The concept of power is central in statistics and pivotal for the correct interpretation of p-values. Power analysis leads to sample size estimation, a most important aspect of experimental design, statistical analysis and, beyond that, of grants application. Estimating the appropriate sample size is also and perhaps above all of ethical importance, for if the experiments involve living organisms, such as mice, not anticipating the correct numbers, may impact quite dramatically on the animal's welfare.

Proper understanding of descriptive statistics such as the mean or standard deviation is essential in statistical analyses as most tests rely on these metrics to be appropriate summaries of the data's behaviour. A good understanding of these basic principles is often all that it takes to understand most of quantitative statistics. Data exploration is a pivotal step in data analysis and a proper graphical exploration of your data often tells us all there is to know and conclude from the data and the relationships between the studied variables.

This short section presents/revisits 3 key concepts: null hypothesis, statistics inference and signal-to-noise ratio. The assumptions for parametric tests are also presented: a stastistical test will always produce a p-value but the extent to which that p-value is useful or valid depends on the data to behaviour in a suitable way for the test to work properly. It is basically about choosing the right tool for the right job.

The Student's t-test is one of the most widely used tests and a friendly one too. We want to know if there is a difference between 2 groups of mice, say WT and KO, the t-test is the one to go for. Well, providing, our data are nicely behaved and providing we take into account the design of our experiment. This video is going through the steps we should always follow before running a statistical, no matter how simple or complex it is. The steps are always the same. Producing a p-value by running a statistic test is dead easy, doing it properly so that we can trust the results, not so much.

Building on what is covered in the video on the Student's t-test, the one-way ANOVA is used when we want to compare more than 2 groups of quantitative values. So, we have one factor, sometimes referred to as a predictor, which has more the 2 levels (e.g. one control and 2 treatments).

In the previous sections, we covered the relationship between categorical (independent aka predictor) and quantitative (dependent aka outcome) variables. In this section, we look at the relationship between quantitative variables using correlation and formalise some of the measurements with linear regression.

Whilst linear relationships are common in data there are other types of pattern observed between variable. In this section we look at non-linear relationships through curve fitting. We cover different types of curves and how to both fit them to your data and check how well they fit.