In the article Association of Arsenic Exposure during Pregnancy with Fetal Loss and Infant: A Cohort Study in Bangladesh
Is there any Recall Bias or Selection Bias and if so how does it take place in the study what is it effect on the measure of association and why
A similar question has recently been answered (http://www.enotes.com/homework-help/association-arsenic-exposure-during-pregnancy-with-513583). The previous question dealt with the presence of any potential bias. The current question seems to be an extension of the same and I will answer the part that was not covered in the previous question. Kindly check the answer to the previous question for any bias in the study. Here I will discuss the effect of such bias on measure of association.
In the study by Rahman et al., potential recall bias may exist, since the pregnancies occurred in 1991-2000 and the water source history was collected later on. Although the authors mention that they tried to minimize such bias, it is entirely possible that subjects with negative outcomes (no infant death or fetal loss) will remember fewer details about their water source (among other details) as compared to subjects with positive outcomes (who are more likely to remember such details more clearly). This will lead to a higher degree of association between arsenic concentration and outcomes.
In the answer to the previous question, I had mentioned the potential bias due to non-inclusion of parameters such as past history of fetal loss (or infant death), availability of appropriate nutrition and healthcare, etc. A subject with a history of similar outcomes (could be totally unrelated to arsenic concentration), lack of nutrition or timely medical care will skew the study towards a positive correlation between independent variables and outcomes.