452 



-10- 



•pp«adLx vtre aoc la fact nslwd until the very end of the fiul recall 

 Interview, efter ell the teat eigerettea had been aatoked and reapondenta' 

 opiniona of thea bad been recorded. However for convenience the Infor^tioa 

 fron theae que^ioai, as it relate* to cluater akeaberahip and general 

 amoklag behaviour^ haa b«ea included in the preaent report. 

 RESULTS AND DISCUSSIOR ■;., 

 I. The Factor Structure . 



An important consideration in deciding between the various Pronax 

 factor solutions which were computed was replication of the factor structure 



m 



found by McKennell in hia aecoad atudy (5). The aolut ion which best net 

 thia requirement waa judged to be the l2-factor solution. Table I ahowi 

 a eomparlfion between the l2-factor solution (co«ifined to items with a 

 loading of 6.4^ or greater) and McKennell 'a lO-fad^C4r solution. The 

 loadings of all «?, items on all 12 factors are shown i^^Appeodix VI 

 (page 74) (this complete set of loadings waa used in the aubseqoent 

 cluster analysis). It will be seen that 9 of McKennell 'a 10 factors 

 have emerged, with much the^flme defining items; for convenience they 

 have been given the same labela aa used by him. His factor DC, Occasional 

 Smoking, has no real counterpart. However two of the items which defined 

 that factor have appeared in two new facfora: "smokes when drinking tea 

 or coffee" ia associated with "smokes when drioking alcohol" in a factor 

 which haa been labelled Refreshment Sotoking; "stookea in a break between 

 jobs" ia associated with "smokes travelling to work" and "smokes in the 

 cinema" in a factor which has been labelled Idle Smoking. An interesting ^ 



point ia the emergence of the item "smokes most at the weekend" as a tsr 



o 

 third new factor all on its own; it is to be noted that although '^'smokes ^ 



