Scalogram analysis of nonunanimous criminal and economic liberties cases in 14 state supreme courts indicates that between 1971–1981 policy preferences structured voting behavior. Preference based voting declines after 1981—with the decline most noticeable in economic cases. This decline may be the result of preference complexity rather than a complete absence of preference based voting. Neither court duration, cases available for scaling, consensus nor three out of four institutional factors (electoral accountability, workload, docket control) explain preference based voting. A fourth institutional factor, opinion assignment procedures, is related to preference based voting— the “lottery” assignment method results in lower levels of preference based voting than an assignment procedure patterned after that used by the U.S. Supreme Court.