```xml <result> <column_name>coverage</column_name> <family>additive</family> <subtype>intensive_additive</subtype> <confidence>0.85</confidence> <requires_user_confirmation>false</requires_user_confirmation> <sum_meaningful>false</sum_meaningful> <signed>false</signed> <reducers> <primary_reducer>mean</primary_reducer> <valid_reducers>mean, min, max, median, distribution</valid_reducers> <excluded_reducers>sum (summing a proportion/coverage level 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. The column "coverage" is a proportion/percentage (bounded [0,1], mean 0.755, median 0.926) representing an intensive level - the fraction of something covered at each entity. Contains zeros (17.7%), which argues against multiplicative. Summing coverage values across entities is meaningless (you cannot add proportions); averaging is the natural rollup. This is a classic intensive_additive measure: an absolute level that must be averaged, not summed, when volumes merge. </reasoning> <table_type_conflict>false</table_type_conflict> </result> ```