Sambandsanalyser Enkla statistiska metoder och exempel

Abstract

This publication is intended to be of help for students writing minor theses, and their mentors. It may also be a part of course litterature in basic statistical methods. Examples are made with the program IBM-SPSS. Introduction to commands for handling SPSS-files is given.Abductive reasoning in order to choose scientific problems is taken up, further to formulate them as research questions, and start building a problem frame for an investigation. Ways to combine qualitative and quantitative methods are discussed. Program theory is illuminated together with cognitive and causal mapping. Operationalising concepts into measures are exemplified in some detail. Validity and reliability are taken up, and also methodological artefacts like regression to the mean. Sampling, and especially sampling for case studies, is discussed. So are norms for how many individuals or cases that should be sampled, and also how to do with missing data. The concept of normality is taken up together with skewness, kurtosis and outliers. Alternatives to the normal distribution are briefly discussed. The debate about significance testing and adequate effect size s is commented, together with group and individual measures on reliable and clinical change. The robustness of regression is discussed, including bootstrapping. Methods like cross-tables and mean comparisons with dependent and independent data are given a specific chapter for about twenty different methods of parametric and non-parametric techniques. Simple regression and different correlation methods have a certain chapter as well as multiple regression (including polynomial regression) and path analysis. Logistic regression in binary, multinomial and ordinal variants are taken up. So are analysis of variance and covariance. In a final chapter three pedagogical trials made by teachers are reported on, together with t-tests, multiple regression, mixed ANOVA, and contrast analysis used.