Cyclic variation in cattle feed intake data: characterization and implications for experimental design.

W. W. Stroup, M. K. Nielsen, J. A. Gosey

Research output: Contribution to journalArticlepeer-review

7 Scopus citations


Feed intake data were collected every 6 h in a 140-d feeding trial involving two pens of 15 bull calves, each using one Pinpointer single-animal feeding stall per pen. Spectral analysis of these data revealed strong cyclic patterns of feed intake. These patterns were unique to each animal and consisted of two or more cycle lengths, some up to 28 d, for each animal. Feed intake behavior is an important factor in many forms of animal experimentation. Animal researchers frequently use row-column designs, where columns represent animals and rows represent periods. Latin squares and crossover designs are common examples. Standard analysis of variance procedures are appropriate for these experiments only if certain assumptions are met; one is that if cyclic variation is present, it is identical for all animals, i.e., there is no row X column interaction. If this assumption is not satisfied, standard analysis of variance procedures will result in upward bias of the mean square error and may result in serious distortion of treatment effect estimates. This occurred using the data from the feeding trial reported in this paper. Time-series analysis of covariance greatly improved the accuracy and efficiency of estimates of treatment effects. Consideration of variations in cyclic behavior should be part of the design process in experiments using feed intake data. Guidelines for the design of experiments to take advantage of time-series methodology are given.

Original languageEnglish (US)
Pages (from-to)1638-1647
Number of pages10
JournalJournal of animal science
Issue number6
StatePublished - Jun 1987
Externally publishedYes

ASJC Scopus subject areas

  • Food Science
  • Animal Science and Zoology
  • Genetics


Dive into the research topics of 'Cyclic variation in cattle feed intake data: characterization and implications for experimental design.'. Together they form a unique fingerprint.

Cite this