DIGITALNA ARHIVA ŠUMARSKOG LISTA
prilagođeno pretraživanje po punom tekstu
ŠUMARSKI LIST 9-10/1997 str. 47 <-- 47 --> PDF |
V. Hitrec : STOHASTIKA U ZNANSTVENIM ISTRAŽIVANJIMA Šumarski list br. 9-10, CXXI (1997), 499-505 Uvijek se mora imati na umu da se modelira pojava a ne skup točaka. 4. Ova bih razmatranja završio jednom misli i jednim upozorenjem. 4.1. Ne upotrebljavajmo složene modele jer su nesigurni, nepraktični, netočni i skupi. 4.2. Ne oslanjajmo se na gotove statističke pakete u računalima, ako nam nisu poznati temelji statističkoga (stohastičkoga) načina razmišljanja. 4.3. Warren je svoj, ovdje nekoliko puta citiran rad nazvao: Statistika: razum ili ritual. Mislim da nam je to odlična pouka. Nemojmo koristiti statistiku ritualno, već Cum grano salis. (hrv. Sa zrnom soli odnosno S malo mozga) LITERATURA - References Host, E. G., 1988: Causal Models, Empirical Data, Tomanić ,S., V. Hitrec & V. Vondra: Sistem and Brief Philosophy of Science. IUFRO S6.02 određivanja radnog vremena sječe i izrade drva. Newsletter No. 16, str. Monografija, Liber, Zagreb, 1978, str. 1-443. Hitrec , V. ,1989: Deterministic, Stohastic, ExplanaWarren , W. G., 1988: Empirical and Explanatory motory and Empirical Models. IUFRO Newsletter dels: A comment IUFRO S6.o2 Newsletter No. S6.02,Nol8,str. 1, str. McRoberts , E. M., 1988: In Defens of Empirical Warren , W. G., 1988: On the presentation of StatistiModels. IUFRO S6.02 No. 16, str. cal Analysis: Reason or Ritual. Can. J. For. Res. Oderwald , G. R. ,1988: More "Comparing Empiri16.( 1185-1191) cal and Explanatory Models". IUFRO S&.02 Warren , W. G. ,1988: Star Games. IUFRO S6.02 Newsletter No .12, str. Newsletter No. 17, str. Rauscher, H. M. 1987: Comparing Empirical and Wannacott, T.&R. Wannacott, 1990: IntroductoExplanatory Models. IUFRO Newsletter No. ? ry Statistics for Buisness and Economics, John Rauscher , H. M. 1988: Comparing Empirical and Wiley and Sons pp. Explanatory Models Revisited. IUFRO Newsletter S6.02, No 16, str. SUMMARY: Scientific research in the field of forestry is mainly stochastics, requiring the use od stochastic models, i.e. the method of inferential statistics. Hardly any theories are available, though in some fields they occured and were studied (Raucher 1980). The paper deals with the problem encountered by the researchers dealing with living organisms, particularly with their activities. Such work (applicative) is compared with the so coled fundamental research resulting in theories. Stochastics, deterministics, empirical, i.e. explanatora models are accordingly discussed. This part of the studies was aroused by Raucher (1987) and a discussion within the IUFRO Group S6.02. The second part of the paper is a short mommentary on the misinterpretations about statistics as an aid in making decisions. Statistics is frequently regarded as indispensable, overpowering, objectiv and accurate. In many cases the results yielding obvious conclusions are encumbering with unnecessary statistical "evidence". It sometimes happens that an evident hypothesis is rejected if refused by statistical test. It is frequently forgotten that the inferencial statistics is based on a great number of conditions which have not always been fulfilled. On the other hand, the conditions are sometimes aproached too rigorously. The autor thinks that the use of statistics by being aware that "we are not doing quite correctly " is less hazardous than the use of the correct procedure which we do not understand (here is the emphasis on the problems of statistical packages). The autor points out the idea (Warren 1986) that our conclusions are always subjective. Another misunderstanding is the opinion that statistics can prove opinion and hypotheses. The third par illustrates inadequate use of mathematical statistics with example which are in the authors opinon still presentin literature, though quite some time has passed since the first warnings known to the author (Utts 1986, Warren 1987). Significantly, with computers playing a major part in it, the same analysis of the same data will not always give the same results. |