What are potential sources of measurement bias in a study?
Measurement bias is any systematic error that occurs in data collection in a study. Measurement bias is also referred to as detection bias. In most cases this bias occurs due to lack of blinding. Potential sources of measurement bias may include the following.
Recall bias. Human beings tend to remember some events and forgetting others. Recall bias occurs whenever data is collected retrospectively. Recall bias is more likely to problematic when the result of interest is likely to influence memory. More often than not, recall bias focus more on the respondent. However, it can also be initiated by the researcher who asks same questions in different ways expecting a specific answer. Observer bias is also another type of measurement bias, especially when the researcher/observer sees what they want to see. This results to a systematic deviation from the truth. For example, when researching on aortic dissection, may choose to say that the pain is severe ignoring other descriptions because that is what they anticipated to see. Interpretation of medical tests is subjective and therefore, avoiding this bias is almost impossible.
What elements of measurement error will you be looking for?
One of the elements to be looked for in measurement errors is systematic error element. Systematic error is constant and predictable and is caused by the error in calibration of measurement instruments, imperfect observation or interference of the measurement process. Imperfect zeroing of the instrument of measurement leading to zero error is a perfect example of this measurement element. Random error is another element to consider in measurement errors. This element is always present measurement. It is caused by fluctuations in measurement apparatus readings or the researcher’s interpretations of the readings. Random errors however can be reduced by averaging multiple measurements.