ブックタイトル明星大学 心理学年報 第32号

概要

明星大学 心理学年報 第32号

Hayashi：Development and Testing of a New Indirect Attitude Measure for Pictorial Objects15ed with attitude scores from 102 participants as thedependent variable (Table 9). As expected, resultsrevealed that the main effect of source of the attitudescoresand theone-wayinteraction effect werenot significant (F(1, 101)＝1.16, n.s,η?＝.00; F(2,202)＝.12,n.s,η?＝.00).Therewasasignificant maineffect of original valence of the target (F(2, 202)＝70.29, p＜.001,η?＝.20).This indicated that the twosetsofaverageattitudescoreswerealmost identicalto each other.Thesplit-halfreliabilityoftheSECT,calculated using these two sets of attitude scores,was slightly lower than that of the other threeexperiments (ρ＝.57). These results indicate thatincluding filler trials when calculating attitudescores is beneficial; however, further researchshould be conducted for a more thorough investigation.General DiscussionThe purpose of the present study was to demonstratea new method for indirect measurement ofattitudes toward pictures, and determine the validityand reliability of this method. The methodproposed in the present study,the SECT,was similarto theFUMIE test proposedbyMori et al.(2008),in that it forces participants to convert the originalvalenceoftarget pictures;however,it was supposedthat the SECT possessed prominent features tomagnifythedifferenceinaverageresponselatenciesfor two main blocks when compared with theFUMIE test.Experiment 1comparedattitudescoresobtained using the FUMIE test and the SECT toexamine whether the SECT would measure attitudestoward pictures more appropriately than theFUMIE test. Results showed that the FUMIE testcould not discriminate at all between the threetarget pictures with different emotional valences,whereas SECT attitudescores differed significantlyin two of the three pairs compared. These resultssupported thehypothesis that theSECT was a moreeffective method with which to measure attitudestoward pictures. Whereas the three target pictureswere fixed, Experiment 2 altered emotionally positiveandnegativefiller pictures between groups andconfirmed that attitude scores were not alteredsignificantly according to the filler pictures. Theresults indicated that the SECT attitude measurementwas not influenced by the peculiarity of thecontent of the filler pictures. Experiment 3 alteredthe three target pictures between groups,and fillerpictures were fixed between groups. The attitudescores obtained for positive, neutral, and negativetarget picturesshowedthesamemagnituderelationshipsbetween groups, and attitude scores for thethreetarget pictures differed from each othersignificantly.The overall results indicated that the attitudescores in Experiments 1 and 2 were not theresult of an artifact caused by the peculiarity oftarget pictures used. However, an unpredicted differencewas found between groups on attitudescores toward neutral target pictures. Although itwas impossibleto fullyexplain this differenceusingthe evidence available from the present study, itcould havebeen caused bydifferences in theparticipants’familiaritywith the neutral target picturesbetween the two groups.Throughout the three experiments in the presentstudy,SECT attitude scores significantly predictedpleasantness ratings for target pictures with adjustedR squared ranging from .25 to .37. In contrast,several previousstudiesusing theIAT asanindirectmeasure of attitudes have reported correlationsbetween IAT scores and attitude scores on self-report measures.For example,Nosek (2005)reportedthat the average correlation coefficient was .34on various measurements.Similarly,Hofmann et al.(2005)reported an averagecorrelation coefficient of.24. Furthermore, many studies examining self-esteem have reported that correlation coefficientsbetween self-esteem scores on the IAT and self-report measures tend to fall below .20,and arenotalways significant (e.g., Dijlsterhuis et al., 2009;Greenwald & Farnham, 2000). Although the calculatedR squared in this study was not smaller thancorrelation coefficients between IAT and self-report measures, it would be difficult to comparethem directly. As noted, the present study did notinvestigate correlation coefficients between theSECT attitude scores and the pleasantness ratingfor each target picturebecausepleasantness ratings