ECON20003 Lecture Notes - Lecture 3: Kurtosis, Skewness, Categorical Variable

• School: University of Melbourne
• Department: Economics
• Course: ECON20003
• Professor: Griffiths

Document Summary

=(cid:4666)(cid:4667) (cid:2779)=(cid:4666) (cid:4667)(cid:2779) (cid:2167)=(cid:4666) (cid:4667)(cid:2780) (cid:2780) (cid:2167)=(cid:4666) (cid:4667)(cid:2781) (cid:2781: kurtosis (cid:373)easures peaked(cid:374)ess of distribution and thickness of its tails, when x is normally distributed sk = 0 and k = 3. Ske(cid:449)(cid:374)ess (cid:373)easures ho(cid:449) ske(cid:449)ed the distri(cid:271)utio(cid:374) is. Testing for normality: estimate sk and k by getting sample value, che(cid:272)k to see if they"re (cid:272)lose to 0 a(cid:374)d 3 if so, x is normal, use a probability distribution to do so. If normally distributed, jb has a chi-squared distribution with 2 degrees of freedom: assu(cid:373)e 5% sig(cid:374)ifi(cid:272)a(cid:374)(cid:272)e le(cid:448)el (cid:4666)(cid:2870), (cid:2868). (cid:2868)(cid:2873)(cid:4667) (cid:2870, hence, reject h0 if (cid:2166) (cid:2782). =5. 99: quantitative or (cid:272)o(cid:374)ti(cid:374)uous (cid:272)a(cid:374) (cid:271)e (cid:373)easured o(cid:374) a(cid:374) i(cid:374)ter(cid:448)al s(cid:272)ale, categori(cid:272)al data that (cid:272)a(cid:374) (cid:271)e ra(cid:374)ked ordi(cid:374)al data, categorical data that cannot be ranked. Also distinguish between independent and matched pair samples. Independent implies no correlation/covariance: matched pairs involves using the exact same sample.