Collecting Data

Solutions to Practice Questions

1. A number of confounding factors were proposed in the article. What were they, and why are they confounding factors?

The "treatment" is the race of the driver.  The response is whether or not the drivers were speeding.  A confounding variable, then, will provide an explanation of why speeds might differ among racial groups.  Let's examine the factors mentioned by the Justice Department:

a) glare on windshields:  for this to be a confounder, it must explain why blacks were speeding at different rates.  Presumably, glare would affect both the ability to classify the drivers' race, but unless this somehow meant that white speeders were more often mistakenly classified as black, it is hard to see how this could be a confounder.
b) weather:  again, unless the weather affects both the classification into a race and the speed, this is not a confounder.  It is possible that different racial groups might respond differently to the weather in terms of their driving, and this might cause a disparity in speed.
c) camera placement:  again, unlikely to be a confounder.  This might be addressing a sampling bias problem;  perhaps the location is more likely to pick up black drivers than non black?
d) age.  This is a potential confounder.  Younger drivers drive faster and demographic research showed that the black population is younger than the white population. Thus, age explains the disparity.
e) out-of-state.  Black drivers are more likely to be from out-of-state and out-of-state drivers drive faster since they have been driving longer distances than local drivers.

2. Do you agree that these confounders are valid?

3. Can you think of other confounders?

4. A claim is made that speeders (however defined) are more likely to be black than white. Does this necessarily mean that blacks are more likely than whites to be speeders? Explain.

Not necessarily.  You might try to concoct some numbers that would demonstrate this.

(from Intro Stats by DeVeaux & Velleman)

Science News reported that depressed individuals cry no more often in response to sad situations than nondepressed people. Researchers studied 23 men and 48 women with major depression, and 9 men and 24 women with no depression. They showed the subjects a sad film about a boy whose father has died, noting whether or not the subjects cried. Women cried more than men, but there were no significant differences between the depressed and nondepressed groups.

5. Is this an observational study or an experiment? Why?

This is an observational study. The "treatment" is whether or not the subjects are depressed, and the response is whether or not an individual cries.  The treatment was clealry not determined by the researchers.

For solutions, click here.