Reading Assignment:
Richard W. Boyd, "Popular
Control of Public Policy: A Normal Vote Analysis of the 1968 Election,"
American Political Science Review, Vol. 65 (June 1972), pp. 429-449.
Objectives
The objectives of this assignment are:
- to write an SPSS program to generate data with which to do a normal
vote analysis;
- to calculate a simple regression equation from which the expected or
estimated vote of a group can be calculated;
- to construct a graph to display data necessary for normal vote analysis;
and
- to write about the data used in normal vote analysis.
Assignment
In previous assignments you selected a dependent and an independent
variable to explain a relationship. In these assignments, your analysis
was based upon the percentage of a selected group who voted for a candidates
in the 1988 or 1992 presidential election. In this assignment you will
be testing the power of an issue or a candidate orientation to produce
a defection from a "normal" vote (i.e., a vote cast on the basis
of party identification). What issues or which candidates have the power
to effect the out come of an election?
Select an issue or candidate orientation that is present in both election
studies. For this assignment, write a brief paper in which you compare
a similar issue or a measure of candidate orientation from both the 1988
and 1992 election study in terms of that variables power to encourage voters
to defect from their party idenetification. To do this, you will need to
calculate a regression equation to predict the expected Democratic vote
for president that would occur if the ONLY factor people based thier vote
on was party identification.
Methodology
This is the formula to calculate the Expected Democratic:
Vote (EDV). V = .483 + .268 M
where
- V represents the Expected Democratic Vote expressed as a decimal,
and must be converted to a percentage; and
M represents the mean of a distribution of party identification
that occurs for a given issue position.
Computing the expected vote for each group of persons holding a different
position on an issue can be accomplished by completing the following steps:
- Create a new variable (PID);
- RECODE responses to the Survey Research Center measure of party
identification (V09) as follows:
|
Old Code
|
New Code |
Party Identification |
|
1
|
2
|
Strong Democrats |
2
3
|
1
|
Weak Democrats
Independents Leaning
Democratic
|
|
4
|
0
|
Independents |
5
6
|
-1
|
Independents Leaning Republican
Weak Republican
|
|
7
|
-2
|
Strong Republican |
Calculating the Mean
- From the recoded scores, a mean score is calculated for the distribution
of party identifiers within each response category of the issue item.
- The mean score (M) is inserted into a regression equation for
each response category of the selected issue.
- The observed Democratic vote (ODV) is obtained by a crosstabulation
of presidential vote (V2) by the selected issue.
- Two lines are then plotted on a graph. One line is a prediction calculated
by the regression equation and it represents the Expected Democratic Vote
(EDV). The second line is obtained from the appropriate percentages
taken from the contingency table. This line represents the Observed Democratic
Vote (ODV). For the 1992 presidential election, a third line for
Perot will also need to be included.
- If desired, the Expected Republican Vote (ERV) can be obtained by subtracting
the EDV from 100. The Observed Republican Vote (ORV) is obtained from the
contingency table.