Study Guide and Practice Problems for Political Science 3000 – Final Exam

 

From the first half of the course:

(Note:  While information from the first half of the course will not necessarily be explicitly tested, some of this information will be needed to analyze techniques or issues involved with gathering or analyzing data).

 

** Look over all the terms located at the end of the chapters.  Although you won’t be asked to identify any of the terms in chapters read before the midterm, this terminology may be used in the problems/questions of the final.  The following terms/issues from the first half of the course are especially important:

Interval level of measurement

Nominal (categorical) level of measurement

Ordinal level of measurement

Value

Variable

Case (observation)

Measurement error

Reliability

Validity

Hypothesis

Causal relationship

Theory

Multiple causation

Spuriousness

Dependent variable

Independent variable

Generalizeability

Operationalization

Research design

Experimental design

Control

Pretest/posttest

Sample/population

N

Representative sample

Sampling error

Confidence level

Margin of error

 

**Also, be sure you know the criteria for causality.  You will be asked on some problems “Does this information/statistical results seem to indicate a causal relationship?”  You must know and understand the four criteria for causality in order to answer this question.

 

Now, information from the second half of the course:

 

**Again, it is vitally important that you understand each of the terms from the end of the chapters you have read since the midterm.  You will be asked to define and state the significance of several of these words on the final (similar to the first page of the mid-term).  All words for this portion of the final will come from those lists at the end of the chapters.

 

Additionally, you will need to know/understand information on the following topics:

 

1.  Survey techniques—mail vs. telephone vs. interviews

                You may be asked to compare the pros/cons of the 3 survey techniques – What are the differences/similarities/problems associated with each of the 3 techniques.

 

               

Example Question: 

                A.  Compare and contrast the benefits of using a mail survey as opposed to a telephone survey?  In general, how would you have to change a question used in a mail survey to make that question applicable for a telephone survey?

                B.  Suppose you are doing research for a potential Alabama gubernatorial candidate.  You want to know whether people know her name and what issues she should emphasize in her campaign.  You have been given a large sum of money and have several months to complete the research.  First, what type (mail, telephone, interview) of survey would you suggest using?  Second, given what you know about that type of survey instrument and implementation, sketch out a couple of questions you might use in the correct format to gain the above information.

 

2.  Focus Groups

                Understand, in general, why and when people use focus groups to gain information.  What are the advantages and disadvantages of using focus groups?  How many people should be in a single group and how many different focus groups would you have ideally?  What is the role of the moderator, and how might that person influence the focus groups.  Define bias as it relates to the moderator’s influence.

 

3.  Content Analysis.

                What is content analysis – when and why is it used?  What are some advantages and disadvantages of this method of observation?  What problems exist with coding?  What is intercoder reliability?

 

4.  Direct observation.

                Compare and contrast obtrusive observation, unobtrusive observation, participant observation.  Distinguish between structured and unstructured observations.  What is an observation schedule and why have it?  Why take field notes?  Define reactivity and bias as it relates to observation.  What ethical questions might you come across as you decide how to go about direct observation?

 

Example Question:

                C.  In example question B, pretend that you chose to do in-depth interviewing.  You and several graduate assistants interviewed about 300 people to talk about issues important to them.  You recorded these interviews and have the transcripts sitting in front of you.  What type of analysis would you perform and what difficulties might you face? 

                D.  Suppose, instead, that you chose to do focus groups instead of a survey to get the above information.  You moderated the focus groups in the northern half of the state, and engaged a University of South Alabama graduate student to moderate focus groups in the Southern half of the state.  How many focus groups will each of you do and why?  What difficulties might occur from having two moderators?  How could you overcome some of these difficulties?

                E.  Suppose you are interested in exploring the reasons why persons might join a “hate group”.  Which of the above techniques would you use to explore this phenomenon?  What problems, including ethical issues, might you run into as you gather this data?

 

5.  Describing data and individual variables.  What are frequency tables?  What kinds of tables/graphs might you use to graphically show frequency distributions?  What are measures of central tendency and dispersion – what do they mean and what do they tell you about your data?  When do you use mean, median, mode, interquartile range, standard deviation, and variation ratios?  What is the “normal distribution” and how does it relate to standard deviation?  What is a contingency table?

 

Example Questions:

                F.  You asked all your classmates to place themselves on an ideology scale, where potential responses ranged from 1 to 7 (where 1 is very conservative and 7 is very liberal).  Their responses are the following data points: 1, 1, 5, 4, 2, 3, 3, 2, 2, 1, 1, 6, 4, 7, 3, 2, 1, 7, 2, 2, 2, 1, 2, 2, 2, 2, 2, 5. 

                Using these data points, calculate the mean, median, mode, interquartile range, variation ratio, and standard deviation.

                Calculate the z-score for the data point of 5.  What does this z-score tell you about the score of 5 with respect to the mean?

 

                G.  The following survey question was included on a recent national survey:

                “Mothers should remain at home with young children and not work”    The distribution of responses to this statement follows:

 

Code       Response                              frequency/# answering this response

1              Agree Strongly                                     247

2              Agree somewhat                                  386

3              Neither agree nor disagree                 293

4              Disagree somewhat                             277

5              Disagree strongly                                                297

 

                                1.  What is the level of measurement for this variable?

                                2.  What is the N for this variable?

                                3.  Calculate the appropriate measure of central tendency and dispersion for this variable.

                                4.  Just for the heck of it, calculate the mean and standard deviation of this variable.

 

6.  Bivariate association

                Define measure of association and statistical significance.  What is the correct measure of association and test of statistical significance for each level of measurement (interval, ordinal, nominal)?  What is a cross-tab?  What is lambda, gamma, somer’s d, Chi-Square?  What is a t-test and why would I use it?  What is correlation coefficient and when do I use it? 

 

Example Questions:

                H.  In the table below are responses by whites and blacks to a statement in the 1996 American National Election Study about equality.  The statement is “One of the big problems in this country is that we don’t give everyone an equal chance”  The responses are as follows:

                                                                                Whites                   Blacks

                Agree strongly                                     181                          91

                Agree somewhat                                  390                          60

                Neither agree nor disagree                 224                          9

                Disagree somewhat                             331                          13

                Disagree strongly                                                145                          6

 

                                1.  Calculate the percentage frequencies for each response category.

2.  Based on your analysis of the frequencies and percentages, do you think there is evidence of a “racial divide” on perceptions of equality in the United States?

                                3.  What statistical technique might you use to see if a “racial divide” exists?

 

                I.  Let us suppose that I was interested in seeing if students who took my poli3000 course did better on the GRE than other liberal arts students who did not take my Poli3000 class.  I was able to determine the following information.

Mean GRE score for my poli3000 students:  1300           Standard dev:  85                 N: 35

Mean GRE score for other students:                1270        Standard dev:  90                 N:100

Using the appropriate statistics, is there a statistically significant difference between Poli3000 students and other students?

 

J.  I am interested in determining if there was a gender gap in the last election.  That is, is there a relationship between gender and what Presidential candidate that person voted for (Gore, Bush, or Nader)

In general (no numbers in the cells) what would a crosstab look like – Identify the cells, marginals, and N (or where they would be on the table).  If the Lambda was .42, what would this mean?  Second, if the Chi-square value for this crosstab was 2.52, what would this say about the statistical significance of any relationship/pattern that emerged in the table?

 

 

 

                K.  Suppose you wanted to know if there is a relationship between the unemployment rate in cities and the number of robberies that occur within the city limits.  You take a random sample of cities and plot the data points for these two variables on a two-dimensional grid (unemployment rate in percent is on the X-axis and number of robberies per month is on the Y-axis)  You then run a regression and fit the line to the data points.  You come up with the following regression line:  Y= 50 + 20X.  Then, the correlation coefficient is calculated – r=.70.

                                1.  Sketch (roughly) the diagram you might expect on the grid.

2.  What is the strength and direction of the relationship between these two variables?  How do you know this and what does it mean?

3.  If the unemployment rate in a given city increases by 1 percent, about how many additional robberies will occur?

4.  If there was no unemployment in the city, how many robberies would you expect to occur?

5.  Does this statistical analysis demonstrate a causal relationship?  Why or why not?  What other factors might influence robbery rates?  How would you rule out or account for these other factors?

 

                L.  Assuming that PARTY is coded 0=Democrat and 1=Republican and AGE is coded 1=less than 35, 2=35-50 years old, 3=51 to 65 years old, and 4= older than 65

1.  What sign would gamma need to have for the following hypothesis to supported by the bivariate data:  “The older a person is, the more likely they are to be a Democrat”.

2.  In comparing gamma and Somer’s d, what difference would you expect between their values?

 

7.  Multivariate relationships/multivariate causal analysis

                Why use multiple regression?  In general, what does this technique do – why can it help rule out spuriousness?  What is the B coefficient?  What is the standardized coefficient (beta)?  What is the r-squared?

 

Example Problem:

                M.  Suppose we are interested in determining what factors are related to the salaries of state agency heads.  We found the following information in a survey from the American State Administrators Project

 

Dependent Variable:  State Agency Head’s Salary (in dollars)

Independent Variables:

                1)  Size of Agency Headed by Administrators (in numbers of employees)

                2)  State in which Agency Head Works (Coded 0=Non-Southern State; 1= Southern State)

                3)  Gender of Administrator (Coded 0=Male; 1=Female)

                4)  Years employed in State government (in years)

5)  Education (coded 1=High school; 2=some college; 3=Bachelor’s degree; 4=graduate study;

 5=graduate degree)

 

You decided to run a regression including these five Independent variables and get the following results:

 

Model                                    B                             Beta                        Sig.

Size Agency                          1.733                       .141                         .000

State                              -1200.456                         .162                         .095

Gender                              164.593                        .002                         .951

Years Employed              226.852                        .059                         .050

Education                       3395.127                        .099                         .002

 

r-squared=.23

 

               

 

1.  Write out the regression equation.

                2.  Analyzing the unstandardized regression coefficients, what variables seem to influence a state agency head’s salary level?  How do you know this?

                3.  Using the standardized coefficients, which variables seem to have the MOST impact on salary levels?  How do you know this? 

                4.  Your father is a state agency head for Alabama Department of Transportation.  He has never told you his salary.  You know that the Department of Transportation has about 12000 employees, your father earned his master’s degree in PUBLIC ADMINISTRATION, and is one year away from retirement.  Using the regression results above, what would you estimate his salary to be?

                5.  What does the r-squared tell you in general?  What does the r-squared of .23 tell you about this model.

 

Good Luck!  I’ll be here for Questions and for that special study session.