Ph 150A - Error In Epidemiological Studies

Front 1

Internal Validity

Back 1

reqs proper selection of study groups, lack of error in measurement, and ctrl for confounding factors

Front 2

External Validity

Back 2

ability to extrapolate from a sample population to a target population

Front 3

Accuracy

Back 3

validity

* how closely does a measurement represent the truth?

Front 4

Reliability

Back 4

reproducibility, repeatability, precision, consistency

* how consistently to repeated measurements agree?

Front 5

Random Error

Back 5

fluctuations around a true value of a parameter because of sampling variability

can occur as a result of poor precision, sampling error, or variability in measurement

Front 6

Systematic Error (Bias)

Back 6

bias: systematic difference or distortion from the truth in a particular direction

illustrated by values placed away from true value in center

resulting estimate less accurate

much more serious problem for study validity than random error

bias can result in findings that are distorted:
- toward null hypothesis
- away null hypothesis

Front 7

Selection Bias

Back 7

study population doesn't represent target population

Front 8

Information Bias

Back 8

occurs during data collection

1) misclassificaton: not classifying people properly in terms of exposure and outcome

2) ecological fallacy: people with the outcome may not be the ones with exposure

3) regression to the mean

* includes recall bias

Front 9

Confounding

Back 9

presence of an extraneous variable which is a risk factor for an effect among unexposed people and is associated with exposure under study