Power Analysis: main sources of statistical error, sample size, effect size and power of a test

Data: 09, 16 May 2022
Luogo: Meeting Hall - Psychology - Porta di Massa road 1, Napoli, Italy

This course has been taught in Italian.

Invited professor at the PhD School “Mind, Gender and Language” of the Department of Humanities as part of the “Statistical Methodologies in Psychosocial Research” lectures for PhD students.

Outline of the course

Any measurement entails a source of uncertainty that cannot be eliminated and can significantly affect the results of a statistical analysis. Every time a survey is carried out, several decisions must be made: among them, the choice of sample size plays a crucial role.

The power of a statistical test allows to control the probability of making incorrect decisions, to assess the reliability of a study and to determine the (ideal) sample size required to obtain statistically significant results. In this context, power analysis is generally used at the beginning of a research project to determine a proper sample size.

This two-week course is divided in two parts. The first lecture will focus on the main concepts concerning the sampling theory (sample collection and selection), with special emphasis on the main sources of sampling error, while the second part of the seminar will focus on statistical test theory and the study of the power of a test. The topics covered will be accompanied by various applications and, in addition, G*Power software will be introduced as a tool for performing power analysis.

Main requirements

Participants are expected to have basic statistical knowledge but no specialized knowledge is required. It is recommended that participants have a PC with the G*Power software previously installed in order to carry out the exercises. In conclusion, some R software packages (especially pwr and simglm) used to apply Power analysis in different contexts will also be introduced for demonstration purposes.

Syllabus