Set 1 Statistical Errors

Simpson's paradox

A statistical phenomenon where a trend appears in several groups of data but reverses or disappears when the groups are combined, often due to a lurking variable.

Lurking variable

A hidden factor not explicitly accounted for that significantly influences results and can flip conclusions when data are aggregated.

Conditional (confounding) variable

A variable that conditions or confounds the relationship between two measured variables; accounting for it can change the observed association.

Aggregated data

Data combined across groups or categories; aggregation can mask meaningful differences within subgroups and lead to reversed conclusions.

Subgroup analysis (stratification)

Breaking data into relevant categories (e.g., by health status or age) to reveal patterns hidden in the overall aggregate.

Survival rate

The proportion of individuals who survive in a group (e.g., number of patients who lived divided by total treated), used in the video’s hospital example.

Weighted average

An average where groups contribute in proportion to their sizes; differing group sizes can drive overall results opposite to each group’s rate.

Correlation vs. causation

A reminder that an observed association in data (correlation) does not guarantee a cause–effect relationship, especially when confounders exist.

Data grouping

The way observations are categorized (e.g., by health status or age). Different groupings can change the apparent trend.

Misleading statistics

Conclusions drawn from data that appear persuasive but are inaccurate or reversed when proper context and variables are considered.