Observation bias for epidemiology study
While selection bias happen because of the way we select the object study to build relationship or association between a suspected causative agent and disease or illness, observation bias is focused to the bias of the means of getting information about the study. This kind of bias is generally involved when you are doing retrospective epidemiological study. I suggest you visit that page to obtain more detail characteristics of retrospective epidemiological study.
This kind of study is fragile because it involves individuals memory. Retrospective study is a study looking in the past, something that had happened. We gather and group individuals who developed disease or illness while others do not. The problem comes when we try to collect information from those groups. Sometimes, we may make mistakes by grouping those who developed disease and those who do not without seeing the exposure information. Because it may possible individuals who do not develop a disease are actually exposed by causative agent thus will give less stress to the causative agent.
It is a common nature that people tend only remember event that related to his activities, accidental or routine. When we ask individuals who develop disease or illness, its prevalent that they will give more thought to the causative agent than those who do not develop disease although these individuals are exposed in similar manner. Individuals who develop a disease will have more power to recall memory associated to the disease, while individuals who do not have disease will less able to recall and the problem will get worsen when these individuals are grouped in an unexposed group while in fact they are exposed. Thus, this will make false association. Stronger association may be resulted since more weight are produced for the cause by individuals having the disease.
This kind of cases will lead to misleading and inaccuracy information. Careful identification of exposure information will ensure the right man is on the right place to produce more valid data and association to establish relationship between a suspected causative agent and disease or illness.
