Jenny H' Qin

1
Helmet Use Effect Phase II Sensitivity Analysis
on Source of the Data

• Jenny H. Qin
• Kentucky Injury Prevention Research Center

2
Source of the Data
3
Regression Model

• Regression Model
• Charge Base Charge ( Intercept )
• AgeP1
• SexP2
• HelmetP3
• Three independent variables Age, Sex, Helmet
• Age is a continuous independent variable
• Age gt 0
• Sex is a categorical independent variable
• Male1, Female0
• Helmet is a categorical independent variable
• Use Helmet1, Not Use Helmet0

4
Regression Results for Different Cases
5
Effects of Helmet Sex
6
Effect of Age
7
Summary

• The regression results were sensitive to the data
  source for Age base charge is around 23,000,
  total charge decreases several hundred dollars
  for male motorcyclists, and decreases with
  respect to the Age from the Crash file, whereas
  base charge is around 16,000, total charge
  increases several hundred dollars for male
  motorcyclists, and increases with respect to the
  Age from the Hospital file.
• The data source for Sex does not have significant
  impact on the regression results.
• The helmet effect is not sensitive to the data
  source for age and sex. The helmet protective
  effect for every case is similar, the value is
  around 5000.
•  
• When the same sensitivity analysis is run after
  dropping cases with high outlying charges, the
  regression results are more stable. Therefore,
  much of the sensitivity found in this analysis
  may be related to cases with high outlying
  charges.

8
Thank You
9
Regression Results w/o Outliers
   
 
     
10
Effects of Helmet Sex (without Outliers)
11
Effect of Age (without Outliers)