The State of Undergraduate Education in Statistics: A Report from the CBMS 2000
Scheaffer, Richard L., Stasny, Elizabeth A., The American Statistician
The Conference Board of the Mathematical Sciences (CBMS) consists of 16 member organizations, including the American Statistical Association and the Institute of Mathematical Statistics. Every five years since 1965 the CBMS has received support from the National Science Foundation to conduct a national survey of undergraduate education in the mathematical sciences in the United States. The purpose of the survey is to track enrollment trends in undergraduate mathematical sciences courses, but additional information on faculty and instructional practices is collected as well. CBMS does not make recommendations on mathematical sciences education based on this survey. Every CBMS survey continues longitudinal studies of fall term undergraduate enrollments in the mathematics programs of two-year colleges and in the mathematics and statistics departments of four-year colleges and universities. The CBMS surveys include departments that offer associate, bachelor's, master's, and doctoral degrees. Every CBMS survey also studies the demographics of the faculty in those programs and departments, and examines the undergraduate curriculum to determine what is taught, who teaches it, and how it is taught. In addition, each CBMS survey includes a set of questions on special topics of current interest. As members of the steering committee, we proposed that CBMS 2000 investigate two special statistics topics: the educational background of faculty members teaching statistics courses in fall 2000, and the impact of the new Advanced Placement (AP) Statistics program on university statistics departments.
A detailed report on the CBMS 2000 survey was published by the American Mathematical Society (see Lutzer, Maxwell, and Rodi 2002). In the following we present summaries of the CBMS 2000 results that are of particular interest to statisticians. Throughout, we make comparisons with previous CBMS surveys. (See, e.g., Loftsgaarden, Rung, and Watkins 1997; Loftsgaarden and Watkins 1998.) We note that previous CBMS results are presented without estimates of uncertainty because no such estimates were obtained before the CBMS 2000.
In this section we provide information on the survey methods for the CBMS 2000. Section 2 presents information related to the undergraduate student population enrolled in statistics courses, showing the increasing trends in those enrollments as well as how they compare to calculus enrollments. We include findings related to the new AP statistics courses and the effects of those courses on undergraduate statistics education. Section 3 gives results relating to the faculty teaching undergraduate statistics, showing that much of the teaching load is shifting from tenured or tenure-accruing faculty to other full-time faculty rather than part-time faculty. Section 4 describes teaching practices in undergraduate statistics courses, documenting the increasing use of computers. Section 5 presents conclusions.
1.1 CBMS 2000 Methods
The Survey Research Unit of the University of North Carolina was hired to help design and carry out the survey and to conduct statistical analyses of the responses. This is the first time in the history of the CBMS that a statistical group was involved in the design and implementation of the survey. The CBMS 2000 survey used stratified simple random samples of three separate populations: mathematics programs in not-for-profit two-year colleges (N = 1.007, n = 300), mathematics departments in four-year colleges and universities (N = 1.430, n = 240), and statistics departments in four-year colleges and universities (N = 70, n = 60). Surveys were mailed to sampled departments in September 2000 and data collection was completed in February 2001. The response rates were 60% for two-year college mathematics programs, 70% for four-year college mathematics departments, and 75% for statistics departments. Because stratified random sampling without replacement was used for choosing the samples within each of the three populations, the stratification also had to be taken into account in calculating summary measures and their standard errors. …