Fixed Effects

Table 1. Type III tests of fixed effects for random effects model
Source	Numerator df	Denominator df	F	Sig.
Intercept	1	157.803	5739.141	.000
COUPVAL	3	1044	4.303	.005
SEQ	3	178.543	.977	.405
CARRY	3	1044	.601	.614
WEEK	2	1044	.014	.986

The tests of fixed effects suggest that, of the four factors, only COUPVAL, with a significance value less than 0.05, has an effect on customer purchasing.

Figure 1. Fixed effect parameter estimates for random effects model

There seems to be good separation between the levels of COUPVAL, with increasing values from the No coupon [COUPVAL=1] level to the 25 percent [COUPVAL=4] level. However, though COUPVAL, overall, is a statistically significant factor, only the No coupon level is statistically different (significance value less than 0.05) from the 25 percent level. A different model might reduce the standard error of these estimates, making them significant.

Judging by the size of the parameter estimates, there might also be a carryover effect that is obscured by the large standard errors.