Final Exam by VISAKH


									APLS 301 Dr. Botsch                   Test 2/Final Exam                          Fall 2004

             Due on the day of the final exam by 5 pm, Wednesday, Dec 8

Honor Pledge: On my honor as a USCA student, I have neither given nor received any
unauthorized aid on this exam.
                                 Name _____________________________

NOTE: 10 of the 100 points on this test are for neatness and clarity.

Use the following data for questions 1-9.
                                          Church attendance
Case # Fundamentalism* Educ:level-Ed. years in last month

 1            f                 low - 9                1
 2            f                 hi - 13                4
 3            f                low - 10               4
 4            f                 hi - 14               3
 5            f                low - 11               1
 6            f                 hi - 13               2
 7            f                low - 12               4
 8            f                low - 12               3
 9            f                low - 12               2
10            f                 hi - 16                4
11           n                 low - 11               3
12           n                  hi - 15               1
13           n                  hi - 16               2
14           n                  hi - 13                2
15           n                 low - 12                0
16           n                  hi - 16               4
17           n                 low – 11                0
18           n                   hi - 13              1
19           n                  low - 10              2
20           n                  low - 12              0
      Note: * f is fundamentalist and n is non-fundamentalist

1. What is the level of measurement of each of the following variables in these data
(nominal/ordinal/interval/dichotomous)? (8 pts)
Fundamentalism: ______ Educ Level: _______ Educ Years: ___________
    Church attendance: __________

 2. The most appropriate kind of statistical measures to use to find out whether
fundamentalists have more years of education than non-fundamentalists would be
______________________. (4 points) (Hint: You're comparing the "typical" cases of both
    a. measures of dispersion
    b. measures of central tendency
    c. frequency distributions
    d. percentage distributions

 3. The most appropriate kind of statistical measures to use to find out which group has
greater equality in church attendance would be ________________________. (4 points)
(Hint: you're comparing the equality of the two groups.)
    a. measures of dispersion
    b. measures of central tendency
    c. frequency distributions
    d. percentage distributions

4. a. Set up a table that compares means in order to test the following hypothesis:
Fundamentalists attend church more than do non-fundamentalists. Ignoring statistical
significance (no chi squares), does the table show support for this hypothesis? Explain (that
is, write the appropriate sentences explaining how to read the table). (6 points)

  b. Explain why comparing means is preferable to testing the relationship using a
percentage crosstabulation. (2 points)

5. a. Suppose we were to use educational level as a control variable for the bivariate
relationship in the previous question. Logically or theoretically, what role is most reasonable
for the control variable to play? _____ (2 points)
     a. confounding (spurious)
     b. conditioning
     c. intervening
     d. either a or b
     e. either b or c
     d. educational level could play no logical role

  b. Create the control tables for the bivariate relationship between fundamentalism and
church attendance using educational level as the control variable. Show all the appropriate
control tables. Interpret each table--tell me how to read it. Draw and explain a conclusion
about what role the control variable actually plays in the bivariate relationship. Show the
path diagram. (Ignore statistical significance--do NOT compute chi squares. Make sure that
each of your control tables have the same form as the original bivariate table with which you
are comparing it, i.e. compare means, not a crosstab!) (6 points)

6. a) Construct a percentage crosstabulation to test the following hypothesis and test the
significance of your conclusion using the chi square test. Hypothesis: Those with higher
education are less likely to be fundamentalists while those with low education are
more likely to be fundamentalists. (Use education level, not years of education.) Write a
sentence explaining your interpretation of the table and how you interpreted the chi square.
Show ALL of your work, including how you calculated the chi square and turned this into a
measure of statistical significance. (10 points)

   b) Your conclusion from this analysis introduces new evidence about the relationship
among the three variables we have been examining (education, fundamentalism, and
church attendance). Does this modify the path diagram you had in #5b? If so, redraw it and
explain your reasoning. (2 points)

7. Compute the variance and standard deviations of church attendance for fundamentalists
and then for non-fundamentalists. Show all of your work in neat tabular form. Round off to
two decimal places (nearest hundredth). I expect you to correctly work out the math on this!
(8 points)

8. Suppose you compared these variances and/or standard deviations for non-
fundamentalists and fundamentalists. What would you conclude from this comparison?
Why? (Hint: look back at questions 2 and 3. One of these should help you. And this should
help you with one of them!) (2 points)

9. Construct the following for the entire sample:
  a. the frequency and percentage distribution of church attendance in tabular form (4
  b. the percentage distribution of in the form of a bar graph educational level (not years) (4

10. a) Construct a scatterplot between education years (as the independent variable) and
the frequency of church attendance (dependent variable). Make sure you label each axis in
the graph. Does the pattern suggest any relationship? Explain. (5 points)

   b) What new evidence does this scatterplot suggest (assuming statistical significance)
about the relationship among our three variables (education, fundamentalism, and church
attendance)? Does this modify the path diagram you had in #6b? If so, redraw it and
explain your reasoning. (2 points) ______________

11. a) How can you tell whether a variable plays a confounding role or an intervening role in
an path diagram? (Hint: the best answer is a single word -- 2 points)

  b) Which of the following should you do first when analyzing a hypothesis? (2 points)
            A. Look at the effect of possible confounding variables that could render a
            relationship spurious
            B. test the bivariate relationship
            C. look for conditioning effects
            D. look at the effect of possible intervening variables.

  c) What should you do second? (2 points)

Use the tables shown below for questions 12-15.

 National Sample
                            Party Identification
 Pres. Preference     Democrat Independent Republican

   Kerry                 80%          55%           30%

   Bush                  20%          45%            70%

                        100% (200) 100% (200) 100% (100)

 Northern Subsample
                            Party Identification
 Pres. Preference     Democrat Independent Republican

   Kerry                 90%           65%           35%

   Bush                  10%           35%           65%

                        100% (100) 100% (125) 100% (75)
 Southern Subsample
                           Party Identification
 Pres. Preference     Democrat Independent Republican

   Kerry                 70%           38%           16%

   Bush                  30%           62%           84%

                        100% (100) 100% (75) 100% (25)

12. Draw an arrow diagram to show the relationship that was tested by these tables. Label
the variables and describe the role that each plays (e.g. independent variable, control, etc.)

(6 points)

13. What conclusions should be drawn from the tables about the hypothesis that was tested
(ignoring statistical significance)? Make sure you discuss the impact of the control variable
on the original bivariate relationship. (6 points)

14. True or False? The number of Southern Independents who voted for Kerry is more than
the number of Northern Independents who voted for Bush. (3 points)

15. Construct a crosstabulation showing the bivariate relationship between region and
voting choice. Show all frequencies, percentages, and marginals. Ignoring statistical
significance (no chi squares), what conclusions would you draw? (8 points)

Bonus (5 points): Suppose you are looking at the effects of education (measured in 5
groups: <hs, hs, some col, col, and col+) on political efficacy (measured in 4 levels: low,
mod, hi and very hi). When you produce a crosstabuluation you have several cells with very
few or no cases in them. What should you do?


To top