Democracy and Social Science II

Professor Jeff Grynaviski

Winter 2003 (Prepared March 7, 2003)

 

 

 

Final Examination

 

 

This examination is open book and open note, but not open friend. Once you have begun the exam you may not discuss this material with anyone other than Professor Grynaviski until it is turned in. There is no limit on the length of time that you may take on the exam, though it is designed to be a three hour exam for students who studied in advanced.

 

The exam is divided into two parts. Part I asks you to demonstrate your understanding of how to interpret and use regression models in the social sciences through a couple of short answer questions.  You may type or handwrite your answers to Part I.

 

Part II contains longer essay questions about The New American Voter and The Macro Polity. You must answer at least three of the four questions. You may not write more than 1500 words for this section and you must report your word count. This section must be typed using the style guidelines reported in the syllabus.

 

A paper copy of this exam is due by Tuesday, March 18 at 5pm in my office (Pick 528). Your paper will be marked down a full letter grade at 5:01pm on Tuesday and every 24 hours thereafter.

 

Good luck!

 

 

Part I. Please answer the following three short answer questions.

 

 

1. A random sample of college students were asked to rate their courses on a scale of 1 to 10 (10 being the best) and to state their grade in the course on a scale of 1 to 12 (A = 12, A- = 11, and so on). The average rating was a 5.53 and the average grade 9.69. A least squares regression of rating (Y) on grade (X) with 120 observations yielded:

 

Y    =   -2.14 + 0.79 X

                                    (2.91)  (0.30)

           

with the regression coefficients’ standard errors in parentheses below the regression equation.

 

Plot the regression line making sure that you include the scale of the variables on your axes (put numbers on your axes). What does a least squares regression tell us that a comparison of the average values 5.53 versus 9.69 does not? How do you interpret the estimate -2.14? How do you interpret the estimate 0.79? If your null hypothesis is that the effect of grades on rating is less than or equal to zero (in other words, H0: B1  £ 0), is the estimated relationship statistically significant at the 1 percent level? How do you know?

 

 

 

 

 

 

 

 

 

 

 

 

2.  Identify three reasons why one might be inclined to disagree with the following statement:

 

In the regression model reported below, I demonstrate that income has a statistically significant positive influence on individuals’ propensity to become Protestant.

 

Dependent Variable is Religion. Religion equals 1 if Protestant and 0 otherwise.

 

Coefficient

Standard Error

t-statistic

 

Intercept

 

 

0.421

 

 

0.032

 

12.99

 

Income

 

 

0.005

 

0.002

 

2.545

N = 1564, R2 = 0.004

 

Range of independent variable: 0 = No Income, 12= $50,000/yr, and 24 = $100,000 plus / year.

 

 

 

3. Suppose that you had the following three hypotheses about the factors that influenced whether someone considered him- or herself a liberal or conservative. First, you hypothesized that blacks would be more liberal than whites. Second, you hypothesized that people with higher incomes would be more conservative than people with lower incomes. Third, you hypothesized that the impact of income on ideology was greater for whites than blacks.

 

a.       Plot the hypothesized relationship between ideology and income for whites. On the same diagram, plot the relationship between ideology and income for blacks.

 

b.      Write down a multiple regression model that would allow you to test each of the three hypotheses.

 

c.       For the multiple regression model created in step a, write down your null and alternate hypotheses about each of the regression coefficients. What is the critical values of the t-statistic that would allow you to reject each of the null hypotheses? (be careful to get the sign of the t-statistic correct).

 

 

 

 

 

 

Part II. Answer three of the following four essay questions. Your answers for the three questions may not exceed 1,500 words in total (approx. 5 pages) and you must report the word count.

 

 

1. Assess the validity of the following statement:

 

Physical scientists are able to use experiments with rigorous controls as a way to test their truth claims while social scientists usually cannot. As a result, causal claims made by physical scientists are more valid than those of social scientists.

 

A “good answer” will have a well-defined thesis and will address at least one objection to that thesis.

 

 

 

2. What is the funnel of causality? Describe an application of the funnel of causality in Miller and Shanks’ The New American Voter and how they made their causal argument. Are you convinced by their explanation? Why or why not? Feel free to draw an illustration if it helps to make your point.

 

 

 

3. What is Erickson et al’s argument in The Macro Polity for how populations can be ignorant at the individual level and informed at the macro level? Are you convinced by their argument? Justify your answer.

 

 

 

4. Compare and contrast the conception of partisanship in The New American Voter with that laid out in The Macro Polity. How do these different conceptions of partisanship lead to different conclusions about the impact of party identification on presidential elections?