To my ear, the term “fearmongering” has a pejorative connotation, but in writing this post I was surprised to find that my dictionary did not indicate this in its definition:

“the action of deliberately arousing public fear or alarm about a particular issue” (New Oxford American Dictionary)

“Fearmongering” would then seem to be a suitable label for the subject of a recent historical review and accompanying commentary in the American Journal of Public Health. The former, by Fairchild and colleagues (2018), examines the use of “fear-based messaging” in public health campaigns and highlights past experience in public health responses to tobacco use and HIV transmission during the early days of AIDS. The subject of discussion for this post is Chapman’s (2018) accompanying editorial. Chapman writes in support of the empirical and ethical justification of fear-based messaging in addressing important public health challenges. In doing so, he presents a set of “five criticisms” of fear-based messaging campaigns (FMC), allegedly common in public health discourse, and provides his rebuttals. I do not find his argumentation compelling, but for various reasons I believe it warrants open commentary here. In this first post I will dissect Chapman’s alleged criticisms and his rebuttals. In the next, I will assess his commentary on the larger issue and explore the nature of the topic and its treatment from an ethics perspective.

Chapman’s alleged criticisms (C#) and his responses are presented (paraphrased with best intentions) in number-order below, with my commentary after each.

C1: FMC is ineffective.

Chapman: (i) FMC is effective, and (ii) sustained criticism based on effectiveness appears to be mostly leveled against FMC in comparison to more “positive” messaging campaigns, “suggesting this criticism is enlisted to support more primary objections about fear campaigns.”

Phelan: I find this criticism relatively uninteresting. Certainly, if a method is wholly ineffective (operationalized as e.g. average net effect ≤ 0), then from a consequentialist perspective its implementation constitutes at best a waste of resources and at worst a net harm to someone involved. However, evaluation of whether FMC has a non-negligible effect on target behaviors is a question for epidemiological study, and I am willing to concede Chapman’s assertion that the criticism is false, both because I’m not interested in debating the substance of evidence on the issue, and because the alternative makes discussion of ethical aspects of FMC unnecessary. If we are justified in believing that FMC offers negligible positive value, then we should not employ it, regardless of our judgment of its ethical acceptability.*

What is more interesting is Chapman’s suggestion that claims of ineffectiveness are selectively mustered against FMC (relative to all types of public health campaigns). Even if selectivity of this kind exists, it need not bear on the truth of the claims leveled. If we consider the hypothetical where FMC is ineffective and stands in contrast to other public health messaging techniques for this reason, we should not be surprised that criticism like C1 is advanced selectively against FMC, for this would be most in line with the truth! I take the implication Chapman draws in (ii) to mean that proponents of C1 primarily intend it to bolster other criticisms which are otherwise ostensibly standalone, rather than that proponents are being insincere in asserting C1. I do not see that this implication bears directly on the truth of C1, for the same reason as the putative selectivity does not. We shall return to the support relations of C1 in discussion of C4.

*I am ignoring here the question of any ethical duty we might have to implement FMC despite an expected lack of effectiveness. I am comfortable considering the question of effectiveness of to be “prior to” the question of ethicality in this case.

C2: FMC’s target victims, not causes, “and so are soft options mounted in lieu of [more substantive measures.]”

Chapman: (i) Virtually all major prevention programs entail both “policy reforms and motivational interventions.” (ii) The implication (C2a) “that until governments are prepared to embrace the full panoply of policy and program solutions to health problems, they should not implement any individual element of such comprehensive approaches” is incorrect. (iii) Messaging to promote positive personal health behaviors “will always be needed.” (iv) C2 falls to a reductio ad absurdum entailing that “attention-getting warning signs and poison labels are unethical.”

Phelan: If Chapman’s assertion in (i) is true, this would constitute a severe blow to C2, but we should note that its relevance to C2 is contingent on the second clause in the latter (i.e. what I will call the ‘alternative adoption assertion’ that FMC is employed as an alternative to comprehensive reforms). The core of C2 appears to be the ‘targeting distinction,’ where it may be morally relevant (if true) that FMC targets victims rather than underlying causes of undesirable behaviors. I do not see that this targeting distinction necessarily entails alternative adoption, and so while I am willing to grant Chapman the truth of (i), I would truncate C2 to only the targeting distinction (C2*). In this form it provides only an implied argument against FMC, which I would state as (C2b) that the ethical character of FMC is thus inferior to alternative approaches which target causes (and that such alternatives are thus ethically preferable to FMC). Evaluation of C2b appears far more interesting than C2, and escapes Chapman’s responses.

Chapman’s assertions in (ii) and (iii) are naturally appealing, though I am skeptical that the proposed implication (C2a) necessarily follows from C2. In any case, the defeat of C2a by (ii) does not bear directly on C2.

Is C2 susceptible to the reductio Chapman furnishes against it? As presented, (iv) in fact appears entirely irrelevant to C2. Neither the targeting distinction alone nor its combination with the alternative adoption assertion constitute an argument which can be reduced in the manner that Chapman suggests. Only C2b (which does not appear in Chapman’s discussion) connects any assertion about FMC to an ethical decision about its use, and even it is not susceptible to the proposed reductio. A susceptible argument must ascribe a judgment of unethicality to some characteristic of FMC, such that this characteristic is shared by more modest and apparently benign interventions such as warning labels. Clearly neither C2 nor any of its implied conclusions fall prey to such a reduction.

If this dismissal of (iv) seems hasty, we can give maximal leeway to Chapman’s intent by attempting to construct a plausible implied argument from C2 to connect FMC and his end scenario of warning labels. Note that the alternative adoption assertion is not relevant to (iv), so we can begin with only the targeting distinction (C2*). We can surmise that “targeting of victims but not causes” is the relevant characteristic of FMC, and that because it bears this characteristic it should be categorically ruled unethical. Such a judgment of course requires justification (as discussed above), but let us ignore that for the moment. If such targeting is the only criterion for categorical rejection, then all other interventions bearing this characteristic can be deemed unethical as well. If we can then ascribe this characteristic to hazard signs, warning labels, and the like, then we can argue that C2* enables the reductio that leads us to conclude that warning labels must be unethical by the same reasoning. Is such an ascription reasonable? I don’t think so. First, do hazard warning labels target victims? Unlike messaging directed at individuals already engaging in a negative behavior (e.g. health warnings on cigarette packs, which are essentially only purchased by smokers), warning labels are neither intended nor primarily received by victims of accidental injury (e.g. poisoning, electrocution, etc.). Second, do hazard warning labels ignore causes? To evaluate this we must assess what causes may be at play and/or targeted. I presume that a major proximal cause of accidental injuries is hazard unawareness, which is clearly addressed by warning labels. Other factors are almost certainly contributory causes (e.g. insufficient security of hazardous materials or equipment, system-level procedural problems amenable to human factors engineering solutions), just as there are many causal factors involved in behavioral promotion of smoking. However, whereas anti-tobacco FMC might be said to fail to target any meaningful causes (where personal action is generally considered a product of addiction rather than choice), warning labels clearly address a major causal factor driving undesirable outcomes. Thus, a brief evaluation finds it difficult to employ the proposed reductio against C2*, and even were it successful, the problems remain that C2* is not the criticism Chapman presents, and more importantly, no justification for the requisite ethical judgment from C2* is given. Thus, (iv) fails even though C2 is weakly constructed.

C3: FMC creates perceptions of criticism, devaluation, rejection, or stigmatization in those targeted; negative impacts on health result, and are both predictable and ignored.

Chapman: Arguments from C3 are “mostly” applied to FMC targeting “personal behavior, as opposed to public health and safety.” Chapman gives a long list of other-harming behaviors as contrast to the personal behavior subjects, and says (sarcastically?) “some people deserve to be stigmatized, apparently.”

Phelan: I should first observe that Chapman’s response to C3 does not constitute a rebuttal, as he provides no reason we should discount this criticism. Instead, he again posits the existence of selectivity in the application of C3, in this case to FMC’s of a certain kind. As with C1, I see no reason to believe that such selectivity should bear on the soundness of the criticism. Furthermore, I find Chapman’s contrast examples puzzling.

Notwithstanding that his suggestion of deserved stigmatization may be sarcastic, the difference between the personal health and other-harming behaviors listed seems highly relevant. Consider that C2 conceives of individuals engaging in self-harming (but not necessarily other-harming) personal behaviors as victims, whereas Chapman’s examples (e.g. sexual predators, domestic abusers) carry strong connotations of victimization of others only. Both diet-induced obesity and domestic abuse are important public health problems, and both are likely influenced by external factors driving personal behavior (e.g. food environments, exposure to or personal history of domestic abuse). In some sense the agents of interest in both cases are suffering the influence of external factors leading to undesirable behavior and consequences. However, the implication that these respective classes of social problems are meaningfully comparable in any sense relevant to FMC seems disingenuous.

Smoking may be a useful example here because it is often both self- and other-harming: an addicted smoker is naturally seen as a victim despite driving self-harm through their own behavior, but likewise we see as victims the others exposed to secondhand smoke. I argue that the idea underlying C3 is that FMC is unjustifiable to the extent that it addresses the self-harming nature of the target behavior, for in this way it ignores that those targeted are invariably victims of external influences driving their behavior. If FMC targeting the other-harming nature of target behaviors does in fact result in negative health effects on those targeted, I should think that these would often be outweighed by sparing others from harm that they would have otherwise suffered (at the hands of those targeted).

In fact, on this note I find Chapman’s (presumed) sarcasm surprising; surely we should desire a public sentiment that other-harming behaviors are deplorable, and what is this to offenders if not stigma? The judgment that some such stigma is deserved thus strikes me as defensible!

C4: Health-related behavior changes (presumably those desired to result from FMC) are difficult and multifactorial beyond the individual.

Chapman: (i) Views of human behavior that essentially deny agency are extreme and relatively untenable, except through the perspective of strict determinism. (ii) “It is instructive, for example, to reflect that today in many nations, it is only a minority of the lowest socioeconomic group who still smokes.”

Phelan: If you find (i) confusing, I’m right there with you. It would seem from Chapman’s phrasing that if I endorse a strict determinist view of behavior, my judgment that FMC is futile (for example) is exempt from his critique of extremity. This leads to the somewhat strange result that I can advance C4 against FMC with no explicit rebuttal, though (ii) must of course be considered.

Perhaps more importantly, C4 requires a closer look to determine exactly how it applies to FMC. One possible interpretation (C4a) is that because individually-driven behavior change is difficult, FMC is bound to be ineffective. This of course implies C1, which Chapman rejects. An alternative interpretation (C4b) is that because individually-driven behavior change is difficult, FMC is leveling an unfair charge at its targets. For example, we know it’s difficult to lose weight, and anti-obesity FMC engenders negative emotions in overweight and obese individuals, so if we employ FMC we are risking emotional distress in people with little chance at benefitting from it anyway. To some extent this also seems to entail C1, but it need not be so strict as to do so. Some people lose weight intentionally, just as some people quit smoking intentionally, so public health initiatives addressing these behaviors certainly have the potential to impact health at the population level. We can recognize this and still understand that the majority of people will fail in their efforts, even if driven by fear. The thrust of C4 thus seems to be that there is something morally significant about fear such that it is wrong to impose it if there is a sufficiently low probability of relieving it through successful efforts. Such a property of fear might cast FMC as unethical such that it harms its targets in these cases, and/or such that we fail in some duty to them by employing it. As presented, C4 does not constitute a complete criticism of FMC in an ethical sense.

Returning to Chapman’s response (ii), is it in fact instructive to consider his observation on the current prevalence of smoking among lowest-SES groups? The implication of (ii) seems to be that, since the lowest-SES groups bear the highest burden of factors favoring smoking (relative to higher-SES groups), the fact that the remaining smokers are in the minority of these groups indicates that smoking cessation (or perhaps abstinence) is feasible for individuals to pursue in even the most challenging circumstances. The most obvious problem with this implication is that there are many potential reasons why the prevalence of smoking in these groups has (presumably) declined other than personal agency driving behavior change. For the sake of argument, I will assume Chapman’s reference to these data is legitimate and the prevalence has decreased because people are quitting smoking and/or never starting (rather than smokers dying proportionally faster/earlier than non-smokers, for example). Even so, why should we attribute this trend to personal agency? What of the many factors that often make behavior change difficult in the first place? Might the circumstances of these particular lowest-SES groups have changed over time to tilt the balance (at least partially, if not overall) in favor of non-smoking? I am thinking here of the various structural interventions intended to decrease smoking at the population level, such as tobacco product taxation, bans on smoking in public places, etc. Thus, I do not find Chapman’s example instructive in refuting C4 (or C4b).

C5: “…we should always avoid messaging that might upset people.”

Chapman: [Chapman states that “two subtexts” underlie this argument, but lists only one] There is an implied premise that we have (roughly) a duty to avoid challenging others’ judgments, but such challenges are common and sometimes lead to change (presumably positive), despite making us uncomfortable.

Phelan: This appears disingenuous; surely we should find a morally significant difference between the slightest negative emotional reaction elicited versus the most severe. The former may merely produce a momentary feeling of discomfort, but the latter may (in the extreme) constitute psychological trauma and produce cognitive and behavioral dysfunction. Furthermore, almost any conceivable message has a plausibly non-zero probability of upsetting someone, so if we must “always” avoid these cases then almost all messaging would be prohibited by this dictum.

Accordingly, I reject the absolutism in Chapman’s construction of the fifth criticism. I find that the phrasing (“always”, “might”) presents a straw man of sorts, an unrealistic criticism against which Chapman can easily advance counterarguments. Consider an emended(*) fifth criticism:

C5*: We have a duty to avoid messaging that we can reasonably expect to be unacceptably upsetting (or distressing, disturbing, etc.) to its recipients, where the weight of this duty is (for example) proportional to the degree of unacceptability.

This construction avoids the original absolutism and allows for distinction between degrees of upset. Here I intend “reasonably expect” to refer to the typical “reasonable person” standard (where further explication is beyond the scope of this discussion) and “unacceptably upsetting” to denote a relative judgment which must be made by those deliberating on the messaging. Ideally such deliberation would include input from (at least) a representative sample of the intended recipients.

To the extent that such a duty requires justification, consider the use of gruesome images (e.g. traumatic injuries) in FMC intended to promote awareness of and disinclination from drunk driving. Presumably the goal would be to inspire fear by forcing recognition of the connection between drunk driving and terrible injuries, and so make potential drunk drivers afraid of injury befalling themselves or others by their own hands. However, fear is only one emotion potentially elicited by such imagery; disgust (as distinct from fear) seems far more likely to accompany depictions of gore than less graphic imagery such as wreckage devoid of blood and the like. This is not to say that fear itself should not qualify as “upsetting” to people (it seems natural that it should), but rather that Chapman’s focus on the broad category of “upset” does not distinguish between the type or source of discomfort. I would argue these distinctions are important, since the potential for psychological harm (e.g. intrusive remembrance, nightmares) may well be greater for imagery with shock value. In any case, these issues are entirely separate from the risk of “off-target” effects: it is logistically near-impossible to target FMC only to members of the public for whom behavior change is desired. In the drunk driving example, public displays (e.g. billboards, posters) are visible to all, including children, non-drivers, non-drinkers, etc. For these individuals, there is far less potential benefit (if any) in their exposure to FMC, whereas the potential harms are not similarly reduced (and in fact may be magnified, as in the case of children). Even in the implausible situation where only actors capable of the relevant behavior are targeted, their risk of the behavior is not necessarily commensurate with their risk of FMC-related harm.

Of course all of this does not imply that any potential harms should disqualify FMC as an option, but rather that they should be examined both in detail and in context. Finally, it should be apparent that C5* escapes Chapman’s responses, as they only apply to the absurd absolutism in his original construction of C5.

Summary - My Appraisals

In summary, I would assess Chapman’s presentation of the “five criticisms” as follows:

C1: Fairly constructed, and Chapman’s response conceded.
C2: Weakly constructed, but Chapman’s responses fail.
C3: Fairly constructed, but Chapman’s responses fail.
C4: Incompletely constructed, and Chapman’s responses fail.
C5: Unacceptably weakly constructed. Chapman’s responses are incomplete, and fail against emended construction.

…to be continued in Part 2.