Kalle Hauss

Statistical Inference from Non-Random Samples. Problems in Application and Possible Solutions in Evaluation Research

Shortlink: https://www.waxmann.com/artikelART102531

Buy article

Abstract

Inferential statistical methods have been well-established in evaluation research and practice for many decades. While their use has been controversially discussed in some areas of science, this discourse has remained largely unnoticed in evaluation research. The paper discusses methodological limitations of making statistical inferences from non-random samples and complete sets of data. Based on a content analysis of 355 journal articles and evaluation reports the paper asks how often inferential statistics is applied and what types of samples are used to make inferences. The aim of the paper is to raise the awareness for restrictions in inferential statistical methods and to contribute to a reflected application in the evaluation. To this end, approaches to valid inferences from non-random samples are discussed.

Keywords
Inferential Statistics, Content Analysis, Non-Random Samples, Significance Testing