Thoroughly respond to each problem below. Responses should be typed, organized, and clearly linked to each problem. 1. During presidential election years, candidates give a great deal of attention to states that are closely split between Democrats and Republicans — often referred to as battleground states — since these are the states that often determine the winner of the election. In addition to being important in determining who becomes president, some people believe battleground states can influence political participation. Analyst 1 argues that because of the large number of political advertisements, campaign visits, and all of the media attention, people living in battleground states eventually grow weary of the election and become disengaged from the political process. Because of this, Analyst 1 believes those in battleground states are less likely to turnout and vote compared with non-battleground states. Analyst 2 argues the opposite is true. Due to the large number of political advertisements, cam- paign visits, and all of the media attention, people living in battleground states are very well informed about the candidates and all of the attention actually mobilizes the public. Analyst 2 suggests those living in battleground states are more likely to turnout and cast a vote com- pared with non-battleground states. To test the two propositions, turnout data from the 2012 presidential election for all 50 states is used. The results of a difference in means test between the turnout rate in battleground states (coded 0) and non-battleground states (coded 1) is shown below . mean vep12_turnout, over(battle12) cformat(%9.3f) Mean estimation Number of obs = 50 ————————————————————————– | Mean Std. Err. [95% Conf. Interval] ————————-+———————————————— c.vep12_turnout@battle12 | 0 | 58.483 0.934 56.606 60.360 1 | 66.556 1.579 63.382 69.729 ————————————————————————– 1 . lincom [email protected] – [email protected], cformat(%9.3f) ( 1) [email protected] – [email protected] = 0 —————————————————————————— Mean | Coef. Std. Err. t P>|t| [95% Conf. Interval] ————-+—————————————————————- (1) | -8.073 1.835 -4.40 0.000 -11.759 -4.386 —————————————————————————— (a.) What is Analyst 1’s hypothesis? (1 point.) (b.) What is Analyst 2’s hypothesis? (1 point.) (c.) Interpret the p-value for the test that the difference does not equal zero. What does it tell us about the difference in means? (3 points.) (d.) According to the results of the test presented above, which analyst is correct? Explain. (5 points.) 2. Therearemanyexplanationsforthedifferencesinincomeinequalityobservedacrosstheworld. One potential cause for differences in inequality is the variation in government institutions from country to country. Expectations about the relationship between political regime types and income inequality, developed by three different researchers, are listed below. Researcher 1: “Presidential democracies create the lowest levels of economic inequality. Be- cause of the checks and balances in countries with separately elected presidents and legislatures, laws are carefully made so there isn’t too much government interference in the economy and everyone has an equal opportunity to succeed. Therefore, economic outcomes are much fairer and inequality is lowest in these countries.” Researcher 2: “The point about checks and balances in countries with presidential systems is well-taken, but in reality these obstacles to lawmaking actually lead to more income inequal- ity. Parliamentary democracies don’t have all of these barriers and passing legislation is much easier. This leads to more policies designed to redistribute wealth, meaning that parliamentary systems have the lowest levels of inequality.” Researcher 3: “Both of you have it wrong. Democracies, regardless of the type, always lead to more inequality because the rich have all of the political power in these systems. Dictatorships have the lowest levels of inequality since most countries with a dominant leader focus explicitly on equal economic outcomes by following communist or socialist ideals.” These arguments are tested below by analyzing the tabulation of each country’s household Gini coefficient and political regime type. The Gini coefficient is a common measure of income in- equality where higher values indicate more inequality. The version used below separates coun- tries into 3 categories ranging from lowest to highest levels of inequality. Political regime type 2 also has 3 categories and places each country into one of the following groups: dictatorships, parliamentary democracies, and presidential democracies. Here are the results: . tab gini3 regime_type3 , +——————-+ | Key | |——————-| | frequency | | column percentage | +——————-+ col chi2 3 quantiles of | gini10 | Dictators Parliamen President | Regime types Total —————-+———————————+——-— Low Inequality | 13 24 2 | 39 | 27.08 63.16 6.67 | 33.62 —————-+———————————+———- Med. Inequality | 22 7 9 | 38 | 45.83 18.42 30.00 | 32.76 —————-+———————————+———- High Inequality | 13 7 19 | 39 | 27.08 18.42 63.33 | 33.62 —————-+———————————+———- Total | 48 38 30 | 116 | 100.00 100.00 100.00 | 100.00 Pearson chi2(4) = 33.0120 Pr = 0.000 Based on these results, discuss the relationship between regime type and income inequality. Which of the three arguments is the most consistent with the results? Finally, consider the possibility that any relationship found in the above analysis is a spurious one. Identify one factor that should be accounted for in order to limit the likelihood of finding a spurious relationship between regime type and inequality. Explain why it is important to account for this factor when studying this particular relationship. (15 points.) 3. Two independent researchers, Dr. Smith and Dr. Jones, are both studying the relationship between the size of social protests and the degree of violence during protest events. The two scholars use the same newspaper database to identify protest events, but they devised different operational measures for gauging the size of protests. Although the researchers operationalize their measures using different criteria, both measures are scaled to range from 1 (very few people at the protest) to 100 (a very large protest). Imagine that you know the “true” average size of a protest event is equal to 45 on a 1 to 100 scale. Smith and Jones each hire five research assistants to code the newspaper reports as a way to assess their measurement procedures. Dr. Smith obtains the following five averages of the protest size measures: 62, 61, 68, 65, 64. Dr. Jones obtains the following five averages from the research assistants: 61, 65, 60, 49, 52. 3 (a) Which operational measurement is more valid, Dr. Smith’s or Dr. Jones’s? Explain how you know. What can you say about the reliability of the two measures? Discuss the importance of measurement validity and reliability in the context of research design. (10 points.) (b) A third researcher is aware of the work being done by Smith and Jones, and argues that both should be concerned about their approach to studying the effect of protest size on violent events. A central problem is their reliance on newspaper reporting to measure protest size and violence. Importantly, it’s likely that news organizations overreport on large protests and violent events, and underreport smaller demonstrations and nonviolent events. If this is true, the studies will seriously suffer from selection bias. Explain the problem of selection bias in this case from the perspective of descriptive inference and causal inference. (10 points.) 4. Consider Hertel-Fernandez’s (2016) “Explaining Durable Business Coalitions in U.S. Politics: Conservatives and Corporate Interests across America’s Statehouses.” Provide an overview of the purpose of this research. Then, explain the approach used by the author to answer his research question. How does the author justify using this approach? Does he test any specific hypotheses? Evaluate the internal and external validity of this study. Is this research valuable? Why or why not? (20 points.) 5. Several state legislatures have recently developed proposals to create a new health insurance pi- lot program for low-income residents, which would expand existing state and federally funded insurance programs. A program in one state, called the Care Advantage Program (CAP), would offer high quality health insurance plans for those with household incomes below $50,000 (about double the federal poverty level for a family of four). Although around 650,000 house- holds in the state would be eligible for CAP, the proposed program would initially have limited funding and would only be able to cover 30,000 households. Since the legislature expects there to be high demand for the program — over 400,000 households are estimated to apply for ben- efits — the state would set up a lottery that would randomly select families who want the new coverage and register with the state to be added to the lottery pool. Each of the 30,000 selected households would automatically be enrolled in the program, while those not selected would not receive any CAP health insurance benefits Researchers are excited about this pilot program because it could potentially allow them to study the effects of the health insurance program on a variety of health-related topics like use of preventative health care, household medical expenditures, and overall family health outcomes. Imagine you are hired by the state running the pilot program to study the effects of the CAP. Develop a research design that would allow you to assess the effectiveness of the insurance program. Make sure you are specific about the details of your approach, including a discussion of what your unit of analysis would be, sampling techniques (if necessary), how you might obtain measures of your dependent variables, what your independent variable would be, the kinds of causal inferences you could make with your design, how you would test the relationship between CAP coverage and the health outcomes, and any limitations your design faces. (35 points)
Sample Solution
