M Maaike Labots1, M C Marijke Laarakker2, D Dustin Schetters3, S S Saskia Arndt4, H A Hein van Lith5. 1. Department of Animals in Science and Society, Division of Animal Welfare & Laboratory Animal Science, Faculty of Veterinary Medicine, Utrecht University, Utrecht, The Netherlands; Brain Center Rudolf Magnus, University Medical Center Utrecht, Utrecht, The Netherlands. Electronic address: M.Labots@uu.nl. 2. Department of Animals in Science and Society, Division of Animal Welfare & Laboratory Animal Science, Faculty of Veterinary Medicine, Utrecht University, Utrecht, The Netherlands; Brain Center Rudolf Magnus, University Medical Center Utrecht, Utrecht, The Netherlands. Electronic address: Marijke.Laarakker@gmail.com. 3. Department of Animals in Science and Society, Division of Animal Welfare & Laboratory Animal Science, Faculty of Veterinary Medicine, Utrecht University, Utrecht, The Netherlands. Electronic address: D.Schetters@vumc.nl. 4. Department of Animals in Science and Society, Division of Animal Welfare & Laboratory Animal Science, Faculty of Veterinary Medicine, Utrecht University, Utrecht, The Netherlands; Brain Center Rudolf Magnus, University Medical Center Utrecht, Utrecht, The Netherlands. Electronic address: S.S.Arndt@uu.nl. 5. Department of Animals in Science and Society, Division of Animal Welfare & Laboratory Animal Science, Faculty of Veterinary Medicine, Utrecht University, Utrecht, The Netherlands; Brain Center Rudolf Magnus, University Medical Center Utrecht, Utrecht, The Netherlands. Electronic address: H.A.vanLith@uu.nl.
Abstract
BACKGROUND: Guilloux et al. introduced: integrated behavioral z-scoring, a method for behavioral phenotyping of mice. Using this method multiple ethological variables can be combined to show an overall description of a certain behavioral dimension or motivational system. However, a problem may occur when the control group used for the calculation has a standard deviation of zero or when no control group is present to act as a reference group. NEW METHOD: In order to solve these problems, an improved procedure is suggested: taking the pooled data as reference. For this purpose a behavioral study with male mice from three inbred strains was carried out. The integrated behavioral z-scoring methodology was applied, thereby taking five different reference group options. The outcome regarding statistical significance and practical importance was compared. RESULTS: Significant effects and effect sizes were influenced by the choice of the reference group. In some cases it was impossible to use a certain population and condition, because one or more behavioral variables in question had a standard deviation of zero. Based on the improved method, male mice from the three inbred strains differed regarding activity and anxiety. COMPARISON WITH EXISTING METHOD: Taking the method described by Guilloux et al. as basis, the present procedure improved the generalizability to all types of experimental designs in animal behavioral research. CONCLUSIONS: To solve the aforementioned problems and to avoid getting the diagnosis of data manipulation, the pooled data (combining the data from all experimental groups in a study) as reference option is recommended.
BACKGROUND: Guilloux et al. introduced: integrated behavioral z-scoring, a method for behavioral phenotyping of mice. Using this method multiple ethological variables can be combined to show an overall description of a certain behavioral dimension or motivational system. However, a problem may occur when the control group used for the calculation has a standard deviation of zero or when no control group is present to act as a reference group. NEW METHOD: In order to solve these problems, an improved procedure is suggested: taking the pooled data as reference. For this purpose a behavioral study with male mice from three inbred strains was carried out. The integrated behavioral z-scoring methodology was applied, thereby taking five different reference group options. The outcome regarding statistical significance and practical importance was compared. RESULTS: Significant effects and effect sizes were influenced by the choice of the reference group. In some cases it was impossible to use a certain population and condition, because one or more behavioral variables in question had a standard deviation of zero. Based on the improved method, male mice from the three inbred strains differed regarding activity and anxiety. COMPARISON WITH EXISTING METHOD: Taking the method described by Guilloux et al. as basis, the present procedure improved the generalizability to all types of experimental designs in animal behavioral research. CONCLUSIONS: To solve the aforementioned problems and to avoid getting the diagnosis of data manipulation, the pooled data (combining the data from all experimental groups in a study) as reference option is recommended.
Authors: Mathias V Schmidt; Jan M Deussing; Iven-Alex von Mücke-Heim; Lidia Urbina-Treviño; Joeri Bordes; Clemens Ries Journal: Mol Psychiatry Date: 2022-09-14 Impact factor: 13.437