Näkökohtia tilastotieteellisten tutkimusmenetelmien käytöstä maatalousekonomiassa

Authors

• Lauri Kettunen Maatalouden taloudellinen tutkimuslaitos

DOI: https://doi.org/10.23986/afsci.71657

Abstract

The article deals with the problems arising in the application of statistical research methods to agricultural economics. As the available statistics that are often used in research work do not meet all the requirements set for statistical inference the results of the research cannot usually be generalized. The article calls attention to the logical content of function models in the use of regression analysis where the first difficulty lies in determining the causal ordering of the variables and in selecting the primary variables of the function to be estimated. Secondly, the interpretation of the results is difficult either because the ceteris paribus clause is not valid or because, though it may be valid theoretically, the correlation between the observed values of independent variables makes the estimates of parameters unreliable. Finally the writer of the article presents some views on the testing of hypotheses and calls attention to the fact that although statistical tests show differences that are significant, the importance of the differences, for instance in regard to their economic and other consequences, depends on the subjective judgement of the scholar.