The determinants of direct mail coupon usage revisited: evidence from count panel data models