```xml <result> <column_name>referential_integrity_pct</column_name> <family>uncertain</family> <subtype>none</subtype> <confidence>0.0</confidence> <requires_user_confirmation>true</requires_user_confirmation> <sum_meaningful>false</sum_meaningful> <signed>false</signed> <reducers> <primary_reducer>unknown</primary_reducer> <valid_reducers>none</valid_reducers> <excluded_reducers>all - no data available</excluded_reducers> <average_semantics>n/a</average_semantics> </reducers> <inner_outer_measure> <inner_measure_reducer>n/a</inner_measure_reducer> <outer_measure_reducer>n/a</outer_measure_reducer> </inner_outer_measure> <multiplicative_partner_derivable>false</multiplicative_partner_derivable> <reasoning> The column has 100% null values (null_fraction=1.0) with no statistics or sample data available. The name "referential_integrity_pct" suggests a ratio_multiplicative measure (percentage token), but this is purely nominal evidence with zero data support. On a DIMENSION table with no sequence axis, multiplicative is already implausible by the gate prior. Without any numeric evidence, I cannot classify this column's behavior under volume union and must abstain entirely. </reasoning> <table_type_conflict>false</table_type_conflict> </result> ```