Procedures used in Doctor of Education (Ed.D.) dissertations at the University of Alabama (Tuscaloosa) were studied. Focus was on identifying: (1) characteristics of the research designs used; (2) sources of the instruments used to collect data; (3) reports of reliability estimates and evidence of validity of the instruments; and (4) types of statistical techniques applied. Seventy-five dissertations were randomly selected from a pool of 209 completed between 1984 and 1988. A revised version of a checklist previously developed was used in the analysis. Interrater reliability was assessed at the beginning of data collection to resolve potential problems. Results indicate that quantitative procedures were used in 88% of the dissertations, and qualitative procedures were used in 12%. Less sophisticated procedures (e.g., "t"-tests, chi-square, one-way analysis of variance) were used more frequently (72%) than were multivariate procedures (28%). Validity information was reported more frequently (80%) than was reliability information (49%) for the instruments used. Pilot studies of researcher-developed instruments were reported in 64% of the dissertations. Recommendations to alleviate some of the weaknesses found in the procedures used in the Ed.D. dissertations include: (1) more coursework in research and statistics in Ed.D. programs; and (2) departmental seminars so that Ed.D. candidates and faculty advisors can become more aware of more complex analysis procedures. The dissertation analysis instrument is appended. (Author/SLD)