PSYC209 Lecture Notes - Lecture 13: Sampling Distribution, Test Statistic, Statistical Hypothesis Testing
Document Summary
Our test statistics for hypothesis testing are all fractions. Inevitably there will be differences. have to account for chance variability. chance variability idea that samples are not populationssample means will be different from population mean just because. Standard error is so important from theoretical sampling dist this is # that tells us what expected chance variability is. We eliminate possibility of type 3 error testing in the wrong tail . Equally impt to test if drug signi cantly decreases or increases life. Testing for possibility that drug works but also that you were wrong and drug signi cantly decreases life. Denominator= noise = chance variability tells us expected variability with no particular reason. Every fraction in statistics is a fraction of signal (what i"m trying to detect, goal of test) and is that above and beyond noise, chance variability t is bigger then more likely to reject null and detect an effect.