Conducting on-farm trials

Ever wanted to know if one method is better than another? To really determine this for your production system, a trial can help. But beware, if the trial is not carried out correctly the result could be just a chance result and not a valid comparison; it could also waste your time, effort and cost you money.

If considering change or a 'research' trial, follow these important steps:

  • work out the aim of the trial
  • design the trial without influence from other factors
  • conduct the trial and gain the right conclusions from data.


Ask yourself these questions before making a start:

  • What are you trying to find out? Test preferably only one or maybe two effects at a time. The more complex a trial is, the larger the sample needs to be and the interpretation is more difficult.
  • What are the possible outcomes? Explore existing knowledge, such as product and technical information and industry magazines.
  • Can AUSPIG or another tool help without doing a trial?
  • Do you have the resources and time available to carry out a trial?
  • Will you be able to implement the result?
  • What sort of difference will make it worthwhile in dollar return? A difference that is less significant statistically may be worth trying if it is not costly to implement, and observe what happens.
  • How will the results be analysed? Data needs to be collected on associated factors, e.g. the breed, to check if this influences the result.
  • Decide on the duration and timing of events. For example, if your primary measurement is growth rate, decide on start and finish dates.


Aim to avoid the effect of other factors that could influence results. This means all pens and their animals are treated the same with identical conditions except for the test treatment e.g. the same climate, stocking density, feeders and position of feeders. This can be a minefield for the inexperienced and all but the simplest comparisons will benefit from specialist advice.

Animal ethics responsibilities

The Animal Care and Protection Act 2001 places a duty of care on those in charge of animals and on the use of animals for scientific purposes. Trials such as feeding an animal differently to normal for a scientific purpose requires Animal Ethics approval. Before any trial starts:

  • You (or the institution where you are employed or where you study) must be registered with us as using animals for scientific purposes. Registration is valid for three years and has an associated fee.
  • You must submit a written application for your proposed animal use to an animal ethics committee. You need a separate application for each different proposed use.

A key consideration for the animal ethics committee is that the use of animals is justified, after balancing the benefits from this use against the potential negative effects on the animals' welfare. Efforts must always be made to find alternatives to using animals. Where this is not possible, the least number of animals necessary should be used to achieve the objective, and the impact on the animals minimised.

Number required

The number of animals (often pens, not individuals) needed in an on-piggery trial varies according to what is being measured and the level of precision required. The smaller the difference you are looking for, the more difficult to measure accurately or the more variable the factor, then the more pens (replicates) are needed to distinguish real differences from chance events. For example, more replicates are needed for a weaner feed comparison than for growers, as weaner performance is likely to be more variable, and a small difference is important.

A guide to the number of pens required for a grower trial (e.g. two different feeds and measuring growth rate and backfat) is to have four pens of 10 pigs for each treatment (one the 'test' feed, the other the control). This means eight pens and 80 pigs. An extra one or two replicates for each treatment (say, six pens per treatment equals 12 total) is useful insurance when analysing results as often some pigs may have been treated for disease or have died and these pens can substitute for the problem pens. The numbers needed for your particular situation can be calculated by a biometrician (research statistician).

Carrying out the trial

Randomly allocate treatments (including the control) to the pens, then, randomly allocate the pigs to the pens; one way is to weigh the pigs of similar weight (3-5 kg range is ideal), age, genetics, previous treatment and same sex) and sort the weights from heaviest to lightest. Then, for the example, with 12 pens of 10 pigs, rule a line after each 12, and randomly allocate one pig from each group of 12 to each pen. Good individual animal identification is essential.

A good idea is to colour-code the treatment and equipment (e.g. feed trolleys, not with the name of the treatment or numbers, which can be misread when worn), to avoid favouritism and to ensure that the correct treatments are given to the correct pens.

Check the equipment is working correctly and consistently before and during the trial; check scales are weighing correctly.

Provide equipment and resources for the trial to run smoothly to minimise mistakes and effort e.g. for an adlib feeding trial have the feed for a period already weighed into bins near the pens and weigh what is left at the end of that period.

Allow one to two weeks for pigs to get used to the treatments and to settle in their new housing before collecting data for the analysis. This helps to avoid initial problems such as feed refusals from any sudden changes.

Record all data in a book (not loose leaf) with appropriate ruled sections and a diary section for any additional observations and keep it in pig-proof container near the trial area. Keep a copy elsewhere.

Be uniform and ensure you work the same way. Keep the same people involved throughout the trial, so they know why tasks need to be done carefully and accurately. Prevent interference and keep going even if one treatment appears better than the other. Do not remove an animal halfway through the trial if it appears a lot larger than the rest of the pen. Record and weigh any pigs that die.

Working out the results

A biometrician can help you design a trial and interpret the results. Remember that no difference in the results may provide valuable information for decision making also.