```xml <result> <column_name>Family</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 an intensive 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> Table-type gate: DIMENSION with no sequence axis makes multiplicative highly implausible. The column name "Family" provides no semantic signal for measure type. Center-of-mass (mean=0.79, median=0.84) is bounded in [0, 1.18], centered away from both 0.0 and 1.0, suggesting an intensive level or normalized score. Contains zeros (0.6%) which argues against multiplicative. Negative log_symmetry_gain (-1.19) indicates the raw distribution is already more symmetric than log-transformed, ruling out ratio_multiplicative. This appears to be a normalized score, index, or categorical encoding where averaging is meaningful but summing is not - classic intensive_additive behavior. </reasoning> <table_type_conflict>false</table_type_conflict> </result> ```