This dataset, previously analyzed by Hochberg (1977), Chen (1989) and Schafer (1997), pertains to the effectiveness of seatbelts for preventing injury in automobile accidents. In a sample drawn from police reports, 80,084 accidents were classified by sex of the driver, degree of damage to the car (low or high), seatbelt use (no, yes) and whether the driver was injured (no, yes). Experience had shown that police tended to overestimate the proportion of drivers who were not using seatbelts and were not injured. To evaluate those potential biases, a followup study was conducted with 1,796 additional accidents. In the followup, investigators obtained more reliable information on seatbelt use and injury from personal interviews and hospital records.
seatbelta data frame with 80 rows and 8 variables:
sourcesource of the data; "sample" indicates
the original sample of police reports, and "followup"
indicates the followup study
sexsex of the driver ("M" or "F")
damagedegree of damage to the automobile
("low" or "high")
belt.pwhether the driver was wearing a seatbelt,
according to the police report ("no" or "yes")
injury.pwhether the driver was injured,
according to the police report ("no" or "yes")
belt.fwhether the driver was wearing a seatbelt,
as determined in the followup study ("no" or "yes")
injury.fwhetherthe driver was injured,
as determined in the followup study ("no" or "yes")