 10 Statistical Mistakes in 5 Minutes This will be a brief visual summary of the ELIFF article, 10 common statistical mistakes to watch out for when writing or reviewing a manuscript. The first mistake is failure to include an appropriate comparison against a control condition or control group when determining the effectiveness of any intervention. The second mistake is making indirect comparisons between groups or conditions. We cannot simply make inferences on the basis that one group or condition has a statistically significant effect and another does not. If we want to compare those two groups or conditions, then we need to make sure that we directly compare their effects. Mistake number three is inferences based on spurious correlations. One example of where this might occur is through the presence of outliers distorting the overall relationship. Another way in which spurious correlations can occur is by pulling the data from two or more groups together, which brings us on to mistake number four, which is combining within and between participant or group analyses. Not only can combining within and between group analyses result in spurious correlations, but it can also falsely inflate the degrees of freedom leading to an increased likelihood of false positive results. Mistake number five is to treat a correlation or association between two variables as a cause and effect relationship. This should only be treated as causal when it is due to a manipulation or intervention, and even then we should be cautious about the role of third variables or confounding factors. Number six is the use of small sample sizes in underpowered studies. The smaller the sample size, the larger the false positive effect sizes observed on average. So just because you've observed an effect that is large in magnitude, that doesn't mean you can be confident that it's a true effect within your underpowered study. For more on sample size calculations, you can check out my previous video giving a five minute demonstration into sample size calculations in G-Power. Mistake seven is to conduct a circular analysis in which a hypothesis is tested on the same data set that led to the generation of the hypothesis in the first place. Number eight is p-hacking. This is where investigators can manipulate the analysis procedures in search for significant result. This could be something as simple as removing individual data points or trying out multiple tests or dependent variables until they get the significant result that they're looking for. Whilst it's difficult to identify after the fact, we can start to address this by avoiding changing hypotheses once the study has begun and ideally pre-registering the analysis plan beforehand. Mistake nine is failing to correct for the fact that multiple comparisons were performed within the study. When we have a large number of variables or statistical tests, we know that even if the null hypothesis is true, then a small number of those tests will achieve significant results simply by chance and so that's why there's a need for a correction to be performed. Finally, absence of evidence is not evidence of absence. Just because we failed to reject the null hypothesis by finding a significant result, that does not mean we are accepting that the null hypothesis is definitely true. For more on that and the use of effect sizes and confidence intervals to make inferences, you can check out our editorial in Sports Biomechanics on recommendations for statistical analysis involving null hypothesis significance testing. For a full lecture on various statistical concepts taught and demonstrated through simulation, you can check out Kristen Sonani's lecture as part of the Sports Biomechanics lecture series. Thank you for watching. For more information, I recommend that you check out the original E-Life article for which I'll make sure the link is in the description below. This includes examples, detailed explanations, how to detect it, and solutions for researchers. All of the other articles and videos mentioned within this video will also be linked in the description. To keep up to date with any more similar videos, then please click subscribe and click on the bell next to it to receive notifications. Thank you very much.