hi, struggling a bit with hypothesiss testing, specifcally when to reject/not to reject null based on t values and critical t values? understand when using intervals, upper/lower limit, but get confused with conclusion based on t and critical t values. thanks, John
calculate a t-value for what you are testing. based on your # of entries in your sample and your level of significance - you have a t-critical value from the tables. if your t_calc > t_crit --> reject the null if t_calc < t_crit accept the null. another way of looking at it – draw a diagram always lower …mean …upper lower = mean - t_crit * std error upper = mean + t_crit * std_error this gives you at your level of significance how much of a play (+/-) can you have on your variable to be able to say at your level of significance that you are safe… (this might answer your question on the other post as well).