Abstract
Statistical methodology plays a critical role in the progression of our scientific understanding by providing a framework which enables hypotheses to be operationalized and tested. This chapter provides an overview spanning study design to reporting results and addresses some of the common advantages and challenges associated with data analysis. Specifically, we cover common methodological misunderstandings, dealing with confounding and interactions, multiple testing, meta-analysis, measurement error, diagnostics and missing data. Approaches for analysing longitudinal data are discussed, including structural equation modelling techniques such as growth curve and trajectory models. In addition, causal modelling approaches, including propensity scores and the use of instrumental variables, are presented. The importance of non-technical factors and of the biases that can be introduced by undertaking science in the real-world are highlighted.
Original language | English |
---|---|
Title of host publication | Rutter's Child and Adolescent Psychiatry: Sixth Edition |
Publisher | John Wiley & Sons |
Pages | 188-200 |
Number of pages | 13 |
ISBN (Print) | 9781118381953 |
DOIs | |
Publication status | Published - 10 Jul 2015 |
Keywords
- Causal analysis
- Confounding
- Growth curve
- Longitudinal data analysis
- Mediation
- Meta-analysis
- Missing data
- Moderation
- Propensity scores
- Regression
- Sensitivity
- Specificity
- Structural equation modelling
- Trajectory models