495 



approval of C aad D oa tccoonc of chtlr •crca(Ch allihcly outcuab«ra4 

 •poQCaaeouf llMMroval for b«Laf coo tcroot. Claarccta B was parecivai 



aa baLag cha tao«i$Me aod cha leaac Irrleaac to cha chroac. 



The covataSf. aate ky Lodlvidual clutcera vera broadly la Hat vitk 

 chose suaaarited li^abta ), aod did ooc fall iaco aay obvious paccera 

 related co Inner Need.O- 



Teat Cigarettes - Preftreoe* 

 ^' The preference results ca'a^be expressed la two ways: (a) the results 



-' of analysis by the Round Robin prog'r%a which increases the effective 

 HB^le sise by including indirecc or ioferred preferences as well as 

 direc^^^^eferences; (b) the stralghtforvaf< preferences without Round 

 Robla analysis. C^ 



(a) Rpaa4 Robin toalysis C^ 



For Cba varC^lcs listed la Appendix V (i.e. the €»al Maple, the 

 11 clusters, aad che'^nokers of specified categories of (^aretta) che 

 Round Robin prograa calculates Che order of preference for the & produces 

 and the statistical slgnificaAoe of the difference between any two 

 products. The result, as applies ;-co question lib (i.e. where respondents 

 were forced to state a prefereoc«) v» shown in Table 4. The order of 

 preference for the four cigarcttaa readi ^oa left to right, and cigarettes 

 have to be two cells apart (or aore) for the difference to be significant 

 at the 951 confidence level. Aa exaaples of how^^ interpret the table, 

 Cluster 1 significantly preferred D co both A and C, With B occupying 

 an inceraediace position; Cluster 7 preferred D, C aod B Co A, differences , 

 anong the first three being snail and statistically uareliaSViv Cluster 11 ' i 



; - 



had no significant preferences, although there were distlocc inds^ations g 



vo 



