POPM 3240 Lecture Notes - Lecture 12: Internal Validity, Randomized Controlled Trial, Recall Bias
Document Summary
Can affect classification of exposure or outcome. Poor questionnaire design (misclassifying e or o) When the participants did not undergo treatment. Error is not related to exposure or outcome, but an inherent problem with the data collection methods. Estimate always get biased towards the null value. Less likely to detect an association, even if it exists. If or shifts from 0. 6 to 0. 9 or from 3. 4 to 2. 5, both are towards 1 (null) When the impact on both groups are equal. Error in classifying e are the same in both d+ and d- or classifying d are the same in both e+ and e- The rate of exposure misclassification differs between d+ and d- or classifying d differs between e+ and e- Direction of bias can be either way (towards or away from the null) Need to think about which direction is more likely. An individual may not remember completely, or accurately, something that occurred in the past.