Program Context Antecedents of Attitude toward Radio Commercials

Program Context Antecedents of Attitude toward Radio Commercials Kenneth R. Lord State University of New York at Buffalo

Myung-Soo Lee Baruch College, City University of New York

Paul L. Sauer Canisius College

A theoretical model ofprogram context effects on attitude toward the ad (Aad) is developed and tested. Involvement in and liking for a program are shown to exert a positive influence on both claim and nonclaim components of Aad by enhancing commercial-processing motivation. Additional analyses replicate earlier findings that Aad mediates program influence on brand attitude and identify claim strength, appeal of nonclaim factors, and number of exposures as moderators of program effects on Aad.

Over the past decade, the importance of attitude toward the ad (Aad) as a contributor to brand attitude has been well established in both the academic and trade literature. Concurrently, an emerging stream of research has highlighted the impact of program context variables on consumer response to television commercials. In a broadcast media environment characterized by fragmented audiences and commercial clutter, it is becoming increasingly important to isolate program context variables that influence Aad and understand the nature of their effects. Yet not until Murry, Lastovicka, and Singh's (1992) recent finding of a significant influence of"program liking" (PL) on Aad (and an unexplained program involvement [PI] effect) did any published research examine the role played by program variables in the formation of Aad. Journal of the Academy of Marketing Science. Volume 22, No. 1, pages 3-15. Copyright 9 1994 by Academy of Marketing Science.

Many unanswered questions remain. Those addressed by this research are: What is the mechanism by which program context variables affect Aad? What is the nature of the PI effect on Aad? Does program context affect the claim or nonclaim components of Aad, or both (Miniard, Bhatla, and Rose 1990)? Experimental results reported here provide supportive evidence for a theoretical model of context effects on Aad in which PI and PLjointly exert a positive impact on both claim and nonclaim components of Aad by enhancing commercial-processing motivation (CPM).

LITERATURE REVIEW Four categories of program variables have been employed to date in attempts to explain context effects on consumer processing of and response to advertisements. Schumann (1986) examined PL. Reasoning that PL should serve as a peripheral cue to affect consumer response to advertisements, he demonstrated a positive relationship between attitude toward the program and ad effectiveness. Several studies of PL have shown an inverse relationship between that construct and the recall of advertising content (Kennedy 1971; Bryant and Comisky 1978; Soldow and Principe 1981; Lord and Bumkrant 1988b), and mixed results have emerged with respect to PI's capacity to facilitate or hinder ad elaboration (Lord and Burnkrant 1988b, 1993; Anand and Sternthal 1992). Program-induced feelings or mood states have been shown to enhance or inhibit consumer receptivity to commercial messages (Goldberg and Gorn 1987; Pavelchak, Antil, and Munch 1988; Batra and Stayman 1990; Yi 1990). Finally, in a

4

JOURNALOF THEACADEMYOF MARKETINGSCIENCE

study of print advertisements, Yi (1990) identified a fourth type of context effect--"cognitive priming." This effect occurs when the specific content of the program/editorial context primes an attribute that has implications for the evaluation of the advertised brand. However, Aad was found to be sensitive only to "affective priming" (contextinduced affect), whereas "cognitive priming" was observed to affect brand attitude and not Aad. Accordingly, cognitive priming was not considered in the present study. The conclusions that one can draw from the above studies are limited by the failure of most to include Aad as a dependent variable (exceptions are Yi [1990] and arguably Goldberg and Gorn [1987], who used a single-item measure of "perceived commercial effectiveness") and by the absence of a parsimonious conceptual framework that can account for the distinctive effects of PL, mood, and involvement. At least two conceptual models have been proposed that partially address these concerns. Lutz, MacKenzie, and Belch (1983) included "reception context" among the rather expansive array of antecedents embraced by their Aad model, arguing that it should induce mood states that would influence Aad through the peripheral route. However, that was not among the antecedents included in the empirical test of their model (MacKenzie and Lutz 1989); nor did they posit any role of PL or PI. Lord and Burnkrant (1991) proposed a program context effects model in which commercial-processing decrements attributable to high PI may be attenuated by programad mood congruity; but again no empirical evidence was offered, and Aad effects were not specified. The most comprehensive model and empirical work published to date is that of Murry, Lastovicka, and Singh (1992). They proposed and tested a model that posited a direct role of program-elicited feelings and PL in Aad formation. They found support for the influence of PL, but not for program-elicited feelings. (When added to a regression model, feelings did not induce a significant increase in R 2. Murry, Lastovicka, and Singh concluded that PL may be a summary construct that accounts for programelicited feelings.) They reasoned that PL was restricted to serving as a peripheral cue, but they lacked the process indicators needed for a direct assessment of the mechanism by which the observed effect arose. Employing PI as a covariate, they found it to account for a significant amount of variance in Aad but did not explain the direction or nature of this result. A comprehensive overview of the Aad literature is impractical and probably unnecessary here. Interested readers can consult Brown and Stayman (1992). Because of its direct relevance to the present research, however, the recent work of Miniard, Bhatla, and Rose (1990) warrants brief consideration. This work demonstrated that Aad can be decomposed into evaluations of claim (Aad-c) and nonclaim (Aad-nc) ad elements. Which of the two components dominates was shown to be a function of audience elaboration levels. To the extent that program context variables influence ad-processing motivation or opportunity, they may be expected to exert differential impact on Aad-c and Aad-nc; however, no prior context effects studies have investigated this possibility.

WINTER 1994

MODEL AND HYPOTHESES We propose a model of program context effects that attempts to: (1) identify the contextual variables most likely to affect Aad; specify the direction of those influences; (2) clarify the mechanism by which they are exerted; and (3) illuminate their distinctive impact on Aad's claim and nonclaim elements. A graphic depiction of the model appears in Figure 1. Murry, Lastovicka, and Singh's (1992) demonstration of a significant role of PL in Aad formation, and their unexplained significant result associated with PI, suggest the need to incorporate these two contextual constructs into a model of program effects on Aad. Tentatively accepting their supposition that PL captures program-induced feeling or mood states, we do not include that construct in the model (although measures are included in the empirical work described later to confirm that variable's relative unimportance to Aad formation).

Program Involvement The theoretical explanations offered for PI effects on ad processing in prior literature include (1) drive for closure (motivation to "fill in" missing or unresolved program details [Kennedy 1971]); (2) information overload (rehearsal of program content concurrent with ad exposure produces "stimulus overload," which precludes rehearsal of the advertising message [Bryant and Comisky 1978]); (3) attention deficits/capacity constraints (ongoing elaboration of program content leaves inadequate attentional capacity available for ad elaboration [Soldow and Principe 1981; Lord and Burnkrant 1988a, 1991]); and (4) ease of counterarguing (PI interferes with counterargumentation [Anand and Sternthal 1992]). Although the explanations vary, they possess a common thread--an assumption that PI affects the amount and/or target of stimulus-based elaboration. The relationship between PI and ad elaboration has typically been assumed to be an in- verse one; that is, ongoing elaboration of program content interferes with ad-relevant thought (Lord and Burnkrant 1988a). However, studies showing a deleterious effect of PI on consumer response to ads have typically used extreme manipulations of PI. Lord and Burnkrant (1988b) found PI to be characterized by a broad-ranging continuum of values (rather than a high-low dichotomy). Anand and Sternthal (1992) anticipated that a deleterious effect on ad elaboration would occur only within a narrow range of high PI values. Indeed, Lord and Burnkrant (1988b) found a consistent positive relationship between PI and adrelevant cognitive response generation across two different ads. Likewise, Krugman (1983) found ads to be more effective when placed in more "interesting" programs than in less interesting program environments (although it is unclear whether program interest in that research was driven by PI or PL). A nonlinear PI-ad elaboration relationship could account for both the positive and negative ad elaboration effects shown in prior studies. Lord and Bumkrant (1993)

Lord et al. / ATTITUDETOWARDRADIOCOMMERCIALS

5

FIGURE 1 Program Context Effects on Aad

-Z

J

x2

x3 x4 x5

I

Program 1 Involvement (PI)

Commercial Processing Motivation (CPM)

y6

A ad Claim (A ad - c )

~

x6 x7 x8

~/2

~'1

y4., (

Program Liking (PL)

[

~ y2

I Nonclaim Aad ~ Y7 y8

1

(A ad - nc )

l

Brand Attitude (A b )

argued for such a relationship, suggesting that at a minimal level of PI the audience may be in such a cognitively passive state that readiness or motivation to attend to a novel stimulus (the ad) may be undermined. A moderate level of PI may increase the likelihood of attention to an ad by elevating the overall level of cognitive arousal, whereas a high level may introduce distraction from ongo-

y9

~4~10 "~ Yll '~ Y12

ing program elaboration. This relationship is depicted in Figure 2. Because the preponderance of television and radio programming probably falls between the extremes of cognitive passivity and attentional capacity overload (the cognitive arousal domain of Figure 2), the model depicts a positive influence of PI on CPM.

6

JOURNAL OF THE ACADEMY OF MARKETING SCIENCE

WINTER 1994

FIGURE 2 PI and Commercial Processing /

~

, r_.uuanccu

9

C

o

P

r

* ~

C

P

:;:

o

a

o ~

.

/

%

~

.

~

c

r

s

C i

S i

a

n

v

/

~

/

%

%

%

%

%

/

I l l l l l l l l l l l l l l l l l l l 0 1 0 1 1 1 1 1 0 1 1 1 1 1 1 0 1 1 l l 0 1 1 1 1 1 1 1I

. . . . . . . . . . . . . . . . . .* ..m. i . g I . o

%

"..n.gil'a

%

. I. .l .l .l . l. l. .l .l .l .e

4 o

l l l l l l ~ l l l lI

~

~

~

~

~

. . o. . . . . . . .o . .

IIIII

I

0

..:.....

,,.al.aoo

....

Distraction

10011BI

R

" ~ , ' ~ , ' , '. ~ ~'..:N i . . . . . . . . . . . . . . . . .

I11

I

IIIiIqlll

I

I

III01

I

I

g010411011111111RIIIIRIIIIRI00111111101

"/,',',',',::::::::..'x:.....................................

~

iIlOIIiIiIiIlOtlllROiIiIiIl~llllllll

,

.

.

t .

/

s, i !!i!iii iiiiiiiiiiiNi ::::::::::::::::::::::::::::::: .-,Arousal s - , , , , , , - , , ~ : i ~ . x : . : . : . . : : : : . : : . : : : : : : : : : :

_ .

%

~

.

~

m

/

,. .~. ,. ~. . ,. ~. . -. .~ i : i : i : i : : : i :':":"i~: 'i":"i ':":':"i':"i":' :"i' :" i' ": ':"i ' " " ' " ' " "

.. Co~nmve .

/

. . , . %. .% . % . %. .% . % .

,

.

-

.

~ ~

~ ~ ~ ~~~

,

. . . . .

. . . . /. % ./ %./ \ . I %. / .% / u

.

.

.

t

.

.

.

,

.

.

.

.

, , , ~ t , ,: , : , : , : ,

.

.

.

.

,

~ ! ~ N i . . . . . . . . . . . . . . . l, .l , l ,l. ,l ,, .l , l ~l .l 1' l7. l 6l .l l l. l l .l l

%

9

.

.

,,,:..:1,,,,,.

.

.

.

.

.

.

,.,

l l l l

.

. '~ , ' - ' , ~ ~ : ~ . ~ i . ' . ' . . . : . : .: , : , ': , : , :., : , , ', , , . . ", ' , . .' ,.: :, . . . ,1 .,., ,.o ,

,,,

, ,

,,,,

iliiiiiiiiililiiiiiiii)!iiiiiiiiiiiiiiiiiiiiiii!iii!i!i iiiiiii!ii•!iii•ii•iiiiii!iii•••iiiiiiiiiiii•i•iiiiiiiiiii•••iiiiiiii!ii•iiiiiii ,

. . . .

%

,

/

%

/

%

I

%

/

%

I

%

/

%

I

%

/

%

/

%

I

%

/

%

/

%

I

%

I

%

/

%

I

I

"Ill l IOBO00

I I I I I I l l l l i l l l l l t l l l l l l , l l l I . l l l l . l l l l l l l l . l l l l l l l l l l l l

Program

Program Liking Prior treatment of PL has assumed that construct to function as a peripheral cue, exerting an impact on Aad or other consumer response constructs by way of conditioning or direct affect transfer (Schumann 1986; Murry, Lastovicka, and Singh 1992)9 However, studies examining that construct have not tested possible mediators of the PL effect9 Goldberg and Gorn's (1987) finding that "happy" programs elicited higher commercial recall and more adrelevant thought than did "sad" programs implies that program-induced affect may exert a positive influence on the cognitive processing of ads. Although Goldberg and Gorn did not measure PL per se, they did use a "pleasant/ unpleasant" semantic differential scale that was highly correlated with the "happy/sad" item, implying that PL may capture program-induced feeling states, as argued by Murry, Lastovicka, and Singh (1992)9 It is at least plausible, then, that the program context effects on commercial effectiveness observed by Goldberg and Gorn (1987) can be explained by cognitive-processing enhancement induced by PL, as opposed to a noncognitive conditioning or affect-transfer mechanism. In view of the potential for program-induced feeling states to affect PL, a brief overview of recent findings with respect to the role of mood or affect in persuasion is in order9 It is important to remember, however, that program mood and PL are not fully equivalent constructs. For example, many well-liked programs induce a variety of

.

..:.:.:.:.

, , , , , , , , , , , , , . .

%

/

o~o,,

, l l l l l l l l t l l 1001441111111

i

I

, *

I

I

I

l I I

I

I

I

* i I l I I *

I

I

I

I

I

,

l

S

1

I

Involvement

affective states, negative to positive, as they move from scene to scene. Isen (1984) suggested that positive affect changes cognitive organization, priming broader or more integrated categories in memory. Silk and Vavra (1974) argued for an intensity effect, suggesting that either positive or negative affective states may serve to heighten cognitive arousal (and by implication processing motivation), and Srull (1983) obtained consistent results in an advertising context. More recently, Petty et al. (1993) demonstrated a positive mood's ability to enhance attitude directly when processing motivation is low, and indirectly (by inducing positive cognitive responses) in high-elaboration conditions. Alternatively, Worth and Mackie (1987) found positive mood to undermine systematic message processing. Consistent results have been obtained by Forgas, Burnham, and Trimboli (1988), Mackie and Worth (1989), and Batra and Stayman (1990). Explanations of the latter effect include a motivation to maintain a positive mood and a reduction in cognitive capacity. Mackie and Worth (1989) reported evidence in support of cognitive deficits, showing that, in the absence of time limits, subjects in a positive mood actually exerted more time and effort in message processing than did those in a neutral-mood condition. Their results suggest that, given the informational simplicity of most television and radio advertisements (which do not require excessive capacity to process and comprehend), positive program-induced affect (and to the extent that it reflects such affect, PL) may be expected to provide a motivational boost to facilitate ad processing9

Lord et al. / ATTITUDETOWARDRADIOCOMMERCIALS 7 Because attendance to radio and television programs is seldom a forced activity, it is unlikely that viewers or listeners would long remain tuned to a disliked program. Thus most advertisements would tend to reach their audiences via a program that is positioned in the neutral or positive ranges of the disliked-liked continuum. In such a constrained setting, the question of whether any CPM and ability enhancement is due to the direction or the intensity of PL would be of minimal concern to advertisers, because any possible enhancement due to program disliking would seldom be encountered. What is relevant is the apparent positive impact of positive affect on stimulus processing. Consequently, our model depicts a positive role of PL on CPM in normal processing contexts. Both PI and PL are thus expected to affect Aad through a process mediated by CPM rather than directly as peripheral cues.

Ad Evaluation Increased ad processing brought about by program context variables should exert a positive impact on viewers' or listeners' evaluation of high-quality ads. Because advertisers are usually cognizant of the importance of making messages appealing in both their claim and nonclaim components, the motivation of increased commercial processing on the part of the listener or viewer should result in both more favorable Aad-c and Aad-nc. However, because increased processing motivation has been shown to result in a disproportionate focus on message arguments (Petty, Cacioppo, and Schumann 1983), the effect should be stronger for Aad-c than for Aad-nc.

Hypotheses The expected effects detailed in the above model discussion are summarized in the following research hypotheses. 1-11: Increasing PI leads to enhanced motivation to process advertisements. 1-12: Increasing liking for the program leads to enhanced motivation to process advertisements. 1-13: Program influence on Aad is mediated by CPM. I-I4: The effect of program-enhanced CPM is positive for both Aad-c and Aad-nc, but greater for Aad-c than for Aad-nc. Empirical evidence supporting H 1 would require a significant positive parameter for the path linking PI to CPM. Support for H2 requires a significant parameter for the path linking PL to CPM. Support for H3 requires evidence that direct paths linking PI and PL with the two Aad constructs are not significant. Significant positive parameters for the paths linking CPM to Aad-c and Aad-nc, in which the parameter value for the CPM ~ Aad-c path is significantly greater than the parameter value for the CPM ~ Aad-nc path, would support H4.

Moderators Although prior investigations into program context effects have to date been largely silent as to potential moderators, it seems reasonable to assume that variables that in other contexts have been shown to moderate the effectiveness of consumer-influence strategies may affect the extent to which a program environment influences processing motivation and attitudes. One message factor shown to have such an effect is ease of message counterargumentation (Lord and Burnkrant 1988a; Anand and Sternthal 1992). In the absence of any prior consideration of how this or other moderators may affect program influence on Aad, no explicit hypotheses are developed. However, the issue of how three variables known to affect counterargumentation and other forms of cognitive response--argument strength, executional (nonclaim) elements, and number of exposures--will moderate the program-Aad relationship is posed as an exploratory research question. This is clearly not an exhaustive set of possible moderators. Among other relevant variables is the message's executional style. For example, the relative cognitive and affective orientations of PI and PL, respectively, and the mediating influence of processing moti- vation, may differentially affect informationally and emotionally oriented ads. Because of tractability, however, the exploration of such effects is left to future research.

DESIGN AND STIMULI The program stimulus, taped directly off the air, was a half hour of conventional programming from a station in a listening area adjacent to that from which subjects were drawn and the experiment took place. Its content included soft rock selections, news feature items, and commercials, interspersed with narration from the radio announcer. To undertake an exploratory assessment of the consistency or variability in program context effects across different claim, nonclaim, and exposure strategies, test ads were created in four versions (all combinations of strong/ weak arguments and liked/disliked instrumental background music) and inserted into the program excerpt either once or three times--a 2 (Argument Strength) • 2 (Background Music) • 2 (Number of Exposures) factorial design. The test ad was a 30-second radio commercial for a fictitious brand of toothbrush (Hygent). Significant differences emerged in pretest data for both argument quality perceptions and musical selection preference. Copy for the message argument conditions is contained in the appendix. Stimuli for all conditions were professionally produced and featured a music lead-in at normal volume, with the music dropped to a background level through the narration (provided by a professional radio announcer) and briefly raised to a normal level at the end of the ad. The ads originally aired with the program were retained as fillers and were inserted into commercial pods at varying levels of exposure (one, two, or three).

8

JOURNAL OF THE ACADEMY OF MARKETING SCIENCE

SUBJECTS, PROCEDURE, AND MEASURES In exchange for extra credit in an introductory marketing course, 328 undergraduate marketing students participated in the study. Instruments for four participants were discarded because of failure to complete all items needed to operationalize the model, leaving a sample size of 324 for analytical purposes. Subjects listened to an audio recording of the radio program and embedded advertisements. Prior to the commencement of the tape, subjects were told that they would be exposed to a half hour of programming recently aired by a radio station in a nearby listening area; the station was interested in learning their feelings about the quality and impact of its programming for a student audience. They were then asked to listen to the program as they would in a normal listening situation, at home or in a car. On completion of the tape, questionnaires were distributed. Included were scales measuring the modeled variables--that is, PI, PL, CPM, Aad-c, Aad-nc, and brand attitude (Ab). Other measures tapped program-induced feelings, recognition memory, respondent demographic characteristics, and perceptions of the purpose of the experiment. Five items from Zaichkowsky's (1985) Personal Involvement Inventory, related to the radio program to which subjects had just listened, were employed as measures of PI, each scaled in 7-point semantic-differential form (important/unimportant, of no concern to me/of concern to me, irrelevant/relevant, beneficial/not beneficial, vital/ superfluous). These and other multiple items were reverse coded as needed to ensure appropriate, consistent polarity across all items in the scale. Cronbach's alpha, used to assess reliability for this and the other multipleitem scales, was .80 for the PI items. Three 7-point semanticdifferential items tapped the PL construct (good/bad, unpleasant/pleasant, favorable/unfavorable; alpha = .89). The same three items, related to the background music in the ad (the major nonclaim characteristic of the test ad) and the test product, were repeated as measures of Aad-nc (alpha = .94) and Ab (alpha = .85), respectively. (Although liking for background music would in many instances be inadequate as an indicator of Aad-nc, the absence of other nonclaim executional elements makes it appropriate in this instance.) Program-induced affect was measured by means of Watson, Clark, and Tellegen's (1988) Positive and Negative Affect Scale (alpha = .78 for positive affect, .89 for negative affect). CPM was assessed through the use of three 7-point semantic-differential items. (While listening to the radio commercials I was very involved/very uninvolved, concentrating very hard/concentrating very little, paying a lot of attention/paying very little attention; alpha = .93). Aad-c was measured by three 7-point semanticdifferential items related to ad claims (persuasive/ unpersuasive, weak/strong, believable/unbelievable; alpha = .72). (Other measures were included as part of the multiple-item scales developed for the modeled variables, but, in the interest of model tractability and measurement model validity, analyses were restricted to those reported

WINTER 1994

above. Information on other items is available from the authors.) Although not directly incorporated in the theoretical model, some other variables were measured for related purposes. To assess recognition memory, subjects responded to a diverse pool of claims, including those that actually appeared in the ad to which they were exposed and an equivalent number of plausible but inaccurate statements. Following Jacoby and Hoyer's (1989) procedure, subjects responded to each statement using a threeresponse scale ("claim made in ad," "claim not made in ad," "don't know"). Correct recognition was coded as +1, incorrect as -1, and "don't know" as 0, with the sum of the recognition scale serving as the recognition memory index for purposes of analysis. Respondents indicated their gender, years in school, and marital status. Because none of the modeled variables showed significant differences between demographic groups, ensuing analyses made no use of those variables. Responses to the query as to the experiment's purpose ranged from a repetition of the cover story to a general sense that the Hygent ad was the topic of interest. (The latter perception was to be expected, because subjects responded to multiple questions about the ad immediately prior to the final item about the study's purpose. Although the placement of the relevant item at the end of the instrument undermines the ability to separate stimulus-induced from instrument-induced hypothesis guessing, this was viewed as less threatening than placing the question at the beginning of the questionnaire, where it could bias responses to the more theoretically relevant items that would follow.) None of the study participants made reference to the specific relationships of relevance to the hypothesis or, even less directly, to the variables involved. Therefore, hypothesis guessing was not viewed as a significant threat to the interpretation of results emerging from this study.

ANALYSIS AND RESULTS Construct Validation Modeled variables. The model depicted in Figure 1 was tested by LISREL 7 (Jrreskog and Sfrbom 1989) using the covariance matrix shown in Table 1. Coefficients of deterruination for the x (program context) and y (CPM, Aad, Ab) variables were .981 and 1.000, respectively, and no modification indices of substantive magnitude emerged for the lambda matrices, implying good fit of the measurement model. Aside from the question of the indicators' ability to validly measure the latent constructs, the objectives of the research required that the stimuli induce sufficient variability in response to observe the changing impact of program context variables across a range of values. For example, if PI was at the same level for all subjects, the effect of that construct on processing motivation would be undetectable, even if it existed in the population. From the descriptive statistics reported in Table 2, it can be seen that

X

, g

II1

(J

~.~ .~

10

JOURNAL OF THE ACADEMY OF MARKETING SCIENCE

WINTER 1994

TABLE 2 Descriptive Statistics for Modeled Variables Variable

Mean

SD

Minimum

Maximum

Coefficient Alpha

Program Involvement Program Liking Commercial Processing Motivation Aad-claim Aad-nonclaim Brand attitude

2.61 -0.64 3.00 3.83 0.16 0.81

1.14 1.43 1.56 1.32 1.82 1.22

1.00 -3.00 1.00 1.00 -3.00 -3.00

6.00 3.00 7.00 6.67 3.00 3.00

.80 .89 .93 .72 .94 .85

the scales of all the theoretical constructs in the model elicited sufficient variability in response to facilitate the detection of meaningful relationships between them. Both PI and PL, although eliciting a mean response slightly to the low/negative side of the neutral midpoint (PI: mean = 2.61, where 1 = low, 7 = high; PL: mean = -0.64 where - 3 = negative, +3 = positive), were characterized by values across the sample that occupied all or most of the scale points (minimum and maximum values of 1 and 6 for PI, -3 and +3 for PL) and demonstrated reasonable variance (standard deviations of 1.14 and 1.43 for PI and PL, respectively). The relatively neutral mean scores for PI and PL suggest that the program stimulus was an appropriate one for capturing context effects likely to emerge from typical radio broadcasting, as opposed to the more intense stimuli used in most prior context effects studies. Because post hoc self-report measures were used to capture CPM, it is important to establish that those items were not contaminated by subjects' inability to accurately make such assessments (Lord, Burnkrant, and Owen 1989) or by self-report bias. After partitioning the sample into high- and low-CPM groups (based on a median split), it was found that subjects high in CPM scored higher on the recognition memory index than did those in the low group (means of 3.99 and 2.90, respectively, p < .01)--a result that could not have occurred without substantive differences in CPM. Moderators. As noted earlier, argument strength and background music manipulations were validated by pretesting in the stimulus development stage of the research. Still, some potential biases need to be investigated. First, whenever a weak-arguments manipulation is used, one must be cautious that it is not of such unrealistically low quality as to damage credibility or render the results meaningless to real-world advertisers. This would not seem to be a problem in this case, because the weak-arguments condition yielded a neutral rather than a negative mean Aad-c score (-.03 on a -3/+3 scale). Second, a concern remains that background music may inadvertently induce attribute-relevant inferences (Miniard et al. [1991] demonstrated the possibility of such a role for executional elements of an ad), biasing the intended exploratory analysis of any possible moderating influence of liking for nonclaim elements on program's influence on Aad. Any such effect would, however, be expected to be observable by way of an effect of the background music factor on

Aad-c, but none emerged (Aad-c = 3.91 and 4.07 for the liked and disliked music conditions, respectively, p >. 10). Similarly, Aad-nc was unaffected by the argument manipulation (0.33 and 0.00 for weak and strong arguments conditions, respectively, p > .10).

Model and Hypothesis Testing Model fit indicators and standardized path parameters are contained in Table 3. Although the chi-square statistic (262.30, d f = 160) remained significant (p < .001), goodness-of-fit and adjusted goodness-of-fit are both greater than .90 (.925 and .902, respectively), and the root mean square residual is reasonably low (. 111), indicating that the data fit the model depicted in Figure i adequately. Following Anderson and Gerbing's (1988) recommended approach for model evaluation, the focal theoretical model was compared with two alternative specifications. In the first (a constrained model), one of the estimated paths in the theoretical model (CPM --~ Aad-c) was fixed at zero. The difference in chi-square values between the theoretical and constrained models was significant, Z2(1) = 14.93, p < .01, suggesting that the hypothesized model had significantly better explanatory power than did the constrained model. The second comparison involved a competing model in which paths constrained to zero in the theoretical model were estimated. For this purpose, paths linking the program context variables with Aad (PI --->Aad-c, PI --+ Aad-nc, PL --->Aad-c, PL---> Aad-nc) were freed. The difference in chi-square between the hypothesized model and this less constrained alternative was not significant, Z2 = 3.10, p >. 10, suggesting that the additional paths provided no significant increase in explanatory power. At the global level, then, evidence exists in support of the conceptual rationale developed earlier, from which the model was drawn. Support for individual hypotheses, however, requires an examination of specific path parameters. H1 predicted a positive relationship between PI and CPM. This relationship was observed in the model through the significant positive PI ~ CPM path, supporting H1. A similar result is seen for PL, supporting H2. Thus increases in both PI and PL, in the constrained ranges associated with typical radio programming were shown to enhance consumer motivation to process commercial messages. The question of whether the program variables' impact on Aad was direct or mediated by CPM (the issue ad-

Lord et al. / ATTITUDETOWARDRADIO COMMERCIALS

11

TABLE 3 Measures of Fit and Parameter Values Measures of fit Chi-square (dr~significance level)

Goodness-of-fit index Adjusted goodness-of-fitindex Root mean square residual

262.30 (190/.00) .93 .90 .11 Path Parameters: Unstandardized (standardized) Measurement Model

CPM

Yl Y2 Y3

1.00" (1,42) 1.04" (1.48) 1,16" (1.65)

Aad-claim

Y4 Y5 Y6

1.00" (1.46) 0.86* (1.25) 0.54* (0.79)

Y7 Y8 Y9

1.00" (1.81) 0.96* (1.74) 0.99* (1.80)

Brand attitude

Yl0 Yll Y12

1.00" (1.16) 0.77* (0.89) 1.07" (1.23)

PI

x1 x2 x3 x4 x5

1,00" (1.07) 1.01" (1.07) 1,06" (1,13) 1.02" (1.08) 0.74* (0.78)

x6 x7 x8

1.00' (1.35) 0.89* (1.20) 1.10" (1.48)

Aad-nonclaim

PL

Structural Model

PI --) CPM

.26* (.20)

PL --~ CPM

.20* (.19)

CPM ---)Aad-c

.73* (.72)

CPM ~ Aad-nc

.69* (.54)

Aad-c ~ Ab

.51' (.64)

Aad-nc --->Ab

-.08. (-0.13)

~PI, PL

.92* (.64)

*p < .05. dressed by H3) required an expansion of the model shown in Figure 1 to allow for any significant direct effect to be observed. The model form required for this purpose is the less constrained model described earlier. As noted, the two models did not differ in their overall explanatory power. However, although the mediated paths remained significant in the expanded model, none o f the four direct paths attained significance (all t values < .80). It is thus clear that program influence on A a d was mediated by CPM, in support of H3. As predicted by H4, program-enhanced C P M exerted a significant impact on both the claim and nonclaim components of A a d (parameters for both the C P M --+ Aad-c and CPM -+ Aad-nc paths were significant and positive). Also consistent with the hypothesis was the relative magnitude of the two paths. C P M ' s influence on Aad-c was about one third greater than on Aad-nc (the respective path coefficients were .715 and .538). However, constraining the two C P M ~ A a d parameters to be equal did not yield a significant difference in chi-square. Thus, despite direc-

tional consistency, the component o f H4 that predicts a stronger C P M influence on A a d - c than on A a d - n c was not statistically supported. Hence H4 was only partially supported. Although not hypothesized to affect A a d either directly or mediated by CPM, program-induced feelings were incorporated into an expanded model to allow for a more comprehensive assessment o f possible program-context effects. Consistent with the findings o f Murry, Lastovicka, and Singh (1992), none o f the paths linking positive and negative program affect to C P M and directly to A a d had significant parameters. Thus PI and PL appear to capture program effects on Aad. Another model revision, adding direct paths from PI and PL to A b (PI ~ Ab, PL ~ Ab), was employed to test the replicability o f the Murry, Lastovicka, and Singh (1992) finding that program effects in A b are fully mediated by Aad. Again, the mediated paths remained significant, whereas the direct paths were not. The consistency of this result with that of Murry, Lastovicka, and Singh

12

JOURNAL OF THE ACADEMY OF MARKETING SCIENCE

WINTER 1994

TABLE 4 Moderator Analysis--Path Parameters AadCPM ----> C P M -->

Aad-Claim ---> Nonclaim --->

P I -->

P L --->

Aad-

Brand

Brand

Moderator Variable

CPM

CPM

Aad-Claim

Nonclaim

Attitude

Attitude

Argument strength Weak Strong

.19 .18

.24* .14

.75* .'~76"

.46* .61"

.70* .58*

Music attitude Disliked Liked

.29* .12

.12 .24*

.44* 1.00"

.54* .63*

Exposure level 1 3

.12 .23*

.41' .10

.37* .97*

.16 .94*

Goodness-

rpI. PL

9C2

of-Fit Index

-.15 -.11

.70* .57*

243.89* 219.45"

.86 .89

,77* .54*

-.27* -.02

.59* .64*

218.46" 234.10"

.88 .88

.54* .71"

-.14 -.15

.61" .66*

232.02* 203.77*

.87 .90

NOTE: PI = Program Involvement; PL = Program Liking; CPM = Commercial Processing Motivation. *p < .05.

(1992) leaves little question about Aad's dominant role as the mediator of program influences on Ab; that is, program context effects on brand attitude are filtered through Aad. One theoretically plausible result that the models specified for previously reported analyses would be incapable of detecting is an interactive relationship between PI and PL. It is possible, for example, that PL would affect CPM only when PI is low. To test for this, or other forms of PI/PL interaction, two 2-group analyses were conducted. In the first, PI was removed from the model, leaving PL as the only program context variable. The two groups were those who were high and low in PI (determined by a median split on the PI scale). A comparable analysis was performed for high- and low-PL groups, with PI retained as the sole program context variable in the model. In each instance, parameters were invariant across groups, implying a lack of dependence of the effect of either program context variable on the other. Thus the premise implicit in the main theoretical model--that PI and PL, although correlated, operate independently to affect CPM---appears to be valid.

Moderator Assessment Some diagnostic value exists in the examination of potential moderating effects of variables within the control of the advertiser that were incorporated into the experimental design: message claim elements, nonclaim cues, and amount of consumer exposure. To accomplish this, the model in Figure 1 was analyzed separately for one- and three-exposure conditions, weak- and strong-argument conditions, and liked- and disliked-music conditions. The decision to run these three 2-group analyses rather than testing the model separately for each cell of the 2 x 2 • 2 data set was based on a multivariate analysis of variance (MANOVA) result that showed significant main effects for the number-of-exposures and background music factors and a marginally significant (p < .10) main effect of argument strength, with no significant interactions. Stan-

dardized parameters for the structural paths are shown for each of the resulting six models in Table 4, along with indicators of model fit. Model fit in the six models was not substantially different from that reported earlier for the sample as a whole. There was some reduction in chi-square values, but they remained significant. The goodness-of-fit index declined slightly, but remained close to the traditional .90 criterion for fit. In view of the exploratory nature of this analysis, the results warrant replication, and inferences must be treated with caution, but they do offer some theoretically plausible and potentially useful insights. With respect to the relationship between program context variables and processing motivation, each of the three manipulated variables appeared to exert a moderating influence. First, listener response to the program was shown to exert a significant impact on CPM when weak arguments were employed, but not in the strong-arguments conditions. Second, program effects were primarily observed in connection with PI when less appealing nonclaim elements were used and with PL in the liked-music condition. A possible explanation is that the combination of a liked program and appealing background music in the ad creates a positive processing environment, but when nonclaim executional elements are less appealing, PI is needed to boost processing motivation. Finally, most program influence was attached to PL in the single-exposure condition and to PI when consumers heard the ad three times. Because the correlation between PI and PL was significant for both exposure conditions, it appears that the number of exposures operates directly on CPM to determine the channel (cognitive or affective) through which program context effects will flow to influence Aad. Perhaps interest in the program is needed to sustain processing motivation after the initial enhancement effect of PL on the first exposure. Correspondingly, the processing benefits of PI may require repeated ad exposure to be fully realized, allowing listeners a better opportunity than a single exposure to elaborate on the message.

Lordet al. / ATTITUDETOWARDRADIOCOMMERCIALS The observed relationship between CPM and Aad in the various conditions is largely consistent with that obtained in the aggregate model (i.e., significantly positive for both Aad-c and Aad-nc, with the CPM ~ Aad-c parameter greater in magnitude than that attached to the CPM Aad-nc path). Exceptions were the disliked-music condition (where the CPM ~ Aad-nc parameter was somewhat greater than that obtained for the CPM ~ Aad-c) and the single-exposure condition (where CPM --4 Aad-nc was positive, but not significant). A possible explanation for the deviation in the disliked-music condition is that the cognitive orientation of the program's influence on CPM (involvement rather than liking) mitigated the perceived relevance of an affective response to the music in evaluating Aad-nc. From the relatively lower magnitude of CPM ~ Aad parameters, both claim and nonclaim, in the single-exposure condition (compared with other conditions), it would appear that the attitudinal consequences of program-enhanced processing motivation require more than one exposure to reach fruition. The relationship between Aad-c and Ab was largely invariant across conditions, matching the result obtained from the pooled sample. The negative impact of Aad-nc on Ab that emerged from the pooled analysis carded through directionally in each of the six conditions but was significant only for disliked music. However, the fact that for four of the six models the parameter value exceeded (in absolute value) that observed in the earlier analysis suggests that the apparent lack of an effect may be due to constrained statistical power. The two conditions with lower Aad-nc ~ Ab values were those in which nonclaim elements were sufficiently positive to preclude any adverse impact (liked music) or message claims were potent enough to minimize nonclaim effects (strong arguments). Based on a reviewer's recommendation, a four-group analysis was also conducted, in which differences in model parameters and fit were examined for the conditions defined by the music and exposure factors. (The argument factor was omitted because of its relatively weaker effect in the MANOVA and in the interest of maintaining adequate cell sizes for the multigroup analysis.) The results revealed an invadant pattern across the three-exposure groups, regardless of the presence of liked or disliked background music. Parameter values and significance levels in both instances varied little from those shown in Table 4 for the three-exposure condition. Among subjects receiving only one exposure, however, significant variation was observed between the liked- and disliked-music conditions, consistent with the results shown for the background music variable in Table 4. It would thus appear that the influence of background music on CPM is exerted primarily in the initial exposure to an ad. Such an inference must be treated with caution, however, in view of the statistical nonsignificance of the exposure/music interaction in the MANOVA.

CONTRIBUTIONS AND LIMITATIONS Most directly, this research makes the following conlributions, consistent with its objectives: (1) it establishes

13

involvement and liking as program variables warranting attention by advertisers and researchers in their effort to understand contextual influences on consumer ad processing; (2) it specifies the mechanism by which they influence Aad--that is, by enhancing consumers' CPM; and (3) it reveals that program-enhanced processing motivation benefits both Aad-c and Aad-nc. By way of replication, the results also confirm Murry, Lastovicka, and Singh's (1992) finding that PL, rather than program-induced feelings or moods, accounts for the affective effects of the program environment on Aad and reinforce their claim that program influence on brand attitude is mediated by Aad. Finally, the analysis of potential moderators suggests that program context influences on CPM and Aad can be managed to some extent by appropriate claim strength, executional, and exposure strategies. The demonstration of PI propensity to enhance rather than reduce commercial processing in a typical programming environment (as opposed to the extreme PI manipulations commonly employed in prior studies) is good news for advertisers. It does not obviate the possibility of attention and elaboration decrements under extreme PI conditions. (If the nonlinear PI-ad elaboration function we espoused earlier is correct, that result is as valid as the enhancement effect observed in this research.) But it suggests that in the more moderate range of PI values that probably characterize a disproportionately large share of programming time in broadcast media, PI stands to benefit rather than to interfere with the active processing of advertising messages. The fact that the two key program variables (involvement and liking) were measured and not manipulated imposes a design limitation. The stimulus program clearly worked well in terms of its ability to induce sufficient variability in listener liking and involvement to render the hypothesized relationships observable. Still, with no systematic difference at the stimulus level, it becomes difficult to rule out intersubject response differences. Accordingly, the research should be considered exploratory. Likewise, one cannot be certain that other program types (e.g., sports events, fine arts programs, different music formats), a different medium (e.g., television), or different test products (e.g., those more cognitively or affectively involving) or ad formats (e.g., slice of life, testimonial) would not have yielded different results. Future replication efforts employing pretests and manipulations of different program, media, product, and ad types are needed to establish the generalizability and boundary conditions of these results. This research demonstrated program context effects on Aad at commonly encountered levels of PI. However, the design does not allow its empirical findings to speak to the effects at more extreme levels. Manipulations of PI at levels capturing the continuum of possible values (very low to very high) would allow a test of the premise of a nonlinear PI-ad processing motivation relationship and its attendant implications for Aad. As a cautionary note, the research undertaking such a comprehensive investigation of the PI continuum would need to employ a different analytical method than the LISREL modeling used in this

14

JOURNALOF THE ACADEMYOF MARKETINGSCIENCE

WINTER 1994

study because of the nonlinearity assumed to exist in the upper- and lower-most portions of the distribution. The potential exists for shared-method bias, in view of the use of a c o m m o n attitude scale to assess PL, Aad-nc, and Ab. It would not appear to be exerting too great an influence, however, because the relationship where one would expect any such bias to be most prevalent (Aad- nc --->Ab) due to c o m m o n measures was consistently weaker than the alternative path (Aad-c --->Ab), which did not have c o m m o n measures. A theoretical construct warranting additional research from a program context perspective is processing motivation. Its importance as a mediator o f program context effects is clear from the results o f this study. However, the exclusive use of an advertising message that was essentially informational in nature limits the confidence one can have in generalizing the observed results to alternative executional styles. It is possible, for example, that the processing o f fear appeals or other affectively oriented approaches may respond differently to program context variables. In that regard, executional style may serve as another moderator of program context effects that are not captured by this study. Furthermore, the moderator variable analysis suggests that the CPM construct may not be strictly unidimensionah A plausible explanation of the results of this exploratory work is that PI affects processing motivation by boosting attention levels, whereas PL does so through a generalized favorability effect. As noted by a reviewer, these two effects could operate quite differently (perhaps even in opposite directions), depending on whether the message would benefit from greater attention or from less critical appraisal. Subsequent research should thus focus on the antecedents or dimensions of the motivational mediator of program context effects. Given the sparseness o f research into the impact o f the program environment on consumer ad response, most research to date has focused on the identification of relevant program variables and the observation of their impact on c o m m o n response measures. As seen in the exploratory analysis o f possible moderating variables, there is a need to expand that focus in a strategic direction. In addition to replicating the message claim, background music, and number-of-exposure effects investigated in this study, consideration should be given to interactions between these variables, as well as to variations in executional style, copy, and media-buying strategies that may serve as catalysts for, override, or otherwise shape program effects on Aad.

also shown that regular brushing with Hygent aids in the prevention of gum disease. And it costs less than competing brands of toothbrushes! So retire that old toothbrush and protect your family's dental health with Hygent.

APPENDIX Advertising Copy Strong ArgumentsVersion Time to retire that old, worn-out toothbrush? Try new Hygent, the toothbrush that allows you to choose from four varieties of bristle hardness, from soft to ultrafirm. Hygent retains its bristle shape and strength longer than competing brands, to ensure maximum brushing comfort and effectiveness. Clinical tests have

Weak Arguments Version Time to retire that old, worn-out toothbrush? Try new Hygent, the toothbrush that captured the European market. Hygent comes in your choice of five designer colors, each with multicolored bristles. What's more, Hygent is designed to prevent the accumulation of toothpaste at its base. And its bristles are 100% nylon. So liven up the aftermeal brushing chore with a splash of color. Retire that old toothbrush and brush with Hygent--the one the Europeans love!

ACKNOWLEDGMENTS The authors acknowledge financial s u p p o r t b y w a y of a summer research fellowship from the State University of New York at Buffalo School of Management and extend a p p r e c i a t i o n to W B F O R a d i o , S u n g - U k Yang, a n d Cassandra Wilson for assistance in stimulus development and data collection, and to the three reviewers for their valued suggestions.

REFERENCES Anand, Punam and Brian Sternthal. 1992. "The Effects of Program Involvement and Ease of Message Counterarguing on Advertising Persuasiveness."Journal of Consumer Psychology 1 (3): 225-238. Anderson, James C. and David W. Gerbing. 1988. "Structural Equation Modeling in Practice: A Review and Recommended Two-Step Approach." Psychological Bulletin 103 (May): 411-423. Batra, Rajeev and Douglas M. Stayman. 1990. "The Role of Mood in Advertising Effectiveness."Journal of Consumer Research 17 (September): 203-214. Brown, Steven P. and Douglas M. Stayman. 1992. "Antecedents and Consequences of Attitude toward the Ad: A Meta-analysis." Journal of Consumer Research 19 (June): 34-51. Bryant, Jennings and Paul W. Comisky. 1978, "The Effect of Positioning a Message within Differentially Cognitively Involving Portions of a Television Segment on Recall of the Message." Human Communication Research 5 (Fall): 63-75. Forgas, Joseph P., Denis K. Burnham, and Cani~lina Trimboli. 1988. "Mood, Memory and Social Judgments in Children." Journal of Personality and Social Psychology 54 (April): 697-703. Goldberg, Marvin E. and Gerald J. Goru. 1987. "Happy and Sad TV Programs: How They Affect Reactions to Commercials." Journal of Consumer Research 14 (December): 387-403. Isen, Alice M. 1984. "The Influence of Positive Affect on Decision Making and Cognitive Organization." In Advances in Consumer Research. Ed. ThomasC. Kinnear.Provo, UT: Association for Consumer Research, 534-537. Jacoby, Jacob and Wayne D. Hoyer. 1989. "The Comprehension/ Miscomprehension of Print Communication: Selected Findings." Journal of Consumer Research 15 (March): 433-443. J~Sreskog, Karl G. and Dag Strbom. 1989. LISREL 7: A Guide to the Program and Applications. Second Edition. Chicago: SPSS, Inc. Kennedy, John R. 1971. "How Program Environment Affects TV Commercials." Journal of Advertising Research 11 (February): 33-38.

Lord et al. / ATTITUDE TOWARD RADIO COMMERCIALS

Krugman, Herbert E. 1983. "Television Program Interest and Commercial Interruption: Are Commercials on Interesting Programs Less Effective?" Journal of Advertising Research 23 (February/March): 21-23. Lord, Kenneth R. and Robert E. Bumkrant. 1988a. "Commercial Processing Efficiency: A Program Elaboration Model." In AMA Educators' Proceedings: Efficiency and Effectiveness in Marketing. Eds. Gary Frazier, Charles Ingene, David Aaker, Avijit Ghosh, Tom Kinnear, Sidney Levy, Richard Staelin, and John Summers. Chicago: American Marketing Association, 112-116, Lord, Kenneth R. and Robert E. Burnkrant. 1988b. "Television Program Elaboration Effects on Commercial Processing." In Advances in Consumer Research. Ed. Michael J. Houston. Provo, UT: Association for Consumer Research, 213-218. Lord, Kenneth R. and Robert E. Burnkrant. 1991. "Program Context Effects on Commercial Processing." In Advances in Information Processing in Organizations. Eds. James R. Meindl, Robert L. Cardy, and Sheila M. Puffer. Greenwich, CT: JAI, 19-30. Lord, Kenneth R. and Robert E. Burnkrant. 1993. "Attention versus Distraction: The Interactive Effect of Program Involvement and Attentional Devices on Commercial Processing." Journal of Advertising 22 (March): 47-60. Lord, Kenneth R., Robert E. Bumkrant, and Robert Owen. 1989. "An Experimental Comparison of Self-Report and Response Time Measures of Consumer Information Processing." In AMA Educators' Proceedings: Enhancing Knowledge Development in Marketing. Eds. Paul Bloom, Russ Winer, Harold H. Kassarjian, Debra L. Seammon, Bart Weitz, Robert Spekman, Vijay Mahajan, and Michael Levy. Chicago: American Marketing Association, 196-200. Lutz, Richard J., Scott B. MacKenzie, and George E. Belch. 1983. "Attitude toward the Ad as a Mediator of Advertising Effectiveness: Determinants and Consequences." In Advances in Consumer Research. Eds. Richard P. Bagozzi and Alice M. Tybout. Ann Arbor, MI: Association for Consumer Research, 532-539. MacKenzie, Scott B. and Richard J. Lutz. 1989. "An Empirical Examination of the Structural Antecedents of Attitude toward the Ad in an Advertising Pretesting Context." Journal of Marketing 53 (April): 48-65. Mackie, Diane M. and Lelia T. Worth. 1989. "Processing Deficits and the Mediation of Positive Affect in Persuasion." Journal of Personality and Social Psychology 57 (July): 27-40. Miniard, Paul W., Sunil Bhatla, Kenneth R. Lord, Peter R. Dickson, and H. Rat Unnava. 1991. "Picture-based Persuasion Processes and the Moderating Role of Involvement." Journal of Consumer Research 18 (June): 92-107. Miniard, Paul W., Sunil Bhatla, and Randall L. Rose. 1990. "On the Formation and Relationship of Ad and Brand Attitudes: An Experimental and Causal Analysis." Journal of Marketing Research 27 (August): 290-303. Murry, John P., John L. Lastovicka, and Surendra N. Singh. 1992. "Feeling and Liking Responses to Television Programs: An Examination of Two Explanations for Media-Context Effects.'" Journal of Consumer Research 18 (March): 441-451. Pavelehak, Mark A., John A. Antil, and James M. Munch. 1988. "The Super Bowl: An Investigation into the Relationship among Program Context, Emotional Experience, and Ad Recall." Journal of Consumer Research 15 (December): 260-267.

15

Petty, Richard E., John T. Cacioppo, and David W. Schumann. 1983. "Central and Peripheral Routes to Advertising Effectiveness: The Moderating Role of Involvement." Journal of Consumer Research 10 (September): 135-146. Petty, Richard E., David W. Schumann, Steven A. Richman, and Alan J. Strathman. 1993. "Positive Mood and Persuasion: Different Roles for Affect under High- and Low-Elaboration Conditions." Journal of Personality and Social Psychology 64 (January): 5-20. Schumann, David W. 1986. "Program Impact on Attitude toward TV Commercial." In Proceedings of the Division of Consumer Psychology. Ed. Joel G. Saegert. Washington, DC: American Psychological Association, 67-73. Silk, Alvin J. and Terry Vavra. 1974. "The Influence of Advertising's Affective Qualities on Consumer Response." In Buyer/Consumer Information Processing. Eds. G. David Hughes and Michael L. Ray. Chapel Hill: University of North Carolina Press, 157-186. Soldow, Gary and Victor Principe. 1981. "Response to Commercials as a Function of Program Context." Journal of Advertising Research 21 (April): 59-65. Srull, Thomas K. 1983. "The Impact of Affective Reactions in Advertising on the Representation of Product Information in Memory." In Advances in Consumer Research. Eds. Richard P. Bagozzi and Alice M. Tybout. Ann Arbor, MI: Association for Consumer Research, 520-525. Watson, David, Lee Anna Clark, and Auke Tellegen. 1988. "Development and Validation of Brief Measures of Positive and Negative Affect: The PANAS Scales." Journal of Personality and Social Psychology 54 (June): 1063-1070. Worth, Lelia T. and Diane M. Mackie. 1987. "Cognitive Means of Positive Affect in Persuasion." Social Cognition 5 (1): 76-94. Yi, Youjae. 1990. "Cognitive and Affective Priming Effects of the Context for Print Advertisements." Journal of Advertising 19 (2): 40-48. Zaichkowsky, Judith L. 1985. "Measuring the Involvement Construct." Journal of Consumer Research 12 (December): 341-352.

ABOUT THE AUTHORS K e n n e t h R. L o r d received his Ph.D. from Ohio State University and is Assistant Professor at the State University of New York at Buffalo. His research has appeared in the Journal of Consumer Research, Journal of Advertising, and Journal of Economic Psychology. Myung-Soo Lee holds a Ph.D. from the State University of New York at Buffalo and is Assistant Professor of marketing at Baruch College, City University of New York. He has published previously in the Journal of Economic Psychology. Paul L. Sauer, Associate Professor of marketing at Canisius College, earned his Ph.D. at Ohio State University. He is the author of articles appearing in the Journal of Consumer Research, Journal of lnternational Consumer Marketing, Journal of Direct Marketing, and EDI Forum.