The Numbers Game and Reproduction
Transcript of Select Sires' Reproductive Moment Program
on DairyLine Radio Which Aired Aug. 28, 2003
With Mel DeJarnette, reproductive specialist,
Select Sires Inc.
Mel, this week you’re talking about the numbers game where producers and scientists need to be careful on how much faith they put into fertility numbers.
When it comes to numbers related to reproductive efficiency, there are few dairymen who wouldn’t suggest their numbers could stand a little improvement. However, many producers often place either too much or too little faith in the ability of new technologies to solve their breeding problems.
Fertility is a binomial variable, which means there are only two possible outcomes to a given insemination: either she is pregnant or she is not. A nature of all binomial traits is that they are very easily biased by random chance, and thus many observations are needed before we begin to have confidence in what the numbers are trying to tell us.
Also, herd by-treatment interactions are very common in fertility studies, which means, what appears to works in some herds does not necessarily work in all herds. Thus, a common mistake by many researchers is that we’re often tempted to place too much faith in the results of single location studies that had too few animals to adequately address the questions we were asking.
In contrast, a common mistake made by many producers is to take a valid, well tested and proven technology, and think they can get a true estimate of how well it’s going to perform in their herd based on the first 10 or 20 animals treated. With such small numbers, the expectations can easily be over or under sold. One hundred animals per treatment are needed to even pretend we have a handle on fertility, and even then, the true answer may be plus or minus 10 percent of the answer we observed.
Dr. Mike Day at The Ohio State University recently conducted a mock fertility trial by randomizing 419 cows in four different herds into two "imaginary treatments," but then he treated all cows exactly the same. This was just a test to see how much random chance may be influencing the results of our "real" fertility studies. In one herd of 83 cows, "imaginary treatment one" had 15 percent advantage over "imaginary treatment two." However in another herd of 130 cows, "imaginary treatment two" had 28 percent advantage over treatment one. Across the four herds, the results were exactly identical, just as they should be, but focusing too deeply on the results of any individual herd could have been very misleading. This reminds us that when scientists say the difference between two treatments is "statistically significant," this simply means they are 95 percent confident the differences are real, however, that still leaves a five percent chance that the observed differences may not be real.
So what does all this numbers talk mean for the average dairyman? Well first, before investing time, dollars and resources in a new "Johnny-come-lately technology," question the quantity of supporting data and the repeatability of results across multiple herds. If it’s a small amount of data and sounds too good to be true, it probably is. Secondly, once you buy into a new technology, you must be willing to stick with it long enough to get at least 100 or more services to even begin to suggest you’re getting a handle on how well it is working in your herd. For more advice on proven technologies that may help to get fertility numbers headed in the right direction in your herd, contact your local Select Sires, Inc. Reproductive Solutions™ Specialist.
™Reproductive Solutions is a trademark of Select Sires Inc.
|
