[diptojeet]

Correlation are of two kinds association and difference.

correlation relates 2 metric variables
spencer relates 2 ordinal variables
kendal w relates 3 or more ordinal variables
cochran Q relates 3 or more dichotomous variables
Multiples regression 3 or more metric variables


if alpha > 0.05, h1 is false

correlation -
test of association: H0 no assoc H1 assoc
test of difference:H0 no difference H1 difference

Independent sample T test is test of difference of means.

two levels of coupon and its dependence on sales, we write T test
3 levels of coupon and its dependence on sales, we write Anova
if we have one more variable "in-store" with 2 levels, we write 2 way Anova.

There was no non-parametric test done.

standardized beta measures correlation one with respect to another


Extraneous Variables:
History
Maturation
Testing Effects
Instrumentation
Statistical regression
Selection
Bias
Mortality