Management topics

Aug 19, 2011

The disastrous effects of biased reporting

There is an old comp u t e r phrase— garbage in, g a r b a g e out. The same can be said for performance data and its use to the seedstock and commercial sectors. Breed associations calculate biannual expected progeny differences (EPDs) (except for Angus, which runs them weekly) each fall for release at the first of the year, which commercial producers use to make their buying decisions. If you believe the WLJ Bull Survey, more and more emphasis is placed on EPDs every year, and rightly so. They have been shown to be by far the best objective measure of an animal’s value—better than ratios, adjusted weights, raw weights, or eyeballing.

The genetic evaluation models that calculate EPDs all assume they are dealing with unbiased data. Unfortunately, we all know this is not always the case. There are only a select few breeds that report more weaning weights than they do registrations something necessary for complete contemporary group (CG) reporting. It is interesting to note that one of these has mandatory reporting while the other does not.

A CG is defined as a grouping of animals born within an age range, managed the same, same sex and weaned at the same time. Sending all the animals from a CG gives a full genetic picture, but still there are many seedstock producers who mistakenly think they will improve the EPDs on their cattle by only sending in their best ones.

Let’s look at an example CG of six calves (Table 1). When all of the calves are reported, the average weaning weight is 488 pounds, and you see the CG’s weight deviations and ratios. Ratios are what breed associations supply producers because they are easy to understand, giving the percent above or below the CG average, with 100 being average. For the purposes of EPD calculations, weight deviations are used in the model. They are given in Table 1 so you can see their magnitude in the example. Despite the difference, the ratios and deviations track together. That is, as one changes, so does the other.

As you can see, in complete reporting, the three highest performing calves all would have their weaning weight EPDs go up because they all have positive deviations and ratios.

If only the top 50 percent of the calves are reported, however, only one animal (1006) has a positive deviation, and only his weaning weight EPD would go up. The other two highperforming calves, when biased reporting occurs, actually would have their EPDs drop.

The news gets worse.

Breeds use weaning weight as a correlated trait for numerous other EPDs and Indexes. Let me give you an example of how yearling ratios are calculated.

It is a given that people are going to cull cattle at weaning; hence, the formula accounts for selection, but it is critical to have complete weaning CGs for it to work correctly. Like the ratio formula’s accounting for selection, the EPD formulas do the exact same thing.

Obviously, if you cull your low-performing calves at weaning, you expect the ones you have left to be better than average. Many producers do not know yearling ratios are adjusted for selection or that adjusted yearling weights actually are a combination of adjusted weaning weight and post-weaning gain.

Let’s go through the example on the same animals in Table 1, assuming only the best three were developed as bulls so they would be the remaining yearling CG, and the other three were steered and sold at the sale barn. The formula for calculating yearling ratios is a tad complicated (see Figure 1), but its application is quite simple.

As for the 160-day postweaning gain that shows up in the formula, a breed association simply gets the actual average daily gain between the weaning date and yearling weight, multiplies that number by 160 days, uses the result for the ratio calculation, and adds it to the 205-day adjusted weight to get a 365day adjusted weight.

This is where it gets easy again. Let’s get back to the ratios. For our example, animals selected from Table 1 were 1006 with a 160-day post-weaning gain of 500 pounds (remember, he was the top-ratio bull at weaning), 1028 of 600 pounds and 1026 of 575 pounds, for an average 160-day postweaning gain for the three bulls of 558 pounds. If you run those bulls through the formula when complete reporting had oc curred, you must remember the average adjusted weaning weight for the six bulls in the weaning CG was only 488 pounds. If you do calculations for the adjusted ratios based on a fully reported weaning CG of six and a yearling CG of three, 1006’s ratio is 100.6, 1028’s ratio is 106.8, and 1026’s ratio is 102.7. Again, the EPD calculations follow the same logic, and like those ratios, account for selection and will be reflected in the bulls’ yearling EPDs.

Let’s see what happens if you only partially reported your calf crop. Now, if you run through the calculations, the ratio for 1006 is 97.3, 1028 is 103.4, and 1026 is 99.4. Those obviously are much lower, with two calves dropping below 100. That false appearance of lower performance will be reflected in lower yearling EPDs than they would have had with complete reporting.

Colorado State University (CSU) did an interesting study (Table 2) that shows how selective reporting affects the milk EPD. In its sample population, two sires each had 10 calves. In the study, CSU ran EPDs and, under complete reporting, Sire 1’s EPDs were 0 for weaning weight (WW) and milk, while Sire 2 had EPDs of 20 WW and -2 milk. When the researchers selectively reported Sire 1’s calves, which were lower performing, however, his WW EPD went up to 6, but his milk EPD dropped to -2. By the same token, Sire 2 dropped in WW EPD 8 pounds but rose in milk by 3.

You see milk react negatively, as is the case for Sire 1, because—when WW EPD is biased like that and daughters come into production—their calves grow slower than expected, and that gets attributed to the milk EPD incorrectly. The opposite occurs with Sire 2. Because of selective reporting, his WW EPD is artificially low, so, when his daughters have calves, his milk EPD will be biased higher.

A seedstock producer must think long and hard about what data they are going to submit this fall. If you give a calf preferential treatment and add it in for short term gain, the milk EPD will boomerang around to get you. There is no way to game the system over the long haul. If you are a seedstock breeder, you should make a decision this fall when you weigh calves—are you going weigh them all or weigh none of them—never a biased sub-sample. As a commercial producer, you really need to decide who you are doing business with if you are going to rely on EPDs. Absolutely, cattle still need to be structurally correct, but the industry works on a small margin and genetic predictions need to part of any informed decision. — Dr. Bob Hough [Dr. Bob Hough has served as the executive vice president of the Red Angus Association of America and more recently as executive vice president of the North American Limousin Foundation from 2009 to early 2011. He is now a consultant, freelance writer and semi-retired.]