Lab 4 – Infections and elections
ANSWER THE FOLLOWING QUESTIONS & DO THESE PROCEDURES. ALL ANSWERS SHOULD BE INSERTED INTO THIS WORKSHEET. THIS IS ALL THAT WILL BE TURNED IN. YOU WILL NOT TURN IN YOUR SPSS OR EXCEL FILES!
STUDY DESCRIPTION
Choices that individuals make in the voting booth, such as whether to support a more conservative or
liberal candidate, may be affected by a number of factors. Individual voting histories, the policy positions
of the candidates, and current events may each shape voters’ preferences. In their research, Beall, Hofer,
and Schaller (2016) sought to examine the role of outbreaks of infectious diseases on voting behavior.
The authors hypothesized that an outbreak of a disease, such as Ebola, may increase support for more
conservative political candidates.
To test this hypothesis, the authors examined the frequency of google searches for “Ebola” during the
weeks prior to and after the outbreak of Ebola that occurred in 2014. The authors also examined support
for the two major political parties in the United States (the conservative Republicans and the liberal
Democrats) by collecting and aggregating polls of voters into a single score called the voter intention
index.
(Kevin P. McIntyre, 2021. Open Science Lab)
ANALYSIS
Question #1.
- How many total observations are in this sample? ____________________ [LIST NUMBER HERE]
- Does each measured variable have the same number of observations? If not, which variable has the greatest number of observations, and which has the least? [INSERT YOUR ANSWER HERE]
- This data set has multiple variables. Based on the description above, which two variables will we focus on in this lab? [INSERT YOUR ANSWER HERE]
- BONUS: why do some variables not have observations for every day? [INSERT YOUR ANSWER HERE]
Question #2. We have several dependent variables in this data set. A description for each is listed below. Let’s determine if any of the dependent variables have a “normal” distribution.
- Month = the month of the observation
- Date = the date of the observation
- Republican.Support = support for republican party based on nationwide poll
- Democratic.Support = support for the democratic party based on nationwide poll
- Intention.Index = voters’ intention leaning more conservative (positive) or more liberal (negative)
- Search.Volume.Index = frequency of Ebola searches on google
- Search.Volume.Index = frequency of ISIS searches on google
- DJIA = Dow Jones Industrial Average
- LexisNexisNewsVolume = number of Ebola related news articles
Click on Analyze -> Descriptive Statistics -> Descriptives. Put all variables aside from Month and Date in the “Variables” box. Click on “Options” and check “kurtosis” and “skewness.” Then hit “continue” and “ok”. The closer the skewness and kurtosis values are to 0 the more “normal” they are. If they are each approximately between -2.0 and +2.0, then the distribution is probably ok (rule of thumb).
Are any of the variables not normally distributed? If so which ones? [INSERT YOUR ANSWER HERE]
Question #3. We will test the relationship between infection scare (Ebola.Search.Volume.Index) and voting intentions (Voter.Intention.Index). We will use bivariate correlation to test this association.
Analyze -> Correlate -> Bivariate. Put both variables in the variables box. Under the “Correlation coefficients” box check “Pearson” IF the data are normal and check “Spearman” IF the data are not normal.
Describe your results in words and a statistical summary that includes the correlation coefficient, degrees of freedom, and p-value (e.g., r(df) = #, p = #).
[INSERT YOUR ANSWER HERE]
Question #4. Next, we will determine if any of the other variables are related to voter intentions.
Analyze -> Correlate -> Bivariate. Put all variables of interest (NOT Month or Date) in the variables box. Under the “Correlation coefficients” box check “Pearson” IF the data are normal and check “Spearman” IF the data are not normal.
Describe your results in words and a statistical summary that includes the correlation coefficient, degrees of freedom, and p-value (e.g., r(df) = #, p = #). Tip: this can be described in one sentence (e.g., height is related to weight (stats) and age (stats), but not the school attended (stats)).
[INSERT YOUR ANSWER HERE]
Question #5. The authors were interested in how the breakout of Ebola announced on September 30, 2014 influenced voter intention. Let’s analyze the data both before and after the breakout to see if this event changed voter intentions. We need to filter the data to only include data that occurred before the announcement. We will use a linear regression to answer this question.
Filtering: Data -> Select cases -> in the “Select” box chose the “If condition is satisfied” option. Click on the IF box below. Type “Month = 9” into the box, then press “continue” and “ok.” You should now see new message in the Output window indicating a new filter has been applied. You will also see rows crossed out in the Data view window that will not be included in the analysis.
Analysis: Analyze -> Regression -> Linear. Put Voter.Intention.Index in the “Dependent” box and Date in the “Independent” box. Click “ok.”
Describe your results in words and a statistical summary that includes the standardized beta, degrees of freedom, t-statistic, and p-value (e.g., β = #, t(df) = #, p = #).
[INSERT YOUR ANSWER HERE]
Question #6. Next, let’s analyze the data after the breakout to see if this event changed voter intentions. We need to filter the data to only include data that occurred after the announcement.
Filtering: Data -> Select cases -> in the “Select” box chose the “If condition is satisfied” option. Click on the IF box below. Type “Month = 10” into the box, then press “continue” and “ok.” You should now see new message in the Output window indicating a new filter has been applied. You will also see rows crossed out in the Data view window that will not be included in the analysis.
Analysis: Analyze -> Regression -> Linear. Put Voter.Intention.Index in the “Dependent” box and Date in the “Independent” box. Click “ok.”
Describe your results in words and a statistical summary that includes the standardized beta, degrees of freedom, t-statistic, and p-value (e.g., β = #, t(df) = #, p = #).
[INSERT YOUR ANSWER HERE]
Question #7. Use SPSS to create a scatter plot of the Voter.Intention.Index before and after the breakout. Make sure to include lines of best fit.
Graphs -> Legacy Dialogues -> Scatter/Dot… -> Simple Scatter -> Put Voter.Intention.Index in the Y-axis variable and Date in the X-axis variable and hit “ok.” To add the line of best fit, right click on the image -> Edit content -> In separate window -> click the “Add fit line at total” button and uncheck the “Attach label to line.” Hit “Apply.”
[INSERT BOTH YOUR GRAPHS HERE]
