EPID 4070 exam 3

find causal links between exposure and outcome

one of the main goals of epidemiology

correlations or associations between exposure and outcome

observational studies measure

selection bias

information bias

confounding

sources of bias

selection bias

information bias

confounding bias

sources of potential bias

selection bias

sicker people are less likely to participate than healthy people in case control study

differential rates of dropout in exposed compared to unexposed cohorts

differential rates of dropout in RCT treatment compared to control groups

information bias

if sick people recall risk factors more accurately than healthy people

if interviewers ask questions differently depending upon the type of participant

confounding bias

additional risk factors are unequally distributed among exposed and unexposed cohorts

additional risk factors are unequally distributed among case and control groups

random error

derives from use of samples to assess association between two or more factors

random error

decreases with larger sample sizes

selection bias

does not decrease with larger sample sizes

statistical inference

process of drawing conclusions about a population based on a sample of that population

p-value

we assess whether an association may be due to chance by the

confidence intervals

used to show the reliability of our estimate of the measure of effect (RR, OR, RD)

increasing sample size, or making more precise measurements

can increase the reliability and precision of an estimate by

p-value

probability of obtaining the estimate you observed, or a more extreme value, if the null hypothesis is true

confidence interval

corresponds to the precision of the parameter based on the study data

statistical power

the ability of the study to find a true difference

state the problem as a null hypothesis

state the research hypothesis

select the level of significance

select the test statistic

do the study and get a result

compute p-value from study data

six steps of testing a hypothesis

type 1 error

incorrectly reject the null hypothesis

type 2 error

do not reject the null hypothesis even though it is false

inadequate sample size

type 2 error is usually caused by

strength of association

consistency of association

specificity

temporality (the only required one)

does-response

plausibility

coherence of explanation

analogy

experiment

Hill’s criteria for causality