|Interpreting Buck Test Data|
|In last month's column I
presented information about the dates, locations, costs, etc. for this
year's performance tests. As you are reading this hopefully many of your
young bucks are at one or more of the performance tests, and you are
wondering how to interpret and use the data you are receiving. As promised
last month, in this month's column I will discuss how to interpret the
performance test data.
Depending on the test your animals are entered in you will receive all or some of the following data: average daily gain, ribeye area, beginning scrotal circumference, final scrotal circumference, and feed efficiency.
The first thing to remember is that you cannot compare results between test sites. Unfortunately the conditions and equipment at each test site are different making it impossible to make a valid comparison of animals between sites.
For instance, the animals at Angelo State almost always gain weight faster and have larger loins than the animals at Langston. Does that make them better? Much as I would like to answer that question with a resounding yes, since my animals are tested at Angelo State, that would be an incorrect answer. In fact many of the animals at Langston are 1/2 (same sire) or full brothers to the bucks at Angelo State. A large number (almost all) of the bucks at Langston the last few years came from two breeders who also test their animals at Angelo State. The results for these breeder's bucks at Angelo State were significantly better than the results for their animals at Langston even though they had similar or identical genetic backgrounds.
A variety of factors could account for this including differences in environment, measuring methods and equipment, feed, etc. Bottomline, don't try to compare results between test sites.
Many meat goat people seem to enjoy the sport of breed bashing, and think performance test results are great for promoting their favorite breed and/or bashing a breed they dislike. I recently ran across an instance of this in a different publication. The writer tried to prove the superiority of one breed over another using the test results from 3 animals from her favorite breed versus the test results for 9 animals from a competing breed. To compound the problem the animals at the test were kept under conditions that were not identical and tended to skew the results in question. Finally, the writer did not know how to interpret and compare results leading to more false conclusions.
The second thing to remember is that you need to deal with statistically significant sample sizes to draw any kind of valid conclusions. The results from three goats of one breed whose members number in the thousands and are genetically diverse tells you almost nothing about the characteristics of the entire breed. From a statistical standpoint those results are meaningless. Similarly, the results from one son of a sire tell you almost nothing about that sire.
Lets take another look at the article I mentioned above. The writer pointed out that at the test her favorite breed had a combined average daily gain of 0.3783 lbs/day versus 0.3711 lbs/day of gain for the other breed. Does that really mean one breed is superior to the other or even comparable? No. Given the sample sizes we can say with 99% confidence that the average daily gains for that writer's favorite breed range between 0.5868 lbs/day and 0.1698 lbs/day, and the average daily gains for the breed she dislikes fall between 0.4915 lbs/day and 0.2507 lbs/day. Maybe one breed is better than the other maybe not, we can't tell from such a small test sample.
To draw the conclusion that one is better than the other with 99% confidence from the test results obtained would have required testing 5,034 animals of each breed. A wider difference in results would require smaller sample sizes, but the samples would still have to be much larger than the number of animals currently being tested. For instance, if one breed could on average outgain the other by .1 lb/day, sample sizes of only 27 animals from each breed would be required to state with 99% confidence that the higher gaining breed was superior to the slower gaining breed from a weight gain standpoint.
I might add that the animals tested would need to comprise a representative and random sample of the gene pool for both breeds for the results to be valid. I don't think anybody would want to put the reputation of their favorite breed on the line against 27 Boer bucks that I handpicked from my herd and/or a few of the other top testing Boer herds. Which brings me to another point, the best animals in just about every breed are going to be superior to the average or worse animals in competing breeds. The key is to select the best genetics of whatever breed you choose to raise.
Performance test results are much more useful for identifying superior lines within a breed than for promoting or bashing breeds. Lets take for example a sire that had 18 sons on test at Angelo State in 1999. His sons averaged 0.679 lbs/day of gain versus the average of 0.575 lbs/day. From a statistical standpoint we can be 99% confident that this sire's offspring gained weight at a better than average rate because at that level of confidence statistically speaking the average daily gains for his offspring surpassed the average for all goats on test by between 0.036 lbs/day and 0.172 lbs/day.
Now your saying great, I don't have 18 offspring to test from my buck. That isn't necessarily a problem. It would be great if you had 18 animals to test since than the confidence window is only + or - .068 lbs/day (meaning that if your animals gain that much more than the average you can be 99% certain that your genetics are above average) when you test 18 animals. However, lets say you only have five animals to test. The confidence interval then expands to + or - 0.122 lbs/day. So if your five animals can outgain the average by at least 0.122 lbs/day you have a winner. Test four bucks and now you need to beat the average (or your competition) by 0.136 lbs/day, bring only 3 and the number grows to 0.157 lbs/day, two bucks requires 0.191 lbs/day extra, and if you only bring one little buck he needs to outgain the average by .269 lbs/day for you to have much to crow about. That's a pretty tall order since that would generally make him the top testing buck most years.
Explaining it another way, if you take one buck to the test and he gains .500 lbs/day, it really means that your sire's offspring gain on average somewhere between 0.231 lbs/day and 0.769 lbs/day on average as a group. That is a huge difference, a wide enough spread to say that your sire might be the best or just as conceivably the worst sire out there. Even if that one lonely buck gained 0.800 lbs/day on test it is possible that your sire is worse than average since the low end of the range of gains for his offspring would be 0.531 lbs/day compared to the average of 0.575 lbs/day. In other words the results are next to meaningless unless you bring a large sample of offspring for each sire you want to test. The more offspring from each sire you test, the more certain you can be of the accuracy of the results obtained.
The third thing to keep in mind is that for the data to be meaningful the test must be designed in a manner that treats all of the test subjects the same. If the goal is to measure parasite resistance the test subjects should ideally be kept in the same pen throughout the test so they are subject to the same conditions. You can't tell much about parasite resistance if one group of six goats is kept in a one acre pen and the other group only has three goats in a one acre pen. It shouldn't be a huge surprise if the goats in the pen with the higher stocking rate has a bigger problem with parasites. Such a result tells you nothing about parasite resistance. When analyzing test results try to find out as much as you can about how the test was conducted so you can determine if the results really say what they appear to say.
Finally, be very careful in analyzing the data. Average daily gains are fairly straight forward to analyze and compare, but you can quickly get into trouble trying to compare scrotal circumferences, feed efficiencies, and/or ribeye areas. All of these measures are size related, and the relationship is not linear (meaning that if the weight of the goat doubled the measurement would double as well). The following table shows a comparison of average scrotal circumference and ribeye areas for different size groups of goats from the 1999 and 2000 Angelo State University performance tests:
Final Scrotal Average # in
Weight Circ. REA Group
71.2 23.3 1.39 10
81.3 24.4 1.66 7
87.5 24.4 1.70 6
92.0 26.9 1.74 5
96.8 27.1 1.73 12
101.6 27.3 2.01 13
107.1 28.1 2.03 15
112.4 27.4 2.19 20
117.4 28.2 2.09 33
121.6 28.6 2.15 23
127.3 28.8 2.11 16
131.9 29.0 2.15 23
136.0 29.1 2.24 8
142.1 29.3 2.11 7
146.0 29.8 2.10 4
161.4 30.6 2.20 9
As you can see when the weight of the goats doubles from 80 lbs to 160 lbs the scrotal circumference only increases about 25% and the ribeye area increases about 33%. This chart isn't perfect since some of the sample sizes are too small, but you get the general idea. If we could get more animals on test I expect you would see some refinement in these numbers until it would resemble a smooth curve when graphed instead of the jagged curve such a graph would appear as with the above data.
Feed efficiency (measured in lbs of feed required per lb of weight gain, thus lower numbers are better here) is another tricky statistic to analyze. The only test site supplying feed efficiencies for individual animals is Langston. The chart below summarizes the feed efficiencies observed at the 2000 Langston test:
Final Feed # in
Weight Eff. Group
81.5 5.9 2
92.4 6.5 8
100.9 7.3 12
109.8 7.5 8
122.8 7.8 2
136.6 9.1 1
Before you get too excited about that buck with the 2.2 square inch ribeye area and 6.5 feed efficiency you need to take into account the size of that buck when the measurement was taken. If those measurements are for a 100 lb buck they would be pretty good, but a 2.2 square inch ribeye is merely average for a 160 lb buck and a 6.5 feed efficiency is below average for an 80 lb buck.
Getting back to the article mentioned previously the author of that piece made a big deal about the superior feed efficiency of her favorite breed compared to the breed she disliked. In doing so she completely ignored the fact that her breed came off test weighing 20 lbs less than the other breed, thus one would expect the breed she was promoting to have a feed efficiency 1.3 less than the other breed. What was the actual difference in feed efficiency? Only 0.82 indicating that had an adjustment been made for the difference in size between the two groups of animals, the conclusion would have been exactly the opposite. The breed being bashed in this case, is on a size adjusted basis actually superior to the breed that this particular author was trying to promote. Be careful when analyzing your test data so you don't draw any false conclusions from it.
Also, remember that the charts above are a summary of the results from specific test sites in past years, they may not apply to the results obtained at this year's tests at those same sites, and will almost certainly be inapplicable to the results past or future from other test sites.
In conclusion, test results are not comparable between different test sites, results are only valid if statistically significant samples are tested and the tests are properly designed, and finally the data must be evaluated accurately taking into account size differences that affect the results. If you keep these things in mind you will find performance test results to be a valuable tool for improving the quality of your commercial meat goat herd.
|This page updated 07/27/01|