December 14, 200
Sociology 2080 
Statistics for Sociologists 
Name_______________________________

I pledge that I have neither given nor received unauthorized help.
 

 Signature_____________________________


 
FINAL
Each part of a problem counts equally. The total number of points possible is 34. Show your equations and calculations: Arithmetic errors will not count if the procedure is correct. This is an open-book, one 4"*6"-notecard-of-notes exam. Feel free to use a calculator. Remember to indicate units of your answers as appropriate. If you have questions about the wording of a problem, please ask me for clarification. Notice that in some cases, you are just asked to interpret results or show how you'd verify them, rather than to do something new. You may want to read over the whole exam and do the problems you find easiest first.
 

1. In one UNC dormitory, the expected number of votes for Gore for the women in the dorm, E{X}, was 250 and for their mothers, E{Y}, it was 200. The signma2 {X} was 49 and sigma2{Y} was 36. The covariance, sigma{X,Y}, in the votes of these women was 30.

(a.) What is the expected number of votes for Gore from these families?
 
 
 
 
 
 
 
 
 

(b) What is the variance of this total?
 
 
 
 
 
 
 
 
 

(c) What is the correlation of the votes of the college women and their mothers? What does this say about voting patterns of mothers and daughters?

 

2. Two presidential candidates are contesting election results from a southern state. Let's call them George and Al. One, George, has a certified lead of 538 votes. The other candidate, Al, hires you as a statistical consultant. You examine data from exit polling of 10,000 people. On the basis of this sample, you estimates that Al would gain, on average, 573 votes from a hand recount, with a variance in the number of votes = 60. Would you say that it was very probable that Al would win if a recount were allowed and urge him to continue to push for this? You want to be pretty confident in what you tell Al, say 95% certain.

(a) What would be your null and alternative hypotheses?
 
 
 
 
 
 
 

(b) What is the region of rejection?
 
 
 
 
 
 
 

(c) What is the value of the test statistic?
 
 
 
 
 
 
 

(d) What is your conclusion about the election results with a recount?
 
 
 
 
 
 
 

(e) What assumptions did you have to make to do this test?
 
 
 
 
 
 
 
 

3. You send your undergraduate methods students out to do a survey of gun ownership. You know from national data that the probability of gun ownership is as follows:
 
own a gun: probability
yes .40
no .60

Your 10 students go out and independently ask one person they meet whether s/he owns a gun.

(a) On the basis of what you know from the national data, how many gun owners would you expect to find in the students' study?
 
 
 
 
 
 
 

(b) What would be the variance?
 
 
 
 
 
 
 

(c) What would be the probability of finding that 6 or 7 of the 10 respondents were gun owners?

Write out the equations you would use and calculate the number, then check the result from the tables.
 
 
 
 
 
 
 
 

(d) What does the probability distribution of number of gun owners out of 10 look like (graphically)?
 
 
 
 
 
 
 
 
 
 

4. Do women feel less successful in their work lives than men? You ask the question, "How successful are you in your work life," with 5 possible responses:
1=not at all successful
2=not very successful
3=somewhat successful
4=very successful
5=completely successful.

Your results from your survey are:
 
 
   X-bar s
women 16 3.35  .83
men 15 3.55  .81

(a) What would be your null and alternative hypotheses?
 
 
 
 
 
 

(b) What is the value of your test statistic? Be sure to show clearly all the calculations it takes to get this.
 
 
 
 
 
 
 
 
 
 

(c) If you were using alpha =.10, what would you conclude based on the p-value for the test statistic?
(Use the closest p-value in your tables.)
 
 
 
 

(d) What assumptions did you make to do this test?
 
 
 
 
 
 
 
 

5. You have data on 11 Western countries. You are interested in the difference in how much countries spend on old age benefits vs family benefits. You have the following data from your sample (in 1980 $s per eligible person):
 
n    X-bar  s
Old age/survivors/disability benefits (per person over 64)  11  5977 1401
Family transfers (per child under 15)  11 599 258

The correlation between the two variables is .07.
 

(a) Construct a 95% confidence interval for the average difference between old age and family benefits. Be sure to show the pieces you calculate to do this.
 
 
 
 
 
 
 
 
 
 
 

(b) In words, what does this mean?
 
 
 
 
 
 
 
 
 
 

(c) What assumptions did you make to calculate this interval estimate?
 
 
 
 
 
 
 
 
 
 
 

6. One concern of sociologists is the effect of family background on social resources. One way this has been studied is by looking at the effect of parental (usually paternal) socioeonomic status on the child's education. You do a regression to check this out. EDUC is years of education the respondent has and PASEI is a measure of father's socioeconomic status, which can theoretically go from 0 (very low status occupation) to 100 (highest status occupation). You get the following Stata output:

. regress educ pasei
Source      | SS         df         MS Number of obs =   2320
------------+------------------------------  F(  1,  2318) =  396.66
Model | 2910.89949          1      2910.89949  Prob > F = 0.0000
Residual | 17010.5419   2318       7.33845638 R-squared = 0.1461
  ---------+------------------------------
        Total | 19921.4414 2319 8.590531

------------------------------------------------------------------------------
        educ |      Coef.         Std. Err.             t          P>|t|        [95% Conf. Interval]
-------------+--------------------------------------------------------------------
         pasei |    .0601458    .0030199      19.916    0.000 .   0542238 .   0660678
        _cons |    10.65368    .1543069      69.042    0.000    10.35109    10.95627
------------------------------------------------------------------------------
(a) What is the correlation between education and father's SEI?
 
 
 

(b) How much variation in education is explained by father's SEI? Is this a little or a lot or what?
 
 
 

(c) What do "Coef." for PASEI and CONS estimate?
 
 
 
 

(d) Is the effect of father's SEI on education statistically significant? At what level, if it is significant?
 
 
 
 

(e) In words, describe the effect of father's SEI on respondent's education.
 
 
 
 
 
 
 
 


7. You know that µ=10. You have two estimators of µ:
 

M1, with E{M1}=9 and sigma2 {M1}=10
M2, with E{M2}= 6 and sigma2 {M2}= 3.

Which estimator do you prefer and why?
 
 
 
 
 
 

8. The number of students who speak in class on any given day is a discrete uniform random variable X., with parameters a=1 and s=11.
 

(a) What is E{X}?
 
 
 
 
 
 
 

(b) What is sigma2 {X}?
 
 
 
 
 
 
 

(c) What is P (2<X<=4)?