A Comparison of Three Interactive Television AD Formats
Summary (7 min read)
Introduction
- The digitization of television introduces new capabilities to television viewing experiences, including interactive formats for advertising (Cauberghe and De Pelsmacker 2006).
- Because the interactivity is so simple, the accompanying messages fit easily on banners superimposed over the regular ad, which plays out normally underneath them.
- Again, this format is widely used on Sky in the United Kingdom.
- Intuitively, interactors should be more interested in the product but after interacting also be more aware of the advertised brand, with more favorable attitudes and intentions toward it, than non-interactors.
- If interaction with iTV ads has positive effects on awareness and persuasion, iTV ads could be designed and tested to maximize these effects and thus occupy an even more important role in the advertising mix.
LITERATURE REVIEW
- Cauberghe and De Pelsmacker (2006, p. 23) define interactive digital television as "a group of technologies that gives users the possibility to take control over their TV experience, enabling interactivity with the content.".
- The key term in this definition is "interactivity.".
- The authors develop seven research questions about what drives viewers to interact with iTV ads and what effects might result from interactivity.
Product Category Involvement
- For advertisers, one of the most useful aspects of iTV ads is their ability to "cherry pick" the viewers most interested in buying the advertised product.
- This capability would spread the potential benefits of interacting to those viewers who could be most affected by it, that is, those not already sold on the advertised brand by information they already have.
- These findings suggest that though there is undoubtedly a positive correlation between product category involvement and the extent of interactivity, minimal interactivity, such as pressing a button on a TV remote control, might occur at a low threshold of involvement.
Effects of Interactivity on Awareness and Persuasion
- Research into interactivity and its effects has been plagued by vagueness and inconsistency in the definition of what, exactly, "interactivity" is (Bucy and Tao 2007; Rafaeli and Ariel 2007).
- Ratings of perceived interactivity appear to measure the distance between the current interaction and the "gold standard" for interactivity: face-to-face conversation (Rafaeli and Ariel 2007; cf. Bucy and Tao 2007).
- Like a persuasive salesperson, interactive stimuli can identify and answer objections, increase the expected value of desired outcomes, bolster the customer's belief the outcomes are possible, and adapt goals to the stage of the behavior-change process (Cassell, Jackson, and Cheuvront 1998).
- In their third research question, the authors consider whether this new rule of thumb applies across iTV ads generally: RQ3: Generating more thoughts about the ad is unlikely to encourage buying if all these thoughts are negative.
- Usually, an excess of positive thoughts is summarized by a positive attitude toward the ad; in the absence of any prior information about the brand, a positive attitude toward the ad generally is predictive of a favorable brand attitude and purchase intentions (Brown and Stayman 1992; MacKenzie and Lutz 1989).
Potential Differences Between iTV Ad Formats
- As well as testing for a generally positive effect of interaction with iTV ads on awareness and persuasion, the authors are interested in testing the relative effectiveness of the three main iTV ad formats.
- But at higher levels of available interactivity, the advantages of access to the right pieces of information may come at the cost of more time and effort (Rogers 1986), as well as the need to split resources across two tasks: comprehension and navigation (Yeung, Jin, and Sweller 1997).
- First, the DAL experience is not "as engaging and genuinely interactive as web advertising" (McLachlan 2009, p. 28).
- An excess of negative over positive thoughts will generate a negative attitude to the ad and therefore a less favorable brand attitude and purchase intentions, especially if the brand is unfamiliar (Brown and Stayman 1992), as all the brands the authors test are.
METHOD
- The authors use a controlled experiment to test the effects of interaction with three currently employed iTV ad formats, using an audience panel recruited through newspaper advertisements and direct mail from the general public in an Australian city.
- This balance of interests helps ensure the independence of the research.
SAMPLE
- This study was conducted in Australia to take advantage of a well-equipped audience research laboratory with eight years of experience in developing and testing interactive TV applications.
- But Australia is culturally similar to the United States (e.g., on Hofstede's [1980] individualism-collectivism index, Australia scores 90 [#2], and the United States scores 91 [#1]), and the main language in both countries is English.
- They were told that because the show had been recorded in the United States, it included U.S. ads in the ad breaks.
- Any effects the authors observe cannot be explained by prior exposure (Campbell and Keller 2003).
- Participants were invited to undertake a one-hour study in return for a AUD$20 department store voucher.
EXPERIMENTAL DESIGN
- For all three iTV ad formats, the direct response offer (the "call to action," or CTA) is a banner ad superimposed over the regular ad.
- On exiting the DAL, interactors returned to where they had left the TV content, just like viewers in the telescopic condition, which reproduced delayed viewing with a DVR.
- Participants in all four conditions saw three test ads: two product test ads, one for a high- and one for a low-ticket product (selection and order counterbalanced), and a third test ad for the TV program.
- The two product test ads always appeared in the middle position in the second and third ad breaks.
PROCEDURE
- All participants viewed the content in individual viewing labs designed to encourage natural viewing.
- Each lab had a regular TV set, a comfortable chair, pictures on the wall, and potted plants.
- They first saw a standard set of video instructions: "Colored buttons on the screen can be selected with the corresponding color button on your remote control.
- Even the control condition participants had to use their remote controls to advance through these instructions and vote electronically at the end of the show, but otherwise, interaction was not forced.
- After the session, participants completed a paper-and-pencil questionnaire in another room, were thanked and compensated, and, if they consented, were phoned the next day to measure their dayafter recall (333 [60%] consented).
Thought Listings
- The posttest survey asked about four of the ads participants had seen (3 test, 1 filler, except for the telescopic condition, which had 2 test, 2 fillers).
- Before answering any other questions on the questionnaire, participants listed all the thoughts they had while viewing these ads, using a separate line for each thought, which they self-coded as positive (+), negative (-), or neutral (N) (Rossiter and Percy 1997; agreement between self-coding and judge coding is usually high, such as 98% in Petty et al. 1993).
- Total thoughts refers to the line count, and net positive thoughts is the number of positive thoughts minus the number of negative thoughts.
Weighted Purchase Intention
- In Table 2, the authors describe how they frame and weight the 11-point Juster (1966) purchase intention (PI) scale to predict purchasing, depending on whether the product is a highticket, planned purchase or a low-ticket, regular purchase.
- The authors also reveal how they estimated the percentage of buyers per cell (Rossiter and Percy 1997).
- For these data, both estimates of purchase incidence are practically identical, which suggests their weighted PI results are not affected by distributional anomalies (e.g., bimodal distributions).
Attitudes
- The questionnaire went on to measure two attitudes, which the authors expected would mediate the effects of the number and valence of thoughts generated by the ad downstream to weighted PI.
- The measure of attitude toward the brand (Ab) uses the mean of four 7- point semantic differential items anchored by bad-good, dislike quite a lot-like quite a lot, unpleasant-pleasant, and poor quality-good quality (Gardner 1985; α = .93 to .96).
- Participants telephoned the next day indicated whether they could recall the ad (Brown 1985).
- To measure the product category involvement levels for several product categories, including the four represented by the five test brands, the authors turn to the sign-up survey that participants completed when they joined the audience panel, an average of 10 days prior to participating in the experiment.
- The authors also measure demographics according to the audience panel sign-up survey: age (date of birth), gender, education level, and occupation.
ANALYSIS
- The authors created a repeated measures data set with one row for each test brand.
- In their telescopic condition, the authors created a within-subjects repeated-ad condition by substituting, in the "holes" left in the second and third breaks by offering one instead of three interactive ads, the second and third repeats of a second test ad, first seen in the first ad break.
- The number of data points for each dependent variable varies, however, because of missing data (e.g., "don't know" responses, not being available for a DAR phone call).
Controlling for Selection Bias
- To test their research questions, the authors must identify how much of the differences between interactors and non-interactors is due to self-selection by already-persuaded participants, and how much might be due to the effects of the interaction.
- With the data from their control group, which saw regular, noninteractive TV ads, the authors can determine whether interaction with iTV ads does no more than select out the high-interest consumers from any group.
- If that were the case, the data from their interactors would be identical to data from an equivalent proportion of the control sample with higher product category involvement, and any differences the authors observe between interactors and non-interactors would be due entirely to this truncating effect of self-selection (Greene 2008).
- If interaction generates effects beyond those observed in the top 40% of the control group, it strongly suggests that interaction has persuasive effects beyond self-selection.
Descriptive Statistics
- The authors list, in Table 3, the descriptive results for their key dependent measures across the four experimental conditions.
- Consistent with the effects of self-selection, the high-involvement controls (the top 40%) provide a closer comparison to interactors in the iTV ad conditions.
- Finally, the authors use Table 5 to list the means, standard deviations, and correlations across all the measures they use to test their seven research questions.
- Net positive thoughts have a positive correlation with Aad, and Aad has a positive correlation with Ab, which has a positive correlation with weighted PI.
- Superscript letters indicate significantly different comparisons (Tukey HSD tests) (p < .05).
Research Question 1
- With RQ1, the authors investigate whether iTV ads can generate interaction from viewers who are not highly involved with the advertised product category.
- The data indicate the answer to this question is yes.
- When the authors divide the three iTV ad format conditions into high and low prior involvement groups using median splits, the proportion of interactors is identical in the low- and high-involvement groups for all three formats (χ2(5) = 3.30, p = .654).
- Participants who saw DAL iTV ads could explore them for as long as they wanted, and time-in-the-DAL correlates significantly with prior product category involvement (r(158) = .39, p < .001).
Research Question 2
- In RQ2, the authors asked whether interaction with iTV ads can increase thinking about the ad and ad recall.
- Interactors generate more thoughts than non-interactors (Table 4), but this effect might be due simply to self-selection, because the authors find no difference between interactors and the top 40% of the control sample (or, for that matter, the bottom 60% or the control sample as a whole; p = .626).
- It is more challenging, however, to use self-selection to explain differences between interactors and non-interactors with low product category involvement prior to this study.
Research Question 4
- With RQ4, the authors ask whether interactions with iTV ads can increase net positive thoughts about the ad.
- Again, the answer is tentatively positive, because the increase associated with interaction could be due to self-selection.
- Interaction also has a positive effect on low-involvement viewers of iTV ads, which cannot be explained well by selfselection.
Research Question 5
- In response to RQ5, about whether interaction with iTV ads can increase the probability of purchasing the advertised brand, the authors find a positive response, though again, perhaps no more than could be explained by self-selection.
- Again, the authors note the difference between low-involvement interactors and non-interactors, which they cannot explain easily with self-selection.
Research Question 6
- Thus far, their results indicate no general positive effects of iTV ads that the authors cannot explain with self-selection, though it is difficult to use self-selection to explain the differences between interactors and non-interactors with low product involvement prior to the study.
- In RQ6, the authors ask whether interactors with DAL iTV ads might generate more thoughts about the ad, as well as have higher levels of ad recall, more net positive thoughts, and a higher probability of buying the advertised brand, compared to interactors who view the two other iTV ad formats.
- The DAL interactors do not generate more total thoughts than interactors with the other two formats, though they are more likely to recall the ad than impulse interactors (Table 4) and the control sample as a whole (χ2(1, N = 219) = 8.30, p = .004).
- This positive effect on DAR may be due to self-selection, in that it is not significantly higher than the score for the top 40% of the control group.
Research Question 7
- Finally, RQ7 asked whether interactors with telescopic iTV ads might generate more thoughts about the ad, as well as have higher levels of ad recall, more net positive thoughts, and a higher probability of buying the advertised brand than interactors with impulse iTV ads.
- Telescopic interactors generate more net positive thoughts than the control sample as a whole (p = .007).
- Moreover, the telescopic format is the only one to yield results that cannot be explained by self-selection.
- Telescopic interactors exhibit higher levels of DAR and weighted PI than the top 40% of the control group.
DISCUSSION
- The results of this study suggest that iTV ads can have effects that go beyond direct response: iTV ads can be persuasive as well as selective.
- Responses to these ads qualify leads from consumers who are highly interested in the product category, but iTV ads can also generate interactions, prior to response, from consumers less interested in the category, and increase their interest in buying the advertised brand.
- Many of their results may reflect self-selection effects, with one significant exception.
- Rather, DAL interactivity generates more negative thoughts than the other two formats, perhaps because the interactivity that their DAL ads offer promises more than it could deliver.
- Pressing navigation buttons on the remote was "too much work," in that these viewers preferred to watch rather than "read TV.".
Implications for Advertisers
- The authors results suggest that iTV ads can generate leads and build purchase intentions, just as online banner ads can have branding effects beyond click-through rates (Hollis 2005).
- The authors confirm prior findings that one interaction with a DAL iTV ad equals three repeat exposures to a regular ad, in terms of generating awareness (Bellman, Pribudi, and Varan 2004) and extend this new rule of thumb for media buyers to telescopic ads.
- But if the production of the long-form video is planned beforehand, it may not add much to the budget for a standard 30-second commercial (e.g., it could consist of "out-takes" which would otherwise end up on the cutting room floor).
- Case studies show that DAL ads have been very successful (Sky Media 2009); additional research to compare progressive levels of DAL information content may find that advertisers can use the highly customizable DAL format to deliver precisely targeted and highly persuasive messages.
- Microsites on IPTV, which users interact with through a mouse, may be especially effective (Loughney, Eichholz, and Hagger 2008).
Limitations and Suggestions for Further Research
- This exploratory study contains several limitations that further research could address.
- Second, the authors generated interactivity among people with low product category involvement, but this finding might be due to characteristics of their study, such as demand effects or random chance; it thus needs further replication.
- High-involvement niche programs like home improvement shows might increase the inclination to interact with relevant product ads.
- The results of this exploratory study produce some limited preliminary findings that could help inform research, especially the suggestion that interacting with iTV ads can enhance awareness and purchase intentions as well as deliver an addressed response.
- More research is needed, as the authors are unable to draw firm conclusions from this study.
ABOUT THE AUTHORS
- Steven Bellman (Ph.D., University of New South Wales) is an Associate Professor and Deputy Director of the Interactive Television Research Institute at Murdoch University in Perth, Western Australia.
- Anika Schweda (Ph.D., Murdoch University) is a former ITRI Post Doctoral Research Fellow at the Interactive Television Research Institute and currently divides her time between writing papers like this one and looking after baby twins.
- Schweda@itri.tv Duane Varan (Ph.D., University of Texas at Austin) is the Director of the Interactive Television Research Institute and holds the inaugural Chair in New Media at Murdoch University, also known as Email.
- He also oversees Beyond :30, a collaborative industry research project focused on understanding the interactive viewer and exploring emerging models for TV advertising, and is the Executive Director and Chief Research Officer for The Disney Media & Advertising Lab in Austin, Texas.
ACKNOWLEDGMENT
- This research was supported by the members of the Beyond :30 consortium and the Australasian CRC for Interaction Design, established and supported under the Cooperative Research Centres Program through the Australian Government's Department of Education, Science and Training.
- The authors thank the editors and three anonymous reviewers for their helpful comments.
Did you find this useful? Give us your feedback
Citations
381 citations
197 citations
178 citations
Cites background from "A Comparison of Three Interactive T..."
...Bellman, Schweda, and Varan (2009) compared advertising effectiveness across three screen types: TV (35 in.), personal computer (10 in.), and iPod (2 in.)....
[...]
176 citations
References
24 citations
20 citations
19 citations
17 citations
"A Comparison of Three Interactive T..." refers background in this paper
...Total thoughts refers to the line count, and net positive thoughts is the number of positive thoughts minus the number of negative thoughts....
[...]
17 citations
"A Comparison of Three Interactive T..." refers background in this paper
...Levy and Nebenzahl (2006, 2008) repeatedly find that higher product category involvement is positively associated with interacting with iTV ads, though they measure product involvement after the interaction, which implies it could have been an effect rather than a cause of interactivity....
[...]
...…advertising services are also likely to feature on emerging platforms such as IPTV (Loughney, Eichholz, and Hagger 2008; Schechner and Kumar 2009) and mobile phones (Nasco and Bruner 2007); they are already present on some video-ondemand (VOD) and digital video recorder (DVR) services (Manly…...
[...]
Related Papers (5)
Frequently Asked Questions (2)
Q2. What have the authors stated for future works in "A comparison of three interactive television ad formats" ?
This exploratory study contains several limitations that further research could address. Third, the additional measures used in further research should include process variables, such as perceived interactivity, which the authors assume increases in the presence of interactive opportunities but do not measure directly ( Tremayne 2005 ). Fourth, though the authors use rigorous controls to rule out alternative explanations for their findings, such as primacy/recency effects, unequal offers across ad models, or differences in demographics, iTV is still a new phenomenon in Australia, and they can not rule out novelty effects. The authors also can not ignore the possibility that cultural factors, such as different preferences for the products advertised or varying experience with certain technologies ( e. g., Teletext ; Schweda, Bellman, and Varan 2005 ), may influence their results.