Donald M. Marshall1 Department of Animal and Range Sciences
SDSU CATTLE 94-1
A typical experimental format involves Several of the reports in this publication
evaluating the response caused by application of refer to least squares means. I n balanced different treatments t o experimental subjects experimental designs, least squares means are (animals, carcasses, pens, pastures, etc.). The often the same as the simple raw means. effect of a given treatment might be evaluated However, when numbers of experimental by comparison t o a control group or t o one or subjects are not evenly distributed across more other treatment groups. However, a treatments, adjustments t o the means are problem with animal research (and other types needed. Appropriate adjustments are made by as well) is that variation not due to treatments least squares procedures. In addition, least often exists among experimental subjects. squares means are sometimes adjusted for extraneous sources of variation through a so- For example, suppose that animals receiving called analysis of variance. ration A grow faster than animals receiving ration B. Was the observed difference in growth Means (averages), correlations and other rates actually due t o differences in the rations or statistics presented in research results are t o other factors (i.e., genetics, age, sex, etc.) or sometimes followed by f some figure known as some of each? Statistical analyses evaluate the the standard error. The standard error provides amount of variation between treatment groups an indication of the possible error with which the relative t o the amount of variation within statistic was measured. The size of the treatment groups. In addition, variation caused standard error of a treatment mean depends on by factors other than treatments can sometimes the animal to animal variation within a treatment be eliminated by the statistical analysis. group and on the number of animals i n the group. The statement "the difference was statistically significant (P = .05)"indicates the All other factors being equal, the greater the probability of a difference of that magnitude number of animals and(or) replications per occurring from chance rather than from the treatment, the smaller the difference required t o research treatment is about 5%. achieve a given value for probability of significance. Stated another way, increasing the A correlation coefficient provides an number of animals or replications increases the indication of the relationship between t w o likelihood of detecting differences due t o factors and can range from -1 to + 1. A strong, treatments when such differences do indeed positive correlation (close to 1) indicates that as exist. one factor increases the other factor tends to increase, also. For example, several studies Several of the research reports i n this have shown a positive correlation between cow publication contain statistical terminology. milk yield and calf weaning weight. A strong Although such terms might be unfamiliar t o negative correlation (close t o -1) indicates that some readers, the statistical analyses allow for as one factor increases the other factor tends to more appropriate interpretation of results and decrease. A correlation near zero indicates the make the reports more useful. t w o factors are unrelated.