ARS | Publication request: USE OF BOX-COX TRANSFORMATION IN ANALYSIS OF FECAL EGG COUNT DATA

Submitted to: World Congress of Genetics Applied in Livestock Production
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: September 26, 2006
Publication Date: October 10, 2006
Citation: Sonstegard, T.S., Marcos, D., Gasbarre, L.C., Van Tassell, C.P. 2006. Use of box-cox transformation in analysis of fecal egg count data. World Congress of Genetics Applied in Livestock Production. 8(15-35):1-4.

Interpretive Summary: A genetic component to host resistance in cattle has been reported and could be used to accelerate genetic improvement. However, the non-normalized nature of the phenotypic data used to characterize the genetic potential of an animal is more likely to cause errors in analysis for estimating heritabilities and genome locations of parasite indicator traits. We applied a transformation method to this data in an attempt to normalize it. Improved normality should improve behavior of the statistical tests applied to this data. Transformation of bovine FEC data utilizing the Box-Cox transformation family was effective in reducing the skewness and kurtosis for all the variables studied, and dramatically increased estimates of heritability.

Technical Abstract: Fecal egg count (FEC) is used to identify and quantify gastrointestinal parasite infestations. However, FEC values are not distributed normally, and a small percentage of the herd is responsible for a majority of parasite transmission. Non-normality is a possible source of error when (co)variance components. A total of 6,378 FEC measures were determined for 409 animals between 1992 and 2003 from BARC Wye Angus herd. In this study, original data were transformed using an extension of the Box-Cox transformation to approach normality and to estimate (co)variance components. Transformation of bovine FEC data utilizing the Box-Cox was effective in reducing the coefficients of asymmetry and kurtosis for all the variables studied, which improved estimates of variance components. These results indicate that after transforming data by the Box-Cox procedure the genetic parameters may be more exact.