497 



-16- 



tb«c C cad D vcrc aor* ftvourcblj viewed ctua B «a4 4. Ic should b« 

 eaph««lie4 that che dUteace tp«rc of «af two products la Teble 4 

 dcaote* tb« •catittie«l tlfalflcAnce of cb« difference becwcea cbea 



end it Boc oee^terily « relUbl* Indlctdoa of cbc caounc hj which 



^^ 

 ooe wet preferrec^ ctw other. For IntCAacc, because of Che l«rs« 



ouaber of respondeat! ^ cb« total saople the standard error of the 



differences was saall acd^eir •Lxolflcancc was thereby increased 



relative to the individual clstccrs and other sub-saaples, wtrlch 



y 



explains why cigarette A for thc^tjtsal aaa^le is placed so far to the 



r^ht. ^^^ 



^ The aost strikias feature of the results ia Table 4 is the fact 



chat ciaarettc A, with the lowest nicotine d*(avery (0.7 m%) , was the 

 ~ r- ^ 



'^ - 



lease ptefer^^ not only by the saaple as a wholr but by alaost every 



■^ . "C- . 



aroup of aaokerSTi Cluster I waa the only genuine except lon^ tyen che 



3 (roups who failed^TO distinguish significantly between/jpny of cIm 



cigarettes placed k dir<i;tioaally lowest. Turning co the cigarette Chat 



was aost preferred, «v«ry d^ter placed either C or at the top of cIm 



Ay 



preference scale although la ao)^ cases there was no significant diffcrenc* 

 between D. C and B. It sight have oe^n expected that sose, at least, of 



the Low Need clusters would hav« ehoseiC'cigarette B (0.9 ag nicotine 



/^ 



delivery) aa first preference, but la oo case did this happen. On the 

 other hand two of the High Keed clusters placc<fvChe cigarettes in the 



expected order of preference (D, C, B, A) and were the only clusters to 



. . .-^ » 



prefer D significantly to B. There is little distioctioti between saokers C 



/^ C 



of brands in the various categories wich regard to preferertt*- for the *f 



-^^ o 



experioental cigarettes. In most cases they failed to difCerca^^ce ot 



f to 



