SASS Teacher Follow-up Survey (TFS)

5. Data Quality and Comparability

Sampling Error

Because the TFS sample is a subsample of the SASS teacher sample, the SASS teacher replicate weights are used to derive the TFS replicate weights. (See the discussion of sampling error and variance estimation for SASS in the SASS chapter.) The base weight for each TFS teacher is multiplied by each of the SASS replicate weights divided by the SASS teacher full-sample base weight for that teacher. To calculate the 88 replicate weights, which should be used for variance calculations, these TFS replicate basic weights are processed through the remainder of the TFS weighting system.

Nonsampling Error

Coverage Error. There is a potential for bias to be introduced into TFS because the TFS frame only includes teachers who responded to SASS.

Nonresponse Error.
Unit nonresponse. The weighted unit response rate for public school teachers who completed the Current Teacher Questionnaire for the 2012–13 TFS was 81 percent. The weighted unit response rate for  public school teachers completing the Former Teacher Questionnaire was slightly lower (75 percent).

The overall response rate represents the response rate to the survey, taking into consideration each stage of data collection. For a teacher to be eligible for TFS, it was necessary to have received the Teacher Listing Form from the school during the previous year's SASS data collection, which provided a sampling frame for teachers at that school, and for the teacher to have responded to the SASS teacher questionnaire.

Item nonresponse. Item response rates indicate the percentage of respondents who answered a given survey question or item. The weighted TFS item response rates are produced by dividing the number of sampled teachers who responded to an item by the number of sampled teachers who were eligible to answer that item, and then adjusting those rates by the final weight.

Measurement error. Reinterviews were conducted for the purpose of measuring response variance in the 1994–95 TFS. The reinterviews were conducted through two reinterview questionnaires—one for mail cases and another for telephone cases. Each questionnaire contained a subset of questions from the original questionnaire. Seventy-eight percent of the questions evaluated displayed high response variance; only 5 percent displayed low re­sponse variance. (All but one of the 54 questions on teaching methods had moderate or high response variance.) This reinterview study again confirmed that "mark all that apply" questions tend to be problematic. See Response Variance in the 1994–95 Teacher Follow-up Survey (Bushery et al. 1998).

Data Comparability

Care must be taken in estimating change over time in a TFS data element, because some of the measured change may not be attributable to a change in the educational system, but due to changes in the sampling frame, questionnaire item wording, or other changes. For example, the definitions of the locale codes based on the U.S. Census were revised in 2000 and again in 2003. Changes in how schools' locales are categorized over time may account for at least some changes that are noted from previous administrations. This impacts the urbanicity variables included in the data files, which are based on the 2000 Census definitions for locale codes.

Additionally, although previous TFS release reports have included information on teachers who taught in public or private schools during the SASS base-year, only public school teachers are included for 2012–13 TFS reporting due to unprecedented low response rates among private school teachers; private school teacher responses were dropped from the 2012 TFS Current and Former Teacher data files. For further information on the comparability of data elements, see Documentation for the 2012–13 Teacher Follow-up Survey (Cox, Thomas, Simon, and Strizek, forthcoming).

Top