Projected values vs. actual (raw) counts in LENA Start data
LENA’s norms for Words and Turns are based on 12-hour recordings; however, a family’s recording can be shorter, or can be up to 16 hours long. Variable durations make a direct comparison to the norms invalid. Statistical interpolation is necessary to compare variable length recordings to the norms and, by extension, to a family's own performance over time as the child ages. To this end, LENA computes and reports 12-hour projected values representing what the Words and Turns counts would have been had the recording been exactly 12 hours long. A recording must be at least 9.5 hours long to be eligible for this projection and comparison to the norms; shorter recordings are not valid. The 12-hour projected values are reported in the percentile-governed daily bars of the LENA report and in the AWC, and CTC columns of the full recording export. Note that on the first valid and non-hidden recording, the daily AWC and CTC values reflect the projected values with application of the Hawthorne effect correction.
The actual counts for Words and Turns in the processing results are just that - the actual number detected in the recording. These raw counts are generally not appropriate for analysis at the full recording level for the reasons stated above. The actual counts can be seen in the hourly bars on the LENA reports, in the AWC and CTC columns in the hourly and 5-minute level exports, and in the ITS file.
- To find the actual counts for a full recording, if desired, sum the hourly data. For recordings >12 hours in duration, the sum of hourly counts will generally be higher than the reported daily values unless there is no data in the recording beyond the 12th hour, in which case the values will be the same; for recordings <12 hours, the sum of hourly counts will be lower than the reported daily values. Note that the resulting values are not adjusted for duration and so are not appropriate for comparison against the 12-hour norms, nor against other recordings of varying durations from the same child or other children.
Note: For recordings that are too short for comparison against the LENA norms (<9.5 hours, E2), the LENA report will show an error on the daily bar, and the full recording export will show the actual count - that is, the sum of the hourly counts.