Posts: 3,675,934
Threads: 734,342
Joined:
Sep 2021
Reputation:
5
To assess the effects of cigarette smoking on rates of automobile accidents, three groups of volunteer subjects were followed for a four-year period. One group was composed of 1,000 current cigarette smokers, a second group was composed of 800 former smokers, and the third group was composed of 750 people who had never smoked cigarettes. All three groups contained participants of both genders and across all ethnic/racial groups. At the end of the four years, the automobile accident rates for the three groups were compared. The results showed an automobile accident rate for the current smokers that was higher than the rate for nonsmokers (p ¡Ü .05), but not statistically different than that for the former smokers (p > .05).
When interpreting the results of this study, one should be most concerned with the effects of what type of bias?
A. Expectancy bias
B. Late-look bias
C. Measurement bias
D. Proficiency bias
E. Recall bias
F. Selection bias
Posts: 3,675,934
Threads: 734,342
Joined:
Sep 2021
Reputation:
5
hmmm..i thought there will be more people undetected in car accident..
Posts: 3,675,934
Threads: 734,342
Joined:
Sep 2021
Reputation:
5
this has weird explanation.. check this out
Posts: 3,675,934
Threads: 734,342
Joined:
Sep 2021
Reputation:
5
The correct answer is F.
This question describes a cohort study in which three different groups of people are followed over time to see if they have different incidence rates of automobile accidents. For this type of study, the most common type of bias is selection, or sampling bias. The problem is that the people who volunteer to participate in this study may be very different from the general population. One would need to check the demographic variables of the participants and compare them to those not participating in order to be sure the results were at all generalizable. Some selection bias is probably inevitable because people get to decide to participate or not to participate in any given study. It is very likely that the different decisions that people make reflect the different types of people that they are.
Expectancy bias (choice A) is when a researcher or physician knows which subjects are in a treatment vs. a placebo group, which may, unwittingly, cause him to interact with them differentially, based on that information. If you think someone is getting a better treatment, you are more likely to think they get better, and perceive effects over and above the physiological effects of the drugs administered. The solution to the expectancy effect is a double blind design, in which neither subjects nor the researchers who have contact with them know which arm of the study the subjects are in. This is not the answer here because there is no intervention, but merely observation.
Late-look bias (choice B) is a problem when gathering information about some types of severe diseases. The problem is that the most severe cases will be dead or inaccessible before you can gather their information. This is not the answer here, because classification as to cigarette smoking was done at the start of the study, and the outcome variable of automobile accidents is public and easy to confirm.
With measurement bias (choice C), something about how the information is gathered affects the information collected. This can be true because survey questions use inappropriate wording that slants respondents to a particular answer, or because just knowing that you are being measured causes people to act differently than they would if they were not observed. Although it is possible that the study participants may all try to be better drivers because they are in the study, this would be true of all of the groups in the study, and would not distort the comparison among them.
Proficiency bias (choice D) is an issue when comparing the effects of different treatments administered at multiple sites. Simply put, the physicians at one site may have more skill with a given procedure than others. This means that the different levels of skill of the physicians delivering treatment might impact patient outcomes more than the treatment selection itself.
Recall bias (choice E) is a problem in retrospective studies (think case-control study) in which people are asked to remember what happened in the past and report it in the present. If people do not remember, and say so, then we have missing data. But often, people will invent answers either from a desire to please the researcher, or because our memory of the past changes over time.
Posts: 3,675,934
Threads: 734,342
Joined:
Sep 2021
Reputation:
5
yes its kinda strange but it make sense in some part..
Anyone else has other comment??
