Chapter 3: Explanations, Hypotheses, and Making Comparisons Preliminaries In previous lectures you learned that political science research starts with defining and measuring concepts. Measurement entails: Defining the concept to be measured clearly. Accurate measurement of the concept. Selecting variables or indices to accomplish that measurement. The measurement process is intended to identify clearly what it is you want to study, and then obtain empirical data which represents that concept. However, the process does not stop here. We need to propose and test explanations for the political phenomenon we want to study. Here the concern is over “why” relationships might exist among our research concepts. Elements of Explanations Any explanation for research concepts entails explaining variation and covariation. If a concept y varies, why does it vary? Does y covary with another concept x? As political scientists, we seek to understand differences, variations, and covariations? Examples: Why do people in the United States turnout/not turnout to vote? Country 1 has more democratic institutions than country 2. Why? Some people are opposed to “abortion on demand?” Why? Some states place more restrictions on abortion than others. Why? Some people turn out to vote, while others do not. Why? Public liberalism and conservatism can be shown to vary through time. Why? Presidential approval ratings rise and fall through time. Why? Some people think it would be a good idea for college students to be able to carry concealed weapons on college campuses. Others do not. What explains these differences? Good explanations describe a connection between a dependent variable and a causal variable. A causal variable is also called an independent variable. Example: Support for concealed weapons on college campuses is a function of people’s partisanship. Good explanations provide a direction for relationships. Example: Support for concealed weapons on college campuses is higher among Republicans than it is among Democrats. Good explanations should also be testable. Example: Do an opinion survey among the general population. Construct a measure of support for concealed weapons on college campuses. Record the respondents’ partisanship during the survey. We should find covariation in the expected direction if the explanation is true. Is this explanation exhaustive? Are there some other reasons people might support or not support concealed weapons on college campuses. Hypotheses An hypothesis is a testable statement about the empirical relationship between an independent variable and a dependent variable. An hypothesis tells us how different values of a dependent variable should be related to values of an independent variable. Examples: Explanation: Support for gun regulation is a function of partisanship. Hypothesis: Support for concealed weapons on college campuses should be higher among Republicans than among Democrats. Explanation: Voter turnout is a function of voter demographics Hypothesis: Voter turnout should be higher among older people than it is among younger people. Hypothesis: Voter turnout should be higher among better educated people than among lower educated people. Hypothesis: Voter turnout should be higher among people with high incomes than among people with low incomes. Why do some people favor candidates or policies, while others disfavor certain candidates or policies? Competing theories. Rationality - People favor policies and/or candidates that promote their own personal self-interest. Hypothesis: Voting Republican should be greater among those with higher income. Hypothesis: Voting for the incumbent candidate should depend on the individual’s personal financial situation in the recent past. Hypothesis: Non-support for efforts to slow global climate change should be greater among those who work in industries which pollute the environment. Sociotropic - People favor policies or candidates that maximize “social welfare”. Hypothesis: Support for environmental regulation should be greater among those whose values are communitarian, rather than individualistic. Hypothesis: People who view the economy as a whole as performing well, regardless of their own financial situation, are more likely to vote for the incumbent candidate. Hypothesis: Cross-nationally, countries having values more supportive of equity should be more supportive of programs to help the poor. Testing Hypotheses with Descriptive Statistics or Comparisons We can easily use the tools you have learned up to now to construct some basic tests of hypotheses. However, note that these are very basic tests which do not provide measures of certainty about the hypotheses. Examples: Hypothesis: Smoking is a function of one’s income level (a proxy for health sophistication). Hypothesis: A higher level of economic development produces greater support for civil liberties. (Or, does the relation run in the other direction?) Hypothesis: Students who study more perform better in their courses. Hypothesis: Voter turnout is higher among older people than it is among younger people. Hypothesis: Political Activists are more likely to turnout to vote. Hypothesis: Collective bargaining bolsters economic welfare for the middle class. Hypothesis: Support for gun control is higher among Democrats than among Republicans. Hypothesis: Smoking is higher among low income groups (a proxy for health sophistication). Hypothesis: A higher level of economic development produces greater support for civil liberties. (Or, does the relation run in the other direction?) Hypothesis: Collective bargaining bolsters economic welfare for the middle class. Same as before using group percentages, rather than a graph.