Transportation Statistics and Microsimulation

Experimental Design    ◾    251

years, we would have many other confounding factors. Changing speed

limits, vehicle and road design changes, weather dierences, and trac

volume changes all would provide challenges to the analysis. At least

for challenges to internal validity, this design provides protection for

regression and selection biases. See Campbell and Stanley (1963) for

more detail.

e nal design that we consider is the pretest-posttest control group.

It is symbolized by

R

R

O

O

XO

O.

e subjects available (perhaps all volunteers) are randomly assigned to

treatment and control (nontreatment) groups. Both groups are observed

before and aer implementation of the program. is design mitigates

most of the threats to internal validity presented, ...