Jennifer part 2

Jennifer is now at the new traffic junction which reported a lot of accidents. She decided to make another count.

"Hmm.... 0.4% accident when turning left, 1.1% if going straight, 0.05 when turning right," she noted and wondered, "This is confounding!" Comparing to part 1, why is it confounding?

She thinks about the theories that she had known, and thought, "Vision theories can explains," continuing, "There is a mountain on the left side of the road, which blocks out sunlight most time of the day. The lack of light is probably the culprit." Is vision theory appropriate?

She took some readings of illumination and found the right side of the road is indeed brighter. She did statistics and found correlation between brightness and accident is positive at 95% confidence. Is this the result we wanted?

Jennifer Part 1

Jennifer is a researcher, and someone from the traffic department approaches her. A particular junction is found to be hazardous.

She took the job and investigated the junction. She made a count, that cars that turn right have 1% chance of meeting an accident. Going straight is 0.5% and left is 0.2%. So her theory goes that 'right turn causes accident.' True?

Jennifer subsequently measured the hardness and gradient of the right turn. Models of the car that came by. And radius of the turn. Has she come closer to reality?

Soon, people starts to loose interests in the junction, but yet another one caught attention. Now, she being an expert in traffic accident investigation, is tasked to look at that junction too. Is she going to do better because of her experience?

How much are we a researcher like Jennifer? What's wrong with her approach?