514 



Th« ntlncs for chrtt of cb« accrlbuttt - taoochnctt. Utct aad 

 ««tt of 4r«v • Ihov « dUciocc rclacioashlp vlch ch« l«v«l of lanor 

 Uichouc ozeopcloC %t High R«od eluttort r«c«4 P «• Uccor chaa f for 

 •II chrco Accri^tes, wtMreai ch« Lov Heod clutctrt raced C •• •qutl t» 

 or better chaa F n^ all three. Phyalcal ■easureaeoci (3) tbowed ch«c 

 F had a loaewhac lover ^esrare drop than E and aighc therefore have 

 been margiaally easier to dMv. A potiiblc explanacioo for thi* apparcac 

 difference la perception betveW Low Meod and High Need naokert aight 

 be that respondents first decided^vbich of the two cigarettes they liked 

 ind then rated that cigarette higher nr, a mabar of attributes, including 

 physHal attribunes for which there was no-r^al difference. On this 



basl* LoW^eed smokers would have liked E bette> thaa P, High Seed 

 snoker* would, E^vc liked F better than E. The scdr^s for overall 

 opinion, 'o« « sca^ ranging fro* "excellent" to "a verjQkoof .«l$jrette", 

 lend support to this'v^w: of the four Low Need clusters 6^ gav* a 

 higher score to B, two tc&rcd tbea equally; all three of the High Nead 

 clusters gave a highoc tcore^^^ P. 



The picture in tcras of ove^ll opinion docs not exactly tally with 

 that based on percentage preference, further analysis, described la 



the next section, was perforaed In order^xa clarify the situation. 



*^ 

 F urther Analysis of Results -" ^ 



When the second product test was inlcfated 0)e computer was prograaeed 



to carry out certain analyses and to print out the <1($vers as a series 



-'a 



of tabulations. The results discussed so far represent V'suciaary of 9 



those tabulations. Attempts to Interpret the results reveate^two areas |^ 



o 

 where additional analysis was required: the computer was progranta^ •• 



accordingly, and the results of those additional analyses are discussed ■* 



below. 



