Changes to Editing System - August 2003


Calculation of Owner-Sampler Percentage

By George Wiggans

More complete reporting of testing characteristics on test day and increasing diversity of testing arrangements has led to interest in reducing reliance on the DHI type-of-test plan code in processing test-day records. In May 2003 the basis for determining owner sampler percentage (OS%) was switched from DHI type-of-test code to test-day supervision code. The number of unsupervised tests divided by the total number of tests determines OS% for the lactation record. A consequence of this change was that labor efficient record (LER) plans for which the supervisor visited every third month became predominantly owner-sampler. Beginning in August 2003, the test data that were electronically collected (supervision code 6) will be counted as only 50% owner-sampler (receiving half weight in determining OS%) as long as the DHI type-of-test code indicates a supervised test plan. The OS% for test-day data with supervision codes 3 and 7, for which both the supervisor and owner collect data on test day, will be reduced from 50% to 25%, recognizing that direct test-day supervision is greater for this testing arrangement. The implemented changes preserve the intent of the LER program to generate supervised records by extending some credit for the use of electronic meters for data collection between supervisor visits.


Calculation of Data Collection Ratings

By Paul VanRaden

A data collection rating (DCR) reflects the amount of information collected and its accuracy. The DCR for owner-sampler data or if only one milking per day is recorded is lower than if a supervisor records all milkings on test day. To reduce the cost of testing, supervision of only one milking on test day has become the most common plan. In some cases, the owner collects the milk weights for the unsupervised milking(s). Because of the reduction for owner sampling, the DCR in such cases was lower than the DCR for the single supervised milking would have been. The DCR weights have been adjusted so that the addition of the owner sampler data to that of a supervised test will not lower the DCR.


Edits Applied to Owner Sampler Herds

By George Wiggans and Paul VanRaden

Previously, all data back to 1996 calvings for owner-sampler herds were included or deleted based on latest herd record. Limits for percentage of milk shipped and percentage of animals completely identified are now applied to owner-sampler records on a yearly basis. The values for the last test date during each calendar year are used for all calvings during that year. To prevent reporting errors from causing a year's data to be excluded, values from the previous test day must support the decision to reject data.

Data from herds with excessive possible outliers are rejected. Previous edits that failed to detect such herds have been corrected. To mitigate the impact of this correction, the threshold of possible outliers per herd-year was raised from 6.4 to 10%.