Survey Methodology

Data Weighting and Standard Errors of Measurement

Data weighting techniques are only applied during sample survey years to estimate data for nonsampled academic institutions. The survey sample design includes four strata of academic institutions. Three strata are certainty strata in which all of the institutions are surveyed. The fourth stratum is a probability stratum in which a sample of institutions that are master's-granting or below are surveyed.

The survey strata include:

  1. A certainty stratum of all doctorate-granting institutions;
  2. A certainty stratum of historically black colleges and universities (HBCUs);
  3. A certainty stratum of institutions in the university systems of Massachusetts, Missouri, the State of New York, and Wisconsin; and
  4. A probability stratum of master's-granting and below institutions.

once the data are collected, sample weighting to universe estimates is performed (during sample survey years) by increasing the individual questionnaire data values by the inverse of the sampling ratio. Thus, in aggregating the data for institutions from the probability sample for tabulation purposes, each datum value was weighted by the inverse of the sampling ratio.

Estimates derived for institutions in the probability stratum differ to some degree from the numbers that would have been obtained if data were collected for the total population of eligible institutions. The probable magnitude of these differences can be estimated using standard statistical techniques. The relative standard error (coefficient of variation) of an estimate is calculated by dividing the standard error by the estimate. For example, in FY 1992 total separately budgeted R&D expenditures was $18.9 billion and the standard error of the estimate was $72.6 million at the 95-percent confidence level, with a coefficient of variation of +/- 0.4 percent. Similarly, for the estimate of $11.1 billion in federally financed R&D expenditures, the 95-percent confidence interval was +/- $35.8 million, with a coefficient of variation of +/- 0.3 percent (see table A-5).