STATISTICAL ANALYSIS OF SOME PHYSIOLOGICAL EXPERIMENTS

Journal/Book: Reprinted from Sankhyä : The Indian Journal of Statistics Vol. 10 Parts 1 & 2 1950. 1950;

Abstract: By ARTHUR LINDER Statistical Laboratory University of Geneva and Swiss Federal Institute of Technology Zurich AND ETIENNE GRANDJEAN Institute of Physiology University of Lausanne Paper received : 25 August 1949 1. INTRODUCTION In the course of investigations into the physiological effects of moderate and high altitudes a certain number of experiments were carried out by several physiologists in Switzerland. Measurements of physiological reactions were taken both at lower altitudes - e.g. at Lausanne 550 m (1800 feet) above sea-level-and at moderate altitudes - Zuoz 1750 m (5750 feet) -as well as at high altitudes - Jungfraujoch 3450 m (11300 feet). In order to feet the effects of change in altitude the ordinary analysis of variance seems not appropriate owing to the fact that these effects are often showing considerable time-lag. For this reason the authors jointly developed a method of handling such date which may be compared to the well-known control-chart technique due to W. Shewhart (Grandjean and Linder : 1947) This method was first applied to experimental date collected several months before the analysis was performed ; later an it was found useful to draw the control-chart during the course of the experiments. 2. STATISTICAL ANALYSIS A typical experiment the analysis of which will be given in some detail was as follows. Measurements of the thresholds of patellar reflex were taken for 10 days at Lausanne and for 12 days at Zuoz an 8 individuals. Every day 5 measurements were taken consecutively at short intervals. It is now well known that the logarithms of the threshold measurements are normally distributed ; for this reason the apparatus was provided with a scale an which the log of the thresholds could be measured directly. Measurements were taken with a precision of 5 hundredths. In order to simplify calculations measurements were converted by the formula (measurement-2.00)/0.05. In this way a log threshold of 2.85 yielded a converted value of 17. These values are given in Table 1. (without Table) ... ___MH

Zurück | Weiter