USMLE Biostats 6: Null Hypothesis, Confidence Interval, P Value and more!

Want to support the channel? Be a patron at:   / lymed   Welcome to LY Med, where I go over everything you need to know for the USMLE STEP 1, with new videos every day. Follow along with First Aid, or with my notes which can be found here: https://www.dropbox.com/sh/an1j9swvjx... This is our last biostatistics video. We'll start with a discussion on the null hypothesis and the alternative hypothesis. The definition of the null hypothesis is "the hypothesis that there is no significant difference between specified populations, any observed difference being due to sampling or experimental error.". The alternative hypothesis means there is likely a link or significant difference. Know that you always start with the null hypothesis. Also know that you can't definitively prove the alternative hypothesis, but you can prove that it's very likely and significant. You can show the data with a 2x2 table. We will show how one can arrive to a type 1 alpha error, in which researchers believe there is a link when there isn't. This is the most common error. Also there are type 2 beta errors, where a researcher doesn't believe there is a significant link when there is. Beta errors are reduced by increased study power. How can we reduce the chances of making an error? One way to do this is with a confidence interval. This creates a range where data points can fall, and you can state that you are confidence that data will fall into this range. We usually go with a 95% confidence interval (CI) and is associated with standard deviation. Know that if the range includes 1 in an odds ratio or relative risk, then there is no link and you must keep the null hypothesis. Same goes for means that contain 0. Next topic: p-value. The p value is the likelihood that the data occurred due to chance. We want a p value less than 0.05. Let's discuss the correlation coefficient: "a number between −1 and +1 calculated so as to represent the linear dependence of two variables or sets of data." A correlation coefficient close to 1 is correlated. If it's 0, then it is not correlated, and if it is negative, it's inversely correlated. To see how much the variables are correlated, that is the coefficient of determination, which is found by squaring the correlation coefficient. Our last topic is on validity and reliability. Reliability is the ability to repeat a test and get the same result. This is associated with precision. Another important concept is validity, the ability to test what you want to measure and it is associated with accuracy. Increased random errors decrease reliability while systematic errors decrease validity. Done with biostats! Let's talk about ethics next!