```xml <result> <column_name>previous_rank</column_name> <family>additive</family> <subtype>intensive_additive</subtype> <confidence>0.95</confidence> <requires_user_confirmation>false</requires_user_confirmation> <sum_meaningful>false</sum_meaningful> <signed>false</signed> <reducers> <primary_reducer>min</primary_reducer> <valid_reducers>min, max, median, mean, mode</valid_reducers> <excluded_reducers>sum (summing ranks is meaningless), product (not a ratio)</excluded_reducers> <average_semantics>arithmetic_mean</average_semantics> </reducers> <inner_outer_measure> <inner_measure_reducer>min</inner_measure_reducer> <outer_measure_reducer>max</outer_measure_reducer> </inner_outer_measure> <multiplicative_partner_derivable>false</multiplicative_partner_derivable> <reasoning> Gate prior: DIMENSION table with no sequence axis makes multiplicative highly implausible. Column name "previous_rank" indicates an ordinal position/level, not a ratio. Statistics show uniform distribution [1, 211] with mean=median=106, centered far from both 0 and 1, confirming this is an absolute ordinal level. Ranks are intensive values where summing is meaningless (you cannot add "rank 5" and "rank 10" to get "rank 15"), but min/max/mean are valid aggregations. No sequence axis rules out ordinal_additive (which requires normalization along a sequence), so intensive_additive is correct. </reasoning> <table_type_conflict>false</table_type_conflict> </result> ```