skip navigation

Thursday, September 2, 2010

[ – ] Text Size [ + ]  |  Print Page

Errata: Livingston Survey

This page shows the changes made to correct errors to the individual observations in the Livingston Survey data set. Errors in the individual observations may affect the mean values for each survey date and may also affect the median values. When we post corrections to the individual observations, we also post the updated mean and median files. The variable names listed below are defined in the documentation. PDF

On June 8, 2005, we made changes to the following observations:

1. ID # 255 in the survey of June 1992

  • Problem: Previously, for observation 255 in June 1992, the category code and all forecast data (except for the base values) for every variable were missing.
  • Solution: The forecast data and category code for this observation in June 1992 were entered for each variable. Please note that the participant did not provide a forecast for every variable (PPI, WMFG, and RTTR forecasts missing), nor did he or she forecast for every time period. The exact details of this forecast are in the raw data file. The addition of the previously omitted forecast data had a minimal effect on the June 1992 mean and median values for most variables.

2. ID # 394 and #404 in the survey of December 1994

  • Problem: Previously, the category codes were missing for observations 394 and 404. This affected the raw data files published on the web page but not the files of means or medians. A glitch in the program that exports the raw data to Excel caused these two observations’ December 1994 base values and forecasts to shift to the right by one cell. This had no effect on the median and mean values published on the web. However, any statistics for December 1994 computed directly from the raw data files available on our web page would have been affected by this shift.
  • Solution: The category codes for these two observations in December 1994 were entered for each variable. The resulting raw data file currently published reflects this change, as the data are no longer shifted to the right for these two observations. The mean and median values remain unchanged for December 1994.