A REEXAMINATION OF THE NRMP MATCHING ALGORITHM - COMMENT

The authors and Williams (in the companion articles) agree on the theoretical and mathematical points regarding matches and, in some respects, on suggestions for further research. However, the authors question many of Williams' interpretations. Williams' description of the basic National Resident Matching Program (NRMP) algorithm is unaccurate, and it ignores match variations that have been incorporated into the match processing for many years. Also, he has misinterpreted NRMP statements about the result to be obtained by each applicant in the Match and ignored NRMP statements describing the ''bias'' in its algorithm. Williams compares the Match with colleges' admission processes; however, the comparison is invalid, especially regarding the issues of overfilling and the supposed benefits that accrue to students when schools make more offers than there are positions available. Regarding Williams' claim that the NRMP misleads Match participants concerning strategies for ranking preferences, the authors disagree and refer to their discussion of the issue in their companion article. They also explain why a matching process free of ''incentives for students to misrepresent their true preferences'' may not be feasible in practice. The authors contend that Williams has probably overestimated the likely impact of changing the NRMP algorithm. They agree with him about the lack of quantitative information to help applicants assess how long their rank-order lists should be, although available information about the relative competitiveness of different specialities coupled with the guidance of school advisers can help an applicant in this regard. Finally, Williams' assertion to the contrary, the authors' experience with program directors indicates that they are as sensitive to their specific Match results as are applicants. As for improvements needed, if most readers misinterpret the NRMP literature in the way Williams does, it should be revised. The authors disagree with Williams' recommendations about what should be stated in the NRMP literature about the Match; debates about the process should be made elsewhere. In conclusion, they report that efforts to improve the Match continue to be made by the NRMP.