<result> <column_name>sleep_latency_min</column_name> <family>additive</family> <subtype>real_additive</subtype> <confidence>0.62</confidence> <requires_user_confirmation>false</requires_user_confirmation> <sum_meaningful>true</sum_meaningful> <signed>false</signed> <reducers> <primary_reducer>sum</primary_reducer> <valid_reducers>sum, min, max, mean, median, count, distribution</valid_reducers> <excluded_reducers>product (not a ratio), geomean (not a ratio)</excluded_reducers> <average_semantics>arithmetic_mean</average_semantics> </reducers> <inner_outer_measure> <inner_measure_reducer>sum</inner_measure_reducer> <outer_measure_reducer>sum</outer_measure_reducer> </inner_outer_measure> <multiplicative_partner_derivable>false</multiplicative_partner_derivable> <reasoning>The table is a DIMENSION with no sequence axis, so multiplicative is highly implausible and the prior is additive. The column is a duration in minutes, strictly positive, with mean 40.7 and median 37.6, far from 1.0, and its range runs from 6 to 123. It has no ratio naming, so there is no multiplicative evidence. The positive log_symmetry_gain only reflects right skew in a duration and does not override the center-of-mass and naming evidence. The gate leans toward real_additive for physical quantities in dimension tables. Confidence is moderate because a per-entity latency could also be treated as an intensive level where the mean is the natural rollup. Mean is therefore kept as a valid reducer.</reasoning> <table_type_conflict>false</table_type_conflict> </result>