High Records Contrasted with Unselected Records and with Average Records as a Basis for Selecting Cows

The high correlation between a cow's highest record (or her lowest) and the average of the other records from which this one was selected results largely from the statistical effects of this selection itself. This high correlation does not indicate superiority of the selected record for predict...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:Journal of dairy science 1939-08, Vol.22 (8), p.607-617
Hauptverfasser: Berry, J.C., Lush, J.L.
Format: Artikel
Sprache:eng
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:The high correlation between a cow's highest record (or her lowest) and the average of the other records from which this one was selected results largely from the statistical effects of this selection itself. This high correlation does not indicate superiority of the selected record for predicting future records or breeding value. When the highest record is correlated with other records from which it was not selected the resulting coefficient (provided all cows have the same number of records) indicates that the high record is of somewhere near the same reliability as an unselected record but almost certainly less reliable than the average of all unselected records. Differences in number of completed records, however, is of so much practical importance in making selected records unfair that the use of the highest record, as an indication of a cow's lifetime producing ability, cannot be recommended. Averages appear to be more dependable than either selected or unselected single records for evaluating differences between cows.
ISSN:0022-0302
1525-3198
DOI:10.3168/jds.S0022-0302(39)92918-8