I understand one is not paranoid if there’s a legitimate fear that something bad can happen. So I guess I’m not technically paranoid about getting the flu this year, but I have to admit I find myself taking more precautions to avoid getting the flu this year. This motivated me to think about the characteristics of people who are more or less likely to get a flu shot to decrease the probability of avoiding the undesired outcome.
I turned to one of my favorite datasets, Behavioral Risk Factor Surveillance System (BRFSS), and found the survey had asked the question, “During the past 12 months, have you had a seasonal flu shot?” How are the following characteristics related to the answer to that question:
- General health
- Educational attainment
- Medical cost (Was there a time in the past 12 months when you needed to see a doctor but could not because of cost?)
As the variable we are interested in understanding, flu shots, is a binary variable, logistic regression is a preferred statistical tool of choice.
I will summarize the findings in graphs, depicting the probability of getting a flu shot based on the predictor variable characteristics.
A. Educational Attainment
Individuals with higher levels of educational attainment are more likely to have received a flu shot in the past year. However, as age increases the probability for each educational group decreases.
White people are more likely to get a flu shot than non-whites, but again age reduces the probability over the lifespan.
C. Medical Cost and Gender
Males and females who said there wasn’t a time in the past 12 months when they needed to see a doctor but could not because of cost had higher probabilities of receiving a flu shot than those who responded with a yes to the question. Females who said there was a time in the past 12 months when they needed to see a doctor but could not because of cost had the lowest probabilities of receiving a flu shot in the past 12 months. Again, increasing age reduces the probabilities over the lifespan.
Based on the above findings it’s not surprising that we observe people with higher incomes have higher probabilities of receiving a flu shot, albeit those probabilities decrease with age.
E. What about a Nonlinear Relationship with Age?
Social scientists have observed nonlinear relationships of age with many dependent variables. To determine if that’s the case in this analysis I added age as a quadratic variable (square of age) to the logistic regression.
The newly constructed variable is statistically significant and it demonstrates that as age increases past 55 the probabilities of getting a flu shot increase for all levels of general health. This pattern holds for all the other covariates.
So now we know who has the most and less likely probabilities of getting a flu shot in the past 12 months.
Here’s another factoid I recently read. In a study to identify where the most likely germs can be found in our daily lives the number one spot was your restaurant menu! The second most likely spot – your restaurant chair!