This research is the first to examine the effects of moral versus practical pro-attitudinal advocacy in the context of self-persuasion. We validate a novel advocacy paradigm aimed at uncovering why moral advocacy leads to polarization and proselytization. We investigate four distinct possibilities: (1) expression of moral foundational values (harm, fairness, loyalty, authority, purity), (2) reliance on moral systems (deontology and consequentialism), (3) expression of moral outrage, (4) increased confidence in one’s advocacy attempt. In Study 1 (N = 255) we find differences between moral and practical advocacy on the five moral foundations, deontology, and moral outrage. In Study 2 (N = 218) we replicate these differences, but find that only the expression of moral foundations is consequential in predicting attitude polarization. In Study 3 (N = 115) we replicate the effect of moral foundations on proselytization. Our findings suggest that practical compared to moral advocacy may attenuate polarization and proselytization. This carries implications for how advocacy can be re-framed in ways which minimize social conflict.
Society is becoming increasingly polarized, and conflict between groups holding opposing views is frequent. A potential source of this conflict is people frequently advocating for their opinions in terms of moral values, which tend to be perceived as universal and sacred. However, we do not know the specific consequences of moral advocacy, and how it can lead to negative outcomes (e.g. political polarization). We also do not know if there are other forms of advocacy which may attenuate these negative outcomes.
In light of the currently fractured and polarized attitudinal landscape, in which advocacy is frequent, we sought to understand: (1) the consequences of moral advocacy on attitude polarization (attitudes becoming more extreme), and proselytization (people’s willingness to persuade others of their own opinion), (2) the specific kinds of moral values or moral content which lead to such negative outcomes, and (3) alternative ways in which advocacy might be encouraged (e.g. advocacy grounded in “practical” or economic arguments as opposed to moral arguments) to reduce the likelihood of such outcomes.
We conducted an online survey in which 588 people were asked to write arguments for their position on migration and climate-change issues. People were randomly divided to advocate in moral or practical terms. We measured their attitudes pre and post-advocacy, and the confidence they placed in their advocacy attempt. We also calculated scores on the kinds of moral content expressed in their arguments, for example, the extent to which they expressed moral values such as harm, care, purity, and emotions such as anger and disgust. Our findings showed that moral (versus practical) advocacy leads to people’s attitudes becoming more extreme post-advocacy, and increased willingness to persuade others of one’s opinion. This was due to increased expression of moral values, as opposed to other types of moral content, such as emotions, or confidence in advocating.
Our findings suggest that re-framing advocacy away from moral values, and towards more “objective” practical arguments such as, economic consequences, leads to less polarization and proselytization. This means that by encouraging practical advocacy, we might be able to minimize social conflict, while simultaneously retaining our right to free speech.
The US federal government shutdown in response to the election of a border wall around Mexico reflects the failure of society to make compromises on morally-charged issues (
Despite the destructive effects of moral rhetoric, little is known about the mechanisms by which moral rhetoric leads to increased social division. Specifically, it is currently not known how relying on moral rhetoric during pro-attitudinal advocacy (advocating for one’s own position; see
Importantly, the current research explores the mechanisms through which pro-attitudinal advocacy leads to polarization and proselytization with the aim of uncovering how advocacy can be re-framed to reduce closed communicative practices. For this purpose, the current research contrasts “moral” versus “practical” expression of one’s attitudes in the pursuit of a) validating a novel advocacy task aimed at changing the language people use to advocate for their views, b) exploring how the extent to which people rely on moral advocacy may polarize one’s attitude post-advocacy, or increase one’s desire to proselytize one’s attitudes, and c) exploring four possible mechanisms – moral value framing (individualizing and binding moral foundations), moral system framing (deontology vs consequentialist reasoning), emotional language use (moral outrage), and meta-cognitive confidence, which may account for polarization and proselytization.
One potential source of political polarization is the introduction of moral rhetoric to justify attitudes. Extant research in moral psychology suggests that when an issue is construed as moral, it evokes moral emotions such as disgust or anger (
This work in moral psychology is consistent with work in standard persuasion (i.e. an external source delivering a persuasive message to a recipient; see
In addition to polarization, research has also demonstrated that moral rhetoric is a strong motivator of proselytization. For example, in a study of articles published in six Chilean outlets,
Thus, while moral framing of attitudes may be particularly effective in disseminating an attitude among like-minded others, and even mobilizing in-group members to advance a group’s position (
Despite research demonstrating the consequences of moral rhetoric on polarization and proselytization, less is understood about the mechanisms by which this occurs. To address this gap in understanding, we focus on the relationships between moral rhetoric, polarization, and proselytization, in the context of
To our knowledge, this is the first set of studies to explore the effects of moral versus practical advocacy in the context of self-persuasion occurring during pro-attitudinal advocacy. We sought to understand if a) people can self-persuade to take a more or less extreme position (polarize or depolarize) after advocating for their opinion, and b) even if attitudes do not change, how advocacy may inform subsequent communicative intentions, such as intentions to proselytize.
The second gap in the literature we aim to address is creating a specific advocacy paradigm (i.e. moral versus practical advocacy) that can be used in the context of self-persuasion to depolarize attitudes and reduce closed communicative intentions, such as intentions to proselytize. Previous research has shown effects of moral framing on attitude polarization, whereby moral framing leads to greater polarization and unwillingness to compromise (
Given the potential of moral rhetoric to stimulate social conflict, we explore an alternative method one can use to justify one’s attitudes. We term this alternative “practical” reasoning. This type of reasoning is based on economic consequences and cost-benefit analyses. It is commonly described by economic theory, whereby the best course of action is the one that allocates scarce resources to the option that “maximizes expected utility” (
The moral versus practical distinction is important, given that attitudes and arguments grounded in these two kinds of reasoning tend to yield different attitudinal outcomes (
Morally framed messages and attitudes tend to be more persuasive, and likely lead to greater polarization and proselytization compared to non-moral attitudes, in the context of standard persuasion. We test whether this effect generalizes to the context of self-persuasion, and predict that the effect of moral advocacy is driven by the saliency of moral content contained in moral compared to non-moral messages. Unlike previous work examining the effects of attitude bases, or the effects of presenting moral versus non-moral messages, the focus of the current study is on moral versus
People use a variety of frames and appeals to both justify and promote their attitudes. A relatively common way in which individuals express their attitudes is through the use of moral rhetoric – that is, framing attitudes in terms of moral values (
Over the last two decades, the field of moral psychology has made substantial advances in the conceptualization and study of mechanisms underlying moral decision making and moral values (
Moreover, MFT proposes that political liberals place greater emphasis on harm/care and fairness/reciprocity (collectively the individualizing foundations), while political conservatives place emphasis on ingroup/loyalty, authority/respect, and purity/sanctity foundations (the binding foundations) (
Within the context of moral psychology research, MFT approaches have demonstrated particular utility in explaining political, ideological and cultural divides (
The finding that highly polarized positions on political issues tend to coincide with strong endorsement of certain moral foundations raises the possibility that moral foundations endorsement may in fact be an underlying mechanism for polarization. Supporting this idea,
Research has also found that people may strategically or intuitively appeal to different types of moral systems when seeking to justify their positions related to moral issues. For example,
Another possibility that we consider, is that moral advocacy leads to increased polarization and proselytization via greater expression of moral emotions. While moral messages tend to be perceived as more emotional compared to practical messages, in general (
Consistent with this notion, recent research conducted by
A final possibility is that moral versus practical advocacy generates different levels of
Moreover, research from standard persuasion provides evidence for a “moral-matching” effect, such that, those who possess attitudes grounded in moral concerns are more likely to be persuaded by moral messages (
Drawing on this literature, we hypothesize that asking one to advocate for one’s position using moral as opposed to practical concerns, likely leads to increased meta-cognitive confidence in one’s attempt. Given the emphasis placed on the universality and correctness on one’s own moral values, advocates are more likely to experience processing ease and fluency when justifying their position in moral terms. Moral values are highly sacralized and ingrained in our minds, and we use these as a basis to view the world (
The main aim of this research is to explore whether moral versus practical advocacy changes levels of moral language and advocacy-related confidence during pro-attitudinal advocacy related to two compelling socio-political issues: migration and climate-change. One possibility is that moral advocacy increases grounding in foundational moral values. Another is that moral advocacy increases grounding in broader moral frameworks (e.g., deontology vs. consequentialism). A third possibility is that moral advocacy increases emotional grounding, or expression of moral outrage. Yet another possibility is that moral advocacy simply increases confidence in one’s advocacy attempt. Thus, we have four possible accounts of the influence of moral advocacy on polarization, which we explicitly test using a new paradigm and text-based analysis.
In Study 1, we consider the extent to which moral versus practical advocacy influences (a) foundational value expression, (b) moral framework language, and (c) moral outrage. In Study 2, we consider the extent to which these factors account for any effects of framing on polarization, by additionally examining the effects of meta-cognitive confidence. In Study 3, we extend our dependent variables to include communicative practices, such as, intentions to proselytize one’s opinions. Across the three studies, we seek to explore which of the four possible mechanisms likely accounts for the effects of moral advocacy on social division.
Two-hundred and fifty-five US residents were recruited via Amazon Mechanical Turk (
Participants completed a survey online via Qualtrics (
Participants were asked to indicate their attitude towards migration using the following item, “Compared to the number of migrants the government usually takes, how many migrants do you think the US government should take?” on a 7-point scale (1
Following initial attitude measurement, participants were randomly allocated to write advocacy appeals grounded in either moral (
Finally, participants were presented with a few demographic questions (age, gender, political orientation). Our primary measure of political orientation was a single 7-point Likert scale (1 =
Traditionally, moral content analyses have often been facilitated by word frequency-based approaches such as the Moral Foundations Dictionary (MFD;
Thus, to measure the moral content of participants’ text responses, we used distributed dictionary representations (
We created dictionary-based measures for three different kinds of moral content as follows. First, we measured semantic similarity of participants’ responses to each of the five moral foundations, using dictionaries from previous research, where the positive and negative poles of each moral foundation are represented by four keywords (e.g., the words
Moral expressiveness was conceptualized using the two broad categories of the five moral foundations: individualizing and binding foundations. A composite score on the individualizing foundations was created by averaging across
A principal components analysis was first conducted to determine if the three moral emotions theorized (anger, contempt, disgust) load onto one component (i.e. a moral outrage component). The analysis revealed that all three moral emotions loaded onto a single component explaining 87% of the variance. This component represents
In order to assess differences between conditions on levels of moral outrage, moral expressiveness, and moral systems framing, a series of One-Way ANCOVAs were conducted with one between-subjects factor (Condition: Moral vs Practical) with initial stance and political orientation as covariates. Significant omnibus tests were followed by Bonferroni-corrected tests of multiple comparisons. Results revealed that our advocacy manipulation was successful - the moral and practical conditions significantly differed on most moral content tested (see
Moral foundation | Moral |
Practical |
η2p | |||
---|---|---|---|---|---|---|
Individualizing | 0.29 | 0.07 | -0.27 | 0.06 | 37.59** | .13 |
Binding | 0.39 | 0.06 | -0.37 | 0.07 | 73.76** | .23 |
Deontology | 0.25 | 0.08 | -0.24 | 0.08 | 16.73** | .06 |
Consequentialism | -0.08 | 0.09 | 0.08 | 0.09 | 1.67 | .01 |
Moral outrage | 0.45 | 0.08 | -0.42 | 0.08 | 58.72** | .19 |
*
Variable | IS | PO | MO | CONS | DNT | BIND | IND |
---|---|---|---|---|---|---|---|
IND | -.03 | -.14* | .52* | .35** | .80** | .87** | – |
BIND | .00 | -.13* | .52** | .15* | .77** | – | |
DNT | -.01 | -.14* | .12* | .55** | – | ||
CONS | .04 | .02 | -.25** | – | |||
MO | .01 | .03 | – | ||||
PO | -.41** | – | |||||
IS | – |
*
The results of Study 1 demonstrate that explicitly requesting advocates to frame pro-attitudinal advocacy in practical versus moral language changes the moral content of their appeal. Using a novel advocacy paradigm, we show that encouraging advocates to frame their appeal in moral versus practical terms increases the level of moral expressiveness, moral outrage, and deontological framing used to construct one’s appeal relating to migration attitudesii. Despite differences in average levels of moral content, we do not know if these differences are consequential in predicting attitudes or communicative intentions post-advocacy.
We attempt to replicate our findings in Study 2 with a few key modifications: a) using a different controversial attitude issue (i.e. issue relating to climate-change), b) a different sample population (college students versus MTurk population), c) including measures of pre and post-advocacy attitudes to assess attitude polarization as a result of advocacy, and d) to examine which of the following mechanisms: moral expressiveness, moral systems framing, moral emotions, or meta-cognitive confidence drives attitude polarization.
Two-hundred and fifty five psychology undergraduates from an Australian university were recruited in exchange for course credit (
Similar to Study 1 participants completed the study on Qualtrics (
Participants were presented with information on a hypothetical carbon emissions policy which was modelled based on energy-saving recommendations in Australia (
Two items were then used to assess the extent to which participants’ attitudes were based on moral and practical concerns (“To what extent is your attitude towards migration based on moral [practical] concerns?”, 1 =
Participants were randomly allocated to either moral (
Participants listed up to eight thoughts they had in response to generating their arguments (taken from
Participants indicated their Time 2 attitude towards the carbon emissions policy (
Participants indicated how much effort they expended in generating their arguments on three 9-point scales, e.g. “How much energy did you put into generating your arguments?", taken from (
Participants rated how strong they perceived they own arguments to be on three 9-point scales (e.g. “How strong are the arguments that you generated?”; adapted from
Participant knowledgeability on the carbon emissions issue was assessed using four 9-point scales (“How knowledgeable do you feel about energy policies?”, “How much information do you feel you have about energy policies?”, “To what extent do you feel you have expertise on energy policies?”, “To what extent do you feel you know the most relevant facts about energy policies?” where 1 =
Finally, participants responded to a demographics questionnaire assessing political orientation on the continuous measure (
Similar to Study 1, a principal components analysis revealed that all three moral emotions (anger, contempt, disgust) loaded onto a single moral outrage component explaining 93% of the variance (component loadings: anger = .97, contempt = .96, disgust = .96), such that higher scores reflect greater moral outrage following one’s advocacy attempt.
Next, we proceeded to examine any differences in meta-cognitions between the two conditions. As with previous research (see
Similar to Study 1 between-subjects ANCOVAs (Condition: Moral vs Practical) were conducted with initial stance and political orientation as covariates. Replicating the results of Study 1, moral expressiveness significantly differed between the two conditions (see
Moral foundation | Moral |
Practical |
η2p | |||
---|---|---|---|---|---|---|
Individualizing | 0.31 | 0.07 | -0.23 | 0.07 | 29.40** | .12 |
Binding | 0.34 | 0.07 | -0.25 | 0.07 | 38.36** | .15 |
Deontology | 0.23 | 0.09 | -0.14 | 0.09 | 9.33* | .04 |
Consequentialism | -0.01 | 0.09 | 0.07 | 0.09 | 0.40 | .00 |
Meta-cognitive confidence | 0.01 | 0.10 | -0.01 | 0.10 | 0.03 | .00 |
Moral outrage | 0.34 | 0.09 | -0.33 | 0.09 | 28.31** | .12 |
*
Variable | POL | IS | PO | MO | MCC | CONS | DNT | BIND | IND |
---|---|---|---|---|---|---|---|---|---|
IND | .12 | -.11 | -.01 | .61** | .08 | .61** | .86** | .88** | – |
BIND | .17* | -.12 | -.04 | .45** | .11 | .61** | .85** | – | |
DNT | .16* | -.14* | -.09 | .34** | .05 | .76** | – | ||
CONS | .16* | -.02 | -.14* | -.05 | .06 | – | |||
MCC | .12 | .06 | .00 | .07 | – | ||||
MO | .03 | -.13 | .16* | – | |||||
PO | -.14* | -.04 | – | ||||||
IS | .25** | – | |||||||
POL | – |
*
Polarization scores for each participant were calculated by multiplying the direction of change (Time 2 attitude in the direction of Time 1 attitude = 1, Time 2 attitude in the opposition direction to Time 1 = -1, Time 1 and Time 2 attitudes unchanged = 0) by the number of scale points moved from Time 1 to Time 2 (maximum 6 scale points). Polarization scores were created based on seminal research investigating (de)polarization in self-persuasion paradigms (see
Interestingly, the two conditions (moral and practical) did not significantly differ on average polarization scores,
To test the effects of moral expressiveness on polarization, we ran a mediation model using 5000 bootstrap samples and 95% confidence intervals (
*
Next, we tested the second proposed explanation for moral versus practical advocacy effects: moral systems framing (deontological and consequentialist reasoning). Results revealed no significant total indirect effect via both deontology and consequentialism on polarization, IE = .03,
Taken together, the findings of Study 2 suggest that relying on moral as opposed to practical advocacy leads to increased attitude polarization via increased moral expressiveness only, and not other advocacy-relevant constructs, such as moral systems, moral outrage, or meta-cognitive confidence. Conversely, the findings suggest that using practical advocacy appeals attenuates attitude polarization by decreasing moral expressiveness.
In Study 3 we sought to explore other downstream consequences of moral versus practical appeals by a) extending our dependent variables to include intentions to proselytize, and b) testing our predictions on the original attitude issue and sample population.
A hundred and sixty-five participants were recruited from Amazon Mechanical Turk (
Participants followed a similar procedure to Study 1, with a few exceptions detailed below. Participants were presented with the attitude issue in Study 1. They then indicated their attitude on the same 7-point scale: “Compared to the number of migrants the government usually takes, how many migrants do you think the US government should take?”; 1 =
Participants were then randomly allocated to advocate for their position on the issue in practical terms (
Participant intentions to persuade others was assessed using five 9-point scales. Three items were taken from previous advocacy research (e.g. “How likely would you be to persuade your [family, friends, stranger] of your own position on this topic?”, taken from
Participant’s perception of their advocacy appeal being effective in persuading others was assessed using a 7-point bipolar scale, “In relation to your own attitude toward the issue of migrant intake, to what extent do you think that the arguments you generated will change your target audience's attitude?” (anchored at -3 = move as far away as possible from your possible, 0 = unpersuaded/retain original position, +3 = move as close as possible to your position). This was recoded into a unipolar scale (scale points 1 to 7;
Finally, participants completed the continuous (
Corroborating the results of Study 1 and Study 2, a principal components analysis revealed that all three moral emotions (anger, contempt, disgust) loaded onto a single component explaining 92% of the variance (component loadings: anger = .95, contempt = .94, disgust = .98). Similar to Study 2, a principal components analysis revealed that all three meta-cognitive variables (argument quality, effort, source credibility) loaded onto a single component explaining 57% of the variance (component loadings: argument quality = .92, effort = .40, source credibility = .85).
Again replicating our results in Study 1 and Study 2, the two conditions differed on moral expressiveness, such that the moral condition generated higher scores on the individualizing and the binding moral foundations compared to the practical condition, controlling for initial stance and political orientation (see
Moral foundation | Moral |
Practical |
η2p | |||
---|---|---|---|---|---|---|
Individualizing | 0.31 | 0.08 | -0.28 | 0.08 | 27.56** | .20 |
Binding | 0.43 | 0.09 | -0.50 | 0.09 | 56.47** | .34 |
Deontology | 0.17 | 0.13 | -0.23 | 0.12 | 5.64* | .05 |
Consequentialism | -0.12 | 0.17 | 0.17 | 0.13 | 2.50 | .02 |
Meta-cognitive confidence | 0.19 | 0.13 | -0.21 | 0.13 | 4.59* | .04 |
Moral outrage | 0.01 | 0.13 | -0.01 | 0.13 | 0.00 | .00 |
*
Variable | PRO | POL | IS | PO | MO | MCC | CONS | DNT | BIND | IND |
---|---|---|---|---|---|---|---|---|---|---|
IND | .32** | .04 | .06 | .15 | -.00 | .20* | .48** | .71** | .77** | – |
BIND | .31** | .01 | .02 | .11 | -.03 | .19* | .28** | .68** | – | |
DNT | .34** | .08 | -.19* | .32** | -.16 | .21* | .59** | – | ||
CONS | .20* | .17 | -.08 | .17 | -.12 | .00 | – | |||
MCC | .49** | .11 | -.14 | .18* | .07 | – | ||||
MO | .03 | .08 | .24* | -.15 | – | |||||
PO | .09 | .26** | -.59** | – | ||||||
IS | .06 | .06 | – | |||||||
POL | .10 | – | ||||||||
PRO | – |
*
Note, unlike in Study 1, although an overall regression model containing moral expressiveness, condition, initial stance, and political orientation was significant,
We now turn to our primary dependent variable in Study 3: proselytization intentions. Given a) the conceptual similarity between intentions to persuade others, and perceptions of audience persuadability, and b) our desire to create a more parsimonious measure of intentions to proselytize, we conducted a principal components analysis to determine if these items were indeed tapping into one construct. The analysis revealed a single component explaining 74% of the variance (.76 < component loadings < .93). Component scores (regression method) were used in the analyses that follow. This component collectively represents the desire to persuade others of one’s position, and captures the belief that others will be persuaded (hereafter, known as proselytization intentions). The two advocacy conditions significantly differed on average proselytization scores,
In order to test our prediction that advocacy condition indirectly influences intentions to proselytize via moral expressiveness, we conducted a mediation analysis on Process (
*
In terms of the moral systems account, results revealed no significant total indirect effect via both deontology and consequentialism on proselytization, IE = .05,
Taken together, the results of Study 3 suggest that moral advocacy may increase proselytization either via moral expressiveness, deontological framing, or meta-cognitive confidence, and not via consequentialism, or moral outrage. Across all three studies, moral expressiveness tends to emerge as the most reliable mediator of the effects of moral advocacy on polarization and proselytization.
A post-hoc “mini meta-analysis” was conducted across Study 2 and Study 3 following recommendations by
Consistent with the results of the mediation analyses, the relationship between the binding foundations and polarization was significant, overall, Mean
Across three studies, we successfully validated a novel advocacy paradigm (moral versus practical advocacy), aimed at understanding how moral language and confidence in advocacy influences post-advocacy attitudes and communicative intentions. While research in standard persuasion suggests that moral framing increases polarization and proselytization, little work has considered why this occurs. Thus, the current research examined the mechanisms through which moral rhetoric may change one’s attitudes or communicative intentions following pro-attitudinal advocacy. We tested four possible mechanisms – moral expressiveness (reliance on the five moral foundations), moral systems framing (reliance on deontology vs consequentialist reasoning), increased emotional language use (moral outrage), and confidence in one’s advocacy attempt (meta-cognitive confidence). The findings suggest that moral expressiveness is the key mechanism through which morally-framed advocacy attempts lead to polarization and proselytization. In particular, the binding foundations appears to account for the relationship between advocacy condition and polarization across both studies.
To our knowledge, this is the first study to examine the self-persuasive effects of moral versus practical messages in the context of pro-attitudinal advocacy. Our findings support previous work showing that moral rhetoric is persuasive, leads to increased proselytization, decreases willingness to compromise (
In Study 1, we show that the practical advocacy condition generates lower levels of moral expressiveness, on average, compared to the moral condition. The practical condition also generates terms which are less semantically similar to words related to deontology, and moral outrage. In Study 2, we show that only the difference in moral expressiveness is consequential for polarization, such that, the practical condition is less likely to polarize post-advocacy, via lower levels of moral expressiveness. In Study 3, we additionally show that moral expressiveness has implications for behavioral intentions, such as, intentions to proselytize one’s opinions. Importantly, those in the practical condition were less likely to proselytize, via lower levels of moral expressiveness. Overall, the findings suggest that using practical compared moral arguments during pro-attitudinal advocacy may lead to more open communicative practices between opposing groups. Further, although previous work in standard persuasion demonstrates effects of moral values, moral systems, and emotions, in the context of self-generated persuasion (see
Although the current research treated the different moral theories (e.g. moral expressiveness and moral systems) as separate candidate explanations for our effects, the results suggest potential overlap both within and between distinct theoretical frameworks. For example, we found high correlations between the individualizing and binding moral foundations (moral expressiveness) and also between deontology and consequentialism (moral systems). In addition, we found high correlations between deontology and both moral foundations, suggesting some overlap between the different frameworks as well. This suggests that moral psychology theory may manifest differently in naturalistic expressions of moral language, compared to moral judgement tasks.
In contrast to the current work, the limited previous research using text-analysis to explore themes in moral justifications, shows no correlation between deontology and consequentialism, for example (
On the hand, corroborating the high correlations observed in the current work, some previous research has shown moderate correlations between particular individualizing and binding foundations (e.g.
Essays created by researchers in standard persuasion typically contain harm related appeals (
This suggests that targeting the binding foundations may be more effective in reducing polarization during pro-attitudinal advocacy. This notion is supported by previous work in moral reframing, which shows that presenting messages grounded in the binding foundations were more effective in changing attitudes towards recycling, than presenting messages grounded in individualizing/harm foundations (
Interestingly, we do not find any moderation by political orientation. Previous work in moral reframing (standard persuasion) consistently shows that liberals are more persuaded when presented with messages containing individualizing foundations, while conservatives are more persuaded by those containing the binding foundations (
Although previous work in standard persuasion suggests “fundamental moral differences separating liberals and conservatives” (
However, it is important to note that political orientation significantly correlated with particular moral content in the current research, although these effects were not consistent between and studies, nor large enough to moderate the effects of content on polarization or proselytization. For example, political orientation significantly negatively correlated with both moral foundations and deontology in Study 1, suggesting a link between conservatism and a lower tendency to express content related to these two moral domains. To consider another example, political orientation significantly positively correlated with moral outrage in Study 2, suggesting a link between conservatism and expression of moral outrage. These findings suggest that political orientation may play a more subtle role in the context of self-persuasion, compared to standard persuasion, and presents a interesting direction of research which is yet to be explored in the context of advocacy.
Potential differences between standard and self-persuasive processes is further highlighted by the lack of “moral matching” effects in the current study. Previous work shows that presenting counter-attitudinal moral messages (compared to practical messages) tend to be more effective in persuading those who have a moral attitude basis (see
In sum, we show that practical reframing of advocacy appeals may depolarize attitudes and decrease proselytization via self-persuasion. Our findings suggest that moral reframing and moral matching effects may be less effective in reducing close mindedness in the context of pro-attitudinal advocacy. Importantly, the difficulties encountered with encouraging collaboration between liberals and conservatives (
A limitation of the current research is that the indirect effect of advocacy type (moral vs practical) on attitude polarization via moral expressiveness failed to replicate in Study 3. One possibility is that the effect of moral expressiveness on polarization was detected in Study 2 but not in Study 3 due to a smaller sample size. The post-hoc mini meta-analysis revealed that the relationship between the binding foundations and polarization was significant overall, while the relationship between the individualizing foundations and polarization was not. Although the relationship between moral advocacy and polarization can be explained by the expression of binding moral values, this is less clear in explaining the effects of advocacy on proselytization, as neither foundation type predicted proselytization on their own. Future work is required to disentangle the effects of advocacy on the various dependent variables using larger samples from diverse populations.
Another explanation for the diluted effect of polarization in Study 3, is differences in the type of attitude issue tested, as issues vary in their controversy and propensity to evoke emotional reactions (
This is congruent with early research in self-persuasion, showing that depolarization of attitudes tends to be difficult to achieve, in general, (
It may be argued that the lack of a true control condition limits the inferences which can be drawn about the self-persuasive effects of advocacy on the outcome variables (depolarization and proselytization). The primary aim of this research was to consider differences specifically between moral and practical advocacy, as opposed to deviation from a control (for similar study designs in pro-attitudinal advocacy which employ two or more experimental conditions without a control condition, see
We note that the correlations observed in the current work do not provide strong evidence for causal mediation. This is an inherent limitation of using cross-sectional mediation analysis as a test of causal process (see
We find evidence that moral advocacy may lead to increased social division, via greater expression of moral foundations content, compared to other types of moral content, such as deontology, or moral emotions. We provide evidence for a novel pro-attitudinal advocacy manipulation, i.e. practical advocacy, which may promote depolarization and reduce proselytization via self-persuasion. Our findings suggest that encouraging advocates to generate practical reasons for their attitude leads to less polarization and lower intentions to proselytize one’s attitude. The findings provide evidence for a novel and effective method of encouraging free speech, while simultaneously minimizing social conflict.
It is important to note that moral discourse, especially about charged political topics, may have its own value in promoting social cohesion and enabling corrective action, depending on the context. For example, recent research finds that moral language is more persuasive than economic language (i.e. monetary arguments) to influence company management on important social issues faced by employees (
For this article, a dataset is freely available (
The data files used for analysis have been made available on the OSF repository. The Supplementary Materials (Appendices) contain descriptions of the attitude issues examined, advocacy task instructions, and further detail on how the dictionaries were constructed (for unrestricted access, see
The authors have no funding to report for research design, execution, analysis, interpretation and reporting. An Australian Government Research Training Program Scholarship was bestowed to the corresponding author for the completion of their doctoral dissertation.
The authors have declared that no competing interests exist.
We thank Manuel José Rengifo and Shaheed Azaad from the Melbourne School of Psychological Sciences, for their useful suggestions regarding the mini-meta analysis.
Dictionaries were constructed using the
We conducted a few additional analyses to test if our effects were driven by demographic characteristics, such as, political orientation. A linear regression containing initial stance, political orientation, Condition, and the individualizing (binding) foundations as independent variables revealed that political orientation was not a significant predictor of binding (individualizing) foundations (β = -.04,
Partially standardized regression coefficient is the default measurement unit in PROCESS, and controls for all variables in the mediation, but does not completely remove variation incurred by the covariates, thus taking into account natural variation in the outcome variable due to confounders (
The total indirect effect of condition on polarization via moral expressiveness was significant in a mediation model without the covariates, IE = .14,
Post-hoc moderated mediation analyses on PROCESS (
One might argue whether this effect is robust controlling for other potential mediators, such as moral outrage and deontology, which also significantly differed between advocacy conditions. Thus, we conducted a secondary simultaneous mediation analysis with the individualizing foundations, binding foundations, deontology and moral outrage as mediators, and political orientation and initial stance as covariates. Results revealed that the binding foundations continued to significantly predict polarization on its own, even after accounting for all other mediators, IE = .32,
The total indirect effect of condition on proselytization via moral expressiveness was significant in a mediation model without the covariates as well, IE = .16,