The Effects of Program Involvement and Commercial Position on Reactions to Embedded Commercials

James H. Watt, University of Connecticut
Keith S. Coulter, Eastern Connecticut State University
Erica K. Wiegel, University of Connecticut
Srinivas Kowta, Yansong Jiang, University of Connecticut, University of Connecticut
ABSTRACT - The present research utilizes a broadcast media setting to examine the manner in which two distinct measures of program "involvement" (i.e., self-reported vs. secondary task response latency) and commercial pod position influence attitude toward the ad (Aad), commercial recall, and commercial attention. Results indicate that program involvement main effects may vary depending upon how the construct is operationalized. Both commercial attention and commercial recall are found to vary depending upon pod position.
[ to cite ]:
James H. Watt, Keith S. Coulter, Erica K. Wiegel, and Srinivas Kowta, Yansong Jiang (1998) ,"The Effects of Program Involvement and Commercial Position on Reactions to Embedded Commercials", in NA - Advances in Consumer Research Volume 25, eds. Joseph W. Alba & J. Wesley Hutchinson, Provo, UT : Association for Consumer Research, Pages: 492-498.

Advances in Consumer Research Volume 25, 1998      Pages 492-498

THE EFFECTS OF PROGRAM INVOLVEMENT AND COMMERCIAL POSITION ON REACTIONS TO EMBEDDED COMMERCIALS

James H. Watt, University of Connecticut

Keith S. Coulter, Eastern Connecticut State University

Erica K. Wiegel, University of Connecticut

Srinivas Kowta, University of Connecticut

Yansong Jiang, University of Connecticut

[This research was partially supported by a grant from the National Association of Broadcasters.]

ABSTRACT -

The present research utilizes a broadcast media setting to examine the manner in which two distinct measures of program "involvement" (i.e., self-reported vs. secondary task response latency) and commercial pod position influence attitude toward the ad (Aad), commercial recall, and commercial attention. Results indicate that program involvement main effects may vary depending upon how the construct is operationalized. Both commercial attention and commercial recall are found to vary depending upon pod position.

Given the enormous sums of money that are spent on advertising each year, it is important that both broadcasters and marketing practitioners be acutely aware of the various factors influencing advertising effectiveness. The factors that may impact ad effectiveness include both characteristics of the ad (e.g., zooms, pans, abrupt changes in audio and video features, and celebrity endorsers)(Alwitt, et. al. 1980; Lang 1990) and characteristics of the context in which the ad is featured (e.g., viewer involvement with the program; commercial placement within a block (i.e., "pod") of commercials) (Murry, Lastovicka and Singh 1992). The present research utilizes a broadcast media setting to examine the manner in which two distinct measures of program "involvement" (.e., self-reported perceived personal relevance vs. measured attention, defined as the amount of cognitive processing) and commercial pod position influence attitude toward the commercial (Aad), commercial recall, and commercial attention. Involvement is investigated while controlling for many other factors which might produce confounding effects, such as commercial execution and prior exposure to commercials.

BACKGROUND AND THEORETICAL DEVELOPMENT

Program context involvement refers to the degree of commitment of cognitive resources to the contextual material (Soldow and Principe 1981). Research has demonstrated a negative relationship between program context involvement and commercial message processing (Bryant and Comiskey 1978). Other studies have found that television viewers demonstrated better recall of commercial content when they were less involved in the program (e.g., Norris and Colman 1992). Watt and Welch (1983) demonstrated the negative impact of complex program execution elements on verbal recall of program material by children. In explaining these effects, Lord and Burnkrandt (1988; 1993), who observed response times prior to and during advertising commercials, concluded that highly involving program content may induce viewers to commit a large proportion of available attention to its processing, thus reducing the efficiency with which they can encode and store information presented by a commercial. Thus, although high program involvement may activate attentional resources, it may hinder viewers’ ability to process an advertisement by directing those resources toward the contextual program stimuli at the expense of a shift in attentional focus toward the advertising message. The result is less ad-relevant thought (i.e., fewer ad cognitions), and therefore less attention to the commercial stimulus.

Coulter and Sewall (1995) demonstrated that the deleterious effects of program involvement may extend to measures of Aad. These researchers noted, however, that the aforementioned results run counter-intuitive to traditional advertising purchase behavior. Media buys are based upon broadcast ratings, and presumably ratings reflect viewer involvement. Indeed, the transfer of involvement (or attention) from programming to subsequent commercials forms the economic foundation of commercial broadcasting. One theoretical explanation for this transfer is based upon the concept of attention inertia (Alwitt, Anderson, Lorch and Levin 1980), which holds that the longer a viewer has been paying attention to a program, the more likely it is that the attention will persist. The implication is that high attention to the program should carry over to high attention to the commercials following programming segments (i.e., a positive program involvement/commercial attention relationship exists).

The reason for the aforementioned theoretical (and practical) discrepancies may be related to the manner in which program involvement is implicitly defined and operationalized. According to Muehling and Laczniak (1988), involvement has both "attentional" and "personal relevance" components. These two components may have different consequences in terms of commercial effectiveness. However, the pervasive use of post hoc paper-and-pencil program involvement measures among marketing researchers (e.g., Zaichkowsky’s (1985) Personal Involvement Inventory) may have failed to distinguish between the two components. While the presumed positive relationship between program "attention" and advertising effectiveness is the primary driver of ratings-driven media purchases, this effect may be diminished or even reversed when a "personal relevance" component is included in the involvement measure.

Researchers studying the processing of messages have defined "attention" as the amount of cognitive processing capacity allocated to decoding and storing an audio-visual message (Kahneman 1973; eiger and Reeves 1993). In essence, it is the "depth" of information processing. On the other hand, "personal relevance" gives direction to this processing (Meuhling and Laczniak 1988). It implies that type (as well as amount) of processing is an important determinant of personal involvement. While amount of processing may carry over from program to commercial (as indicated by traditional media buying habits), type of cognitive processing may not (e.g., Soldow and Principe 1981). Thus attentional involvement and personal relevance involvement can be independent of each other. Because the present study utilizes two distinct measures of program involvement that may tap different aspects or components of the construct, the following nondirectional hypotheses are proposed:

H1a: Program attentional involvement will have a non-null impact upon attention to the commercial.

H1b: Program personal relevance involvement will have a non-null impact upon attention to the commercial.

H2a: Program attentional involvement will have a non-null impact upon commercial recall.

H2b: Program personal relevance involvement will have a non-null impact upon commercial recall.

H3: Program personal relevance involvement will have a non-null impact upon attitude toward the commercial (Aad).

Commercial Position Within a Pod

Evidence suggests that the position of a commercial within a pod influences ad effectiveness, although research in this area has been sparse and results are often contradictory. For example, Webb and Ray (1979) found that 70% of subjects attended more to the first commercial than to subsequent commercials within a pod (i.e., exhibited a primacy effect). Brand name recall was also highest for the first commercial in 19 of 40 commercial breaks. Kim and Zhao (1993) found both primacy and recency effects in terms of subjects’ recall of SuperBowl commercials (i.e., recall was highest for commercials occurring at the beginning and ending of a pod). Wiegel and Daly (1994), using a commercial rotation strategy to control for commercial characteristics, found attention to be lowest during the second commercial and highest during the third commercial in a four commercial pod. Their explanation for this effect is that viewers may not accurately anticipate the start of a commercial pod, therefore attention is maintained at a fairly high level until the second commercial. The researchers also posit that viewers’ attention may peak during the third position of a pod because they may inaccurately predict when the program will resume.

The literature suggests that commercial position within a pod may affect both the attention paid to the commercial and, independent of attentional effects, the recall of the commercial. Variations in recall due to structural features of television programming, independent of levels of attention, have been routinely reported (cf. Watt and Welch, 1983). Because of the inconsistent results, the following nondirectional hypotheses are presented:

H4: Commercial attention will vary depending upon commercial position within a pod.

H5: Commercial recall will vary depending upon commercial position within a pod.

Program Involvement x Pod Position Interactions

As mentioned earlier, a number of studies have found a negative relationship between program involvement and measures of advertising effectiveness (e.g., Norris and Colman 1992). The theory behind this finding is that highly involving program content may induce viewers to commit a large proportion of available attention to its processing, thus reducing the efficiency with which they can encode and store information presented by a commercial. Although high program involvement may effectively activate attentional resources, it may hinder viewers' ability to process an advertisement by directing those resources toward the contextual program stimuli at the expense of a shift in attentional focus toward the advertising message. Presumably, increased proximity to the program material would exacerbate this debilitating effect. Therefore the negative relationship between program involvement and commercial attention, recall, or attitude should be greatest for the first commercial within a pod, effectively negating or reducing any pod position primacy effects. Recency effects should not be impacted. The opposite effect would be observed if a positive relationship exists between program involvement and ad effectiveness. In the latter case, primacy effects should be enhanced. The following nondirectional hypotheses are therefore proposed:

H6: Program involvement will moderate the effects of commercial position on commercial attention.

H7: Program involvement will moderate the effects of commercial position on commercial recall.

MEASUREMENT

Independent Variables

Attentional (Program) Involvement. As mentioned earlier, the attention component of program involvement has been defined as the amount of cognitive processing capacity allocated to decoding and storing an audio-visual message (Kahneman 1973). We utilized secondary task response latency (i.e., secondary task reaction time-"SRT") as a measure of processing capacity (i.e., attention) (Meadowcroft and Reeves 1989; Geiger and Reeves 1993). Subjects' primary task was to direct their attention to the television presentation. Subjects' secondary task was to press a hand-held button in response to a randomly occurring audio "probe" stimulus. Secondary task response latency was expected to vary directly with the amount of processing capacity allocated to the primary task.

The use of SRT as a measure of attention is based on Kahneman's (1973) attention allocation theory, which suggests that if a person is given two tasks to perform, there is a tradeoff between the amount of cognitive resources the person can allocate toward the two tasks. If the primary task is consuming low amounts of processing capacity, there will be sufficient capacity for the secondary task so that response latency times will be small (i.e., reaction times will be quicker). If the primary task is allocated large amounts of processing capacity, SRT's will be slower. Of course, increasing fatigue during programming might systematically increase reaction times, too. SRT fatigue was investigated in a prior study (Menelly,1991). Menelly presented 20 subjects with only the audio probes from one channel of a tape of commercials used in another study. The tape was approximately 25 minutes in duration, similar to the length of the test tapes used in this research. She found no significant trend in SRT's over the duration of the tape, indicating no detectable fatigue effect occurred. Based on this finding, we do not expect fatigue to confound the SRT measurements.

For each subject, a baseline reaction time (BRT) was computed by averaging 6 SRT values early in the program. Subsequent SRT values (in milliseconds) were subtracted from each subject's BRT to provide the final SRT measures used in our analyses. This procedure effectively removes between-subject variation in physical reaction times and general information processing capacity, by calibrating the attention measure to zero for each subject at the beginning of the program. Overall attention to the program was computed by averaging the two lead-in attention periods 30 seconds before each commercial pod (i.e., 12 attention probe SRT's).

Personal Relevance (Program) Involvement. For this measure, program personal relevancy involvement was defined as the self-reported orientation toward, and absorption in, the program material. Subjects provided ratings of their overall program involvement on three, 7-point semantic differential items ("How much attention did you pay to the program?" "How important was paying attention to the program to you?" and "How involved were you in watching the program?") adapted from Zaichkowsky's Personal Involvement Inventory (1985). The scale formed by the unweighted sum of the three items had a Cronbach's alpha of.80.

Commercial Position Within a Pod. In this experiment, each commercial pod had four slots, so commercial position was operationalized as the slot number within a pod in which a commercial appears (chronologically ordered as numbers 1-4).

Attitude toward the Commercial (Aad). Attitude toward the commercial was measured using a set of four, 7-point semantic differential items: positive/negative, good/bad, favorable/unfavorable, like/dislike (MacKenzie, Lutz, and Belch 1986). The scale(s) formed by the unweighted sum(s) of the four items for the eight commercials had an average Cronbach's alpha of .86.

Commercial Recall. Commercial recall was defined as the ability of the subject to produce unaided verbal responses describing the content of the ad. Recall of advertisement information was measured by a verbal protocol which asked subjects to list as many of the commercials as possible, and to write down as much information as they could about each ad. Three coders used these written protocols to dummy code (O=no recall; I=recall) brand name recall, product class recall, and message-related recall (i.e., slogans, sales messages). Average intercoder reliability for brand, product, and message was .87. Prior attempts at coding transcripts showed decreases in reliability occurred when attempts to code amounts of each subcategory of recall were made, so simple-dummy codes for each subcategory were used.

Self-reported commercial recognition was also measured by the response to a single prompting item: "Do you remember seeing a commercial about ." These prompted recognition items were presented after the unaided recall protocols were completed by subjects. This item was also dummy coded.

The final Recall measure was then formed by the product of the recognition dummy code, and the summation of the three recall dummy codes. It was reasoned that failure to recognize having seen a commercial, after prompting, was equivalent to zero recall. Prompted self-reported recognition without substantiation in the verbal protocols was also treated as no recall. The degree of recall indicated in the verbal protocols increased with each element (brand name; product class; message) provided by the subject.

Experimental Control Variables

Many variables may confound the relationship between involvement, program structure, and outcomes like commercial attention, recall and Aad. While the systematic effect of confounding variables can be eliminated by randomization or by fixing their levels (manipulative control), these strategies have negative consequences in terms of the statistical power (randomization) and external validity (manipulated control) of the study (Watt and van den Berg, 1995). This research used the strategy of measurement and statistical control of potentially confounding variables, a strategy that maximizes both statistical power and external validity (at the expense of increased complexity of data collection and statistical analysis). The variables used in statistical control are briefly described below:

Commercial Execution. ne script, characters, etc. used in a commercial message may affect the attention paid to the commercial and its subsequent recall. This was controlled by creating seven dummy codes representing the eight commercials. The set of dummy codes thus encapsulate all differences between commercials.

Commercial Pod Position within Program. Because of attention inertia and/or fatigue, systematic differences inattention can be expected according to the elapsed time from the beginning of the program. A dummy code for pod position was used to control for this effect.

Program/Commercial Affect Congruity. Emotional responses to commercials embedded in programming that is consistent or inconsistent in emotional tone have been shown to be involved in processing of the commercial message (Schumann and Thorson 1990).

Attitude Toward the Brand (Ab). Attitude toward the brand was measured using a set of four, 7-point semantic differential items: positive/negative, good/bad, favorable/unfavorable, like/ dislike (MacKenzie, Lutz, and Belch 1986). The scale(s) formed by the unweighted sum(s) of the four items for the eight commercials had an average Cronbach's alpha of .89.

METHOD

Experimental Stimuli

Off-the-air videotape recordings of two programs (Seinfeld; In the Heat of the Night) and eight commercials (involving cat food, a delivery service, boxer shorts, greeting cards, a war video, pharmaceuticals, a hospital, and milk) were edited to form the experimental stimuli (see Appendix A). Each tape began with a 3-minute practice period (an excerpt from the cable series Wings) to familiarize subjects with the secondary task procedure for measuring attention. For each of the two experimental programs, all eight commercials were placed in two commercial pods (four in Pod 1 and four in Pod 2). Commercials were rotated through all eight positions, so that 16 experimental tapes were created (8 versions of each of the 2 experimental programs). Each tape was approximately 20 minutes in length.

Audio tones for the secondary task attention measure were inserted in programs at random intervals that ranged between three and 15 seconds. For commercials, an audio probe was randomly placed within the first, second, and third 10-second division of each 30-second commercial. This insured SRT measurement near the beginning, middle, and end of each commercial. The attention measurements were made at the same time points in the commercial, regardless of the commercial's position in the experimental program segment. During the entire experimental session, subjects responded to approximately 120 audio probe tones. Secondary task response times were recorded in milliseconds by computer.

Procedure

Three hundred fifty three students from a major Northeastern University were recruited as subjects. Subjects were randomly assigned to view one of the 16 tape versions described above, and were instructed that the experiment would "measure their reactions to some television programming." They were also instructed on use of the push-button SRT device. After a three minute practice session, subjects viewed one of the experimental tapes. One hundred ninety-six of the subjects were given a written questionnaire which included the: 1) program involvement items, 2) free recall verbal protocol instructions, and 3) commercial recognition items (in that order). The remaining students completed only the SRT attention measurements for both program and commercials. For this reason, hypotheses involving only attentional processes have a base N of 353; while those involving the other variables have a base N of 196.

Analysis

Commercial attention (CA), recall (CR), and Aad are affected by factors other than those discussed above [i.e., Program Personal Relevance Involvement (PPRI), Program Attentional Involvement (PAI), and Commercial Position Within Pod (CPWP)]. For example, commercial pod position within a program (CPPP), commercial executional elements (CE), and attitude toward the brand (Ab) may affect attention, while the emotional congruity between program and commercial (PCAC) may affect both commercial attention (Goldberg and Gorn 1987) and Aad (Coulter and Sewall 1995). To control statistically for these potentially confounding variables, all hypotheses were tested within the context of a series of linear (regression) equations. Reported statistics are partial coefficients, with the effect of all other independent variables held constant. In addition, the research design requires that some hypotheses be tested as between subjects effects (for example, those involving program personal relevance), while other must be tested as within subjects effects, with corrections for repeated measures (for example, those involving commercial pod position)(Watt, et al., 1997). The specific equations utilized are outlined below:

Linear Between-Subject Hypotheses:

CA =PAI (H1a) + PPRI (H1b) + PCAC + Aad + Ab

CR =PAI (H2a) + PPRI (H2b) + Aad + Ab + CA

Aad=PPRI (H3) + PCAC

Linear Within-Subject Hypotheses:

CA =CPWP (H4) + PCAC + CPPP + CE + Aad + Ab

CR =CPWP (H5) + CA + Aad + Ab

Within Subject Interaction Hypotheses:

CA =(CPWP x PPRI)(H6) + (PPRI x PCAC)

CR =(CPWP x PPRI)(H7) + (PPRI x PCAC)

TABLE 1

ATTENTION TO PROGRAMMING AT TWO POINTS IN THE PROGRAM

RESULTS

Program Involvement and Commercial Attention Measures

For the SRT measure, there was no difference in overall mean attentional involvement with the two experimental programs [Mean for Seinfeld=19.46; Mean for Heat of the Night=21.50; t(346)=.31, p>.05]. Both means were significantly higher than the average baseline SRT measured 3 minutes into the program. However, attention varied significantly within programs at different time points (see Table 1). In the Seinfeld program, attention 30 seconds before the first commercial pod is not significantly different than attention at a baseline point 3 minutes after the beginning of the program. However, attention to the Heat of the Night program at the same point is significantly higher than it was at the baseline point in the program several minutes earlier, and it is significantly higher than attention to the Seinfeld program at a similar point [t(346)=3.41, P<00 1 ].

At a point after the first commercial pod and 30 seconds before a second pod, the attention to Seinfeld is significantly higher than it was in the baseline period at the beginning of the program and also at the 30-second lead-in period before the first commercial pod [t(173)=4.96, p<001]. It is not significantly higher than the attention to Heat of the Night at a similar point before the beginning of the second pod [t(346)=1.76, p=.08]. For the program personal relevance involvement measure, mean program involvement was significantly greater for the Seinfeld (x=17.7) than for the Heat of the Night (x=16.3) programs [t(193)=3.50, p<.001]. Commercials showed clear differences in ability to elicit viewer attention, regardless of pod or pod slot [F(7;353)=9.27, p<01]. Five of the eight commercials produced attention significantly higher than that produced by the average of the two programs (see Table 2) and seven of the eight differed significantly from the mean for all commercials.

Hypothesis Tests

Effects of Program Attentional Involvement on Commercial Attention. Hypothesis la predicted that program attentional involvement would be directly related to commercial attention. The SRT Attention measure was used to operationalize both program attentional involvement and commercial attention. This hypothesis was tested in two ways: by computing the Pearson correlation coefficient between subjects' Lead-in Attention measures 30 seconds before a commercial pod (program attentional involvement) and their Attention measures to both the first commercial in the pod and the average Attention for the pod; and by computing the regression coefficient between average program lead-in attention and commercial attention, controlling for program personal relevance involvement, commercial execution, commercial position in pod and in program, affect congruity, Aad, and Ab. Both analyses provided strong support for Hypothesis la (see Table 3). Although significant for the first commercial pod (combined programs r=.13, p<05), the attention carryover effect was much stronger in the second pod of the Seinfeld and Heat of the Night programs (combined programs r=.74, p<0001. The first pod of both programs showed only weak correlation of program attention with commercial attention, with the correlation for Heat of the Night failing to reach significance, while the second pod attention carryover was strong and significant for both programs. In the more controlled analysis, average program lead-in attention was also strongly related to average attention to all commercials (Beta=.70, 4193)=13.77, p<0001).

TABLE 2

COMMERCIAL ATTENTION COMPARED TO MEAN PROGRAM ATTENTION

TABLE 3

CORRELATIONS BETWEEN PROGRAM ATTENTIONAL INVOLVEMENT AND COMMERCIAL ATTENTION

Mean attention to all commercials was not related to the three-item questionnaire measure of program personal relevance involvement (Zero-order Pearson r=.07, n.s.; Beta =.06, t(193)=1.12, p>.05). Therefore, our findings do not support H1b. The relationship between program involvement and commercial attention differs, depending upon the manner in which program involvement is defined and operationalized.

Effects of Program Involvement on Recall and Aad Hypotheses 2a and 2b were tested using the regression equations outlined above. Neither the SRT measure of program attentional involvement (Beta= -.06, t(193)=.64, p>.05) nor self-report measure of program personal relevance involvement (Beta= -.03, t(193)=.48, p>.05) were related to average Recall of commercials. Therefore Hypotheses 2a and 2b are not supported. Self-reported program personal relevance involvement did have a strong influence on Aad, however, with stronger involvement related to more positive Aad values (Beta=.34, t(193)=4.81, p<.01). Therefore, H3 is supported.

Commercial Position Within a Pod. Hypothesis 4 predicted that commercial attention would vary depending upon position within a pod. This hypothesis was tested as a repeated measures within-subjects design, using the regression procedure outlined earlier, which controlled for commercial execution differences, commercial pod position, affect congruity, Aad and Ab. Our findings indicate a significant effect for commercial slot (F(3; 1553)=3.65, p<02). In particular, there is a strong recency effect. Significantly lower attention levels (i.e., quickest secondary reaction times) were found in the first commercial slot (mean SRT=4.23, compared to mean for all commercial slots, t(1555)=1.97, p<05), while the highest attention levels were found in the last commercial slot within a pod (mean SRT= 6.78, 41555)=3.15, p<002; see Table 4). Thus H4 is supported.

Hypothesis 5 predicted that commercial recall would vary depending on pod position. This hypothesis was also tested as a within-subjects design. Pod position significantly affected recall, even when attention to the commercial, position of the commercial pod in the program, affect congruity, Aad and Ab were controlled (F(3; 1 563)=29.42, p<.0001). Commercial recall was significantly greater for commercials in the first and last positions within a pod (see Table 4). Thus, in contrast to our commercial attention findings, both primacy and recency effects were observed in the case of commercial recall.

TABLE 4

COMMERCIAL ATTENTION AND RECALL FOR DIFFERENT POD POSITIONS

Program Involvement x Pod Position Interactions. Hypotheses 6 and 7 predicted that program personal relevance involvement would moderate the effects of commercial position on commercial attention (H6) and commercial recall (H7). Both interactions were found to be nonsignificant (F(3; 1555)=1.75, p>.05 for commercial attention; F(3;1555)=1.02, p>.05 for commercial recall).

SUMMARY AND DISCUSSION

Results indicated that the programs differ in their ability to elicit attention at different points in the program. Seinfeld shows a pattern of lower initial attention, followed by building attention levels. Heat of the Night shows higher initial attention that is more consistent over the duration of the program. This may be due to the differing nature of sitcoms and dramas. The sitcom spends the early portion of the program setting up the comedic premise, then exploits it later in the program, while the drama uses early action to capture attention and maintains a more consistent dramatic tension over the whole program. Since conventional paper-and-pencil (i.e., self-report) measures of program involvement fail to distinguish between different attention-eliciting points in a program, the use of these measures to predict commercial involvement or attention (even given an accurate knowledge of the direction of the relationship) may be tenuous at best.

Results indicated a positive relationship between program attentional involvement and commercial attention. The attention produced by the program immediately before the commercial pod (Lead-In Attention) was critical to producing attention to the commercials. This finding agrees with traditional ratings-based media purchase behavior. The self-reported program personal relevance involvement measure was unrelated to both commercial attention and commercial recall. Thus, the widely reported (e.g., Soldow and Principe 1981; Lord and Burnkrandt 1993) debilitating effects of program involvement on commercial message processing were not observed in this study. Further, a positive relationship was observed between program personal relevance involvement and Aad. This directly contradicts earlier findings (Coulter and Sewall 1995).

The ability of programming material to capture a viewer's attention has traditionally been perceived as a desirable trait by marketing practitioners; our research seems to support this premise. The commercial value of programming that engrosses the viewer is quite evident from this analysis. This value is frequently mentioned in the context of attracting viewers to make the choice to view a particular program. However, this analysis indicates the value extends to the actual viewing situation, after the choice to view has been made. It appears that the carryover effect may be intensified by an attention inertia process (Alwitt et.al. 1980), as the strong carryover correlations occurred after subjects had been watching programming for over 12 minutes.

The pattern of commercial attention associated with position within the pod clearly supports the idea of increased attention for the last pod position, possibly because of anticipation of the beginning of the following program segment. Similarly, average commercial recall scores demonstrate a recency effect, but the first pod position is also found to result in higher commercial recall. This latter finding underscores the importance of understanding the difference in attention vs. recall as measures of commercial effectiveness. The lack of a strong program attention/commercial recall relationship may have been due to the fact that the experiment was conducted under the conditions of forced attention. Therefore, the size of the relationship may have been depressed, as compared to a realistic viewing situation that permits much wider ranges of attention. A second reason that this relationship may have appeared smaller than one would expect has to do with the measure of recall used. This study used a written verbal recall protocol that is sensitive to individual verbal abilities and the motivation to produce a written report on the part of the subjects. The measurement is thus error-prone, and the error will reduce the size of the relationship.

Thus it is possible to call into question the idea that recall, as measured by verbal protocols, is the only good measure of commercial message outcome. Commercial attention, measured in this study as "depth of processing," may itself be an alternative to measuring subsequent recall, as the processing of the message may have an impact on subsequent persuasion processes beyond the encoding and storage of commercial content. Further research into this important area of investigation appears warranted.

APPENDIX A

FORMAT OF EACH TAPE VERSION

REFERENCES

Please contact the second author for a complete reference list.

Coulter, Keith S. and Murphy A. Sewall (1995), "The Effects of Editorial Context and Cognitive and Affective Moderators on Responses to Embedded Ads," Advances in Consumer Research, Frank Kardes and Mita Sujan, eds., Association for Consumer Research, October 1994.

Geiger, Seth and Reeves, Byron (1993), "We Interrupt this Program... Attention for Television Sequences," Human Communication Research, 19 (Fall), 368-387.

Kim, Tae Young and Zhao, Xin Shu (1993), "The Effect of Serial Position on TV Advertisement Recall: Evidence from Two Years of Super Bowl Advertising Data," Presented to the Advertising Division of the Association for Education in Journalism and Mass Communication, Kansas, MO.

Lord, Kenneth R. and Robert E. Burnkrandt (1993), "Attention Versus Distraction: The Interactive Effect of Program Involvement And Attentional Devices on Commercial Processing," Journal of Advertising, 2 (March) 47-60.

Meadowcroft, Jeanne and Byron Reeves (1989), "Influence of Story Schema Development on Children's Attention to Television, Communication Research, 16(3), 352-374.

Meuhling, Darrel D. and Russell N. Laczniak (1988), "Advertising's Immediate and Delayed Influence on Brand Attitudes: Considerations across Message- Involvement Levels," Journal of Advertising, 17 (4), 23-34.

Murdock, B.B. (1962), "The Serial Position Effect of Free Recall," Journal of Experimental Psychology, 64, 482-488.

Soldow, Gary F. and Victor Principe (1981), "Response to Commercials as a Function of Program Context, Journal of Advertising Research, 21 (2), 59-65.

Watt, James and Welch, Alicia (1983), "Effects of Static and Dynamic Complexity on Children's Attention and Recall of Televised Instruction," in J. Bryant and D.R. Anderson (Eds.) Children's Understanding of Television: Research on Attention and Comprehension, New York: Academic Press, 1-34.

Wiegel, Erica and Daly, Thomas (1994), "Celebrities, Commercial Position, and Liking of a Program: The Roles they Play in Viewers' Processing of Television Commercials," International Communication Association Conference, Sydney, Australia.

----------------------------------------