ECON 3210 Lecture 15: Econ 3210 Lecture 15
Document Summary
This t-statistic wouldn"t be helpful if it weren"t for the the central limit. Clt tells us that, regardless of the underlying distribution of yi, in a large enough sample, the t-statistic will have an approximately normal distribution. Moreover, regardless of the underlying distribution of yi, y n( y, 2 y ) as n . Therefore, y y se( y) n(0,1) as n . This is why the t- statistic is so useful. Each graph consists of 10,000 y estimates, one for each of 10,000 randomly drawn samples let"s see what happens as we increase size of each sample: If our guess about is correct, then the distribution of the t-statistic will be normal and centred on 0 with a standard deviation of 1. Thus, getting a t-statistic greater than 2 (in absolute value) is highly unlikely if our guess is correct in other words, a t-stat above |2| means our guess is probably wrong or we were extremely unlucky.