Multiple Choice Questions. Choose the one best answer and circle it on the test form.

• Stratified sampling: a. requires that we know the actual proportions of each stratum in the population b. helps us to learn the actual proportions of each stratum in the population

• Simple random sampling: a. requires that we know the actual proportions of each stratum in the population b. helps us to learn the actual proportions of each stratum in the population
• PPS stands for: a. Pied Piper Sampling b. Population Parameter Sampling c. Probability Proportional to Size d. Purposive Population Sampling e. Parametric Probability Sampling
• The range within which we estimate the statistic to depart from the parameter is known as the: a. confidence interval b. confidence level c. sampling range d. statistical range
• In stratified sampling, the proportion of each group sampled is determined by the size of that group in the population: a. true b. false
• A sample assembled by referral as persons having the characteristic of interest identify others is called a: a. self referential sample b. self assembling sample c. auto assembling sample d. characteristic driven sample e. snowball sample
• A sample in which all members of the population have an equal chance of being selected is a: a. systematic sample b. quota sample c. stratified random sample d. simple random sample e. cluster sample
• Suppose you stood on a street-corner and interviewed people as they came by. What type of sampling is this? a. simple random sampling b. cluster sampling c. quota sampling d.systematic random sampling e. none of the above
• A researcher selects a random sample of 1,000 persons from a city of 300,000 people, a random sample of 1,000 persons from a city of 1,000,000 people, and a random sample of 1,000 persons from a city of 3,000,000 people. Which of the following statements is most correct?

a. The sample of the city with 300,000 people will be much more accurate than the other samples.

b. The sample of the city with 1,000,000 people will be much more accurate than the other samples.

c. The sample of the city with 3,000,000 people will be much more accurate than the other samples.

d. There won't be much difference in the accuracy of the three samples.

• Which of the following would be the best (the most practical) type of sample to select for a national face-to-face survey? a. a simple random sample b. a stratified random sample c. a systematic random sample d. a cluster sample


Page two:

 
1 1.25 In a college class with 125 students, 68 of whom are male, the mean on the midterm was 75. The standard deviation was 7. What is the margin of error for this mean?

2. 400 A researcher wants to obtain a margin of error of no more than 5% in a survey of a county with a population of 30,000. How large a sample is needed?

3. 2.92% In a survey of 1000 voters, 600 were Democrats, 300 Republicans and 100 Libertarian. 85% of the Democrats favored Al Gore in the primary. What is the margin of error for this percentage?

4. 2000 or 5 times 400 A survey is to be conducted of attitudes residents of five counties in southern New Jersey. The researcher wants to achieve a 5% margin of error for the estimates for each county. How large a sample is needed?

In a survey of community residents, the mean income was $38,745 with a standard deviation of $3,345. There are 270,000 residents in the community. 1000 were sampled in the survey.
 

5. $211.56 What is the margin of error for this mean score?
 

6. $38533.44 What would the lower bound of the confidence interval be for this mean?
 

7. $38956.56 What would the upper bound of the confidence interval be?

8. .252 There were 85 students in a survey research class, and they completed a mean number of 4.76 interviews, with a standard deviation of 1.16. Treating this class as a sample of the population of survey research classes, what would the margin of error be on this mean?

Page three:
Consider the following results from a regression analysis:

Dependent Variable: INFANT MORTALITY
N: 50 Missing: 0
Multiple R-Square = 0.516 Y-Intercept = 4.266
Standard error of the estimate = 1.026
LISTWISE deletion (1-tailed test) Significance Levels: **=.01, *=.05
Source Sum of Squares DF Mean Square F Prob.
REGRESSION 51.718 3 17.239 16.378 0.000
RESIDUAL 48.418 46 1.053
TOTAL 100.136 49
           Unstand.b Stand.Beta Std.Err.b t
CIGSMOKERS 0.003     0.007     0.056     0.062
POV LINE   -0.191   -0.439     0.062     -3.100 **
TEEN MOMS   0.417   0.928     0.069     6.052 **
 

Answer the following questions:
 

1. CIGSMOKERS, POVLINE and TEEN MOMS What are the independent variables in this last analysis? Note: one point item, all three must be given to get the one point.

2. INFANT MORTALITY What is the dependent variable in this analysis?

3. TEENMOMS Which variable is the best predictor of the dependent variable?

4. 51.6% What percentage of the variance in the dependent variable is explained in this analysis?

5. Fill in the blanks in this formula: Infant Mortality = 4.266 + ( .003 * Cigarette Smokers ) + ( -.191 * Percent of Population Under the Poverty Line) + ( .417 * Percent of Teen Moms). Note: four points for this item, one for each answer.
 

6. 8.11 What would the predicted infant mortality rate weight be for a state with 25% cigarette smokers, 13% under poverty and 15% teenage mothers? Note: two point item.

IMPORTANT NOTE: When using a regression equation, you keep the independent variables in the format that was used to compute the equation. If the variables were percentages, you use them in percentage form. if they were rates, e.g., infant mortality rates, you keep them as rates. This is different from the "margin of error" computations when percentages are converted to proportions before using the formulas. The results of the regression equation are in whatever unit of measurement was used when the equation was computed. In these equations, the results are infant mortality rates, which are rates per 1000. They are not percents, and your answer is really wrong if you add a % sign to it.
 

To get the answer to question 6: Infant Mortality Rate = 4.266 + (.003 * 25) - (.191 * 13) + (.417 * 15). Note: the percents are left as whole numbers. Also, note the minus sign for the .191, because this coefficient is negative. You might also note that this equation makes little sense if treated as a causal analysis, why would the effect of poverty on the infant mortality rate be negative? It works out that way in the multiple regression, probably, because the poverty rate and the percent of teen mothers are so highly correlated with each other.
4.226+
7. 7.34 What would the predicted infant mortality rate weight be for a state with 15% cigarette smokers, 6% under poverty and 10% teenage mothers? Note: this is answered in the same way as the previous item, only the percents are different. Again, they must be left as percents since that is how the equation was calculated.

IMRate = 4.266 + (.003 * 15) - (.191 * 6) + (.417 * 10)
                  4.266  + .045  - 1.146 + 4.17 =

8. 50 How many cases are there in this analysis?