The sleeper effect is a psychological phenomenon whereby a highly persuasive message, paired with a discounting cue, causes an individual to be more persuaded by the message (rather than less persuaded) over time.

 


Figure A: Normal Decay
Figure B: Sleeper Effect

 

When people are normally exposed to a highly persuasive message (such as an engaging or persuasive television ad), their attitudes toward the advocacy of the message display a significant increase. Over time, however, their newly formed attitudes seem to gravitate back toward the position held prior to receiving the message, almost as if they were never exposed to the communication in the first place. This pattern of normal decay in attitudes has been documented as the most frequently observed longitudinal pattern in persuasion research (Eagly & Chaiken, 1993). In contrast, some messages are often accompanied with a discounting cue (e.g., a message disclaimer, a low-credibility source) that would arouse a recipient’s suspicion of the validity of the message and suppress any attitude change that might occur with exposure to the message alone. Furthermore, when people are exposed to a persuasive message followed by a discounting cue, people tend to be more persuaded over time; this is referred to as the sleeper effect (Hovland & Weiss, 1951; Cook & Flay, 1978). For example, in political campaigns during important elections, undecided voters often see negative advertisements about a party or candidate running for office. At the end of the advertisement, they also might notice that the opposing candidate paid for the advertisement. Presumably, this would make voters question the truthfulness of the advertisement, and consequently, they may not be initially persuaded. However, even though the source of the advertisement lacked credibility, voters will be more likely to be persuaded later (and ultimately, vote against the candidate in the advertisement). This pattern of attitude change has puzzled social psychologists for nearly half a century, primarily due to its counter-intuitive nature and for its potential to aid in understanding attitude processes (Eagly & Chaiken, 1993). In addition, it has been the most widely studied phenomenon in persuasion research (Kumkale & Albarracín, 2004; see also Cook & Flay, 1978).

 

Controversy surrounding the existence of a “sleeper effect”

One of the more challenging aspects that the sleeper effect posed to some researchers in early studies was that the mere difficulty in obtaining the effect (e.g. Capon & Hulbert, 1973; Gillig & Greenwald, 1974). After attempting to replicate the effect and failing, some researchers went as far as suggesting that it might be better to accept the null hypothesis and conclude that the sleeper effect does not exist (Gillig & Greenwald, 1974). However, Cook and his associates (Cook, Gruder, Hennigan, & Flay, 1979) responded by suggesting that previous studies failed to obtain the sleeper effect because the requirements for a strong test were not met. Specifically, they argued that the sleeper effect will occur only if: (a) the message is persuasive, (b) the discounting cue has a strong enough impact to suppress initial attitude change, (c) enough time has passed between immediate and delayed post-tests, and (d) the message itself still has an impact on attitudes during the delayed post-test. Experimental studies conducted did, in fact, provide support for the sleeper effect occurring under such theoretically relevant conditions (Gruder, Cook, Hennigan, Flay, Alessis, & Halamaj, 1978). Furthermore, the sleeper effect did not occur when any of the four requirements were not met.

 

Past hypotheses on how the sleeper effect occurs

Because the sleeper effect has been considered to be counter-intuitive at face value, researchers since the early 1950s have attempted to explain how and why it occurs.

 

Figure A: Forgetting
Figure B: Dissociation
Figure C: Differential-Decay

 

Forgetting and dissociation

Hovland, Lumsdaine, and Sheffield (1949) first discovered the effect in a well-known study that demonstrated the delayed impact of a World War II propaganda film on American soldiers. In a subset of conditions that caused participants to question the credibility of the source in the film, participants later reported a slight increase in persuasion (much to the researchers’ surprise). After examining the results, they initially hypothesized that forgetting of the discounting cue (in this case, the non-credible source) was driving the effect. However, this premise turned out to be incorrect, because the recall measures indicated that recipients of the message were remembering the source of the communication. Consequently, Hovland and Weiss (1951) modified the forgetting hypothesis to one of dissociation. According to this reasoning, the sleeper effect occurs because the association between the discounting cue and the message in one’s memory becomes severed over time; hence, when the message is recalled for purposes of producing an attitude, the source is not readily associated.

 

Differential decay

Years later, Pratkanis, Greenwald, Leippe, and Baumgardner (1988) offered an alternative hypothesis that differed from Hovland and his colleagues. They argued that the conditions under which the sleeper effect is more likely to occur were not highlighted under the dissociation hypothesis. In addition, the requirements for a sleeper effect laid out by Gruder et al. (1978) did not detail the empirical conditions necessary to observe the sleeper effect. Based on a series of 17 experiments, the researchers proposed a theory of differential decay; that is, they suggested that the sleeper effect occurs because the impact of the cue decays faster than the impact of the message. Consequently, an overall increase in attitude change is observed at a later time. Moreover, they found that a critical requirement needed to observe the sleeper effect included the discounting cue following (rather than preceding) the message. This relatively complicated literature has been synthesized recently in a meta-analysis (see Kumkale & Albarracin, 2004).