Can Messaging Produce Durable Changes in Behavior?

Table of Contents

• Quick Summary & Introduction
• What is Your Guess About Message Durability?
• What the Data Tells Us
• Important Caveats
• Summing Up:  Good News for Marketers

Quick Summary

In this installment, we look at how effective intensive messaging is at creating persistent behavior change.  The good news for marketers is that, even for very complex, hard-to-change behaviors, such as exercise, diet, cancer screening and tobacco use, messaging can produce effects that last for extended time periods.  Behavior-change effects due to persuasive messaging appear to persist for at least six months.  At one year, the effects can be expected to decay by ~50%.      

Introduction

As a follow up to our prior summary of the incremental effects of repeated exposure to messages, we thought people might also want to know about the durable power of message exposure on customer behavior.  Both questions relate to changes in message impact across time, but they are almost opposite sides of a coin, which is clear when we state the following fundamental questions.

• What additional lift do you get when people have more exposures to a message?  (see prior post “How Many Message Exposures are Needed?”)
• Do changes in behavior resulting from persuasive messaging persist over time?

As with prior posts, we will look to the now enormous body of experimental work on message effectiveness that has accumulated in the last two decades to answer this question.  

What is Your Guess About Message Durability?

A reasonable ingoing expectation here might be that message durability is likely to be pretty bad.  Many of us learned in undergraduate psychology that recall decays fairly predictably according to a power function.  We also have various studies that show relatively sharp decay effects from things like attitudes shaped by political advertising (see for example, Hill et al, 2013).  But there are good reasons to anticipate messaging effects on health behaviors could operate differently.  First, neuroscience and cognitive science are revealing that patterns of retention and forgetting are much more complex, and often strikingly more durable than traditional models have suggested (see Radvansky et al, 2022).  So, it’s entirely possible that information could stick with a message recipient. 

2 Key Drivers of Retention:  Attention and Salience

One key driver of retention is attention, where we are more likely to forget things that we are not really engaged in.  Another huge factor is content salience.  Specifically, we should parse the effects of persuasive communication on the basis of content that is innately salient to the recipient versus content that is either innately non-salient or only momentarily salient.  Salience can be established in various ways but is often driven by elicited emotion (Hamann et al, 1999) or by the degree of alignment between the message content and the momentary context the person is in (Simola et al, 2013). 

Different Types of Salience

To get a feel for why some content is salient and other content is not, consider information presented in a political advertisement about a candidate running for a public office.  The odds of us retaining that information or having it shape our behavior on an extended basis are low for the simple reason that the salience for most of the ad viewers is low.  After all, you are talking about a person that most people will never meet, and you are focusing on a single behavioral event:  voting.  In contrast, some health behavior messaging can be more naturally salient, for the simple reason that we are sometimes talking about the effective functioning of a person’s own body. 

What the Data Tells Us

There are several healthcare-specific meta-analyses that can inform the question of message durability. 

Health-Related Messaging Data Example #1:  Immediate vs. Time-Delayed Effects

The first is a 2013 study by Mia Lustria and colleagues, which looked at the use of web-based message delivery on both immediate and downstream effects of health-related messages on complex health behaviors.  This included measures of exercise/physical activity, smoking cessation/restriction and diet – all of which are incredibly challenging to influence (as discussed in our post called “Why We Can’t Expect the Same Level of Impact from Healthcare Messaging”).  Their study covers many topics, but the most important for the purposes of this discussion are shown in the table below.  Keep in mind that these effect-size statistics (r_ES) are derived from studies that compared tailored messages to control conditions involving generic, population-based messages, so they are incremental to effects you would expect to get from a generic campaign.  Their analysis showed that “follow-up” (i.e., time-delayed) behavioral effects were essentially identical to effects measured immediately after the web-based messaging intervention.  This certainly suggests that intensive messaging can actually have durable effects, which may run contrary to both intuition and some experimental work on consumer attitudes (see e.g., Hill et al, 2013). 

Table 1:  Impact of Messages on Behavior Change – Immediate vs. Follow-Up Effects

The Lustria et al analysis does not summarize the average timing of the follow-up measures; however, based on my own reading of many individual papers, the follow-up measures are rarely captured at time intervals beyond 1 to 3 months.  Also, the authors describe each study manipulation in some detail.  Many of them include multiple touch-points, follow-up communications and represent a time-extended experience in which the participants are getting message exposure.  This is not necessarily unrealistic, as client organizations certainly work to create similar immersive experiences for patients whenever they can.  Nevertheless, we need to be realistic about expecting anything like these kinds of effects with single-exposure messaging. 

Health-Related Messaging Data Example #2:  Effects of Time-Since-Exposure 

So far, so good, in that we have some evidence that complex health behaviors can persist after message cessation.  But it would be good to get a better sense of the actual relationship between time-since-exposure and persistent behavior change.  An excellent perspective comes from a 2010 meta-analysis by Krebs et al, which includes an impressive analysis of the durability of messaging effects on a similar set of healthcare behaviors.  Analogous to the Lustria et al paper, their synthesis focused on time-based assessments of messaging and its impact on complex health behaviors, such as eating fruits and vegetables, exercise, alcohol consumption and cancer screening. They were only interested in studies that measured messaging effects across time, so their report does not include any dependent variables measured immediately after the message intervention.  Messaging was variable, but all involved web-based delivery tailored to some aspects of patient psychology or demographics. 

Table 2:  Sustained Impact of Messages on Health-Related Behavior Change Across Time Intervals

The Krebs et al data suggest that tailored health messaging can produce relatively impressive durability, with the mean effect size essentially unchanged for up to 6 months, and a stepwise decay thereafter.  Past the 12-month mark, the effect size has dropped to about 50% of its original post-intervention level.  In reviewing their findings, please keep in mind that you have different mixes of studies at each time interval.  Also, the observed effect sizes are in the perfectly normal range for tailored healthcare messaging interventions (cf. Joyal-Desmarais et al’s healthcare r_ES = .125), and it should be noted that I’ve converted the Krebs et al reported effect size format (Cohen’s g) to the more standard r-based effect size to make this easily comparable to what we have reported in other posts. 

Other Types of Messaging Data Example #3:  Effects of Time-Since-Exposure

A final perspective comes from a 2022 meta-analysis of messaging effects by Joyal-Desmarais et al.  I place this analysis last because it is not exclusively focused on healthcare, and because it includes endpoints like attitudes and intentions, rather than exclusively focusing on behavior change.  Still, as a synthesis of over 700 experiments, I think it has something to tell us.   I’ve excluded the “day of” evaluations because our goal for this summary is to focus on lasting rather than immediate effects.  Although the time increments classified by the authors diverge from those of Krebs et al, their data suggest that the same kind of retention effects are observed generally wherever persuasive communication is used.  It is possible that the slightly faster decay of message impact seen in this analysis is due to the fact that things like attitudes and intentions do not have the same kind of self-reinforcing properties that overt behavior can have.  Additionally, many of the experiments in this meta-analysis use less intensive and less immersive messaging. 

Table 3:  Sustained Impact of Messages on Various Endpoints Across Time

Important Caveats

Two things are worth keeping in mind here.  First, experiments designed to measure time-based effects of messages are set up in diverse ways.  Many studies of message impact look at “day of” effects – that is, what impact did the message have on an attitude or intention more or less immediately after reviewing the content.  We want to restrict our assessment only to studies that (a) look at multiple time points and (b) are explicitly looking at behavior change.  Second, in many cases (but not all), when researchers study message effects in academic settings, they are often thinking about stimuli differently than many marketers do.  A message stimulus in a formal experiment will often involve comprehensive written or multi-media interventions, often on a time-extended basis (e.g., a 45-minute video).  Additionally, in some cases, the meta-analyses include studies that involve multiple message exposures over time.  For the present discussion, it is rare for the study to examine the long-term impact of exposure to a small amount of text (i.e.., a message element) because it is assumed that such exposure would not have much durable effect.  As such, when we are looking at the data for this posting, please imagine that participants in these studies have been exposed to something analogous to a complete drug detail by a trained sales rep using visual aids and that the core messages have been reinforced on at least a few occasions. 

Summing Up:  Good News for Marketers

Collectively, these findings should be taken as very positive for marketers.  They show that even notoriously difficult-to-change behaviors have durable responses to well-developed message campaigns.  It is reasonable to expect comprehensive messaging to yield effects that persist for 6-12 months after discontinuation, with roughly a 50% decay in impact after one year.  Further, we think it is reasonable to suspect that easier-to-execute behaviors would be both more responsive to messages and longer-lasting.  This assertion stems from the fact that the endpoints summarized here represent a very tough benchmark.  We feel it is reasonable to expect that patients enrolled in programs designed to support behavior change would likely have similar effects, as they would typically use similar messaging content and delivery systems (e.g., Web-supported, SMS text follow-ups).  Similar results would seem reasonable to expect from physicians exposed to comprehensive detailing, either directly via a rep or in an immersive web-based setting. 

Relevant Topic:  We also look at how messaging gives us the power to change a variety of customer endpoints, including product attitudes, beliefs, intentions, and behaviors in the post “Persuasive Communications Can Shape Any Endpoint that Matters to Marketers“. 

To learn more, contact us at info@euplexus.com.

About euPlexus  

We are a team of life science insights veterans dedicated to amplifying life science marketing through evidence-based tools.  One of our core values is to bring integrated, up-to-date perspectives on marketing-relevant science to our clients and the broader industry.