# Statistical Significance, the Null Hypothesis and P-Values

in LeoFinancelast year

let's assume John tells you he discovered a special dice rolling technique that enables him to roll a sex more frequently than the average person to prove it he rolls the dice in front of you and it actually ends up being six however it's not enough to impress you because it might simply be a matter of luck as there's one in six or roughly sixteen point seven percent chance of that happening regardless of which dice rolling technique you use a researcher who wants to determine if John is right or wrong starts with a so-called null hypothesis which states that no there is no relationship between the technique you use and how often you roll a six in other words you start by assuming that John is wrong John hopes the null hypothesis will be rejected because it would mean it's very likely even if not 100% certain that his right he ultimately rolls the dice thousand times and it lands on six in twenty percent of the cases so higher than the previously mentioned one in six odds is this enough to negate the null hypothesis well no because we also have to determine whether the findings are statistically significant to do this we calculate the so called p-value which is a number between zero and one that lots of applications easily generate for us based on the data we provide generally speaking if the p-value is lower than point zero five it's a good argument in favor of rejecting the null hypothesis and making John happy in our case we'll assume the p-value is point zero four which is low enough to reject our null hypothesis and kind of like a saying John's twenty percent success rate is better than a standard one-in-six odds but there's a four percent likelihood this only happened by chance therefore while we cannot be 100% certain John's technique works we can say that the lower the p-value is the more certain we are all at all here's how things work one start with a null hypothesis to gather and interpret data three calculate the p-value to determine if your findings are statistically significant four if your data supports John's claim and the p-value is lower than point zero five you will reject the null hypothesis and John will be pleased