Data, not only in aerobiology: how normal is the normal distribution?

Autor: Jonathan West, Carmen Galán, Eckhard Limpert, María del Mar Trigo, Werner A. Stahel, James Burke
Rok vydání: 2008
Předmět:
Zdroj: Aerobiologia. 24:121-124
ISSN: 1573-3025
0393-5965
DOI: 10.1007/s10453-008-9092-4
Popis: At the 8th International Congress on Aerobiology (8th ICA) an ad hoc ‘‘Working Group Data Analysis’’ formed as a result of the concern about data analysis and the present use of the normal distribution. The Group was asked to consider these points further. Here we present first results and a recommendation to use, in general, other distributions than the normal since data are more adequately described by them. Frequencies of allergens and micro-organisms in the air as well as their effects on humans and the environment are of major concern. To characterize and summarize the data, several distributions have been used, among which the normal distribution is most popular. In fact, this model has become the standard of quantitative variation across the sciences. Over the recent past, considerable doubts have been raised on the suitability of this probability density function in aerobiology, and other methods have been applied for that purpose (Moseholm et al. 1987; Comtois 2000; Galan et al. 2000; Trigo et al. 2000; Angosto et al. 2005; Sanchez-Mesa et al. 2002). Nevertheless, quantitative variation of original data quite often continues to be thought to be symmetrical and normally distributed. Accordingly, data are again and again summarized, using the arithmetic mean and the standard deviation, by x ± s to indicate an interval which is meant to characterize the statistical variation. However, frequently, the mere summary data immediately demonstrate that the normal distribution would not be a good fit, as in the case of 500 ± 500 further explained below. The description gives a wrong impression of the data and their distribution. Moreover, the sign ± points to additive effects. By contrast, the processes that affect the variability of the number of air spora are usually multiplicative, as are natural laws in general (Limpert et al. 2001; Limpert and Stahel 2006). E. Limpert (&) ELI-O-Research, Scheuchzerstr. 210, 8057 Zurich, Switzerland e-mail: eckhard.limpert@bluewin.ch
Databáze: OpenAIRE
Externí odkaz: