Much campaigning is about trying to change attitudes, and, ultimately, behaviours. With a variety of messengers, channels, and messages, campaigns probe various motivators to influence people’s ideas.
Yet what influences the effect of a message on an attitude? In my experience there is little explicit understanding of how to influence attitude-formation. While effective tactics seem to be used more often than ineffective tactics, without a model for how to persuade people, it is hard to refine a message’s potency and target it appropriately.
The Elaboration Likelihood Model of Persuasion (ELM) is one model for how attitudes are changed: “a fairly general framework for organizing, categorizing, and understanding the basic processes underlying the effectiveness of persuasive communications.” Richard E. Petty and John T. Cacioppo outlined it in a 1986 paper of the same name. Before this, psychologists faced an incoherent set of findings that lacked a model to explain why particular message variables had impacts that mutated not only in magnitude but also in direction, depending on unclear circumstances. The ELM sought to explain these observations.
In this post I will briefly outline the model’s most relevant postulates, then document the findings that are most significant for campaigning. I will discuss how these findings support existing practice and highlight some places where practice might be changed.
The Elaboration Likelihood Model
1. People process information in one of two ways: through either a central route or a peripheral route.
When people are presented with information, their brain may treat it in one of two ways. The information may be carefully reviewed on its merits, “elaborated”, the “central route”. Or it may influence attitudes more subconsciously, without scrutiny, through some other cue – the “peripheral route”.
This is where the model derives its name. The extent to which the brain processes the information on the basis of its “issue-relevant arguments” is the extent of the “elaboration”.
2. Whether the central or peripheral route is used depends on the motivation and the ability to process the information.
People may feel motivated to process the information if: it’s relevant to them, they enjoy complex thought, or they alone are responsible for acting on the information. Higher motivation inclines people towards the central route. Greater ability to process ensues when distraction is reduced, when people have prior knowledge, and when comprehension is facilitated by repetition or clearer communication. Greater ability to process inclines people towards the central route.
Thus, when motivation and ability are present, the “elaboration likelihood” (EL) is elevated, and it’s more likely the audience will consider the argument on its merits.
3. When the EL is elevated, an argument is processed on its merits, and attitude change depends on the merits of the argument.
On the other hand, when the EL is low, factors irrelevant to the actual argument – “peripheral cues” – tend to dominate in determining how information affects attitudes.
These peripheral cues can be a number of things: the authority of the source, the familiarity/likeability of the source, the number or length of arguments, or the emotional affect.
4. Processing can be “biased” in such a way that it is inclined to a particular attitude regardless of the evidence.
Bias can exist due to “prior knowledge” or can be created by “forewarning” of either a message’s content or its persuasive intent.
5. Attitude change that occurs via the central route lasts longer, has more influence on behaviour, and is more resistant to counterpersuasion.
And there you have it!
Campaigns & the Elaboration Likelihood Model
The ELM gives us a number of variables that may apply in a “persuasion context” like campaigning: the strength of the message, the level of audience motivation, level of audience ability, the audience bias, and the presence/absence of peripheral cues. Only some of these variables can be influenced by campaigners. For example, the cognitive or logical strength of the message is an independent variable which the campaign can control – although we can assume that a campaign would always aim for the strongest possible message.
While some campaigns might be able to choose to address different audiences which vary in their motivation and ability levels, others may be restricted to a particular audience which might be fixed in terms of its motivation, ability to process, or bias. In this case, other variables can be manipulated to optimise influence.
A note on message quality
For their testing, Petty and Cacioppo determined message strength by asking subjects to rate messages for how persuasive they were. For example, “we should pay more tax to fund better hospitals” would be rated as more persuasive than “we should pay more tax to have more skate parks”. This allowed them to test the effectiveness of strong messages against weak messages.
Notably, weak messages aren’t just less effective than strong messages, they can be counter-effective. When EL is low, message strength has little influence. But when EL is high, weak arguments turn people against an attitude. The argument “we should pay more tax to have more skate parks” wouldn’t only create less positive impact, it would in fact create negative impact.
This lesson thus over-rides all others: if your messages are not persuasive, they will work against you. Before you think about how to wrap the parcel and when to deliver it, first make sure the gift inside is good.
Now, before we go on, I’ve tabulated the postulates from above, and included this warning, for your easy reference while reviewing the points below.
So what are the most important implications of the ELM?
Increasing personal relevance increases the elaboration likelihood which makes strong messages more effective.
As noted above, increased relevance increases motivation to process a message via the central route, which enhances its impact relative to weak messages (which can be equally effective if the peripheral route is taken). In addition, the attitudinal change lasts longer, influences behaviour more, and is less mutable.
Personal relevance increases when the audience expects the issue “to have significant consequences for their own lives”. In their experiments, Petty and Cacioppo increased relevance by telling subjects that a proposed policy would take effect at their university (as opposed to a distant one), or that it would take place next year (as opposed to in ten year’s time).
For campaigners, the more relevant your message is, the more powerful it will be. If people think your message is of significant consequence to their own lives, they’ll think about it more, be more influenced by it (if it is strong), and be influenced for longer. This feels very relevant to climate campaigning, in which consequences have often been framed as distant in time (2 metres sea level rise by 2100!) or distant in space (the Maldives could be submerged). More recently however, climate communicators have done more to situate impacts locally, both in space and time: the Climate Council’s work emphasises how climate change is already affecting Australia. The next step is perhaps to further link these contemporary climatic changes to people’s everyday existence – how much they pay for insurance, or what bananas will cost them.
This finding is also relevant to electoral campaigns. During the ACT Greens 2013 Federal election campaign, I doorknocked extensively. A key talking point for the campaign was public sector job cuts, which threatened the livelihoods of easily more than ten thousand Canberrans. Yet it remained a challenge to make this relevant to each individual audience, with APS employees tending to think it wouldn’t happen to them, and others workers blind to the flow-on effects of such mass sackings. While the campaign honed messages to better convey the relevancy of this issue, it remained apparent that many people didn’t feel the issue of consequence in their own lives.
Forewarning of persuasion attempts enhances resistance to persuasion
“Forewarning” is a way of biasing processing to reduce the efficacy of persuasion efforts. In effect, it introduces ‘attitude inertia’, making it more likely that someone will stick with, or in fact feel more strongly, their original beliefs. It can be achieved by warning of the message content, or of the message intent (ie, to persuade.)
Generally, it’s understood that forewarning of message content works by affording an opportunity for people to muster arguments in support of their initial positions and thus be more prepared to resist persuasion. Warnings on the intent of messages motivate active counter-argument against the persuasion attempt, weakening its influence. It’s also suggested that people resent persuasion attempts as an attack on their freedom to hold attitudes, and they react to the message with hostility as a way of asserting their own autonomy (Hass & Grady, 1975).
Good union organising practice is consistent with this finding. Part of a recruitment conversation is “inoculation”: telling people to expect bosses to argue against unionism and to use certain messages. The new union member, alerted to an upcoming assault by the boss on their attitudinal freedom, is motivated to assert their attitudinal freedom by rejecting the boss’s arguments. The forewarning also provides the chance to mobilise counterarguments in favour of union membership. Of course, bosses practise the same technique: many a union organiser has walked into a staff induction session to behold a wall of incredulous faces, doubtless forewarned by the boss of the persuasion attempt that the organiser would make.
Electoral campaigns can also apply this finding to reduce the effectiveness of opponents’ persuasion attempts. Materials or voter contact scripts could include a warning (which, in the case of personal contact, could be given only to voters of a certain inclination) that the opponent will attempt to persuade them and will use certain messages. In itself this will have an inoculating effect; the effect could be increased by supplying counter-arguments or, especially powerfully, asking a person to generate their own counter-arguments (Elms, 1966; Watts, 1967).
A last note – inexpert activists are at risk of forewarning of their own persuasion attempts and thus hamstringing themselves. Doorknockers might open a conversation: “I’m here to tell you why you should vote for my candidate.” Someone making fundraising calls might say: “Hey, how’re you going? I’m just ringing to ask for money for our campaign.” Once forewarned, the audience resents that an attempt is being made to influence them and reacts with greater resistance. Fortunately, with a modicum of awareness and training, this tendency can be mitigated in a wholly authentic fashion.
Increasing the elaboration likelihood increases the effect of attitude on behaviour
Ultimately, most campaigns don’t just want to change attitudes, they want to change behaviours: they want people to divest from their bank, to vote a certain way, or to call-out sexism. The good news: behaviour change can be precipitated by attitude change (though it doesn’t have to be – see Chris Rose’s “VBCOP” for an alternative). The tricky news: this relationship is stronger when the attitude change occurs through a situation of high elaboration likelihood. That is, it happens via the central route, when distraction is minimised, relevancy is maximised, and a message is repeated. In such contexts, write Petty and Cacioppo, “attitude-behaviour correlations were higher than when elaboration was low”. (Section VIII. B)
This news is tricky because I feel that campaigns could fairly easily get this right, but they fairly often get it wrong. A case in point: the “Say Yes Australia” TV advertising campaign in favour of pricing pollution. To be effective, this campaign needed to be able to influence voting behaviour, not just attitudes – otherwise the political calculus wouldn’t be shifted in favour of climate action. As we have seen, the attitude-behaviour link is strengthened by increasing elaboration likelihood. To an extent, a TV campaign does achieve this by repeating a message and increasing audience’s ability to process. However, it also takes place in a high-distraction environment. Furthermore, peripheral cues like celebrity advocates enhance persuasion only in the case of low elaboration likelihood (see point 3, above). So, while celebrity advocates may have made the advertisment more effective, they also corroborated its use of the peripheral route to persuasion – which is less likely to motivate behavioural change and is less lasting and less durable.
A positive example is the divestment campaign initiated by Bill McKibben’s Rolling Stones article, “Global Warming’s Terrifying New Math“. This campaign increases elaboration likelihood to make its messages effective in motivating both individual divestment and activism in the divestment movement. McKibben’s article was in a written form (which has been shown to increase ability to process) and rather long (targeting an audience with a high “need for cognition”, which is more likely to elaborate on arguments).
Subsequently, McKibben spoke in Australia (as well as other places) as part of the “Do the Maths” tour. This format further increased elaboration likelihood by providing information through more direct experience, in a focused, low-distraction environment. The structure of the article and talk also emphasises the urgency of the climate crisis and portrays the fossil fuel industry as the villain before casting divestment as the heroic response. This increases the personal relevancy of the message, again increasing EL. In combination, these create a message with a high likelihood of not only changing attitudes, but influencing behaviour. In fact, the current fossil fuel divestment movement is the fastest growing in history.
This point should influence how electoral field campaigns are run. It makes it clear that changing someone’s behaviour in the form of changing their vote is best achieved in a certain context – that in which the EL is high. While something like “maximise relevancy” hopefully doesn’t need to be said, a lesson like “minimise distraction” or perhaps “target people who are less distracted” may be worth sharing. This could be an argument for manipulating tactics to reach people when they are less distracted. This could favour certain tactics: canvassing and phonebanking have the advantage of reaching people when they are less likely to be engaged in other tasks. It could also favour certain contexts for tactics: one might flyer at quiet suburban shopping centres, where people have time to stop and chat, instead of at a busy commercial exchange. Finally, noting that high EL implies more lasting persuasion effects, a field campaign might tend towards higher EL encounters further out from election day (for example, targeting areas with a higher rate of tertiary education), before using tactics closer to election day that might be expected to offer less lasting results.
Yes, it is an elaborate model
Look, this is a long post. But it’s much shorter than the sixty-eight pages I threshed to harvest these kernels of wisdom.
Allow me summarise:
On the Elaboration Likelihood Model, let me paraphrase the authors’ own words:
- There are “two basic routes to persuasion.”
- “One route is based on the thoughtful (though sometimes biased) consideration of arguments central to the issue,
- “the other is based on affective associations or simple inferences tied to peripheral cues in the persuasion context.
- “When variables in the persuasion situation render the elaboration likelihood high, the first kind of persuasion occurs (central route).
- “When variables in the persuasion situation render the elaboration likelihood low, the second kind of persuasion occurs (peripheral route).
- “Attitude changes via the central route appear to be more persistent, resistant, and predictive of behaviour than changes induced via the peripheral route.”
On its implications for campaigning: the persuasiveness of a message is affected by many variables operating simultaneously, only some of which are within your control. These variables include the strength of your message, the audience’s motivation to process it, the audience’s ability to process it, and the presence or absence of peripheral cues. Using the Elaboration Likelihood Model of Persuasion, a campaign can identify the state of different variables for a given context, and change or adapt other variables accordingly. This increases the chance that a message will influence the attitude in the desired way, as well as its prospects for creating lasting change, for affecting behaviour, and for resisting counter-argument.