4.7.5 Analysing data The proposal should indicate that you will use adequate experimental design, sampling methods and replicate your trials where possible. Statistics is a discipline within its own right and there are numerous good statistics books and reference to guide you. It is essential BEFORE preceding an experiment to ensure your design is robust i.e. that it captures the hypotheses that you are aiming to address. It is wise to consult a statistician or biometrician to verify your design when subject to analyses will be valid. A common mistake made by inexperienced researchers is the failure to decide in advance how they will analyse the data that they will collect. The types of analysis
57 possible will be constrained by data you have obtained. The data you have will be determined by the research design and by the types of measurement you have chosen. Theory based assessment the internal logic of your research question will also tell you what sort of data you need (e.g. continuous and discontinuous events. Using the correct analytical method is vital in making the most of your data. Incorrect analysis can lead to the wrong conclusions being made and you may miss important and exciting findings. Check your data (Normality of data) Analyse your data using statistical packages. For example • Detecting differences between samples (ANOVA; multivariate analysis • Representing differences between samples (PCA; cluster analysis • Detecting a trend (linear regression establish correlations (Pearson.