Choosing an Inferential Test

Statistics for Psychologists – Inferential Statistics

In this course, Dr Alison Wadeley (Bath Spa University) explores the challenging subject of inferential statistics. In the first lecture, we think about the decision tree required to establish which non-parametric test should be used, based on the experimental design and the data being collected. In the second lecture, we think about four example research questions and work through the process of deciding which non-parametric test is appropriate for each, before moving on to parametric tests. In the third lecture, we think about how the sign test is used to assess a research question on sleep quality and social media use. Next, we think about probability, critical values, and significance; helping to unpack the inner workings of these tests. In the fifth and final lecture, we think about the errors that can occur when interpreting results from these tests, how you might recognise them, and what can be done to reduce their prevalence.

Choosing an Inferential Test

In this lecture, we think about how to decide which statistical test to use, focusing in particular on: (i) some key assumptions of knowledge, so make sure you’ve brushed up on those before continuing (see below); (ii) utilising a visual decision tree to work through how to decide on which of the four main non-parametric test groups you should use to assess whether there is a difference between two reported data sets; (iii) how, in experimental design, there is an independent design which requires a Mann Whitney test and a related design which requires either a Wilcoxon or Sign test; (iv) how, in a non-experimental design, there is a correlation analysis which requires a Spearman’s Rho and a Chi-Square analysis which requires a Chi-Square test.

Assumed knowledge:

Null and alternative hypotheses – See ‘Statistics for Psychologists – Data and the T-test’ by Dr Andrew Bell.

Levels of measurement: nominal ordinal, interval, and ratio – See ‘Statistics for Psychologists – Measurement Levels’ by Professor Dominic Dwyer.

Experimental design: independent and related – See ‘Research Methods – Building an Experiment’ by Dr Eoin O’Sullivan.

Correlational and Chi-Square analysis – See ‘Research Methods – Data Presentation’ and ‘Research Methods – Probability’ by Dr Ayoub Bouguettaya.