Investing wisely in on-farm trials

An on-farm trial is an investment in your business, so before starting a trial it is important to truly understand the question you are trying to answer. Most on-farm feed trials will never give the result they were set up to provide. In many cases they could have easily been decided by a coin toss, saving a lot of time, labour and expense.

The most practical feed trials for pork producers are those involving nursery or finishing pigs. In the breeding herd, feed trials are often not feasible to conduct because it is difficult to get adequate replication (i.e. to accurately detect 0.5 pigs per litter difference between two sow feeds you will need 388 sows per feed type).

There are two main considerations when conducting a meaningful feeding trial:
1) minimize differences in pig performance that could be caused by factors other than feed and,
2) provide a basis for concluding the results are statistically sound and valid for making a business decision.

So what is required to set up a statistically valid on-farm feed trial? Take the example of a producer who wants to look at two different grow-finish programs. One feed program is the control (currently in use), and the other is the test program. To be statistically valid the following items must be able to be measured and controlled:

  1. Weighing pigs and feed
  • It is essential to use a reliable set of scales to weigh pigs and feed. Estimating pig weight by sight and feed by volume gives an estimate but is unreliable.
  • Use the same set of scales for the entire trial and ensure they are calibrated before each use. A 25 kg bag of commercially prepared feed or premix can serve as a practical check weight.
  • At minimum pigs should be weighed at the start and end of the trial or ideally at the end of each phase.
  • Feed usage can be calculated in a number of different ways. At minimum the total amount of feed used during the trial should be recorded and any feed left over in feeders and feed bins subtracted from the total.
  1. Having adequate replications
  • Replication is important to minimize mistakes and to ensure correct conclusions are made. A replication means observing at least two pens of pigs per feed type.
  • In this case it means having four pens, with their own feeders, two for the control program and two for the test program. Suppose only two pens of pigs were available to conduct the feed trial. If the test program outperforms the control program can you conclude that the new feed is better? No, you cannot be certain that the difference in pig performance was due to feed. It could have been due to other factors, such as a malfunctioning water nipple or simply chance variation.
  • Table 1 provides guidelines for the number of pens per feed program needed to detect a difference between two feed programs with a 95% degree of confidence.
Table 1: Required Number of Pens Per Feed Program for Nursery and Grow-Finish Feed Trialsabc
Percent improvement in daily gain or feed/gain Number of pens/feed type

a Reese et al. 2000
b CV=7% and alpha = 0.05
c Applicable to >5 but <30 pigs/pen

  • Repeat the trial if you cannot get enough pens at one time for adequate replication. Ensure that the trial is set up the same way each time.
  1. Selecting Pen Location
  • To reduce variation all pens used in the trial need to be the same size and contain identical equipment (i.e. flooring, feeders, water nipples, etc.).
  • Trial pen location within the building should be randomized so location does not influence the outcome of the trial (i.e. provide a better/worse environment).
  • Use adjacent pens (with separate feeders) to create a block of pens where both the test and control program can be evaluated under the same conditions.
  1. Minimizing Pig Differences
  • The effects of variation in weight, genotype and sex should be equalized across pens.
  • Initial pig weight is considered the same between pens if the difference between control and test pig weights is less than 5% of the average weight of all pigs in the trial. For example, the average initial weights of the control and test pigs are 25.2 kg and 25.8 kg, respectively. The average weight of all pigs in the trial is 25.3 kg. Therefore, the difference in average initial weight of both groups is 2.4% of the average weight of all the pigs in the trial. If the difference had been greater than 5% then pigs would have to be reallocated to reduce the weight variation.
  • The number of barrows to gilts does not need to be the same in each pen. However the ratio of barrows to gilts and the total number of pigs must be the same in each pen.
  1. Starting with Suitable Animals
  • Start with healthy animals that have received the same vaccination and pre-trial treatments.
  • In the event a pig dies before the trial is completed, record its weight and the date it died so the feed and growth data can be adjusted.
  1. Controlling the Timing and Duration
  • Pigs fed the control and test program must begin the trial on the same day.
  • Determine where the end point of the trial will be. Ideally, for grow-finish pigs feed trials should be terminated when the pigs achieve a predetermined market weight.

Once the data has been collected it needs to be tabulated and analyzed statistically to help make valid conclusions. There are a number of programs, including Excel® that can simplify the analysis. Enlisting the assistance of your feed representative or nutrition consultant can also help to make the whole process easier.

On-farm trials involve a lot of time, effort and dedication but designed properly they help you make sound business decisions. As an added incentive, certain types of scientific research that you do to improve your business may qualify for the Canada Revenue Agency’s Scientific Research and Experimental Development (SR&ED) tax incentive ( making on-farm trials well worth the investment.


Reese, D.E and Stroup, W.W. 2000. Conducting Pig Feed Trials on the Farm. University of Nebraska Cooperative Extension.