This page is an edited version of the section on presentation of results in Festing and Altman, 2002. Guidelines for the design and statistical analysis of experiments using laboratory animals. ILAR Journal 43:233-243.
Summary statistics such as means, proportions and standard deviations
Graphical presentation of data
Specification of the animals, their husbandry and health status
Three C57BL/6 mice heterozygous for the viable yellow gene (Avy). These mice are all genetically identical. The variation in coat colour is not inherited, and is due to non-genetic effects.
Summary statistics such as means, proportions and standard deviations
- Where individual means are quoted excess decimal places, often produced by the computer, should be eliminated. Usually means can be quoted to three significant digits. Percentages can often be given with a single decimal place or even rounded to whole numbers.
- The standard deviation should be given to describe the variation among individuals which contribute to the mean. Avoid using the ± sign. It is better to use a designation such as “9.6 (SD 2.1)” because it avoids any confusion between standard deviation and standard error.
- A confidence interval is used to show the precision of the mean. (e.g. Mean 9.6, 95% CI =7.2-12.0). Alternatively a standard error (e.g. Mean 9.6 (SE 1.2)) can be used.
- Actual observed p-values should be quoted wherever possible, rather than using < > signs, though if the p-values are very low a < sign can be used.
- Where two means are being compared, the difference between them should be quoted together with a confidence interval.
- Lack of statistical significance does not mean that an effect does not exist. Non-significance may be due to the experiment being too small or the experimental material being too variable.
- When non-parametric analyses have been done, it is more sensible to quote medians and centiles, such as the inter-quartile range .
- Where proportions or percentages are given, a standard error or confidence interval and “n” should also be given. When proportions are compared the confidence interval for the difference (or ratio) should be supplied.
- Tabulated means should be shown in columns rather than rows as this arrangement makes it easier to compare values.
- If the means have been compared using a t-test or ANOVA and the standard deviations have been found not to differ materially between groups, use of a pooled standard deviation may be more appropriate than showing the standard deviations separately for each mean.
The number of observations should always be indicated in graphs and tables.
Graphical presentation of data
- Graphs should be used to illustrate points which would be difficult to explain in writing. Presentation of a small number of means can often be done more clearly and using less space using a table rather than a bar diagram. This also means that the reader needing numerical values does not have to read them off a graph.
- Graphs showing individual points rather than bar charts or graphs with error bars are strongly encouraged as these give a much clearer impression of the nature of the data.
- Where means have been compared statistically it may be better to indicate significant differences on the diagram, rather than putting in error bars. Where error bars are shown on graphs or bar diagrams, there should be a clear indication of whether these are standard deviations, standard errors or confidence intervals (preferred), and the number of observations (for each point) should be clearly indicated in the text or figure caption.
- With more complex graphs it may be better not to use error bars, but to summarise the data in an accompanying table.
- Regression lines should never be shown without the data points, and preferably should be shown with a confidence interval and/or prediction interval.
Specification of the animals.
This section is based largely on the paper: Festing, M. F. W. and van Zutphen, L. F. M. 1997. Guidelines for reviewing manuscripts on studies involving live animals. Synopsis of the workshop. In van Zutphen, L. F. M., ed., Animal Alternatives, Welfare and Ethics, pp. 405-410. Elsevier Science B.V., Amsterdam, Lausanne, New York.
The following should be recorded:
- Source, species (with Latin name if not a common laboratory species), conservation status if wild, age and/or body weight, sex.
- Transportation: Length of acclimatisation period
- Genotype: The breed, strain or stock name. Inbred strains, mutants, transgenes and clones should be described using internationally accepted nomenclature where this is available. Brief rules are given in the companion web site. Official rules are maintained by the Jackson Laboratory Any genetic quality assurance verifying the genotype should be mentioned.
- Microbiological status: Whether conventional, specified pathogen free (SPF), germfree/gnotobiotic. Where possible reference should be made to some agreed standards for microbiological characterisation such as the FELASA standards
- Environment: Type of housing including whether conventional, barrier, isolator, individually ventilated cages. Room temperature (with diurnal variation), humidity, ventilation, light/dark periods, light intensity. Cage type, model, material, type of floor (solid/mesh), type of bedding, frequency of cage cleaning, number of animals per cage, cage enrichments.
- Diet: Type, composition, manufacturer, feeding regimen (ad-libitum, restricted, pair fed), method of sterilisation.
- Water: ad-libitum, bottles or automatic, quality, sterilisation.