Assignment Three
Objectives:
Students will continue to become familiar with basic features of SPSS
v.11.5. After you have created an SPSS data set by inputting and
saving
data, you will use linear regression analysis to predict outcomes of
dependent
variables.
Part One: Creating an SPSS Data Set
·Use the assignment sheet from the first assignment
to create an SPSS data set based on data from the handout. Click
here if you need a copy of the data
sheet.
·Make certain you understand what each variable
means.
·For this assignment, the dependent variable is
"voter turnout." The independent variables are: unemployment in
1993,
percentage supporting Perot, percentage of seniors in the population,
and
percentage living in a metropolitan area as determined by the US
Census.
·After creating your data set, answer the following
questions:
1. What might be the "theory" behind considering each of these independent variables? That is, why might they have something to do with voter turnout?
2. What is the "level of analysis" of the data here? Does this make a difference for testing the "theory" in part (1)?
Part Two: Using Linear Regression to Predict Outcomes (Dependent
Variables)
Now we will use the independent variables given to predict outcome
effect on the dependent variables. We will do this by first
completing
a regression model.
·Go to "Analyze" and then to "Regression.
"Choose the option "Linear" since we are doing a linear regression
analysis.
Plug in your dependent variable and all of the independent variables
and
compute.
·For the following questions, do a simple
regression
for each independent variable alone with the dependent variable
following
the same steps. Save all output to turn in (remember to
save
data files and output files separately).
·Answer the following questions.
1. Try a regression model with all the independent variables included. Which variables are statistically significant? Why? For those that are statistically significant, discuss the "practical significance" of the b from the output.
2. What statistic helps show how well the model "fits" the data? What is its value here?
3. Try each independent variable in a SEPARATE simple regression
with
the dependent variable (so this is 4 separate
regressions). Explain any differences between results in these
4 separate regressions and in the one multiple regression done
earlier.
4. Look back at the output for the multiple regression. One state that was not included in the sample was Rhode Island. RI had 7.7% unemployment, 23.2% Perot vote, 15.6% elderly, and 93.6% living in Metro areas. From the regression model, what is the predicted turnout for RI? Use a calculator. (The real turnout was 59.0%.)
5. Sum up the substantive conclusions of your analysis.
Turn in: