We have become very adept at reducing the complexity of the reality around us to facilitate understanding it, explaining it and managing it. If I were to ask you what the pig price was today, you would probably quote me the average market price reported on the Internet or the radio for your area. In fact, most pork processing plants have a set of discounts and premiums (SEUROP and many others in the EU) to a base price each day depending on the weight and quality measurements of an animal at slaughter. When I ask you for today’s pig price, I am not expecting you to sit me down and quote to me all the prices paid for every weight range and quality measurement across all the plants you may sell a pig to today.
In most cases, reducing to the average or some other representational attribute is just fine. If I ask you how you are feeling today, I would most likely hear back, “fine” or “ok” or maybe “a little tired”. I would not expect you to say, “my feet hurt from these shoes, and my arms are a little itchy from the dry weather, and I’m a little queasy from missing breakfast and could really use a short nap.”
The important thing to realize from this little exercise is that there is a large amount and variety of things going on at any one time and we get by the complexity by simplification or reducing the complexity to a single descriptor. Sometimes it is important to know the individual things that are going on. For instance, very few teachers would give every student in the class the class average for the last test, though it would be a representation of how everyone did. In this case the variety of grades received and accounted to each person is important. One way to represent the performance of the class without reducing it to a single measure like the average is to organize the individual grades received on the last test as a distribution.
For instance, if we were using a grading scale like A, B, C, D and F, we could give the number or the percent of the total class which received each grade. If we graph the percent of students receiving each grade using a bar graph, we have created something called a histogram. With the histogram of grades, we have a much richer understanding than a simple class average, of how the class performed and each student can see how they did in comparison to all the others. For instance, with the histogram, a student can quickly calculate what percentage of the class received a higher and/or lower score than them.
We know that when animals get sick, not all the them are affected in the same way. Some of the animals may be very ill if their immune systems were naïve to the infection or if they were weakened by poor nutrition or dehydration etc. Others in the same pen may seem completely unaffected by the sickness of the others. Diseases often have a very diverse effect on a group of animals and understanding how to represent that diversity with tools like a histogram, can deepen awareness about what is going on and what the real cost of it might be.
Veterinarians know that disease is the biggest single cause of variation in the weights of animals in a contemporary group, like a finishing barn containing animals from the same week’s weaning. The economic cost of the disease will vary across each animal depending on how ill it is and how that illness is affecting key economic measures.
The four big economic measures affected by diseases are growth rate of the animal (often measured by the average daily gain at close of the barn, ADG), feed conversion (often measured at closeout by the average kg of feed used per kg of meat produced), mortality rate (though collected as deaths occur, these numbers are often not reported daily, but averaged and reported as an “annualized rate”) and quality of the carcass at harvest (which is a combination of size and factors such as lean meat percent, the absence of lesions and abscesses, sufficient intramuscular fat, and minimum measurements of certain key muscles such as the loin or loin eye muscle).
One of the things we are discovering is that when the cost of a disease is calculated using the average metrics for all animals in the group, like before and after the onset of the disease, the cost is almost always underestimated. We will discuss this in a future article in detail but the problem comes when the average for each of the big four metrics (ADG, FCR, Mortality Rate and Carcass Quality) listed above are imputed to a single representative animal to simplify the math.
The economic difference associated with the change of that average is calculated and the result is multiplied by the number of animals in the building. Until we get to the details in our upcoming article on this error, imagine that the minimum grade needed to pass to the next level in school was a B. If we used the technique above, we would average all the students grades together and give each of them the class average. If it were a B average, we would very likely be passing several students who did not achieve the minimum score, but by reducing the complexity of grading each one separately by imputing to each student the class average, we send many underprepared students into the next grade level. By analogy, we have overestimated their performance or underestimated their readiness to go on.