Let’s talk statistics. Many people know that a p-value is used to determine statistical significance-- but what does that mean? A p-value is a statistical estimation that helps researchers come to a conclusion about their study results, but looking at the p-value alone may be misleading. Hypothesis testing includes 2 hypotheses: the null hypothesis, which assumes that both placebo and treatment group will have the same outcome, and the alternative hypothesis which proposes that there will be differences between the two groups. Researchers typically hope that the data will allow them to reject the null hypothesis, and in doing so they must account for multiple inputs: ➡️P-value, which is a number that tells us how unusual our results are if we assume that the null hypothesis is true; ➡️Significance level: a threshold that is set BEFORE data are collected and cannot be manipulated to fit desired results; ➡️Sample size: directly influences p-value. Note: you can obtain a statistically significant p-value using either a large or small sample size, but if the sample is too small, the results might not be representative of the entire population even if they appear significant. ➡️Effect size: shows the magnitude of the differences between groups. Sometimes a very small difference is clinically meaningful but sometimes an observed statistical significance may not equal clinical significance. ➡️Statistical power: shows the probability that the statistically significant results are not obtained by chance alone. ➡️Underlying assumptions are not only statistical, but also clinically meaningful. For example, whether the study population matches the disease profile and whether the measured outcome is arbitrary or is it clinically meaningful. These factors must be considered when appraising a study. Per Ronald Fisher, the statistician who introduced p-values: even if results are deemed statistically significant, the study must be repeated to see if this significance persists prior to making any policy changes, in order to reduce the likelihood that observed findings were due to chance. There's lots more we want to cover on biostatistics and data science, so stay tuned!