Class #13: Bias and Misclassification Flashcards
What is the definition of bias?
Bias is a systematic (non-random) error in study design or conduct leading to erroneous results.
-bias distorts the relationship (interaction) between exposure and outcome.
What relationship (association) does bias distort if present in your study?
The relationship between exposure and outcome.
Once a researchers study is over, and bias is found, can the study be fixed or changed?
NO, there’s nothing that can be done to fix bias once the study is over. The data will have to be thrown out and the study re-done.
Learning Objective for lecture 13:
Be able to define and recognize each of the biases offered in this lecture.
*Measurement-related biases (information/observation) –> any aspect in the way the researcher collects information, or measures/observes subjects (and their characteristics/variables) which creates A SYSTEMATIC DIFFERENCE BETWEEN GROUPS in the quality/accuracy of their information.
>subject-related variation
-Recall (Reporting) Bias
-Contamination Bias
-Compliance/Adherence Bias
-Lost to Follow-up Bias
*Differential vs. Non-Differential
>Observer-related variation
- Interviewer (Proficiency) Bias
- Diagnosis/Surveillance (Expectation) Bias
- Misclassification Bias
*Selection-related biases –> any aspect in the way the researcher selects or acquires study subjects (cases/controls or exposed/unexposed; even study sample) which creates A SYSTEMATIC DIFFERENCE IN THE COMPOSITION BETWEEN GROUPS.
>Healthy-Worker Bias
>Self-Selection/Participant (Responder) Bias
>Control Selection Bias
To control for bias, what responsibility does the researcher have to commit to when designing their study?
They have to select the most precise, accurate, and medically-appropriate measures of assessment and evaluation/observation.
Of the 4 listed steps in controlling for biases, what are the 2 most important?
1) Blinding/Masking
2) Randomly allocate observers/interviewers for data collection (and train them!; use technology!).
What is a misclassification bias?
A misclassification bias is where there’s error in classifying either disease or exposure status, or both.
-source of measurement (information/observation) bias.
What effect does non-differential bias have on measures of association (RR and OR)?
It moves them both closer to 1.0
e. g: RR of 0.3 moves to 0.7
e. g: OR of 1.9 moves to 1.2
What effect does differential bias have on the measure of association (RR or OR) if present in a study?
Differential bias can move the measure of association in either direction in relation to 1.0; it can inflate or attenuate your effect estimates of association.
e. g: RR of 0.8 moves to 0.2 or 1.4 moves to 2.1
e. g: OR of 2.3 move to 1.1 or 0.6 moves to 0.9
Is differential or non-differential bias considered the worst Misclassification to have in your study?
Differential Bias is considered worse than non-differential.
What type of bias is the most common type in a research study?
Selection Bias
When assessing for bias and its impact, what are the 3 components that investigators evaluate?
1) source/type (2 ‘families’)
2) Magnitude/Strength
- bias can account entirely for a weak association (small RR/OR) but is not likely to account entirely for a very strong association (large RR/OR).
3) Direction
-bias can over- or under-estimate the true measure of association.
>bias can have an enhancing or minimizing effect on the true measure of association (towards or away from 1.0)
What are the 3 aspects that a researcher must evaluate before they can declare a statistical association between “exposure” and “outcome” of their study?
1) Check for bias.
2) Check for confounding or effect modification (interaction).
3) Check for statistical significance.
