Skip Navigation

Frequently Asked Questions About BTLS


Is the sample nationally representative?

While BTLS is a national sample, it is not nationally representative of all public school teachers who began teaching in 2007 or 2008. BTLS included all public school teachers who reported that they began teaching in 2007 or 2008 in the 2007-08 SASS. The SASS sample was designed to provide state-level estimates for new teachers. This survey included only first-year teachers.


Why does the first wave sample size differ from the number of beginning teachers in the 2007-08 SASS?

All teachers in public schools who reported that their first year of teaching was in 2007 or 2008 were initially eligible for BTLS. However, during the second and subsequent waves of BTLS, some cases were determined to have begun teaching prior to 2007 or reported that they were never a teacher. These cases were removed from the BTLS data file and their weights were redistributed among the remaining cases. In general, cases that later reported never being a teacher were those who were teaching a regularly scheduled class in any of grades K-12 during the SASS school year but were actually a principal or a postsecondary teacher who was only temporarily in this position. These individuals did not self-identify as a teacher and were not the intended participants in BTLS.


Why does the BTLS sample size decrease over time?

The BTLS sample of eligible cases has changed over time. As discussed above, cases that are later deemed ineligible for BTLS are removed from all waves—their data are not included on the data file. In addition, cases may become out of scope because they are permanently incapacitated (e.g., imprisoned) or have died. These out-of-scope cases remain on the data file.


How long will NCES follow this cohort of BTLS?

NCES will continue to follow the 2007-08 BTLS cohort through the 2011-12 school year, or for 5 years.


Is BTLS fully imputed?

No, BTLS is not a fully imputed dataset. Several variables in each BTLS wave were identified as “key variables” and were imputed for missing data. As surveys went through the different stages of imputation, a numerical flag corresponding to the type of imputation was assigned to each imputed item. By looking at the value of the imputation flag variable associated with a given survey variable, data users are able to identify which items were imputed and how the imputations were performed. Data users can use this imputation flag to decide whether or not to include imputed data in their analysis and which types of imputed data to employ.