16.7 In Section 16.4.3 we considered a conditional logit model of choice among three brands of soda: Coke, Pepsi, and 7-Up. The data are in the flle cola.dat.

(a) In additionto PRICE, the data file contains dummy variables indicating whether the product was "featured" at the time (FEATURE) or whether there was a store display (DISPLA\. Estimate a conditional logit model explaining choice of soda using PRICE, DISPI-LY and FEATURE as explanatory variables. Discuss the signs of the estimated coefficients and their significance. (Nore: In this model do not include alternative specific intercept terms.)

(b) Compute the odds of choosing COKE relativeto PEPSI andT-UP if the price of each is $1.25 and no display or feature is present.

(c) Compute the odds of choosing COKE relatlveto PEPSI andT-UP if the price of each is $1.25 and a display is present for COKE, but not for the others, and none of the items is featured.

(d) Compute the change in the probability of purchase of each type of soda if the price of COKE changesf rom $1.25t o $1.30,w ith the priceso f the PEPSI and 7-UP remaining at $1.25. Assume that a display is present for COKE, but not for the others, and none of the items is featured.

(e) Add the alternative specific "intercept" terms for PEPSI and 7-UP to the model in (a). Estimate the conditional logit model. Compute the odds ratios in

(c) based upon these new estimates.

(f) Based on the estimates in (e), calculate the effects of the price change in

(d) on the choice probability for each brand.