In: Statistics and Probability
1. In the following observational studies, describe changes that could be made to the data collection process that would result in an experiment rather than an observation study. Also, offer suggestions about unseen biases or lurking variables that my be present in the studies as they are describe here.
a. In a sample of 50 members of a local health club you find that 12 of these members meet weekly with a physical fitness trainer and that the average body mass index (BMI) of these 12 members is less than the average BMI of the other 38 club members in your sample.
b. In a sample of 12 bank tellers at a local branch office, the 7 tellers who have completed the advanced training program offered by the bank have a lower average error rate in the processing of transactions than the remaining 5 tellers.
a.
Experiment:
We have to divide the members into two groups: a treatment group and a control group. It should be ensured that the initial BMI distribution should be the same. This means that both groups should have equal number of members of low BMI, normal BMI, and high BMI. The treatment group will get the physical fitness trainer while the control group will get to exercise without a trainer. The BMI of the two groups will be compared after say 8 weeks.
Biases:
1. Sample selection may have been biased. Those having a high BMI despite having a trainer may have been removed as members.
2. Also, those with low BMI and no trainer may have stopped coming to the local health club.
Lurking variables:
1. It is possible that those who met the physical fitness trainer were already healthy and active and are already having a lower BMI before they started the program.
2. Also, they may also be having more time to excercise than those who did not meet the trainer.
3. Also, they may be having more money to afford a personal trainer and get better results.
4. Those who had a trainer may be more goal-oriented and had a timebound target to achieve.
b.
Experiment:
Divide the tellers into two groups: a treatment group and a control group. Ensure that both groups have people with similar level of understanding of the work to be done. For example, both could have equal number of freshers, juniors, and seniors. Also, both could have equal number of undergraduates, graduate, and postgraduates. Their error rate should be noted before the start of the experiment.
Let the treatment group go through the advanced training program. Their error rate should again be compared at the end of the experiment.
Biases:
1. Only those who had a lower error rate were allowed to go to the advanced training program.
2. Those who had a higher error rate inspite of going through the advanced training program were transferred to a remote branch office.
Lurking variables:
1. Only persons with high IQ were allowed to go through the advanced training program. Also, these persons with high IQ had a lower average error rate.
2. Only those persons who went through an orientation program were aware of the advanced training program. Also, they were also given information on how they could lower their average error rate.