institutions, for a total of seven groups.10 It may be desirable to have more than one for-profit segment and to create separate sub-categories for public and private research universities that do and do not have medical schools, but this is an open questions.
The simplest approach is to base the productivity calculations on aggregate data for all or a sample of institutions in a given segment. Such aggregation is standard practice in sectoral productivity analysis and we see no reason not to use it in higher education. Given that IPEDS reports data for individual institutions, it is almost as easy to do the calculations separately by segment as it is to do them on national aggregates. With the segment indices in hand, it is straightforward to aggregate them to produce a sector-wide index.
Finally, the productivity index for each segment should be normalized to 1.0 in the base period, before the aggregation proceeds. This is a natural step within most any indexing methodology. Moreover, the normalization will emphasize that it is the productivity trends that are being measured and not comparisons of absolute productivity across segments.
The panel’s charge states that we should consider productivity measures at “different levels of aggregation: including the institution, system, and sector levels.” Our proposed model is designed to operate at the sector or subsector (segment) level. Given the IPEDS dataset and the specifics of the calculations, however, there is nothing to prevent researchers or administrators from applying the formulas to individual campuses and, by extension, to any desired set of campuses—say, within a system or state. Indeed, the illustration in Tables 4.1 to 4.4 is based on a single institution.
Two methodological caveats must be noted. First, trend comparisons should be made only with institutions in the same segment as the one being studied. Second, state-level or similar indices should themselves be disaggregated by segment. It makes no more sense to combine the apples and oranges of different segments at the state level than it does at the national level. Aggregation to a single state-level index should use the same methodology as the one described for segment aggregation. However, a question arises regarding the weights to be used for aggregation: should they be the same as for the national aggregation, or
10We indicate for-profit higher education as a separate category because its production methods often differ substantially from those in the nonprofit sector. The for-profit sector has been growing rapidly; additionally, recent concerns about the performance of these schools—including questions about their heavy revenue reliance on federal student loans and issues of quality—make them well worth consideration as part of any serious appraisal of higher education performance. We are not proposing that the productivity of for-profit higher education be measured differently, but rather that it be placed in its own segment for comparison purposes.