The data in this report come from many sources, including surveys conducted by the National Science Foundation (NSF) and other Federal agencies, and by non-Federal organizations. Many methods of data collection are represented, such as universe surveys, sample surveys, and compilations of administrative records. Users should thus take great care when comparing data from different sources. These data often will not be strictly comparable dueamong other thingsto differences in definitions, survey procedures, and phrasing of questions.
Survey accuracy is determined by the joint effects of sampling and nonsampling errors. In all of the surveys that are sources of data for this report, efforts are made to minimize these errors. Sampling errors arise because estimates based on a sample will differ from the figures that would have been obtained if a complete census had been taken.
All surveys, whether universe or sample, are also subject to nonsampling errors; these can arise from design, reporting, and processing errors as well as from errors due to faulty response or nonresponse. Nonsampling errors include respondent-based events, such as some respondents interpreting questions differently from other respondents; respondents making estimates rather than giving actual data; and respondents being unable or unwilling to provide complete, correct information. Errors can also arise during the processing of responses, such as recording and keying errors.