Data analysis and interpretation allow you to test your predictions and interpret your results. This is an exciting time and can be daunting because it’s a big change from data collection. It’s very unlikely that you will have collected exactly the data you set out to collect, but your analysis plan will keep you on track and avoid the dangers of aimlessly exploring your dataset. You will probably need further statistical advice at this stage. This chapter guides you through data preparation, initial data analysis, hypothesis testing, calculating your effect sizes and confidence intervals, interpreting your results and extrapolating from them.

We use statistical analyses to test our predictions using the measures we collect for our sample. Like all aspects of study design, we need to think carefully about our choice of analytical approach. Planning our data analysis in detail, before we collect your data, helps to determine what data we need to collect. It is very common to rush past the analysis plan and dive straight into collecting data. This is partly because statistics are not intuitive and can be intimidating. However, statistical analysis is an integral part of study design. We must understand statistics to understand the strengths, limitations, and potential biases of any research. This may seem daunting, but our understanding of statistics determines the quality of a study. The more we think about this now, the better our study will be. I begin this chapter with how to determine what sort of analyses we need and the need to consult a statistician when we design a study. Next, I cover problems associated with multiple testing and assessing multiple predictor variables. I explain how to prepare an analysis plan and suggest pre-registration.

Statistical evidence is fundamental to science. Understanding statistics helps us to understand the literature and assess it critically, refine our research questions into testable hypotheses and predictions, design studies that are appropriate to test these predictions, evaluate whether our findings support our predictions, and derive appropriate conclusions. The dominant paradigm in primatology and allied disciplines is to test whether patterns we observe in our observations are due to more than random variation in our data. However, the statistical analyses we use to do this are very often misinterpreted. In this chapter I distinguish different kinds of variables, then introduce relationships between variables. I explain how we use statistical analysis to infer something about a theoretical population based on a sample. I introduce null hypothesis significance testing and explain common misunderstandings of this approach. I review the two types of error that arise in NHST and the concept of statistical power. I explain the need to assess and report effect sizes and confidence intervals, briefly introduce alternatives to null hypothesis statistical testing and end with how to interpret statistical results appropriately.