Elsevier

Experimental Gerontology

Volume 32, Issues 1–2, January–April 1997, Pages 39-47
Experimental Gerontology

Experimental design and husbandry

https://doi.org/10.1016/S0531-5565(96)00032-0Get rights and content

Abstract

Rodent gerontology experiments should be carefully designed and correctly analyzed so as to provide the maximum amount of information for the minimum amount of work. There are five criteria for a “good” experimental design. These are applicable both to in vivo and in vitro experiments: (1) The experiment should be unbiased so that it is possible to make a true comparison between treatment groups in the knowledge that no one group has a more favorable “environment”. (2) The experiment should have high precision so that if there is a true treatment effect there will be a good chance of detecting it. This is obtained by selecting uniform material such as isogenic strains, which are free of pathogenic microorganisms, and by using randomized block experimental designs. It can also be increased by increasing the number of observations. However, increasing the size of the experiment beyond a certain point will only marginally increase precision. (3) The experiment should have a wide range of applicability so it should be designed to explore the sensitivity of the observed experimental treatment effect to other variables such as the strain, sex, diet, husbandry, and age of the animals. With in vitro data, variables such as media composition and incubation times may also be important. The importance of such variables can often be evaluated efficiently using “factorial” experimental designs, without any substantial increase in the overall number of animals. (4) The experiment should be simple so that there is little chance of groups becoming muddled. Generally, formal experimental designs that are planned before the work starts should be used. (5) The experiment should provide the ability to calculate uncertainty. In other words, it should be capable of being statistically analyzed so that the level of confidence in the results can be quantified.

References (20)

  • K.E. Muller et al.

    Increasing scientific power with statistical power

    Neurotoxicol. Teratol.

    (1992)
  • D.G. Altman

    Practical Statistics for Medical Research

  • V.A. Benignus et al.

    Bad statistics are worse than none (letter)

    Neurotoxicology

    (1982)
  • J.D. Biggers et al.

    Variance control in the animal house

    Nature

    (1958)
  • J. Cohen

    Statistical Power Analysis for the Behavioral Sciences

    (1988)
  • D.R. Cox

    Planning Experiments

    (1958)
  • G.J. Eaton

    Intestinal helminths in the mouse

    Lab. Anim. Sci.

    (1972)
  • M.F.W. Festing

    Phenotypic variability of inbred and outbred mice

    Nature

    (1976)
  • M.F.W. Festing

    The scope for improving the design of laboratory animal experiments

    Lab. Anim.

    (1992)
  • M.F.W. Festing

    Reduction of animal use: Experimental design and quality of experiments

    Lab. Anim.

    (1994)
There are more references available in the full text version of this article.
View full text