Health Articles

 

Cerge-ei.cz

Getting into Your Head(ache): Advertising Content for Simon P. Anderson,y Federico Cilibertoz and Jura Liaukonytex Marketing Science Institute Working Paper The over-the-counter analgesics market is characterized by substantial advertising expenditures by the top brands. Advertising-to-sales ratios in this industry range from20-50%, and are more than 7 times higher than the average for all industrial sectors.
We use a novel dataset where we code the product characteristics featured in adver- tisements. For comparative advertisements, we record the rival products mentioned aswell as the characteristics used for comparisons. We provide a comprehensive, …ve-yeardescription of the number and types of characteristics mentioned in ads as well as thead expenditures spent on emphasizing particular characteristics. The content analysishighlights the role of di¤erent types of advertising in enhancing product di¤erentiation.
We then consider some theoretical hypotheses and test them with our data. First, we …nd that leading brands will be targeted most. Second, the fraction of compara-tive advertising is highest for new brands, and those with low market shares. Third,comparative advertising targets rival brands rather than stablemates. Fourth, there isonly limited support for the idea that each characteristic will be advertised by onlyone product. However, we …nd that nationally advertised brands tend to specialize inwhich characteristics they advertise the most. Fifth, comparative advertising claimsare more likely for experience rather than credence attributes. Finally, we …nd thatusually comparative advertising targets products with active ingredients di¤erent fromthat of the attacking product.
Keywords: comparative advertising, advertising, content analysis, OTC pharma- JEL Classi…cation: D12 D43 L15 M37 We thank David Mick, Ross Rizley, Columbia Warren and Ken Wilbur for useful comments and gratefully acknowledge funding of the Marketing Science Institute under MSI Research Grant #4-1364.
yDepartment of Economics, University of Virginia, Charlottesville VA 22904, USA. [email protected] of Economics, University of Virginia, Charlottesville VA 22904, USA. [email protected] of Economics, University of Virginia, Charlottesville VA 22904, USA. [email protected] Approximately 84% of the United States population regularly uses over-the-counter (OTC) analgesics for headache, backache or other minor pain relief (Simmons Market Research, 2000). Drug stores, supermarkets and mass merchandisers sold more than $2 billion worth of such medicine in 2005. Selling pain relief is a huge business in the United States and around the world. What makes this market especially interesting to study is the signi…cant discrepancy between true di¤erentiation of almost homogeneous medical products and the perceived di¤erentiation among di¤erent brands. The gap is arguably generated by advertis- ing: the OTC analgesics market is characterized by substantial advertising expenditures by the top brands. Advertising-to-sales ratios in the OTC analgesics industry range anywhere from 20-50% and are more than 7 times higher than the average value for all industrial sec- tors.1 According to OTC analgesics industry practice, substantial media budgets are built to reach around 80% of the target audience over 4 times for an average 4 week period (The OTC Drug Seminar Series (2004)). Moreover, this market is characterized by a high prevalence of comparative advertising, where one brand explicitly mentions a competitor while making comparative claims about competing product features. Some brands spend as much as 30% of their revenue on comparative advertising.
Most of the previous empirical studies on comparative advertising have focused on cross- industry analysis, bundling together comparative advertising across di¤erent industries with diverse market structures (Chou et al (1987) and Harmon et al (1983)). To our knowledge no research has been done to analyze what television ads claim within an industry, and, more speci…cally, what product characteristics …rms tend to highlight and how much they spend on emphasizing particular attributes. This study also is the …rst to systematically examine the nature of comparative advertising claims within a speci…c industry.
In this paper we only code the objective content of advertisements as quanti…ed through their mentioning of speci…c characteristics and competitors. We recognize indeed that ad- vertising may persuade through other channels than pure information, and act on emotional factors. However, we have not attempted to code such e¤ects, for several reasons. First, the primary content of this advertising appears to be informative. Second, we want to …rst address (what is arguably) the purely informational content, along the lines of traditional content analysis: adding the subjective side is a very interesting topic for further extensions.
Third, given the subjective nature of the coding and the large sample size of ads, coding 1 The OTC analgesics products fall within the "Pharmaceutical Preparations" sector (SIC 2834), which has a 4.8% advertising-to-sales ratio. The average advertising-to-sales ratio of all industrial sectors is only3.2% (Schonfeld and Associates, 2005).
would also involve some noise, and necessitate frequent consistency checks.
There are a number of important managerial issues to which this research is relevant.
What methods are used to di¤erentiate seemingly quite homogeneous products? Do market leaders use comparative advertising and why? What types of ads and what types of claims do …rms make about their rival products? Can aggressive advertising be used to restore brand name value and market share following a negative exogenous news shock? These are some of the issues we try to address in this paper.
The paper is organized as follows. In the next section we review the literature on content analysis and comparative advertising. Section 3 provides useful U.S. OTC analgesics market background. In sections 4 and 5 we contrast the "true"and perceived product di¤erentiation existing in this market. The data and coding methodology are described in Section 6. Results of the advertising content analysis are presented in Section 7. In Section 8 we construct several hypotheses and test them against our data. Section 9 concludes. Finally, medical terminology relevant to our research is explained in the Appendix.
Literature Review The theoretical economics literature on comparative advertising is very limited, although marketing research documents the phenomenon and analyzes its e¤ectiveness thoroughly.
Much of this literature (see Bagwell (2007) for a recent survey) has concentrated on normative economic questions (e.g., do …rms reach the socially optimal number of consumers?). The …rst paper to address the issue of product heterogeneity by modeling consumer imperfect information about product characteristics was Anderson and Renault (2007). There are several theoretical attempts to model why comparative advertising arises in equilibrium and how it a¤ects consumer and …rm behavior. Barigozzi, Garella, and Peitz (2003) show that comparative advertising has a di¤erent signaling potential than generic advertising, since the comparative claim will be perceived by rational consumers to be more credible. Aluf and Shy (2001) model comparative advertising using a Hotelling-type model of product di¤erentiation as shifting the transport cost to the rival's product. Anderson and Renault (2007) show that if the product is low quality, the …rm will advertise detailed product information that enables consumers to determine their matches, while the …rm with the high quality product will not.
Shy (1995) argues that in the case of di¤erentiated products, comparative advertising informs consumers about the di¤erence between the brand they have purchased in the past and their ideal brand. In this case, it is likely that comparative advertising is meaningless for the inexperienced consumer because she would not comprehend an ad involving a comparison of The marketing literature indicates that consumers in behavioral studies pay more atten- tion to, and are generally more aware of, products after viewing comparative advertising relative to generic advertising (Grewal et al (1997)). Pechmann and Stewart (1990) suggest that 60% of all ads are indirectly comparative and 20% contain direct comparative claims.
Whether this implies more competition is not clear. Rose et al (1993) point out that it is di¢ cult to di¤erentiate between whether consumers are making better informed decisions or are simply more persuaded by comparative advertising.
In the marketing literature, "content analysis" compares content across cultures (e.g., Madden, Caballero, and Matsukubo, 1986), over time (e.g., Bruce L. Stern and Alan Resnik, 1991), across media (e.g., Avery M. Abernethy and George R. Franke, 1996), and across di¤erent regulatory regimes (e.g., Abernethy and Franke, 1998). The seminal paper by Resnik and Stern (1977) postulated 14 categories of "information cues". The information cues include price, quality, performance, availability, nutrition, warranties, etc. In a survey of other papers, Abernethy and Franke (1996) present the results for an average over 4 studies of U.S. television advertising and show that the mean number of cues was 1.06, with only 27.7% having two or more cues, and 37.5% having no cues.
The marketing studies mentioned above, however, do not address any comparative ad- vertising content, which is a signi…cant component in some advertisements. Our paper aims to use the methodology of content analysis extended to comparative advertising. In addi- tion, our research isolates a single industry, enabling us to consistently set product attribute categories and test which ones are most advertised in a comparative way.
Overview of the OTC Analgesics Industry The OTC analgesics market covers pain-relief medications with four major active chemical ingredients. These are aspirin, acetaminophen, ibuprofen and naproxen sodium. The nation- ally advertised brands for each of these segments are such familiar brand names as Tylenol (acetaminophen), Advil and Motrin (ibuprofen), Aleve (naproxen sodium), Bayer (aspirin or combination), Excedrin (acetaminophen or combination), Midol and Pamprin (varying formulas for menstrual pain relief).
The industry is highly concentrated - the top 6 brands account for approximately 71% of the dollar market share. Generic store brands account for another 26%. The main brands (with national advertising), their market shares (calculated from our data), their ingredients, and ownership are given in Table 1 below.
*average price of a representative 24 pill bottle of regular strength tablets.
ACT-Acetaminophen; IB-Ibuprofen; NS-Naproxen Sodium; ASP-Aspirin TA-Total Advertising; CA-Comparative Advertising Table 1. OTC Analgesic Brands and Market Shares, 2001-2005 data.
Although each of the drugs generally treat pain, fevers and headaches (hence implying that they are close substitutes), there are some di¤erences between analgesic types. While as- pirin, naproxen sodium and ibuprofen are non-steroidal anti-in‡ammatory drugs (NSAIDs), acetaminophen is not. In general, ibuprofen and naproxen are more potent pain relievers, i.e., they reduce more pain than the same dose of acetaminophen or aspirin. On the other hand, acetaminophen is considered to be the safest pain reliever because it does not block prostaglandins, and therefore does not cause any gastrointestinal (GI) bleeding. However, even though acetaminophen reduces pain and fever, it does nothing for in‡ammation. Addi- tionally, high doses of acetaminophen may damage the liver. Aspirin is the only pain reliever shown to reduce the risk of heart attack.
As we discuss in more detail below, all active ingredients deliver pain relief, but their e¤ectiveness and safety pro…les di¤er greatly. Choosing an OTC pain reliever involves bal- ancing e¤ectiveness for a particular ailment with its side e¤ects, but these are not necessarily known to the buying public.
"True" Di¤erentiation From a consumer's perspective, …nding simple relief for a headache or aching back is trou- blesome. The myriad of choices in the drug store aisle is enough to exacerbate the headache.
The product shelf is a maze of confusing labels, diverse pill forms and claims of body-part speci…c relief.
As noted above, almost all pain relief products are based on four active ingredients but packaged and advertised in numerous ways to appeal to di¤erent consumer needs. This observation leads us to emphasize the role of advertising in this industry, and hypothesize about perceived product (and quality) di¤erentiation in consumers'eyes.
Clinically, all four main active ingredients have varying degrees of side e¤ects –including gastrointestinal, cardiovascular, kidney and liver problems – which can become dangerous if the products are taken at more than the advised dosage. Because people react to each ingredient di¤erently, clinical pain researchers are hesitant to assign superiority to any single drug. Active ingredients di¤er in potential medical risks that they entail and in the e¢ ciency of pain relief. The analysis of inherent drug characteristics is important for our research, because we will use these measures in evaluating the credibility of comparative advertising Quantitative Measures of OTC Analgesic Characteristics The medical literature provides objective risk and e¢ ciency measures for each product, based on its active ingredient (or combination of ingredients), strength and recommended dosage.2 There are de…nitive maximum doses and durations of therapy for each active ingredient.
Di¤erences exist across di¤erent active ingredients in terms of the important safety issue of the potential for gastrointestinal (GI) toxicity and cardiovascular (CV) risk.
We attempt to quantify or rank all the true characteristics that were used in advertising associated with each active ingredient. First, we interpret claims concerning strength (e.g., "strong,""stronger,""tougher on pain") as the maximum level of pain relief achieved. Sec- ond, we interpret "fast"as the time taken to achieve a perceptible or meaningful pain relief (in medical literature terminology: onset to perceptible pain relief). Third, such claims as "long lasting" are interpreted as a duration of meaningful pain relief. Fourth, we use NNT (Number Needed to Treat) measure to approximate analgesic e¢ ciency claims3.
The relative risk4 and e¢ ciency (as expressed by NNT) di¤erences are summarized in 2 The material in this section is based on Hersh, Moore and Ross (2000), Circulation (2006), and Oxford League Table of Analgesic E¢ cacy (http://www.jr2.ox.ac.uk/bandolier) 3 NNT is a standard e¢ ciency measure used in the pain relief evaluation literature. See Appendix B for an explanation of how NNT is calculated.
4 See Appendix B for an explanation of how RR is calculated.
Dosage, mg.
(maximum daily, mg)
325-1000, every 4-6h (4000) 325-1000, every 4-6h (4000) 200-400, every 4-6h (1200) 220-440, every 8-12h (660) Table 2. Clinical Characteristics of OTC Analgesics (Hersh, Moore and Ross (2000); Oxford League Table of Analgesic E¢ cacy).
The locations of the main active ingredients in the space of characteristic (data from Table 2) which are most frequently used in comparative advertising claims are plotted in Figure 1. Location of Active Ingredients in the Characteristics Space.
After extensive review of the medical literature we were able to assign each active ingre- dient with a relative e¢ ciency rank in one of 3 categories discussed above.5 Note that this ranking is su¢ cient for us to verify comparative claims and establish relative advantage of speci…c agents. These rankings were created based on maximum recommended dosage (single dose): Ibuprofen –400mg; Naproxen Sodium–440 mg; Aspirin –1000mg; Acetaminophen – 5 We reviewed a number of medical journal articles trying to come up with the unique quantitative mea- sures for the other three e¢ ciency characteristics (maximum level of pain relief achieved, onset to perceptablepain relief and duration of meaningful pain relief). The most common occurence in medical articles was acomparison of only two or three active ingredients. We could not aggregate information from multiple articlesdue to the fact that di¤erent studies were based on di¤erent AI dosages or type of pain cured. See AppendixC for a list of references that we used to rank analyzed active ingredients.
1000mg. These measurements enable us to identify the relative locations of all the products in the characteristics space.
Time to Perceptible
Maximum pain relief
Duration of meaningful
Pain relief (Fast)
achie ved (Strength)
time relief (Longevity)
Soluble Ibuprofen Table 3. Relative Ranking of E¢ ciency Measures (starting from the most e¤ective).
Competing brands, as directed by the FDA, contain the exact same dosage of active ingre- dient. Advil, for example, contains 200 milligrams of ibuprofen, as does competing brand Motrin IB. Extra Strength Tylenol contains 500 milligrams of acetaminophen, as does Extra Strength Excedrin, which also includes a small amount of ca¤eine. There are even di¤erent products with the same brand name that have identical medical formulas. In 2000, the FDA permitted drug companies to label pain products as "migraine relievers". Migraine products carry di¤erent instructions and labels, but that labeling does not guarantee any di¤erence in actual pain medicine. A closer look at the …ne print on the labels con…rms that each "mi- graine"product is the same as the regular version. Each tablet or caplet of Motrin Migraine contains the same 200 milligrams of ibuprofen, the active ingredient, as regular Motrin. The same is true for the ibuprofen-based Advil and Advil Migraine.
The only di¤erence between Excedrin Migraine and Excedrin Extra Strength or Tylenol 8 Hour and Tylenol Extra Strength is the box. The tablets in each contain the same amounts of aspirin, acetaminophen and ca¤eine– the pain reliever's active ingredients. During our analyzed period, Tylenol is the leader in creating product di¤erentiation, o¤ering its Extra Strength Tylenol in six di¤erent pill forms as well as a liquid. There are also Tylenol versions for menstrual pain, migraine, arthritis and pain relief with sleep aid. Advil and Aleve are packaged in three pill forms, and Bayer o¤ers a safety-coated aspirin pill. (See Table 8 in the Appendix for the complete list of advertised products and their ingredients).
Building brand trust via the channel of advertising is also an essential part of building and restoring perceived product quality. In late 2004, the FDA publicly announced that clinical study evidence indicated that patients taking naproxen sodium (Aleve) may be at an increased risk of su¤ering heart attack or stroke. The public press widely and immediately reported negative news about Aleve, unleashing a media frenzy that caused one of the largest crises in the OTC analgesics industry. Headlines like "Aleve –the latest drug to cause Heart attacks" were found in almost every news stand. By the end of January 2005, sales of Aleve plummeted by over 50% su¤ering the largest decline in brand history.6 The aggressive "Good News"and personal testimonials advertising campaigns were designed to demonstrate Aleve's safety and e¢ cacy in a way that would restore con…dence in the brand. The shock was softened and eventually turned around by advertising.
Additionally, in this market we observe a lot of products of similar or identical quality, where cheaper generic products do not advertise and have lower market shares. There exists a lingering perception that generics are somehow inferior to their brand-name counterparts.
However, all drugs are subject to the same regulatory scrutiny by the FDA. They must contain the same active ingredient as the brand-name product, and generic drug makers have to prove their product is just as safe and e¤ective with the same quality, purity, and Firms concede that there is little to di¤erentiate the medicine in the pills, especially when there is a legal FDA restriction concerning strength and safety for over-the-counter use of certain analgesic agents. This con…rms our suspicion that the market is saturated with products that from a true quality perspective are almost identical, yet there is perceived di¤erentiation in this market.
In this section we underscore that advertising generates perceived product di¤erentiation and quality in this market. If all of the characteristics of a good were readily measurable and veri…able, then their value would be known to the consumer and advertising should not a¤ect demand for the good. In some cases, the product is so intricate that its characteristics and its worth are not fully understood by the average consumer. OTC analgesics products are indeed characterized by a quality aspect that is su¢ ciently complex and involves medical terminology. In this case, assuming that all consumers are aware of the existence of all available OTC analgesics in the market, advertising a¤ects the consumer's perception of the quality of goods by supplying information about the product. Consumers derive from advertising an improved perception of the quality of the product, which heavily depends on the value of the attributes advertised and additional information transmitted by advertising.
6 Aleve Case History. Real People Campaign (2006)7 http://www.fda.gov/cder/consumerinfo/generics_q&a.htm The data that we use for the analysis consist of (1) product level data and (2) advertising data. In addition to these 2 datasets we also use (3) consumer level data prepared by Simmons Market Research. The product level data were purchased from AC Nielsen and consist of 4 week observations8 of average prices, dollar sales and dollar market shares of any OTC pain reliever sold in the U.S. national market during the 5 years from 2001 through 2005. Data is disaggregated down to the individual UPC (Universal Product Code) level.
The majority of products in the dataset fall under the category "headache medicine," but pain remedies in the arthritis and menstrual analgesics categories are included in the dataset as well. We have data on essential product attributes: active ingredient, strength (regular, extra strength, etc. as regulated by the FDA), pill type (caplet, tablet, gelcap, etc.), number of pills contained in the product, and purpose (menstrual, migraine, arthritis, general, children, etc.).
Advertising data were purchased from TNS-Media Intelligence. This novel dataset con- tains not only monthly advertising expenditures and units for 2001-2005 for each product advertised in the OTC analgesics category, but also video and image …les of all advertise- ments. While the advertising numbers include expenditures on multiple media, almost all of the advertising budgets were spent on broadcast television advertising, including network, cable, and spot TV.
The availability of such a rich dataset enables us to include advertising content (focusing on comparative advertising) in the analysis of this market. The characteristics of each prod- uct revealed in the advertisements are coded and, in the case of comparative advertisements, the rival products mentioned are recorded as well as characteristics on which the comparative statements were based. Such coding generates matrices of advertised characteristics and the cross targeting of brands via comparative advertising.
Finally, consumer level data comes from the Survey of Media and Markets conducted by Simmons Market Research Bureau. Simmons collects data on individual media habits, product usage and demographics from about 20,000 households annually. This dataset has rich information on the brand choices of OTC analgesics, frequency and purpose of use, exposure to media and demographic characteristics (age, sex, income, etc.). An example of Simmons data is given in Table 4.
These data suggest that there are several consumer types in the market: those with 8 Product level data was normalized to monthly frequency to match the advertising data frequency.
infrequent needs, (headache, muscle ache, etc.) those with moderate but not constant needs (backache, menstrual pain), and those with constant needs (arthritis). This trichotomy alerts us to expect price discrimination across groups with di¤erent valuations among groups, sorted by di¤erent pill bottle sizes.
Purpose of Use
Other Muscle/Body Aches Cold s, Fl u and Fever Arthri tis or Rheumatism Table 4. Purpose of use of OTC Analgesics.
Coding of the Advertisements There were 4503 individual commercials during 2001-2005. Out of 4503 commercials 346 had missing video …les. We should note that usually each individual video was shown multiple times. The total number of commercials shown over the 5 year period in all types of media is a staggering 595,216. We watched all included ads, and recorded mentioned characteristics.
We also decided whether the characteristic could be classi…ed as an experience or as a credence characteristic.9 We elaborate on the types of characteristics and the importance of their distinction in Section 8. The following characteristics were recorded: (1) Whether the commercial had any comparative claims – explicitly comparing to any other products.
(2) If a commercial was comparative, what was the comparative claim: a) Faster (experience) b) Less drug interference/safer (credence) c) Stronger, As strong (experience) d) More E¤ective (experience) e) Longer Lasting (experience) f) More Popular (credence) g) Fewer Pills (credence) 9 The distinction follows that developed by Nelson (1970): Experience characteristics - consumers cannot determine them before they buy the product, but consumption allows consumers to assess them; Credencecharacteristics - consumers cannot check the presence or evaluate credence characteristics, even after theconsumption experience. They must trust the seller or the quality indicators.
h) Better for Arthritis (experience) i) More doctor/pharmacist recommended (credence) j) Also against heart attack (credence) k) Against all menstrual syndromes (experience) (3) If a commercial was comparative, which brand (class of drugs) was it compared to: f) Other Ibuprofen/NSAID (4) What characteristics were mentioned in ad (in a noncomparative way): I Purpose of drug a) Legs/muscle (experience) b) Menstrual (experience) c) Back (experience) d) Arthritis (experience) e) Heart attack (experience) f) Sinus (experience) g) Sleep (experience) h) Headache (experience) i) Children's (experience) II Characteristics indicating e¢ ciency and/or safety a) Trusted/Safe (credence) b) Long lasting (experience) c) Strong (experience) d) Fast (experience) e) Gentle on stomach (experience) f) Doctor Recommended (credence) III Other characteristics a) Every pain relief (credence) b) Non habit forming (credence) c) New product (credence) d) Warning: overdose, label (credence) Classic content analysis in marketing describes the informative content of advertisements by counting the number of "cues" (or characteristics) mentioned. We use a set of cues described in the previous section that are speci…c to the OTC analgesics industry. We …nd that during the sample period 2001-2005, the average number of cues per ad (both comparative and noncomparative) is 2.7 and the median is 3. We then analyze comparative and noncomparative ads separately. Figure 2 juxtaposes comparative and noncomparative ad histograms. Consistent with previous research (Chou et al (1987) and Harmon et al (1983)) we …nd that comparative ads have a slightly higher number of cues (the average number of cues is 3.13 and the median is 3) than noncomparative ads (the average is 2.47 and the median is 2).
This type of distribution is common in standard content analysis studies. Compared to the classic study, using the traditional 14 information cues proposed by Resnick and Stern, it indicates on average more information cues than is usual. However, this is probably due to the much more narrow classes of information cues used in the diagram (and the current analysis). There is no particular reason at this point to suppose there is either consistently more or less information content in pain reliever ads than ads in other sectors.
Number of Characteristics per Ad Number of Characteristics per Ad Figure 2. Histogram of Number of Advertising Characteristics for Comparative and Noncomparative Ads (2001-2005).
Many ads mention several characteristics, or have several targets (see Tables 5 and 8; Figures 3 and 4 for a breakdown along these lines). The e¤ectiveness of an advertising message in any single dimension is presumably diluted if it addresses other dimensions within the same message (for example, for information congestion within the consumer's memory). Put di¤erently, a message with a single theme is likely to have a bigger impact on that theme than if there are several themes within the same message. Even if di¤erent themes got the same airplay within the same message, it is unclear what the trade-o¤ is between e¤ectiveness per theme and number of themes per message. For example, the congestion e¤ect might be dominant and one pure theme in each of two messages might dominate two messages of two themes each. The relation between number of themes and e¤ectiveness per theme remains an interesting topic for future research, and the content analysis is a very useful place to start an empirical investigation of this topic.
One way to code a congestion e¤ect is to assume that each characteristic generates an impact in proportion to the fraction of the message devoted to this theme. Even this method would involve presuming perfect fungibility between contents (so 3 messages with each divided into 3 themes would perfectly substitute for 3 messages with each on a di¤erent theme). However, it is often hard in practice to attribute airtime within a particular message to speci…c components, and the coding time increases dramatically. The simplest approach, and the one we present the data for below, is to assume that the e¤ectiveness of any ad to any characteristic dimension is proportional to the inverse of the number of characteristics mentioned. That is, if n characteristics are mentioned, each is weighted as 1/n of the ad.
This approach deals with the double counting problem mentioned below, but it may err on the side of underestimating the impact of a characteristic in a message, for two reasons.
First, there might be weak congestion e¤ects. Second, the other characteristics might be less powerful. For example, we will in practice divide the weight per attribute in three if an ad targets two rivals for being slower to act, but also mentions the particular drug is for arthritis. This means we will code the ad as being one-third attack on slowness for each of the two targets, and one third a positive ad for own arthritis characteristic. In that sense, the "equal-weighting" approach we have just described might tend to underestimate the expenditure on comparative advertising characteristics.
Figure 3 depicts the histogram (weighted by dollars spent on each ad) of types of charac- teristics mentioned in ads during our analyzed period. E¢ ciency characteristics are the ones with the highest expenditures (Fast, Strong, Long Lasting). We will return to the analysis of these histograms in Section 8 of this paper. The breakdown of advertising expenditures (using the equal division method) by 8 major brands across di¤erent characteristics is given (in restudi 100enpexr 50 Figure 3. Histogram of Types of Characteristics Mentioned in Ads.
Gentle on stomach Table 5. Who Advertises What and How Much. Cumulative 2001-2005 data (in $000).
The observed patterns indicate quite an informative role of advertising for brands. Adver- tising tells consumers that the brand delivers the characteristics as opposed to a persuasive advertising scenario where advertising increases desirability of the particular characteristic.
However, advertising here may actually play both roles: …rst indicating that the particular product has an advantage in one characteristic, while simultaneously enhancing consumers' valuations of that characteristic. We return to this discussion in section 8.
Content Analysis of Comparative Advertisements One of the major contributions of our paper is the consistent analysis of comparative ad- vertising ads in one particular industry. Analyzing the subset of advertising videos that were comparative reveals a lot of interesting patterns. Table 6 presents a summary of the frequency of the number of brands targeted and the number of characteristics mentioned.
A single competing brand is targeted in a little over half of all the comparative advertising cases. Two brands were mentioned in almost a quarter of the ads. There were even a few cases where an ad was targeting and naming 3 di¤erent competitors. The remaining 23% of the comparative ad cases were comparisons against a general class of competing products, such as prescription, NSAIDs or other regular pain relievers. As can be seen from the second column in Table 6, more often than not (around 68% of the time) two or more characteristics were mentioned in a comparative statement (e.g. "faster" and "stronger" coupling).
Table 6. Frequency of Direct Comparative Ads Targeting Speci…c Number of Brands and Figure 4 shows which characteristics were mentioned in comparative statements (and how many dollars were spent on such characteristics). Again, the e¢ ciency characteristics such as Stronger, Faster and Longer lasting are used most in comparative claims. We return to the more detailed comparative ad analysis and the issue of who targets whom in the sections Figure 4. Histogram of Characteristics Mentioned in Comparative Ads.
Targeting in Comparative Advertising One of the most interesting aspects of the coded data is the information about who targets whom. The complete picture of cross targeting (listing both the targets and the comparative characteristics) is presented in Table 8. Table 7 summarizes the comparative advertising target matrix. The important observation from these tables is that all nationally advertised brands used comparative advertising during the sample period. However, the brands against which comparisons are made are only a subset of the nationally advertised brands. The targets are Tylenol, Advil, Aleve, and Excedrin.
Direct CA
Total $122,444 $145,805
Table 7. Comparative Advertising Targeting Matrix. (T- Target Brand, A- Advertiser) The …gures in the left block of Table 7 are the amounts spent on any advertising which mentions the speci…ed target. However, since a particular ad may mention two or more targets, a single ad expenditures may be counted more than once. This is why the sum of the numbers on the left do not add up to the total Direct CA …gure in the …rst of the two right columns: this …gure is the sum of spending conditional on the ad having an explicit target. The second (and last) column on the right, Total CA, also includes comparative ads that do not specify the target by name, but may be more generic (e.g., Tylenol attacking NSAIDs). Likewise, Midol attacks Tylenol every time it attacks Aleve; Motrin almost always attacks the pair Advil and Aleve. Bayer is the other attacker with some ads against a brand other than Tylenol. Tylenol attacks Aleve in every direct attack ad, and half of these also attack Advil.
Target >
Advertiser v
$7,974 Total
Stronger/As strong $3,987 Stronger/As strong $5,056 Stronger/As strong $560 Total
$280 Long Lasting Stronger/As strong Stronger/As strong Stronger/As strong Midol All menstrual syn
All menstrual syn tronger/As strong $14,414 Stronger/As strong $1,462 Total
All menstrual syn $1,108 All menstrual syn All menstrual syn $4,061 More Popular Stronger/As strong $1,158 Stronger/As strong Table 8. Characteristics of Brands Targeted by Comparative Advertising: Dollar Expenditures (in Thousands) by Advertiser Using Equal Division Approach.
Using the relative e¢ ciency and safety information discussed above we positioned most brands in a product characteristics space and marked the "attack" patterns depicted in Figure 5. Most of the arrows point to Tylenol –the leader brand. We analyze these attack patterns in section 8.
Less safe
More safe
Figure 5. Targeting in Comparative Advertising. "E" refers to "attack" on general e¢ ciency grounds, "L" –on length, "S" - on strength and "F" –on speed of pain relief.
In this section we construct several theoretical hypotheses and test them against our data.
These hypotheses do not require rigorous statistical techniques since they can be seen quite clearly from simple correlations in the data. For this reason, we do not need sophisticated statistical models in this paper for our current purposes. That said, we intend to carry out a more nuanced and elaborate econometric analysis with the data, using a structural econometric model. Of course, that approach in turn can be criticized because the structural model imposes strong behavioral assumptions. We therefore keep the structural work sepa- rate from the evaluation of the simpler empirical hypotheses about comparative advertising that are quite evident in the aggregate data.
H1. Targets: Leading brands will be targeted most by comparative advertising.
The conjecture that the brands with higher market shares will be most heavily targeted in comparative advertising stems in part from the recognition that higher market shares most likely re‡ect higher perceived quality or superior performance in some characteristic categories. Other brands can then try to in‡uence consumer perceptions of their own wor- thiness by indicating stronger performance in some categories, and hence suggesting quality comparable to (or higher than) the highest one available. Targeting small brands (with, by extension, low perceived quality) would not provide as much potential boost in perceived quality. The ‡ip side of this hypothesis analyzes which brands will actually instigate the targeting, which is the topic of the next hypothesis.
In the data we see only 4 targeted brands: Tylenol, Aleve, Advil and Excedrin. These brands are also the top four brands by market share. Table 9 lists the targeted brands, their market shares, and comparative ad expenditure against those brands. Tylenol, as the market leader, is attacked the most. The second most attacked brand is Aleve, even though it is only third in terms of market share. Advil and Excedrin are third and fourth respectively.
The data strongly support the hypothesis.
the Target
Table 9. Targeted Brands and Comparative Ad Expenditure (in Thousands of Dollars) H2. Targeters: Brands with low market shares will engage in more comparative advertis- ing as a fraction of their total advertising (than brands with higher market shares).
As noted in H1, one possible reason for comparative advertising is to bring the consumer perception of one's brand up to the level of the target. Thus, if we view market shares as re‡ecting perceived qualities, as argued above, then brands will want to target those "above" them, and not those below them. There is another reason why brands may not wish to target those with lower market shares. Comparative advertising gives at least some publicity to rival brands, indicating the existence of other products. Top brands are unlikely to wish to acknowledge other brands, thus comparative advertising may be used as a tacit admission that the brand is not the market leader. Brands with low market shares may also signal a rising market position through an aggressive marketing strategy of comparison. In Table 10 we clearly see that the brands which use the most comparative advertising account only for very small market shares, whereas Tylenol, being the market leader, spends the least amount on comparative advertising.
Direct CA
Market share
Table 10. Relative and Absolute Comparative Advertising Expenditures and Market H3. Comparative advertising will target rival brands and not "stable-mates." As comparative advertising is implicitly "negative" in terms of denigrating the perfor- mance of other brands, one would not expect one brand to extensively cannibalize another brand owned by the same parent company. This suggests a corollary that parent companies with fewer "stable-mates" will be more likely to engage in comparative advertising.
In the OTC analgesics market this is clearly true: McNeill owns both Tylenol and Motrin IB brands and Bayer owns Bayer, Aleve and Midol. None of these brands target their "stable- mates" in comparative advertising. (See the matrix of cross targeting in Table 8).
H4. A speci…c characteristic will only be advertised by one brand.
Each characteristic is unlikely to be advertised by more than one brand. However, it is possible (and consistent with the statement) that one brand could advertise more than one characteristic. This idea comes from parallel theoretical research by Anderson and Renault (2008). The research models the advertising of particular product attributes by …rms, and (positive) advertising is assumed to heighten the consumer appreciation of the characteristic(s) that are advertised. This idea is termed the "sheening"of the characteristic, i.e., strengthening the consumer valuation of the good by strengthening the appreciation of its strongest constituent characteristics. Therefore, such advertising of a characteristic is likely to "raise all boats" for the brands that are strong in that characteristic and will thus most bene…t the …rm that is perceived to best embody that characteristic. Thus, the product that best embodies that speci…c characteristic will advertise it. The hypothesis is tested by looking at the matrix of information cues in advertising in Table 5.
It is clear that the strong form of this hypothesis is rejected (each …rm advertises sig- ni…cantly more than one characteristic). However, the weak form of the "sheening" model holds: the highest expenditure characteristic is di¤erent for most of the brands – Advil spends most money to emphasize that it is strong, Aleve, that it is long lasting, Excedrin that it is a headache medicine. Midol and Pamprin stress that they are for menstrual pain; Tylenol emphasizes speed, Bayer notes it is also good for heart attack prevention and treat- ment. Motrin, however, follows Advil in stressing strength. Hence, the top brands specialize in which characteristic they advertise the most.
H5. Comparative and general advertising claims will tend to involve experience attributes as opposed to credence ones.
Experience attribute claims are veri…able only after purchase/use, implying that verifying such claims involves considerable cost (actual consumption of a product). Alternatively, credence characteristics cannot be fully veri…ed, both prior to and post consumption. As a result, the buyer has to rely on third-party judgments (e.g., FDA) or on the seller's credentials (i.e. the undisputed record of honesty, competence and determination with respect to the quality). Hence, it is logical to assume that, other things being equal, consumers may demand more information for (or pay more attention to) experience characteristics than for In our case, out of 11 attributes that involved comparative claims, 6 of them can be classi- …ed as experience characteristics. Those 6 attributes are exactly the ones that are advertised most frequently. Indeed, the fraction of experience characteristics in total comparative ad- vertising expenditures was nearly 80%.10 Additionally noncomparative claims also involve higher frequency of experience characteristics compared to credence characteristics.
H6. Comparative advertising will target products with active ingredients di¤erent from that of the attacking product.
Given the distinct properties of active ingredients (AIs), it is reasonable to predict that products will not target other products with the same AIs, at least not on the basis of medical e¢ cacy and safety claims. By inspecting Table 8, we observe that the four highest comparative advertising expenditures are spent against products with di¤erent AIs: Advil (Ibuprofen) vs. Tylenol (Acetaminophen), Aleve (Naproxen Sodium) vs. Tylenol (Acetaminophen), Tylenol (Acetaminophen) vs. Aleve (Naproxen Sodium), Advil (Ibuprofen) vs. Aleve (Naproxen Sodium).
From the 6 highest comparative ad expenditures, only one (the …fth one) is made against the product containing the same AI (Motrin vs. Advil – both contain Ibuprofen as an AI; however, the claim there was that Motrin is "as strong" as Advil).
In this paper we investigate the advertising content of the OTC analgesics industry. This industry is characterized by high advertising-to-sales and comparative advertising levels.
We have described the actual comparative advertising patterns used in the OTC analgesics industry, and linked this to the important characteristics of the products'active ingredients.
We …nd that larger brands tend to be most targeted by comparative advertising. It is 10 Using data from Table 6, the total amount spent (in $000 over the 5 years) on comparative advertising was $423,797; of this $334,364 was on experience characteristics, which constitutes 79% of the total.
smaller brands which tend to use relatively more comparative advertising in their advertising mix. Contrary to the conventional understanding that leading brands are not involved in comparative advertising, we do observe Tylenol comparing itself against Advil and Aleve.
We also …nd that nationally advertised brands tend to specialize in which characteristics they advertise the most. This is consistent with the weak version of "sheening" theory where positive advertising of a speci…c characteristic is likely to "raise all boats" for the brands that are strong in that characteristic and will thus most bene…t the product that is perceived to best embody that characteristic. We also show that the most heavily advertised characteristics, and especially the characteristics that were advertised in a comparative way can be deemed to be experience attributes. Lastly, we note that comparative advertising targets products with active ingredients di¤erent from that of the attacking product.
However, it is worth emphasizing that the patterns observed in this industry do not nec- essarily re‡ect what would be optimal in other contexts. For example, products have distinct and chemically di¤erent active ingredients, and advertising in the industry is governed by a self-regulating code of conduct by which …rms usually provide evidence to substantiate the e¢ cacy and safety claims. Nor, indeed, is it necessarily true that the products in this sample were choosing the best pro…le of advertisements. Even if we see the most successful brand (Tylenol) pursuing a particular advertising strategy, we cannot say whether that strategy was the cause of its success, or whether it might not have done better using another strategy.
Our research does suggest future research directions which we hope to pursue to get a tighter grasp of the extent and type of advertising which is most successful for which …rms.
With these caveats in mind, we can already draw some preliminary and tentative conclu- sions on comparative advertising. First, the most successful …rm (Tylenol) does not engage much in comparative advertising. The question on causality is important here. It is not because Tylenol does not advertise comparatively that it is so successful. Its dominant mar- ket position goes back to before comparative advertising was widely used. Rather, …nding itself in a position of market dominance, with high perceived product quality, it does not want to comparative advertise against its rivals in the OTC industry. To do so would draw attention to them, increase consumer awareness, and might even give more credibility to them. Instead, Tylenol compares itself (favorably, of course) to prescription drugs, which are widely perceived to be more potent. And presumably, by extension, Tylenol is safer because a prescription is not needed.
Second, …rms push their strengths in their own positive advertising, and emphasize rivals' weaknesses in comparative advertising. However, according to FTC comparative claim veri…- ability and substantiation program, …rms need to carefully pick and reason their comparative Third, aggressive advertising can be used to staunch lost sales following bad news. This happened during our sample with the late 2004 news shock to Aleve (and Naproxen Sodium in general). Aleve responded with a strong campaign, which successfully restored public con…dence and turned around the plummeting sales.
This last episode leads us to suggest how future research can help identify the crucial issue of how e¤ective comparative advertising is. In particular, with data on the timing and severity of shocks, we can address how …rms responded, and how successful the response was. News stories are a good source of such data. A second useful source of information is the history of investigation or prosecution of misleading advertising by the FTC. This serves as a good proxy for the cost of running comparative ads as opposed to standard ads.
More research needs to be done on the theoretical side too, in terms of formulating models that describe the costs and bene…ts of raising one's own product quality (positive advertising) or denigrating a rival's quality (comparative ads). A full equilibrium model in this vein would enable us to …nd theoretical correlation between the amount and type of advertising done, and market share. This model could also indicate the advertising responses of …rms to news shocks. The full picture would base the bene…ts from advertising on the underlying product characteristics (i.e., the chemical properties of the active ingredients).
Explanation of Medical Measures Relative risk (RR) is the risk of an event (or of developing a disease) relative to exposure.
Relative risk is a ratio of the probability of the event occurring in the exposed group versus the control (non-exposed) group.
Relative risk is used frequently in clinical trial data, where it is used to compare the risk of developing a disease, in people not receiving the new medical treatment (or receiving a placebo) versus people who are receiving an established (standard of care) treatment.
Alternatively, as it is the case with GI and CV relative risk numbers used in this paper, it is used to compare the risk of developing a side e¤ect in people receiving a drug as compared to the people who are not receiving the treatment (or receiving a placebo). Thus, for example, a CV RR of 1.44 means that CV problems arise with 44% higher likelihood using the drug (vs. placebo).
Number-Needed-to-Treat (NNT) is computed with respect to two treatments A and B, with A typically a drug and B a placebo. If the probabilities PA and PB under treatmentsA and B, respectively, are known, then the NNT is computed as: The NNT for a given therapy is simply the reciprocal of the absolute risk reduction PA) for that treatment. For example, in hypothetical migraine study, risk decreased from PB = 0:30 without treatment with drug M to PA = 0:05 with treatment withdrug M, for a relative risk of 0.17 (0.05/0.3), a relative risk reduction of 0.83 ((0.3-0.05)/0.3), and an absolute risk reduction of 0.25 (0.3-0.05), the NNT would be 1/0.25, or 4. In concrete clinical terms, an NNT of 4 means that you would need to treat four patients with drug M to prevent migraine from recurring in one patient. Typically, the lower the NNT number, the more potent and e¢ cient the treatment is.
Additional Active Ingredients and the List of Advertised Products Some of the marketed pain relievers contain additional active ingredients that either help speci…c pain relief (as is the case with most menstrual pain relief formulas) or a¤ect the e¢ ciency of pain relief. The most common additional ingredients are the following: Ca¤eine. Analgesic active ingredients combining ca¤eine with aspirin, acetaminophen, or both are also available as OTC drugs (Midol and Pamprin for menstrual pain, Excedrin for headache and migraine, Bayer for body aches). Ca¤eine is considered an analgesic adjuvant, that is, it enhances the analgesic e¤ects of aspirin, acetaminophen and ibuprofen; yet it is ine¤ective when used alone. Ca¤eine appears to shift the analgesic dose-response curve upwards, thereby providing more pain relief for a given dose of analgesic.
Aspirin and acetaminophen combination en- hance e¢ ciency of pain relief by adding active ingredients that produce the same e¤ect but by di¤erent channels. These combinations are expected to have lower side e¤ects than aspirin alone. Excedrin, Midol and Pamprin use this mix.
Other Formulas. Night formulae (e.g., Tylenol PM) contain an additional ingredient, Diphenhydramine HCl, which is a sleep aid. Generally, any drug marketed for children (e.g., Tylenol Meltaways) contains a smaller dose of the active ingredient (as little as one-third of the adult dose). Drugs marketed for menstrual purposes (such as Midol or Pamprin) with acetaminophen based formulae often contain diuretics such as Pamabrom and Pyrilamine Table 11 shows all the advertised brands and their products together with the list of active ingredients, recommended dosage, and maximum number of pills one is allowed to take within 24 hours.
Brand and Product
Max in 24 h
Sol. 200 mg ibuprofen Advil (tablets, caplets, gel Aleve (tablets, caplets) Naproxen Sodium 220 mg 1 every 8 to 12 h Bayer Back and Body Aspirin 500 mg.
Caffeine 32.5 mg.
Bayer Extra strength 1 or 2 every 4 to Excedrin (extra strength) Acetaminophen 250 mg, Aspirin 250 mg,Caffeine 65 mg Excedrin (tension headache) Acetaminophen 500 mg, Excedrin (Migraine) Acetaminophen 250 mg, Aspirin 250 mg,Caffeine 65 mg Excedrin (sinus headache) Acetaminophen 325 mg Phenylephrine HCI 5 mg Acetaminophen 500 mg, Diphenhydramine citrate 38mg Midol (menstrual complete) Acetaminophen 500 mg, Caffeine 60 mg,Pyrilamine maleate 15 mg Midol (premenstrual) Acetaminophen 500 mg, Pamabrom 25 mg,Pyrilamine maleate 15 mg Midol (cramps and body aches) Midol (extended relief) Naproxen sodium 220 mg 1 every 8 to 12 h Midol (menstrual headache) Acetaminophen 500 mg, Midol (teen formula) Acetaminophen 500 mg s Motrin, Motrin Jr.
Pamprin (multi symptom) Acetaminophen 500 mg, Pamabrom 25 mg,Pyrilamine maleate 15 mg Pamprin (all day) Naproxen sodium 220 mg 1 every 8 to 12 h Acetaminophen 250 mg, Magnesium salicylate 250 mg,Pamabrom 25 mg Tylenol Arthritis Acetaminophen 650 mg Acetaminophen 325 mg Table 11. Advertised Brands and Their Products (2001-2005).
Determining the E¢ ciency Ranking of Active Ingredients We reviewed a number of medical journal articles in attempt to rank the three e¢ ciency measures (maximum level of pain relief achieved, onset to perceptible pain relief and duration of meaningful pain relief) of the analyzed active ingredients The most common occurrence in medical articles was a comparison of only two or three active ingredients. If the article X said that the drug A is more e¢ cient than the drug B (A>B) and the article Y said that the drug B is more e¢ cient than C (B>C), then we conclude that A is more e¢ cient than B and C (A>B>C). Below we also present the numbered list of references that were used to infer relative rankings. Table 9 lists all those relative relationships, references of medical articles (in parentheses), and gives the resulting ranking presented above in Table 3.
Time to Perceptible Pain relief
Maximum pain relief achieved
Duration of meaningful time relief
Sol Ibuprofen>Ibuprofen (1,6) Naproxen>Aspirin (3) Naproxen>Aspirin (3) Ibuprofen> Acetaminophen (2,6,5,7,10) Ibuprofen/Sol Ib>Acetaminophen (6) Naproxen>Asprin (3) Resulting Ranking (Table 3):
Soluble Ibuprofen Table 12: Relative E¢ ciency Ranking of Di¤erent Analgesic Agents (1) Packman, B., Packman, E. Doyle, G, Cooper, S., Ashraf, E., Koronkiewicz, K. and Jayawardena, S. 2000. Solubilized Ibuprofen: Evaluation of Onset, Relief, and Safety of a Novel Formulation in the Treatment of Episodic Tension-type Headache. Headache: The Journal of Head and Face Pain. Volume 40 Issue 7 Page 561-567 (2) Hyllested, M., Jones, S., Pedersen, J.L., Kehlet, H. 2002. Comparative e¤ect of paracetamol, NSAIDs or their combination in postoperative pain management: a qualitative review. Br J Anaesth. 88:199–214.
(3) Forbes, J.A., Keller, C.K., Smith, J.W., Zeleznock, J.R., Sevelius, H., Beaver, W.T.
1986. Analgesic e¤ect of naproxen sodium, codeine, a naproxen-codeine combination and aspirin on the postoperative pain of oral surgery. Pharmacotherapy. Sep-Oct 6(5):211-8 (4) Milsom, I., Minic, M., Dawood, Y., Akin, M., Spann, J., Niland, N. and R. Squire, A. 2002. Comparison of the e¢ cacy and safety of nonprescription doses of naproxen and naproxen sodium with ibuprofen, acetaminophen, and placebo in the treatment of primary dysmenorrhea: a pooled analysis of …ve studies. Clinical Therapeutics. Vol 24(9) pp. 1384- (5) Cooper, S.A., Schachtel B.P., Goldman, E., Gelb, S., and Cohn, P. 1989. Ibuprofen and acetaminophen in the relief of acute pain: a randomized, double-blind, placebo-controlled study. Journal of Clinical Pharmacology. 29: 1026-1030 (6) Olson, N.Z., Otero, A.M., Marrero, I., Tirado, S., Cooper, S., Doyle, G., Jayawardena, S., and Sunshine, A. 2001. Onset of analgesia for liquigel ibuprofen 400 mg, acetaminophen 1000 mg, ketoprofen 25 mg, and placebo in the treatment of postoperative dental pain.
Journal of Clinical Pharmacology. Nov 41(11):1238-47.
(7) Ong, C.K.S., Lirk, P., Tan, C.H. and Seymour, R.A. 2006. An Evidence-Based Update on Nonsteroidal Anti-In‡ammatory Drugs. Clinical Medicine & Research. Volume 5, Number 1 : 19 -34 (8) Miller, D., Talbot, C., Simpson, W., Korey, A. 1987. A Comparison of Naproxen Sodium, Acetaminophen and Placebo in the Treatment of Muscle Contraction Headache.
Headache: The Journal of Head and Face Pain. 27 (7) 392–396.
(9) Lee, C., Straus, W.L., Balshaw, R., Barlas, S., Vogel, S., Schnitzer, T.J. 2004. A comparison of the e¢ cacy and safety of nonsteroidal antiin‡ammatory agents versus aceta- minophen in the treatment of osteoarthritis: a meta-analysis. Arthritis Rheum. 51:746— 754 (10) Hersh, Elliot, Moore, Paul A. and Ross, Gilbert L. 2000. Over-the Counter Anal- gesics and Antipyretics: A Critical Assessment. Clinical Therapeutics. 22(5).
[1] Abernethy, Avery M. and Franke, George R. 1996. The information content of adver- tising: a meta-analysis. Journal of Advertising. 25 1-17.
[2] Abernethy, Avery M. and Butler, Daniel D. 1992. Advertising information: services versus products. Journal of Retailing. 68 398-419.
[3] Abernethy, Avery M. and Franke, George R. 1998. FTC regulatory activity and the information content of advertising. Journal of Public Policy and Marketing. 17 239-256.
[4] Aluf, Yana, and Oz Shy. 2001. Comparison advertising and competition. Mimeo. Uni- versity of Haifa.
[5] Anderson, Simon P. and Renault, Régis. 1999. Pricing, Product Diversity and Search Costs: a Bertrand-Chamberlin-Diamond Model. RAND Journal of Economics. 30 719- [6] Anderson, Simon P. and Renault, Régis. 2006. Advertising Content. American Economic Review. 96(1) 93-113 [7] Anderson, Simon P. and Renault, Régis. 2007. Comparative Advertising. Mimeo. Uni- versity of Virginia.
[8] Anderson, Simon P. and Renault, Régis. 2008. Sheening Characteristics in Advertising.
Mimeo. University of Virginia.
[9] Anderson, Simon P., de Palma, André, and Thisse, Jacques-François. 1992. Discrete Choice Theory of Product Di¤erentiation. M.I.T. Press, Cambridge, Mass.
[10] Bagwell, Kyle. 2007. The Economic Analysis of Advertising. Handbook of Industrial [11] Barigozzi, F., P. Garella, and M. Peitz. 2006. With a Little Help From My Enemy: Comparative vs. Generic Advertising. Mimeo. University of Bologna.
[12] Belch, George E. 1981. An Examination of Comparative and Noncomparative Television Commercials: The E¤ects of Claim Variation and Repetition on Cognitive Response and Message Acceptance. Journal of Marketing Research. 18 333-49.
[13] Chou, L., Franke, G. R. and Wilcox, G. B. 1987. The Information Content of Com- parative Magazine Ads: A Longitudinal Analysis. Journalism Quarterly. 64 (1): 119-24 [14] Chan, Andrew T., Manson, JoAnn E., Albert, Christine M., Chae, Claudia U., Rexrode, Kathryn M., Curhan, Gary C., Rimm, Eric B., Willett, Walter C. and Fuchs, Charles S. 2006. Nonsteroidal Anti-in‡ammatory Drugs, Acetaminophen, and the Risk of Car- diovascular Events. Circulation. 113 1578-1587.
[15] Cooper, S.A., Schachtel B.P., Goldman, E., Gelb, S., and Cohn, P. 1989. Ibuprofen and acetaminophen in the relief of acute pain: a randomized, double-blind, placebo- controlled study. Journal of Clinical Pharmacology. 29: 1026-1030.
[16] Droge, Cornelia. 1989. Shaping The Route to Attitude Change: Central Versus Periph- eral Processing Through Comparative Versus Noncomparative Advertising. Journal of Marketing Research. 26 193-204.
[17] Droge, Cornelia and Rene Y. Darmon. 1987. Associative Positioning Strategies Through Comparative Advertising: Attribute Versus Overall Similarity Approaches. Journal of Marketing Research. 24 377-88.
[18] FDA and Consumer HealthCare Products Association. 2004. The OTC Drug Semi- nar Series: Over-the-Counter and On-the-Air The Advertising of OTC Drug Products.
[19] Forbes, J.A., Keller, C.K., Smith, J.W., Zeleznock, J.R., Sevelius, H., Beaver, W.T.1986.
Analgesic e¤ect of naproxen sodium, codeine, a naproxen-codeine combination and as- pirin on the postoperative pain of oral surgery. Pharmacotherapy. Sep-Oct 6(5):211-8.
[20] Ford, Gary, Smith, Darlene B, Swasy, John. 1990. Consumer Skepticism of Advertis- ing Claims: Testing Hypotheses from Economics of Information. Journal of Consumer Research. 16(4) 433-41 [21] Golden, Linda L. 1979. Consumer Reactions to Explicit Brand Comparisons in Adver- tisements. Journal of Marketing Research. 16 517-32.
[22] Goodwin, Steven and Michael Etgar. 1980. An Experimental Investigation of Compara- tive Advertising: Impact of Message Appeal, Information Load, and Utility of Product Class. Journal of Marketing Research. 17 187-202.
[23] Grewal, Dhruv, Sukumar Kavanoor, Edward F. Fern, Carolyn Costley, and James Barnes. 1997. Comparative Advertising: A Meta-Analysis of the Empirical Evidence.
Journal of Marketing. 61 1-15.
[24] Harmon, R. R., Razzouk, N. Y. and Stern, B. L. 1983. The Information Content of Comparative Magazine Advertisements. Journal of Advertising. 12 (4): 10-19.
[25] Hersh, Elliot, Moore, Paul A. and Ross, Gilbert L. 2000. Over-the Counter Analgesics and Antipyretics: A Critical Assessment. Clinical Therapeutics. 22(5).
[26] Hyllested, M., Jones, S., Pedersen, J.L., Kehlet, H. 2002. Comparative e¤ect of parac- etamol, NSAIDs or their combination in postoperative pain management: a qualitative review. British Journal of Anaesthesia. 88:199–214.
[27] Lal, Rajiv and Matutes, Carmen. 1994. Retail Pricing and Advertising Strategies. Jour- nal of Business. 67 345-370.
[28] Lancaster, Kelvin J. 1966. A New Approach to Consumer Theory. Journal of Political Economy. 74 132-157.
[29] Lee, C., Straus, W.L., Balshaw, R., Barlas, S., Vogel, S., Schnitzer, T.J. 2004. A com- parison of the e¢ cacy and safety of nonsteroidal antiin‡ammatory agents versus aceta- minophen in the treatment of osteoarthritis: a meta-analysis. Arthritis and Rheumatism.
51:746— 754.
[30] Levine, Philip. 1976. Commercials That Name Competing Brands. Journal of Advertis- ing Research. 16 7-14.
[31] Madden, Charles S., Caballero, Marjorie J., and Matsukubo, Shinya. 1986. Analysis of information content in U.S. and Japanese magazine advertising. Journal of Advertising.
[32] Miller, D., Talbot, C., Simpson, W., Korey, A. 1987. A Comparison of Naproxen Sodium, Acetaminophen and Placebo in the Treatment of Muscle Contraction Headache.
Headache: The Journal of Head and Face Pain. 27 (7) 392–396.
[33] Milsom, I., Minic, M., Dawood, Y., Akin, M., Spann, J., Niland, N. and R. Squire, A. 2002. Comparison of the e¢ cacy and safety of nonprescription doses of naproxen and naproxen sodium with ibuprofen, acetaminophen, and placebo in the treatment of primary dysmenorrhea: a pooled analysis of …ve studies. Clinical Therapeutics. Vol 24(9) pp. 1384-1400(17).
[34] Miniard, Paul W., Randall L. Rose, Michael J. Barone, and Kenneth C. Manning. 1993.
On the Need for Relative Measures When Assessing Comparative Advertising E¤ects.
Journal of Advertising. 22 41-57.
[35] Nelson, P. 1970. Information and consumer behaviour. Journal of Political Economy.
Vol. 78. pp. 311-29.
[36] Neiman, Janet. 1987. The Trouble With Comparative Ads. Ad Week. BR4-5.
[37] O'Connell, Theresa and Wilson, Ryan. 2006. Aleve Case History. Real People Cam- [38] Olson, N.Z., Otero, A.M., Marrero, I., Tirado, S., Cooper, S., Doyle, G., Jayawardena, S., and Sunshine, A. 2001. Onset of analgesia for liquigel ibuprofen 400 mg, aceta- minophen 1000 mg, ketoprofen 25 mg, and placebo in the treatment of postoperative dental pain. Journal of Clinical Pharmacology. Nov 41(11):1238-47.
[39] Ong, C.K.S., Lirk, P., Tan, C.H. and Seymour, R.A. 2006. An Evidence-Based Update on Nonsteroidal Anti-In‡ammatory Drugs. Clinical Medicine & Research. Volume 5, Number 1 : 19 -34.
[40] Oxford League Table of Analgesic E¢ cacy. 2007. http://www.jr2.ox.ac.uk/bandolier/ [41] Packman, B., Packman, E. Doyle, G, Cooper, S., Ashraf, E., Koronkiewicz, K. and Jayawardena, S. 2000. Solubilized Ibuprofen: Evaluation of Onset, Relief, and Safety of a Novel Formulation in the Treatment of Episodic Tension-type Headache. Headache: The Journal of Head and Face Pain. Volume 40 Issue 7 Page 561-567.
[42] Pechmann, Cornelia and Gabriel Esteban. 1994. Persuasion Processes Associated With Direct Comparative and Noncomparative Advertising and Implications for Advertising E¤ectiveness. Journal of Consumer Psychology, 403-32.
[43] Pechmann, Cornelia and David W. Stewart. 1990. The E¤ects of Comparative Advertise- ment on Attention, Memory and Purchase Intentions. Journal of Consumer Research.
[44] Pechmann, Cornelia and S. Ratneshwar. 1991. The Use of Comparative Advertising for Brand Positioning: Association Versus Di¤erentiation. Journal of Consumer Research.
[45] Rao, Akshay R. and Bergen, Mark E. 1992. Price Premium Variations as a Consequence of Buyers'Lack of Information. The Journal of Consumer Research. Vol. 19.
[46] Resnik, Alan J. and Stern, Bruce. L. 1977. An analysis of information content in tele- vision advertising. Journal of Marketing. 41 50-53.
[47] Rose, Randall L., Paul W. Miniard, Michael J. Barone, Kenneth C. Manning, and Brian D. Till. 1993. When Persuasion Goes Undetected: The Case of Comparative Advertising. Journal of Marketing Research. 30 315-30.
[48] Simmons Study of Media and Markets. 2000. Choices III Database.
[49] Shy, O. 1995. Industrial Organization: Theory and Applications, Chapter 11. Cambridge and London MIT Press.
[50] Stern, Bruce. L. and Resnik, Alan. J. 1991. Magazine Advertising: an analysis of its information content. Journal of Advertising Research. 21 39-44.
[51] Stewart, David W. and David H. Furse. 1986. E¤ective TV Advertising: A Study of 1000 Commercials. Lexington, MA: Lexington Books.
[52] Swinyard, William R. 1981. The Interaction Between Comparative Advertising and Copy Claim Variation. Journal of Marketing Research. 18 175-86.
[53] Thompson, Debora Viana and Hamilton, Rebecca W. 2006. The E¤ects of Information Processing Mode on Consumers' Responses to Comparative Advertising. Journal of Consumer Research. 32 530-540 [54] Wernerfelt, Birger. 1994. Selling Formats for Search Goods.
Marketing Science. 13

Source: https://www.cerge-ei.cz/pdf/events/papers/080303_t_2.pdf