IN ARRIVING AT DECISIONS ABOUT WHICH RESEARCH PROJECTS TO PURSUE, THE CORRECTIONS ADMINISTRATOR MAY CONSIDER THE PRIMARY MISSIONS OF THE AGENCY, AGENCY REQUIREMENTS FOR SELF-MAINTENANCE AND SELF-IMPROVEMENT, THE PERCEIVED ROLE OF CORRECTIONS RELATIVE TO OTHER AGENCIES OR SYSTEMS IN THE COMMUNITY, AND CONSIDERATIONS RELATED TO CORRECTIONS THEORY AND EXPERIENCE. THE CORRECTIONS ADMINISTRATOR IS A KEY FIGURE IN THE EVALUATION PROCESS, REGARDLESS OF THE SPECIFIC FORM IT TAKES. THE EVALUATION PROCESS CASTS THE ADMINISTRATOR IN CONTRADICTORY ROLES: THE ADMINISTRATOR IS RESPONSIBLE FOR THE SUCCESS OF THE AGENCY'S OPERATIONS, INCLUDING EVALUATION; YET EVALUATION CARRIES THE POTENTIAL FOR DISCREDITING AN ADMINISTRATIVELY SPONSORED PROGRAM OR FOR UNDERMINING A POSITION TAKEN BE THE ADMINISTRATOR. TO ADDRESS THIS DILEMMA, THE ADMINISTRATOR NEEDS TO DEVELOP A STANCE CHARACTERIZED AS EXPERIMENTAL, AS OPPOSED TO COMMITTED ('TRAPPED'). AN EXPERIMENTAL ATTITUDE ON THE PART OF THE ADMINISTRATOR IS AN IMPORTANT STEP IN CREATING A CLIMATE CONDUCIVE TO RESEARCH. APPROACHES TO OBTAINING STAFF TO PERFORM EVALUATIVE RESEARCH INCLUDE HIRING AN IN-HOUSE RESEARCH STAFF; CONTRACTING WITH A PRIVATE, NONPROFIT ORGANIZATION; EMPLOYING UNIVERSITY FACULTY MEMBERS; OR SEEKING THE SERVICES OF A FOR-PROFIT CONSULTANT. EACH APPROACH IS APPROPRIATE UNDER DIFFERENT CIRCUMSTANCES. FOR THE PUBLICATION FROM WHICH THIS ARTICLE WAS ADAPTED, SEE NCJ-46227. (LKM)