KnowraSimpson's paradoxLinked fromLinked fromThe 22 pages that link to Simpson's paradox, each with the reason it gives.All 22Broader topic 3Related 16Compared with 3Causal inferenceRelated: The reversal illustrates how group composition can distort apparent causal comparisons.ConfoundingRelated: Confounding can produce the aggregate-versus-stratified reversal associated with this paradox.Conditional probabilityRelated: Different conditioning choices can reveal or conceal associations in the same data.Ecological fallacyRelated: It demonstrates how an aggregate pattern can differ sharply from patterns inside groups.CorrelationRelated: It shows how an overall correlation can differ sharply from within-group patterns.Independence (probability theory)Related: It highlights that dependence patterns can change when a third variable is introduced.Regression analysisRelated: Unmodeled group structure can make an overall regression relationship misleading.A/B testingRelated: Uneven traffic composition can make aggregate A/B results conflict with subgroup patterns.Spatial autocorrelationCompared with: Aggregation across space can reverse observed relationships, complicating autocorrelation-based interpretations.StatisticsBroader topic: It demonstrates how aggregation and hidden group structure can reverse a statistical comparison.Descriptive statisticsRelated: Aggregated descriptions can suggest a pattern opposite to the one within subgroups.Mortality rateRelated: Aggregated mortality rates can reverse comparisons made within age or other strata.Pearson correlation coefficientRelated: Aggregated correlations can conceal or reverse patterns present within subgroups.Gini coefficientRelated: Aggregate Gini comparisons can conceal subgroup patterns when population composition changes.Scatter plotRelated: A pooled scatter plot can conceal relationships that differ across subgroups.Line chartRelated: A visible aggregate trend may differ from the trajectories within the chart’s subgroups.Coefficient of determinationRelated: Aggregation and omitted group structure can make an overall fit statistic obscure subgroup patterns.Statistics (numerical data)Related: Aggregation can make a valid numerical comparison misleading.Correlation does not imply causationBroader topic: Changing which groups are combined can reverse an observed association.Eyeball theoremBroader topic: An overall graph can hide subgroup structure that changes the apparent trend.Frisch–Waugh–Lovell theoremCompared with: Control adjustment can change a coefficient's interpretation, though the theorem itself is an algebraic equivalence.Modifiable temporal unit problemCompared with: It illustrates how changing the units of analysis can reverse an apparent relationship.
KnowraSimpson's paradoxLinked fromLinked fromThe 22 pages that link to Simpson's paradox, each with the reason it gives.All 22Broader topic 3Related 16Compared with 3Causal inferenceRelated: The reversal illustrates how group composition can distort apparent causal comparisons.ConfoundingRelated: Confounding can produce the aggregate-versus-stratified reversal associated with this paradox.Conditional probabilityRelated: Different conditioning choices can reveal or conceal associations in the same data.Ecological fallacyRelated: It demonstrates how an aggregate pattern can differ sharply from patterns inside groups.CorrelationRelated: It shows how an overall correlation can differ sharply from within-group patterns.Independence (probability theory)Related: It highlights that dependence patterns can change when a third variable is introduced.Regression analysisRelated: Unmodeled group structure can make an overall regression relationship misleading.A/B testingRelated: Uneven traffic composition can make aggregate A/B results conflict with subgroup patterns.Spatial autocorrelationCompared with: Aggregation across space can reverse observed relationships, complicating autocorrelation-based interpretations.StatisticsBroader topic: It demonstrates how aggregation and hidden group structure can reverse a statistical comparison.Descriptive statisticsRelated: Aggregated descriptions can suggest a pattern opposite to the one within subgroups.Mortality rateRelated: Aggregated mortality rates can reverse comparisons made within age or other strata.Pearson correlation coefficientRelated: Aggregated correlations can conceal or reverse patterns present within subgroups.Gini coefficientRelated: Aggregate Gini comparisons can conceal subgroup patterns when population composition changes.Scatter plotRelated: A pooled scatter plot can conceal relationships that differ across subgroups.Line chartRelated: A visible aggregate trend may differ from the trajectories within the chart’s subgroups.Coefficient of determinationRelated: Aggregation and omitted group structure can make an overall fit statistic obscure subgroup patterns.Statistics (numerical data)Related: Aggregation can make a valid numerical comparison misleading.Correlation does not imply causationBroader topic: Changing which groups are combined can reverse an observed association.Eyeball theoremBroader topic: An overall graph can hide subgroup structure that changes the apparent trend.Frisch–Waugh–Lovell theoremCompared with: Control adjustment can change a coefficient's interpretation, though the theorem itself is an algebraic equivalence.Modifiable temporal unit problemCompared with: It illustrates how changing the units of analysis can reverse an apparent relationship.