Behavioral Science Flashcards
Cross sectional study
collects data at particular point in time from group of people o assess freq of disease and related risk factors.
Case control study
Compares group of people with disease/condition with group that does not. Purpose is to determine risk factors. Retrospective. Good for rare diseases.
Cohort study
Examines large group and watches it evolve over time based on different exposures. Prospective.
Types of Bias
Confounding, Sampling, Recall, Selection, Late-Look
Confounding bias
Variable closely related to another. E.g. people who drink coffee more likely to have certain types of cancer. (Confounder–people who drink coffee are more likely to smoke.)
Sampling bias
Sample of people chosen for study not representative. (E.g. study that only looks at men or healthy patients)
Recall bias
When people asked to recall information, might be basied based on new knowledge, e.g. if they were told they were on the placebo, less likely to say they felt a difference.
Selection bias
Investigators chose how to group participants for purpose of study. E.g. disproportionately assigning patients with more severe disease the experimental treatment.
Late-look bias
Results recorded at wrong time, skewing outcomes.
Main ways to reduce bias
Blinding and randomization
Sensitivity
How often the test will detect the presence of the disease in people who truly have the disease, i.e. how reliable it is in identifying people who have the disease.
Specificity
How often the test will detect the absence of a disease in those who do not have the disease, i.e. how well it identifies disease-free individuals
Positive predictive value
Likelihood that a positive test result truly means that the patient has the condition
Negative predictive value
Likelihood that a negative test result truly means that a patient does not have a given condition
Absolute risk
Likelihood of outcome over time without comparison to another group
Attributable risk
How much risk is actually due to the condition being studied. Attrib risk=Absolute risk (if exposed)-Absolute risk (not exposed)
Relative risk
Comparison of risks between two different conditions or groups of people
Odds ratio
Probablility of event happening/probability of event not occuring
Precision
Reproducibility
Accuracy
How close a measurement is to the true value
Systematic error
Errors that occur the same way every time a measurement is taken. Precise, but not accurate.
Random error
Unavoidable. Different each time a measurement is taken.
Positive skew
Tail on right
Type I error
False positive
Type II error
False negative