in schweser page 204, when asking about the multicoll. it says the F Test is statistically significant. what are we looking at to determine the significance?
p-value of 0.001 < 0.10 Hence reject Null and F test is statistically significant.
Never mind. Didn’t read the question.
if we’re given the critical F value, we can use test significance using the 34.6 F stat?
yeps for sure - I prefer the F-critical method only. The p-value confuses me to death. I would go to the F-table grab the F-crit and then compare that to the F-stat.
there’s several issues i have with this question. the 10% significance (alpha) wasnt given in the stem, and CAIA says p val is rejected at 5% or higher.
Multicol happens when two or more independant variables residual values are correlated. Think trying to regress gdp by both inflation and nominal interest rates, inf & nom are similiar indep variables so the resifuals are correlated. But they are both good explainers so high r^2 and high f-test. T-test are low. Ftest is always one tail and follows some non-normal distibution. Use f table. The higher the F stat the better the regression (bur should always check example r^2, test, anova table). Swap, for pvalue just remember the smaller the value the better the variable.
For p value, remember its test is reverse of t statistic t-stat: Reject null if t-statistic GREATER THAN critical value p-value: Reject null if p-value LESSER THAN significance level and in p-value, you dont need to look up values from the table, just compare directly with signifiance level … MUCH EASIER