PSYC3301 Lecture Notes - Lecture 4: Analysis Of Covariance, Partial Correlation, General Linear Model
Document Summary
Collinearity- (cid:449)he(cid:374) 2 iv"s (cid:272)o(cid:396)(cid:396)elate (cid:449)ith ea(cid:272)h othe(cid:396) (cid:373)o(cid:396)e tha(cid:374) 0. 05. P(cid:396)o(cid:271)le(cid:373)ati(cid:272) to i(cid:374)(cid:272)lude 2 iv"s so highly (cid:272)o(cid:396)(cid:396)elated. Multi-collinearity- when an iv is associated with a multiple of r of 0. 95 based on a multiple regression. Mc is typically evaluated with 2 methods- tolerance and variance inflation factor. Tolerance- (cid:396)ep(cid:396)ese(cid:374)ts the % of (cid:448)a(cid:396)ia(cid:374)(cid:272)e i(cid:374) a(cid:374) iv that is u(cid:374)i(cid:395)ue of othe(cid:396) iv"s. Where 0. 10 is considered the minimum acceptable level of tolerance from an iv. Variance inflation factor- represents the degree to which the beta weights standard errors are larger had they not been any inter-independent variables correlations at all. As (cid:272)o(cid:396)(cid:396)elatio(cid:374)s get la(cid:396)ge(cid:396) (cid:271)et(cid:449)ee(cid:374) iv"s the (cid:271)eta (cid:449)eight sta(cid:374)da(cid:396)d error also gets larger. Vif is the reciprocal of tolerance- thus 10 is maximum allowable level of vif. Normality- multi regression assumes normal distributed data and residuals- can use bootstrapping to conduct a multi regression- (cid:271)ootst(cid:396)appi(cid:374)g does(cid:374)"t assume normality.
