4.Bias, Sample Selection & Confounding

Vocabulary flashcards covering key terms from the lecture notes on bias, sampling frameworks, and confounding.

Bias

A systematic error in study design, conduct, or analysis that leads to an incorrect estimate of the association between exposure and risk of disease

Chance (random error)

Random fluctuations that cause variation in estimates due to sampling variability.

Confounding

A distortion of the exposure–outcome relationship caused by a third variable related to both exposure and outcome.

Selection bias

Systematic error from how participants are chosen, resulting in a non-representative sample.

Information bias

Systematic error in measuring exposure or disease status, leading to biased estimates.

Non-response bias

Bias due to differences between those who participate and those who do not respond.

Loss to follow-up bias

Bias from differential dropout, affecting incidence or risk estimates.

Sampling bias

Bias arising from non-random sampling, causing non-comparable groups.

Randomization

Assignment of participants to groups by chance to balance known and unknown confounders.

Matching

Design approach that pairs participants on specific characteristics to control confounding.

Restriction

Limiting the study to a homogeneous subgroup to reduce variability and confounding.

Stratification

Evaluate association within subgroups of confounding variable

Multivariate analysis

Technique that takes into account many variables at the same time that may confound the effect of exposure

Stratified analysis

Examining associations within each level of a confounding variable.

Odds ratio

A measure of association (odds of outcome with exposure vs without) often used in case-control studies.

Target population

The entire set of individuals of interest to the research question.

Sample

Subset of individuals in target population from which researchers will draw their conclusions about target population. following process of statistical interference

Probability sampling

Sampling where each member has a known probability of selection; includes random, systematic, stratified, and cluster sampling.

Non-probability sampling

Sampling without known probabilities, often leading to biased population estimates (e.g., snowball, convenience, purposive).

Simple random sampling

Each subject in a population has an equal chance of being selected.

Systematic random sampling

Selection of sampling has a known and equal predefined interval such as select every 4th animal in a herd from a random starting point.

Stratified random sampling

Dividing the population into strata and taking random samples from each strata.

Cluster sampling

Population is split into clusters, and then a random sample of clusters is selected

Snowball sampling

recruited subjects refer others into the study.

Convenience sampling

Cheap and quick used for user opinion polls or pilot testing for a survey

Purposive sampling

Researcher chooses sample most useful to the purposes of the research and has clear criteria and rationale for inclusion

Recall bias

The ability of respondents to accurately remember exposure of pets

Interviewer bias

Interviewer interjects his or her bias into interview. the way questions are asked and perceived by respondents affects their response

Surveillance bias

One group is monitored much more closely than another group

Observer bias

Incorrect measurement technique and recording of values

Confounder

A variable in a study that distorts true relationship between an exposure and an outcome to seem more/less associated

Confounding variables

Age, sex, breed