Glossary- We include selected terms and will build on this 

(Please ask for more if you can’t find what you want here)

Coefficient of Variation (CV) – Reported as the ratio of the standard deviation over the mean  and generally reported as a percentage. A CV reported for an experiment gives a measure of overall variability to compare different variables or different experiments.  It is dimensionless.

Factor – In a one factor experiment we may be looking to compare two breeds of goat. In this case  the Factor is ‘breed of goat’. The two breeds are the  two levels we have chosen to compare. We may be interested in comparing three different fertilizer types and at 4 different rates, this would have two Factors namely  ‘Fertilizer Type’ (A,B,C) and ‘Fertilizer Rate’ (0, 20kg, 40kg and 80 kg per ha). The factor is a categorical variable that is used to denote part of the design of the study. the other aspect of design is the replication and any restriction on randomisation such as blocking.

Level – within a factor it is the named particular treatments  we are looking at. In a fertilizer experiment we may have three rates of fertilizer. In an irrigation experiment  we may have four levels (no irrigation, early, late  and fully irrigated).

Parameter – A quantity such as a mean, total  or variance of the whole population. We use sampling theory to design studies that allow us to estimate the true mean. We calculate statistics such as means, standard deviations and confidence intervals from our experiments. We cannot usually get the actual parameters for  a population as it would cost too much or it is a theoretical concept.

Population – The whole either real population (such as all dairy cattle in a region) or a theoretical population. We may consider the population of grain production from one variety –  we would grow the plants to be representative of all that particular variety. We can have totals for a population. The total population size is often denoted N, and we use n for denoting a sample from the population.

Statistic – A value calculated from a sample; it provides an ‘estimate’ of the population parameter. Each time a sample of a known size is taken it will give a slightly different mean. That is why we need a well designed sampling scheme in a survey  (or a designed experimental study) to ensure there is no bias in our sample statistics.

Treatment – within an experiment this is the reason you are looking at measuring  differences. By applying a treatment the scientist is deliberately changing the biological, physical, or management of  some experimental units in the study. We may have two treatments looking at two varieties of crops. We may be interested in comparing three different fertilizer types and at 4 different rates.  Sowing date, plant population and timing of fertilizer are all examples of different management treatments.  In a factorial experiment we may have two factors and then we can consider the number of treatment combinations.  We try to keep all other aspects of a study similar and as constant as possible – hence the name of  experimentation  is also known as a controlled study. As we keep other aspects constant we are only comparing the treatments of interest.


Leave a Reply