ABSTRACT
Although when you read papers you might be struck by an overabundance of statistical terms, the truth is that what you actually need to know is thankfully a fairly small amount. In this chapter we cover:
1. Signicance 2. Hypothesis testing, p-values and condence intervals (CIs) 3. Error – or what if the researchers are wrong? 4. Statistical power and how many people are needed in a study 5. Looking out for error in studies 6. Statistical tests
We talk a lot in scientic studies about ‘signicance’ but it’s important to realize there are two types of signicance: statistical signicance and clinical signicance. Imagine a study looking at a new antihypertensive that showed a drop in blood pressure of 2 mmHg in the treatment group. We could perform a statistical analysis and might nd that the result is statistically significant. By this we mean that we think the result demonstrates a difference between the two groups that is unlikely to have occurred by chance. However, it would be reasonable for you to ask whether, in the real world, a drop of 2 mmHg in blood pressure was of any real value, and in doing so you would be questioning the clinical significance of the result.
