Beha,

Res. & Therap).

1976. Vol. I.4, pp. 365-368.

Per@mm

Press. Prmted ,n Great Bnta,n

S~~JE~IVE RESPONSES TO ALCOHOLIC AND NON-ALCOHOLIC BEVERAGES BY ALCOHOLIC RESPONDENTS ALAN FRANKEL Veterans Administration

Hospital. Salem, Virginia. U.S.A

(Receired 24 Sepremher 1975) Summary-The semantic differential ratings of real alcohol has been used as a pre-post measure of the effectiveness of behavioral treatment of alcoholism. This attitudinal measure has yieided predominantly negative results. Evidence has been presented that the semantic differential used to rate alcoholic and non-alcoholic slides was not unifactorial. The present study tested whether the semantic differential ratings of real alcohol were unifactorial. Eighty alcoholics tasted two alcoholic and two non-alcoholic beverages and rated them on 20 adjectives derived from IO supposedly bipolar adjective pairs. For the alcoholic drinks there was some concept scale interaction and delinite evidence for a splitting of adjective pairs into two evaluative factors, one of which was mixed with a potency factor. The negative findings of previous research may not reflect a lack of attitudinal change. but a methodological error.

Changes in the subjective response to alcohol following electrical aversive conditioning have been studied through the use of the semantic differential (Costello, 1974; Costello, et ul., 1974a, 1974b; Hallam, ef al., 1972: Miller, et ul., 1974; Miller er al., 1973). As pointed out by the Costello group and by Kaplan (1972) there are several problems with this technique. The mid-point of the 7 point scale may be psychologically impossible to define if the subject does not perceive the adjectives pairs as mutually antagonistic. It is frequently assumed that the adjective pairs are bipolar, based on Osgood et al.‘s (1957) original work. Recent evidence (Costello, 1974; Costello et al., 1974a, 1974b) has demonstrated that the scales used by Hallam et al. (1972) were in fact, not bipolar. The stimuli used by Costello and coworkers were an alcoholic and a non-alcoholic slide. following the work of Hallam er al. (1972). Recent studies (Miller er al., 1974; Miller er al. 1973; Wilson, 1973) have utilized rating the taste of real alcohol. as opposed to slides, in order to measure the effectiveness of behavioral therapies in alcoholism. The subject is asked to taste several beverages, some of which have alcohol, and rate the taste of each drink on a set of semantic differential adjective pairs. Miller ef al. (1974) found that while the actual amount of pre-therapy alcoholic beverage consumed predicted which subjects did better at 6 months outcome, the semantic differential ratings did not. One reason might have been that the scales were not bipolar and thus the semantic measurement was highly unreliable. The purpose of the present study was to replicate the Costello studies, using real beverages, alcoholic and non-alcoholic as the stimuli. SUBJECTS

AND

METHOD

Subjects were 80 inpatient males recruited predominantly from an alcohol treatment unit. They had an average age of 45.5 (s.d. = 9.7); had been drinking for 14.5 years, (s.d. = 9.8); had been hospitalized for drinking a median of 14 days. They were asked to rate the taste of four beverages, whose order of presentation was randomized: 100 cm3 of water, 100 cm3 of coke, 30 cm3 of 86 proof Scotch with 70 cm3 of water, and 25 cm3 of 100 proof Bourbon* with 70 cm3 of water. Ratings were done on the following 20 adjectives (from IO presumably bipolar adjective pairs) on a 14 scale from ‘not at all’ to ‘extremely’: strong, bad, fragrant, sweet, awful, light, dirty, hard, beautiful, unpleasant, clean, nice, weak, pleasant, ugly, good, soft, foul, sour, Presented at the meetings of the American Psychological Association. Chicago, 1975 * The equivalent of 30 cm’ of 86 proof Bourbon. 365

ALAN FRAWEL

366

heavy.* The ratings of all 80 subjects for each beverage were intercorrelated. analyzed, and rotated to normalized varimax solutions.

factor

RESULTS

The first seven adjective pairs in Table I are evaluative: while the last three are potency (Osgood, et al.. 1957. p. 45). For Bourbon and for Scotch. the correlations were significant and negative for good-bad. nice-awful. clean-dirty. beautiful-ugly and pleasant-unpleasant. For the same stimuli. fragrant-foul and sweet-sour were not significantly different from zero. The same pattern held for the Coke stimulus. except that beautiful-ugly did not reach significance. None of the correlations for water were significant. except sweet-sour. Table

1. Correlations

of apparently bipolar beverage

Water

Coke

Adjective

03 01 29 -13 -03 -18 -04 * 10 - 14

-41f 15 -11 -31 -25 -18 -51 -33 -02 -19

Bad-Good Fragrant-Foul Sweet-Sour Awful-Nice Dirty-Clean Beautiful-Ugly Unpleasant-Pleasant Strom-Weak Light-Heavy Hard-Soft

adjective

Patrs

pairs

Bourbon -50 09 01 -48 -50 -50 -61 -31 -49 -51

for each

Scotch -61 -11 -03 -45 -45 -39 -66 -28 -51 -50

*This correlation was indeterminate as the variance of ‘strong’ was zero. t All decimal points omitted: p < 0.05, r = 0.22; p < O.OJ, r = 0.29

Tables 2, 3, and 4 present the results of the factor analyses for Bourbon, for Scotch and for Bourbon and Scotch combined. While the initial factor structure yielded four factors, using the criteria of eigenvalues greater than one, visual inspection of the results strongly suggested a 3 factor structure. Table 2 presents results of the factor analysis for the Bourbon stimulus. Factor I appeared to be a bipolar evaluative factor while factor II appeared to be a mixed potency and evaluative factor loading on the negative pole of evaluative and the potency pole. Factor III was a potency factor. For Scotch. the first factor had bipolar loadings on good-bad, dirty-clean, pleasantunpleasant, with potency loadings on heavy and soft. It appeared to be a bipolar evaluative factor. Factor II had three bipolar loadings, good-bad, pleasant-unpleasant, and fragrant-foul which are independent of factor I, suggesting that factor I for Scotch may be an ‘approach’ factor while factor II may be an ‘avoidance’ factor. Factor III was a potency factor. The combined ratings for Bourbon and Scotch are presented in Table 4. Factor I had three bipolar loadings: good-bad, awful-nice, and pleasant-unpleasant; three unipolar evaluative loadings: foul, dirty, and ugly; and three unipolar potency loadings: heavy. hard. and strong. This combination of evaluative and potency might be termed an ‘avoidance of hard liquor factor’. Factor II appeared to be evaluative with loadings on good-bad, dirty-clean, beautiful-ugly, pleasant-unpleasant. but again combined with loadings on soft. Factor III was a potency factor. The factor structures for Coke and water did not in any way approximate the factor structures either for Bourbon or Scotch and Bourbon or Scotch combined, confirming the concept scale interaction as found by Bynner and Romney (1972). DISCUSSION

The finding that the supposedly evaluative adjectives split off into two independent factors for Bourbon and for Scotch suggest that the reason for negative results (Miller *I

am grateful

to Peter

Miller for supplying

a copy of his semantic

differential.

Subjective

responses Table

to alcoholic

and non-alcoholic

2. Factor Loadings tives for Bourbon

Adjectives

I

Bad Good Fragrant Foul Sweet Sour Awful Nice Dirtv Clean Beautiful Ugly Unpleasant Pleasant Strong Weak Light Heavy Hard

66* -82

beverages

367

of Adjec-

Factors II

III

60 31 48

64

48 63

56 -79 61 -58 -61 73 71 -81

41 39 51 44 73 -45

47 65 -45 -41 78

68 73

* All decimals omitted. ing 20.30 are shown.

Only

load-

et al., 1974; Miller et al., 1973) using the semantic differential as a dependent variable in the taste test may be due to the complexity of a given response. While some of the evaluative scales were bipolar when considered one factor at a time, the results of bipolar loadings on factor I and bipolar loadings on factor II suggested that when these bipolar adjective pairs are used in conjunction with real alcoholic stimuli, the dimensions underlying the ratings are not simple. The finding of an ‘avoidance of hard liquor’ factor coincides with the suggestion of Costello et al. (1974b) that while aversive therapy may modify cognitive avoidance of liquor, it may or may not modify the cognitive approach to liquor. Both dimensions need to be assessed. Table

3. Factor Loadings tives for Scotch

Adjectives Bad Good Fragrant Foul Sweet Sour Awful Nice Dirty Clean Beautiful USlY Unpleasant Pleasant Strong Weak Light Heavy Hard Soft

I 76* -43 78

of Adjec-

Factors II - 40 65 31 -32

III

31

51 81 8’ 46 70

73 -33 85 70 -38

-50 76

33 33

* All decimals omitted. ing 20.30 are shown.

35 Only

58 -64 -68 7’ 74 -68 load-

368

ALAX FRANKEL Table 3. Factor loadings of Adjectives for Bourbon and Scotch Factors

I

Adjectives Bad Good Fragrant Foul Sweet sour Awful Nice Dirt! Cleai Beautiful ug1> Unpleasant Pleasant Strong Weak Light Heav! Hard Soft

76’ -32 71 47 77 -32 67

79 68 -35 44

II

III

-30 75 -38

30 76 -35 52 63 -40 -50 76

43 44

* All decimals omitted. ing ~0.30 are shown.

34 Only

48 -66 -67 69 69 -69 load-

Further, within the alcohol beverages there was a concept scale interaction. In this sample, few of the subjects had abused Scotch, whereas most of the subjects had abused Bourbon. Few subjects reported drinking Scotch in their past histories. In observations of subjects during the actual taste test it was not infrequent for subjects to screw up their face indicating displeasure at the Scotch while no such reactions were noted towards the Bourbon. On the basis of these data, it is recommended that each investigator analyze his own semantic differential scales that are used and not assume that the bipolar evaluative scales originally mentioned by Osgood can be generalized to any particular study. The options available are to eliminate the attitudinal measures or to weight the scores using factor weight coefficients as presented by Costello et al. (1974a).

REFERENCES BYNNER J. and ROMNEY D. (19721 A method for overcoming the problem of concept-scale interaction in semantic differential research. Rr. .1. P \~.c,/lir~. 63, 229-234. Cosr~ LLO R. M. ( lY74) .I/~plicd tr~rrl~.si\ q/ r&u/ h&rrior yf chronic alcoholics I~/~L’u~~rul~crrrr~q 011 cr/co/~~/ rclurcd .stwmltr.s. Paper presented at ‘The First Mexican Congress on Behavior Analysis‘. X&pa. Veracrur. Mexico. COSTELLO R. M.. RI~L D. P. and SCHOENF~LI~ L. S. (1974) Measurement of subjective responses to alcohol and non-alcohol slides b) alcohohc responses. Brhu~‘. Rcs. ad 7hrrapy 12. 3540. COSTELLO R. M.. RICL D. P. and SCHOENFELU L. S. (1974) Attitudinal ambivalence with alcoholic respondents. J. comtrlt. cliu. P.s~rhol. 42. 303-304. HALLAM R.. RACHMAN S. and FALI(OWSL;I W. (1972) Subjective. attitudinal and physiological effects of electrical aversion therapy. Brhur. Rex urd 7hcrap~~ 10. I-1 3. KAPLAN K. J. (1972) On the ambivalence-indifference problem in attitude theory and measurement: A suggested modification of the semantic differential technique. Ps~~kol. Bull. 77. 361-372. MILLER P. M.. HERSI:& M.. EISLFK R. M. and ELKIN T. E. (1974) A retrospective analysis of alcohol consumption on laboratory tasks as related to therapeutic outcome. Behac. Res ard 7hwupy 12. 73-76. MILLI.K P. M.. HI.KS~N M.. EISLEK R. M. and HEMPHILL D. P. (1973) Electrical aversion therapy with alcoholics: an analogue stud?. E&K. Rcs. ud Thrrupjs II, 491497. Oscoo~ C. E.. SKI G. J. and TANNENBAUM P. H. (1957) 771e Meusuremwtof rlfcur~~y.University of Illinois Press. Urbana. Illinois.

Subjective responses to alcoholic and non-alcoholic beverages by alcoholic respondents.

Beha, Res. & Therap). 1976. Vol. I.4, pp. 365-368. Per@mm Press. Prmted ,n Great Bnta,n S~~JE~IVE RESPONSES TO ALCOHOLIC AND NON-ALCOHOLIC BEVERA...
326KB Sizes 0 Downloads 0 Views