As we review various production indices, it has become evident that there are two main methods of summarizing such indices. The first is a classical method, of providing an average. The second is what we call a “proportion” or a “management by specification.” In the case of measuring the wean-to-service interval, PigCHAMP provides two alternatives. The first is the average and the second is the proportion of sows bred by seven days. Many producers analyze both, and recognize the utility of the latter, but there is a place to consider the use of proportions in more detail and in more applications.

Less Emphasis on Averages

The use of proportions can be considered part of management by specifications. For management purposes, I believe we should stop monitoring averages and instead monitor specifications. The question at each stage of production, whether it is reproductive performance or transfer of pigs from farrowing room to nursery, is not a measure of the average performance but rather the proportion of pigs that meet the specifications for ease of production. In other words, what proportion of sows with poor performance is acceptable? Management by specifications, rather than averages, is driven by a number of factors:

• Distributions of performance are often not bell-shaped distributions. This is especially true for wean-to-service interval. Averages can be driven by a small population of sows that are at extreme distances from the mean. In the case of this variable, they always exist on one side of the main and therefore create what we call a skewed distribution.
• Measurement that focuses on good and bad sows also is relatively easy. Classification of sows may allow us to focus on the correct population. Not only is the proportion a useful number, but the classification of sows that exceed the specification can allow us to identify the correct animals to manage.
• It’s a good business practice. Production that is out of specification is simply a bad business practice. Such production has extraordinary costs that need to be identified in more detail. In the case of sows coming into estrus beyond seven days, there can be costs in management of the breeding space and extra costs of continued monitoring. There can also be a fatigue in that monitoring so that the likelihood of estrus detection goes down.

An Industrial Model

Most industrial monitoring systems focus on specifications. Performance is viewed in terms of value and out-ofspecification product is viewed as being detrimental to the production flow and the time required for management of the sow. Quality manufacturing guidelines always focus on meeting specifications derived by the next stage of production.

Figure 1: Cumulative Sum Graph of the Proportion of Sows Bred by Seven Days (after weaning by week over 65 sow herds)

Figure 1 shows the cumulative sum of weeks with the proportion of sows bred by seven days. This figure exemplifies a great deal of opportunity for improvement and also emphasizes that until this variable is under control, producers must make specific plans for its management.

Saying that, it should be emphasized that this is not a perfect variable. The main problem is actually in the denominator. In other words, which sows are included in this analysis? If a sow is culled because of a lack of estrus, it is not included in this analysis. Moreover, the pressures upon the burden may result in variation from week to week and season to season, and the decision to retain sows for breeding may also vary. Nonetheless, consider this as an important management variable in improving the success and manageability of the sow herd. Through such management by specifications, producers will see real changes.