Errata for NLSY97 Round 18 Release

Errata for NLSY97 Round 18 Release

NEWEST ERRATA [posted 6/25/2020]

Corrections from a Review of CV_HRLY_COMPENSATION and CV_HRLY_PAY Variables

A comprehensive review of the created wage variables [CV_HRLY_COMPENSATION and CV_HRLY_PAY] led to updates to these created variables and select underlying raw data through all rounds. These updates have been included in the most recent version of the public-use data available on the NLS Investigator.

A summary of the issues for the created wage variables follows:

  1. The text fill in YEMP-97300 and YEMP-97400, which should have been worded ‘per job’ or ‘per item,’ was incorrectly worded ‘per week’. The values for the created variables were recalculated to reflect the respondent’s answer for per week.
  2. Respondents reporting a time unit of "semi-monthly" were asked about "monthly" amount for the YEMP-19200 and YEMP-83100 branches; the created wage variables were recalculated with the new time unit.
  3. The created wage program incorrectly calculated the variables for respondents who reported an ‘other specify’ time unit option in their current/most recent wages (YEMP-38014, YEMP-38107), overtime (YEMP-24502, YEMP-88502, YEMP-34404, YEMP-38202, YEMP-38003, YEMP-98404A), or compensation (YEMP-100205, YEMP-21600, YEMP-38329B, YEMP-38407) as weekly. These cases were reviewed individually to ensure that the created values match the respondent’s answer in the ‘other-specify’ verbatim. Cases in which there wasn’t a verbatim or in which the time unit rate of pay didn’t match the initial reported time unit rate of pay were assigned a -3.
  4. An error in the created wage program caused the compensation amount to be double-counted when respondents reported overtime and any other type of compensation in YEMP-21200 and YEMP-38329D. This occurred in select cases when the respondent reported a start wage with overtime and additional compensation and then updated the compensation for the current or most recent wage.
    1. In rounds 8 and 9, a coding error in the created wage program caused the number of hours worked per week for compensation by hour or by per job/per item to be missing and the created compensation variables were then coded as an invalid skip. The created variables were recalculated using the actual hours.
    2. Longitudinal inconsistencies were corrected in the created wage program when the reported number of hours worked or the number of weeks reported were zero or an invalid answer. These instances were standardized across all rounds. In addition, in Round 11 a programing error caused all annual salaries to be divided by 52 weeks regardless of the number of weeks reported; these values were corrected using the number of weeks reported when available.

Changes to raw data as well as created variables:

  1. In round 8, raw data wage items (YEMP-33600, YEMP-97500, YEMP-100250, YEMP-22629, YEMP-22630, and YEMP-22632) were not adjusted to include 2 implied decimals. This impacted the created wage variables. Both the raw data and the created variables were adjusted.
  2. A relatively small number of cases in rounds 3, 5, 6 and 7 had a duplication of jobs on their employer (YEMP) rosters. In the initial cleanup work done prior to each data release, these jobs were deleted from the YEMP roster, but the raw data in the employer supplement loops was left unedited; these have now been removed.


Corrections to the Created Incarceration Variables and Event Histories [posted 6/5/2020]

Updates have been made to the incarceration monthly arrays and incarceration created variables after a review of cases where the created variable indicating the number of months incarcerated (INCARC_TOTMONTHS) did not match the number of months incarcerated in the incarceration monthly history arrays (variables beginning with INCARC_STATUS). In almost all cases, INCARC_TOTMONTHS was longer than the number of months incarcerated in the arrays. The most common reasons for this difference were reports of overlapping incarcerations and duplicate reports of the same incarceration.

Updates were made to the incarceration created variables and, in some cases, incarceration event history arrays for 75 cases. The variables most affected were INCARC_TOTMONTHS (75 updates), INCARC_TOTNUM (52 updates), INCARC_LENGTH_LONGEST (32 updates), and INCARC_LENGTH_FIRST (27 updates). Updates were also made to INCARC_FIRST (11 updates) and INCARC_AGE_FIRST (9 updates). Updates to the incarceration status array were made for 28 cases. These updates have been included in the most recent version of the public data.

Invalid Dates for Freelance Job Start Dates [posted 2/10/2020]

29 cases in 1998 have invalid freelance job start dates of 01/79 for the variable R2478300/R2478301 (FREELANCE_STARTDATEC.01). These dates should be set to missing as invalid skips (-3/-3).