рдЖрдЧреЗрдЕрдЧрд▓реА рдЧрд╛рдЗрдб
рдореЛрд░рд╛рд╡реЗрдХ рдХрд╛ рд╡рд┐рд░реЛрдзрд╛рднрд╛рд╕
рдмреБрдирд┐рдпрд╛рджреА рдмрд╛рддреЗрдВ
рдмреБрдирд┐рдпрд╛рджреА рдЧрд╛рдЗрдб
SimpsonтАЩs paradox occurs when a trend seen within subgroups reverses or disappears after the subgroups are combined.
рдЗрд╕ рдЧрд╛рдЗрдб рдХреЛ рд╕реБрдиреЗрдВрдПрдЖрдИ-рдиреИрд░реЗрдЯреЗрдб ┬╖ ~3 min
The reversal can result from different group sizes or compositions, but deciding which comparison is meaningful requires understanding the data-generating and causal question. Stratifying data is informative; it does not automatically prove bias, remove confounding, or identify a causal effect.
SimpsonтАЩs paradox describes a reversal or disappearance of an association when data are aggregated over subgroups. For example, a treatment can show a higher success rate than a comparison within each severity group, yet a lower overall rate if the treated group contains more high-risk cases. Differences in the weights assigned to subgroups can change the aggregate. The pattern is an arithmetic property of the rates; it does not by itself say which comparison answers the decision question. The Berkeley graduate-admissions analysis by Bickel, Hammel, and OтАЩConnell is a classic case. Aggregate acceptance rates appeared to favor male applicants, but department-level analysis showed much of the overall difference reflected application patterns across departments with different admission rates. The authors analyzed department-level data and discussed the limits of drawing a discrimination conclusion from aggregate proportions. This is not resolved simply by declaring that aggregate or stratified data are always correct. The relevant comparison depends on the question and causal structure. When a reversal appears, inspect denominators, subgroup composition, and the process assigning treatment or exposure. Consider whether the subgroup variable is a confounder, mediator, collider, or simply descriptive. Conditioning on a variable can reduce, create, or distort bias depending on the data-generating process. Report both aggregate and stratified estimates, explain their weights, and use study design or causal assumptions to choose an estimand. The paradox warns against inferring a causal story from a single summary table.
A treatment appears more successful overall but less successful within both severity groups because assignment and group sizes differ.
A product conversion rate reverses after segments are combined because traffic volume differs across device types.
A reviewer compares aggregate and department-level admission rates and avoids treating either table alone as a causal conclusion.
A researcher uses a causal diagram and study design to decide whether a subgroup variable is a confounder, mediator, collider, or descriptive factor.
Simpson-type reversals remain important in dashboards, clinical studies, hiring, and A/B tests as datasets combine populations with different compositions. Better analytics can make subgroup views easier to inspect, but causal interpretation still requires domain knowledge and design assumptions. Teams should predefine meaningful strata and report denominator weights alongside aggregate outcomes. No single aggregation rule answers every question. Researchers should preserve analysis plans and report how alternative stratifications change conclusions, especially when findings inform high-stakes decisions. Define important subgroup comparisons before results are examined.
рдореБрдлрд╝реНрдд рдиреНрдпреВрдЬрд╝рд▓реЗрдЯрд░
рд╣рд░ рдХрд╛рд░реНрдпрджрд┐рд╡рд╕ рдХреА рд╕реБрдмрд╣ рддреАрди рд╕рддреНрдпрд╛рдкрд┐рдд рдПрдЖрдИ рдХрд╣рд╛рдирд┐рдпрд╛рдБ, рд╕рд╛рджреЗ рдЕрдВрдЧреНрд░реЗрдЬреА рдореЗрдВ рд▓рд┐рдЦреА рдЧрдИрдВред рд╣рдореЗрд╢рд╛ рдХреЗ рд▓рд┐рдП рдореБрдлрд╝реНрдд, рдХреЛрдИ рд╡рд┐рдЬреНрдЮрд╛рдкрди рдирд╣реАрдВред
рдкреНрд░рддреНрдпреЗрдХ рд╕рдкреНрддрд╛рд╣ рдХреЗ рджрд┐рди рдПрдХ рдИрдореЗрд▓. рдПрдХ рдХреНрд▓рд┐рдХ рдореЗрдВ рд╕рджрд╕реНрдпрддрд╛ рд╕рдорд╛рдкреНрдд рдХрд░реЗрдВ. рд╣рдо рдЖрдкрдХрд╛ рдкрддрд╛ рдХрднреА рдирд╣реАрдВ рдмреЗрдЪрддреЗ рдпрд╛ рд╕рд╛рдЭрд╛ рдирд╣реАрдВ рдХрд░рддреЗред
рд╕реНрд╡рдпрдВ рдХреА рдЬрд╛рдВрдЪ рдХрд░реЛ
рдкреНрд░рддреНрдпреЗрдХ рдЙрддреНрддрд░ рдкрд░ рддреНрд╡рд░рд┐рдд рдкреНрд░рддрд┐рдХреНрд░рд┐рдпрд╛, рдФрд░ рдкрд╛рдареНрдпрдХреНрд░рдо рдкрд╛рд╕ рдХрд░рдиреЗ рдХреЗ рдмрд╛рдж рдПрдХ рд╕рддреНрдпрд╛рдкрди рдпреЛрдЧреНрдп рдЖрдИрдбреА рдХреЗ рд╕рд╛рде рдПрдХ рд╕рд╛рдЭрд╛ рдХрд░рдиреЗ рдпреЛрдЧреНрдп рдкреНрд░рдорд╛рдгрдкрддреНрд░ред
рдирд┐рдГрд╢реБрд▓реНрдХ рдПрдЖрдИ рд╢рд┐рдХреНрд╖рд╛ рдХрд╛ рд╕рдорд░реНрдерди рдХрд░реЗрдВред AI Understanding рдПрдХ 501(c)(3) рдЧреИрд░-рд▓рд╛рднрдХрд╛рд░реА рд╕рдВрд╕реНрдерд╛ рд╣реИ - рдХреЛрдИ рд╡рд┐рдЬреНрдЮрд╛рдкрди рдирд╣реАрдВ, рдХреЛрдИ рдкреЗрд╡реЙрд▓ рдирд╣реАрдВред рджрд╛рди рдХрд░реЗрдВ
SimpsonтАЩs paradox occurs when a trend seen within subgroups reverses or disappears after the subgroups are combined. The reversal can result from different group sizes or compositions, but deciding which comparison is meaningful requires understanding the data-generating and causal question. Stratifying data is informative; it does not automatically prove bias, remove confounding, or identify a causal effect.
The paradox concerns changes between subgroup and aggregate associations.
Aggregate rates weight subgroup rates by their denominator composition.
Bickel et al. showed aggregation can be misleading and interpreted department-level data with context.
Direction and interpretation depend on weighting and data-generating structure.
The pattern itself does not identify the causal explanation.
рд╕реАрдЦрддреЗ рд░рд╣реЛ
рдЗрд╕ рд╡рд┐рд╖рдп рдХреЗ рд▓рд┐рдП рдЕрдзрд┐рдХ рдорд╛рд░реНрдЧрджрд░реНрд╢рд┐рдХрд╛рдПрдБ рдЪреБрдиреА рдЧрдИрдВ
рдЖрдЧреЗрдЕрдЧрд▓реА рдЧрд╛рдЗрдб
рдореЛрд░рд╛рд╡реЗрдХ рдХрд╛ рд╡рд┐рд░реЛрдзрд╛рднрд╛рд╕
рдмреБрдирд┐рдпрд╛рджреА рдмрд╛рддреЗрдВ