ECON 2500 Lecture Notes - Lecture 2: Percentile, Sampling Distribution, Inference

• School: York University
• Department: Economics
• Course: ECON 2500
• Professor: Andrei Semenov

Document Summary

Their shape and spread re e(cid:272)t the (cid:272)hara(cid:272)teristi(cid:272)s of the sa(cid:373)ple a(cid:374)d (cid:373)a(cid:455) (cid:374)ot a(cid:272)(cid:272)uratel(cid:455) esti(cid:373)ate the shape a(cid:374)d spread of the sampling distribution: bootstrap inference from a small sample may therefore be unreliable. Bootstrap inference based on samples of moderate size is unreliable for statistics like the median and quartiles that are calculated from just a few of the sample observations: bootstrap versus sampling distribution. Most statistical software includes a function to generate samples from normal distributions. Set the mean to 8. 4 and the standard deviation to 14. 7. You can think of all the numbers that would be produced by this function if it ran forever as a population that has the n(8. 4,14. 7) distribution. Samples produced by the function are samples from this population: the effect of non-normality. The populations in the two previous exercises have the same mean and standard deviation, but one is very close to normal and the other is strongly non-