«Consumer Reaction to Product Recalls: a Review and Extension Autoria: Celso Augusto de Matos Abstract Even though the number of product recalls has ...»
Consumer Reaction to Product Recalls: a Review and Extension
Autoria: Celso Augusto de Matos
Even though the number of product recalls has been increasing in the last years worldwide,
few studies have investigated consumer behavior toward the companies conducting these
recalls. In order to evaluate which factors significantly affect consumers’ responses to product
recalls, two surveys were conducted among Brazilian automobile owners, with 415
participants in the first study and 158 in the second one. Two models were estimated and tested by the multivariate regression analysis to predict which variables would affect the corporate image and behavioral intentions that consumers showed in relation to the company
conducting the recall. Results indicated that (i) corporate image was significantly affected by:
corporate social responsibility, blame attributed to the company and whether or not consumer owned a car made by the brand considered; and that (ii) behavioral intentions were significantly affected by: corporate social responsibility, involvement with the message, impression of danger, corporate image and whether or not consumer had a car made by the brand considered. Next, these results are discussed and compared with previous findings in the literature and, finally, the managerial implications, limitations and indications for future research are presented.
INTRODUCTIONNegative publicity can have a harmful effect on consumer perceptions (Dean, 2004, Ahluwalia, Burnkrant and Unnava, 2000). One common source of this kind of information for consumers is the product recalls, which companies make, whether compelled by the law or spontaneously, when there are potential risks for the consumers (Mowen, Jolly and Nickell, 1981).
The number of recalls has been increasing in the recent years worldwide. In the US, almost 19 million vehicles were recalled in 2002 and annual auto recalls have more than doubled since the early 1990s (Consumer Reports, 2004a). In spite of this, a large percentage of products subject to recall still remain on the road or in the home (e.g., one-third of all recalled vehicles). It was also found that more than 900 times between 1994 and 2004, products that violated mandatory federal safety standards were exported, according to commission records (Consumer Reports, 2004b).
In Brazil, 6.1 million cars were sold between 1996 and 1999 and about 734,000 (12%) of this total had some problem and were recalled (ABRAC, 2005). The percentage of consumers who responded to the recall message was 70%, which is similar to the one in the US market.
Despite this increasing in the number of product recalls and its potential negative effect on consumer perceptions, only a few studies have investigated empirically how consumers process this kind of negative information and what factors influence this process. Indeed, most of the marketing literature about product recall has a managerial perspective and a case study approach, with few exceptions (e.g. Mowen, 1980, Mowen, Jolly and Nickell, 1981, Jolly and Mowen, 1985, Siomkos and Kurzbard, 1994).
The purpose of this study was to investigate which factors influence consumer responses to product recall information. In this way, data were collected in two samples of car owners in the Brazilian context. Using regression analysis technique, two models were analyzed in order to test which of the predictors significantly would affect corporate image and behavioral intentions.
Most of the marketing literature about product recalls comes from the Public Relations arena, focusing on the managerial aspects of how to implement a product recall (e.g. Schoeny, 1992, Smith, Thomas and Quelch, 1996) or how to deal with product harm crisis (e.g.
Siomkos and Kurzbard, 1994). Few empirical studies deal with how consumers react to product recall information and what variables influence this process, with the exception of the studies by Mowen and colleagues.
Mowen (1980), using an experimental approach, manipulated variables as consumer knowledge of the company making the product recall, whether or not the company was compelled to make the recall by a consumer product safety commission, and whether or not other manufacturers had had a similar defect. Results indicated that participants regarded a familiar company as significantly less responsible for the defect than an unfamiliar company.
Contrary to expectations, when the company acted prior to intervention, it was considered more responsible.
In a non-experimental approach, Mowen, Jolly and Nickell (1981) surveyed two hundred consumers in order to investigate their perceptions of four companies that had recalled products (Ford, Firestone, Corning Glass Works, Conair). They found that consumer reactions were influenced by the knowledge that a recall had been made, the impression of danger of the defective product, the corporate social responsibility, the knowledge of recalls by other companies, and the responsibility the company had for the defect. In the multiple regression models, only the variables regarding the length on time to make recall and whether or not the company had had previous recalls did not influence significantly the dependent variable (i.e., consumer perceptions) in any of the four companies.
These authors also predicted that the number of product recalls would continue to increase for reasons such as (i) the consumer activist movement started in the early 1960’s, (ii) the activist movement taken by government agencies such as the Consumer Product Safety Commission and the Food and Drug Administration, and (iii) the increasing complexity of products. Publications such as Consumer Reports (www.consumerreports.org) have also contributed to increase the information consumers have about defective products.
Product recalls, like other types of negative publicity, can severely damage the corporate image. Previous studies have already investigated the heavier weight that negative, as compared with positive information, has on the impression formation process (Fiske, 1980, Mizerski, 1982). In order to minimize this potential negative effect of the product recall message on consumer behavior, the company making a product recall should emphasize that it is taking action in a socially responsible manner (Mowen, Jolly and Nickell, 1981).
Other factors can also influence how consumers process the product recall information, including the source of the information (i.e. the own company or an external agency) and the media used (i.e., printed, radio or TV). Jolly and Mowen (1985) found that the recall was regarded as more objective when it was presented by the government, and not by the company; and also that the print medium was believed to be more trustworthy and objective than the sound medium.
Siomkos and Kurzbard (1994) considered the product recall as one of the possible company responses in a product-harm crisis context. Although they did not test consumer reactions to product recalls, they tried to explain which factors affect consumer impression of danger in the presented problem and the behavioral intentions toward the company’s other products. They found that the degree of danger associated with the defect is small when (i) the company has a high reputation, (ii) the external effects by the press and regulatory agencies are positive to the company’s response during the crisis and (iii) the company responds to the crisis by a voluntary product recall or by being socially responsible and demonstrating concern with consumer welfare.
In a more recent research, Dawar and Pillutla (2000) have shown that consumer evaluations of the brand responding to a product-harm crisis are moderated by the consumer expectations. They found that consumers with stronger prior expectations about a firm affected by a crisis, such as the purchasers of this brand, were more likely to be aware of the crisis and sensitive to the firm response.
Involvement is usually regarded as the personal relevance given to an object, situation or action (Petty and Cacioppo, 1981, Zaichkowsky, 1985, Celsi and Olson, 1988). In the Elaboration Likelihood Model (ELM), persuasion occurs in two manners: (i) the central route when the message recipient attempts to understand the message arguments and then the person integrates all of this information into a coherent and reasoned position; or (ii) the peripheral route, when “…attitude change is determined by such factors as the rewards or punishments with which the message is associated, or the judgmental distortions that take place in perceiving the message” (Petty and Cacioppo, 1981, p.255).
It is important to note that these two routes are the ends of the continuum in the ELM and also that the level of elaboration is dependent on two conditions: motivation (involvement) and ability. This article assumes that ability is met and only addresses motivation. So, it is assumed that in situations of high (low) involvement, the central (peripheral) route will predominate.
One distinction must be made between two types of involvement (Celsi and Olson, 1988): (i) enduring or intrinsic involvement, which is related to the personal interests of the consumer, derived from past experience and stored in long-term memory; and (ii) situational involvement, more heavily influenced by situational factors which arise when making decisions on purchases. Overall, both types have been considered as important moderator in the consumer research (Petty, Cacioppo and Schumann, 1983, Poiesz and de Bont, 1995), and because of this the involvement that a consumer has with his/her car (i.e. enduring involvement) and the involvement with the product recall message itself (i.e. situational involvement) were considered as predictors of the consumer reactions to the recall information.
Method Procedures. A survey was conducted among car owners located in a capital in the southeast of Brazil. Participants in this first study were undergraduate students from two colleges. The product used as target for the product recall information was the car. A company that had announced a product recall in one of the main Brazilian newspapers was chosen, considered hereafter as XYZ.
The message of product recall affirmed that the brake system could not function properly if heated by the severe use. This message was valid for three models of car manufactured in the year of 2002 by the company researched. The message was presented in its original language (i.e. Portuguese) in the same way as it was printed in the newspapers.
A sample of 622 participants was obtained. Of this total, 415 (67%) were car owners and were used for the analysis. In the final sample, 42 (10%) owned a car made by XYZ, 372 (90%) were male and 321 (77%) were between 26 and 45 years.
Variables. Before reading the product recall message, participants answered 4 items in 7 point Likert scale, measuring involvement with cars, namely I consider the car an important product, I am interested in the car product, The car is a necessary product and The car means a lot to me. These items were chosen in order to represent the overall meaning of involvement used in this study: “a person's perceived relevance of the object based on inherent needs, values, and interests" (Zaichkowsky, 1985, p.342). Other facets for the involvement construct are also found in the literature, such as risk importance and risk probability (see for example Laurent and Kapferer, 1985, and Jain and Srinivasan, 1990, for more details).
Next, they were asked whether they owned a car or not and, if so, which brand. After that, they were asked to read the product recall message.
Just after this message, participants were submitted to the following questions: (i) whether or not they had read or heard about that information before, (ii) how important was the information presented (two items in 7 point Likert scale), (iii) how dangerous would be the car with the defect presented (one item with anchors not at all dangerous – very dangerous), (iv) how socially responsible was the XYZ company (one item with anchors not responsible at all – very responsible), (v) how much blame consumer placed on the company (two items in 7 point Likert scale). These questions were adapted from Mowen, Jolly and Nickel (1981), Pullig (2000) and Siomkos and Kurzbard (1994). See the appendix for details.
After that, five items were used to measure the corporate image, adapted from Maheswaran and Sternthal (1990), in a semantic differential scale: inferior image – superior image, obsolete company – modern company, socially irresponsible – socially responsible, low quality products – high quality products, unreliable products – reliable products.
Four items were used to measure behavioral intentions, adapted from Zeithaml, Parasuraman and Berry (1996), with anchors not at all likely – very likely and items: consider XYZ as a brand option, buy a car made by XYZ company, recommend to friends and relatives that they buy a car made by XYZ company and say positive things about XYZ company.
Finally, questions regarding gender and age were employed, mainly because in Brazilian society it is usually considered that men are more involved with cars than women.
Statistics from the governmental agency in 2003, for instance, showed that 73% of all car drivers were male. Because of this, we included gender as a dummy variable in the models.
Dimensionality and reliability of scales. Scales with more than two items were submitted to the analysis of dimensionality and reliability. Using factor analysis, it was found that all scales were unidimensional, with explained variance ranging from 63% to 86%.
Reliability analysis was conducted by computing Cronbach’s alpha for each scale that had more than two items. The values found were:.766 (involvement with the product),.905 (corporate image) and.953 (behavioral intentions). In the case there are only two items in a scale, it’s preferable to report the correlation between the items (Hulin et al, 2001). The two items measuring involvement with the message had a Pearson correlation of.367 and the ones measuring blame attribution had a correlation of.744, both significant at p.05.