)-- Individuals were “tracked” as they moved throughout Kenya and even into Uganda-- Direct contact with 84.3% of target respondents: 81.9% actually surveyed; the rest were dead or refused -- Balance across program (81.7%) and control (82.1%) group individuals-- Two step procedure: “regular tracking”, and “intensive tracking” (follow-up a representative sub-sample)WGAPE 12/2008 Education and Politics 17The 2005-2007 Tracking Survey• Life outcomes: educational attainment, demographic information (marriage, fertility, migration), labor market• Cognitive tests were administered in: arithmetic, reading, vocabulary (English, Swahili), spatial reasoning (Ravens matrix)• Adapted from AfroBarometer, World Values Survey:-- Political knowledge (officials’ names), interest-- Attitudes: democracy, government, voting, violence-- Gender, ethnic and religious identity, “social capital”WGAPE 12/2008 Education and Politics 18Tracking Survey questions• Some sample questions:Political knowledge“Do you know who Kenya’s current Minister of Health is?”Democratic attitudes and participation“We should choose our leaders in this country through regular, open and honest elections.” (1=Strongly agree …5=Strongly disagree)Do you intend to vote in the next presidential election, to be held in 2007?Satisfaction / Skepticism“Overall, how satisfied are you with the way democracy works in Kenya?” (1=Very satisfied … 4=Not at all)WGAPE 12/2008 Education and Politics 19Tracking Survey questionsWomen’s rightsA. Women have always been subject to traditional laws and customs, and should remain so. B. In our country, women should have equal rights and receive the same treatment as