Boolean data types have only 2 possible values (true/false), which creates severe data skew - all rows distribute to just 2 SPUs instead of spreading evenly across all available SPUs. Good distribution keys need high cardinality to ensure uniform data distribution.