This is an open access article distributed under the terms of the Creative Commons
Attribution License (Politicians occasionally adapt their communication to different audiences. For example, in an interview with the business-friendly newspaper Handelsblatt, Philipp Rösler (chairman of the German libertarian party FDP, 2011-2013) said that the government’s energy policy was giving German companies a “huge opportunity” to succeed internationally (FDP, 2013). In an interview with the Christian newspaper Christ und Welt, on the other hand, he emphasized that, most of all, the government’s energy policy served to “secure jobs” (FDP, 2012). This begs the question what kind of outcomes, goals, or values can be addressed in a political message to increase its persuasiveness for a certain audience. In the following, we will argue that recent research on the motivational bases of two distinct dimensions of conservatism might shed some light on this question.
Two Dimensions of Conservatism
Although some researchers support a one-dimensional concept of political conservatism (Jost, Federico, & Napier, 2009; Jost, Glaser, Kruglanski, & Sulloway, 2003), a number of empirical studies suggest that at least two factorially distinct dimensions of political conservatism can be differentiated – a cultural and an economic dimension (Feldman & Johnston, 2014; Van Hiel & Kossowska, 2007; Zumbrunnen & Gangl, 2008). Cultural conservatism is defined as a preference for traditional lifestyles and personal constraint as opposed to new and alternative lifestyles and personal freedom. Thus, a cultural conservative ideology favors issues such as “authoritarian parent-child relationships, traditional work ethics, and conventional female roles” (Kossowska & Van Hiel, 2003, p. 508; see also Middendorp, 1978). On the other hand, economic conservatism is defined as a preference for free markets and free competition as opposed to social justice and economic egalitarianism. Therefore, people high in economic conservatism “adhere to capitalist ideology and believe that private initiative and competition should not be limited” (Kossowska & Van Hiel, 2003, p. 508), while they would reject governmental economic intervention or concerns about worker participation.
While low to moderate correlations are usually found between economic and cultural conservatism (Feldman & Johnston, 2014; Jost et al., 2009), there is reason to assume that the two dimensions are in fact psychologically distinct. In a study by Crowson (2009), cultural conservatism was positively correlated to personal need for structure, fear of death, and dogmatic aggression (i.e., aggression against individuals whose beliefs differ from one’s own), while economic conservatism was largely unrelated to these variables. In line with this, Feldman and Johnston (2014) identified authoritarianism and religiosity as substantial predictors of cultural conservatism, while economic conservatism was again unrelated to these variables. For once, these results suggest that cultural but not economic conservatism is motivated by needs for certainty and security. Further, Feldman and Johnston (2014) demonstrated that a unidimensional treatment of conservatism may also lead to misinterpretations with regard to psychological correlates. Religiosity, appeared to be a predictor of conservatism, when ideology was treated as a unidimensional construct, although religiosity was only related to cultural but not to economic conservatism when the two dimensions were differentiated.
Motivational Bases of Economic and Cultural Conservatism
A theoretical framework suitable to characterize the different motivational underpinnings of cultural and economic conservatism is Schwartz’ model of basic human values (Schwartz, 1992, 1994). The model differentiates between two orthogonal value dimensions. Schwartz (1994) assumed that cultural conservatism should correspond to the conservation versus openness to change value dimension because this dimension is concerned with the conflict between individual freedoms (stimulation, self-direction) and protecting the societal status quo (security, tradition, conformity). Empirical support for this assumption comes from Schwartz, Caprara, and Vecchione (2010). In their Italian sample, three core political values that should relate to cultural conservatism, namely traditional morality, blind patriotism, and law and order, were positively correlated to conservation values and negatively correlated to openness to change values. Similarly, Malka and colleagues (Malka, Soto, Inzlicht, & Lelkes, 2014) found associations between the conservation versus openness to change dimension and cultural conservative attitudes concerning homosexuality, abortion rights, gender equality, immigration and the severe punishment of criminals. In the study by Malka et al. (2014) these associations were reliable across 51 national samples. These results are consistent with the assumption that cultural conservative individuals strive for security and stability and try to prevent loss. Preserving traditional lifestyles or a pre-existing social order allows one to maintain what is familiar and known, while social change is a potential risk (i.e., a risk of losing something valued; Jost et al., 2003; Jost et al., 2009). This is also reflected by positive correlations between cultural conservatism and fear of death (Crowson, 2009), preferences for order, structure and predictability (Cornelis & Van Hiel, 2006; Kossowska & Van Hiel, 2003) and dogmatism (Crowson, 2009). Cultural conservatism is also related to right-wing authoritarianism (Cornelis & Van Hiel, 2006), which is based on a deep-rooted belief that the world is a dangerous and threatening place (Duckitt, 2001; Duckitt, Wagner, du Plessis, & Birum, 2002). Past research employing measures of RWA also provide hints as to what kind of outcomes can be addressed in a political message to increase its persuasiveness for authoritarian audiences. In a study by Lavine et al. (1999) participants high or low in RWA were presented with persuasive arguments that emphasized either the potential rewards of voting or the potential threats (or losses) of not voting (e.g., “Not voting allows others to take away your right to express your values”, p. 342). As expected, participants high in RWA rated the threat- or loss-related arguments to be higher in argument quality than the reward-related arguments.
In contrast, security-related motives are not as important for economic conservatism (Crowson, 2009; Malka et al., 2014). Instead, economic conservatism should correspond to the self-enhancement versus self-transcendence dimension from Schwartz’ model of basic human values (Schwartz, 1994) because this dimension is concerned with the conflict between gaining personal wealth and status (power, achievement) versus promoting the welfare of others (benevolence, universalism). In line with this, the aforementioned study by Schwartz et al. (2010) found that free enterprise and equality, two core political values that should relate to economic conservatism, were most strongly correlated to self-enhancement and self-transcendence values. While self-enhancement values were positively related to free enterprise and negatively related to equality, this pattern was reversed for self-transcendence values. Thus, economic conservatism appears to highlight an individualist achievement orientation where people focus on material gains. In line with this assumption, Kossowska and Van Hiel (2006) found positive correlations between economic conservatism and materialism (e.g., admiring wealthy people) and extrinsic aspirations (e.g., being financially successful) across different national samples. Further, economic conservatism was also shown to be positively correlated with internal-individualistic attributions for poverty and wealth and negatively correlated with external-structuralistic attributions for poverty and wealth (Bobbio, Canova, & Manganelli, 2010). Generally, fewer studies appear to be concerned with the motivational bases of economic conservatism than with the motivational bases of “classical liberalism” (Schwartz, 1994, p. 39) versus classical conservatism (i.e., cultural conservatism). However, economic conservatism is also related to a social dominance orientation (Cornelis & Van Hiel, 2006), which is based on a competitive jungle worldview where only the fittest survive and succeed (Duckitt, 2001; Duckitt et al., 2002). A study by Pratto and Lemieux (2001) also provided hints as to what kind of outcomes can be addressed in a message for people high in SDO. In their study, participants high in SDO actually evaluated an immigration policy more positively after they were presented with arguments emphasizing the policy’s hierarchy-enhancing (or achievement-related) benefits (e.g., “This policy will ensure that only those who can really succeed here can join our society”; p. 420).
Message Framing
To be persuasive for specific audiences, a message should address outcomes that are motivationally relevant for the targeted individuals (also referred to as outcome framing; Cesario, Corker, & Jelinek, 2013). If cultural conservatism is motivated by needs for security and a fear of loss, while economic conservatism is motivated by aspirations to achievement and material gains, then messages addressing loss and/or security should be motivationally relevant for cultural conservative individuals, whereas messages addressing gains and/or achievement should be motivationally relevant for economic conservative individuals. A number of past studies can be found that have systematically varied gain/loss or achievement/security frames to create persuasive messages for different audiences. These studies usually draw on the theoretical framework of regulatory focus theory.
Regulatory Focus and Message Framing
The Regulatory Focus Theory is a self-regulatory approach to the study of human motivation (Higgins, 1997, 1998). According to the theory, there are two separate and independent self-regulatory orientations: promotion and prevention. In a promotion focus, people are oriented towards growth; that is, they are concerned with advancement and accomplishment, and they are sensitive to the presence and absence of positive outcomes (i.e., gains). In a prevention focus, on the other hand, individuals are oriented towards security; they strive for safety, and they are sensitive to the presence and absence of negative outcomes (i.e., losses). Interestingly, regulatory focus theory is most often applied in the field of communication. Research on message framing has repeatedly shown that the same message can be framed to be persuasive for either promotion-focused individuals (by addressing gains and/or achievements) or for prevention-focused individuals (by addressing loss and/or security).
A classic study was conducted by Cesario, Grant, and Higgins (2004). In their study, a gain versus loss framing was employed to create different persuasive messages advocating a new after school program for children (Study 2). The gain-framed message stated that the new after school program would “advance children’s education and support more children to succeed” (p. 403), while the loss-framed message stated that it would “secure children’s education and prevent more children from failing” (p. 403). As predicted, predominantly promotion-focused participants judged the message as more persuasive when they had received the gain-framed message, whereas predominantly prevention-focused participants judged the message to be more persuasive when they had received the loss-framed message.
The effect of an achievement versus security framing was demonstrated by Quinn and Olson (2011). The authors studied the effects of persuasive messages on women’s motivation to participate in collective actions aimed at improving the status of women. Their all-female participants read arguments in support of collective action that were either achievement-framed (e.g. “activism […] will open doors and let more women make it to the top”; p. 2466) or security-framed (e.g. “women need to protect themselves and their rights”; p. 2466). As hypothesized, prevention-oriented women reported a higher action motivation when they had read arguments framed in terms of security, while promotion-oriented women reported higher motivation after reading arguments framed in terms of achievement.
The Influence of the Attitude Object
While all of the effects discussed above are concerned with a match between an individual’s motivational focus and the argument frames presented, characteristics of the attitude object (i.e., the after-school program argued for) need to be taken into account as well. Past research suggests that a match between attitude object and message frame can serve to increase a message’s effectiveness as well. Zhou and Pham (2004), for example, hypothesized and found that different financial products are associated with a different motivational focus. Individual stocks can yield the highest financial gains, whereas mutual funds and retirement accounts are considered safe financial products. In consequence, gains are more relevant when people make decisions about investing in individual stocks, and potential losses are more relevant when people make decisions about investing in mutual funds and retirement accounts (Zhou & Pham, 2004). To the best of our knowledge, this concept has never been tested in a political context. However, to the extent that political issues are associated either with concerns for security or with hopes for gains and achievements, it is reasonable to assume that different argument frames should be most effective. Gain and achievement frames should be most effective for economic conservative issues, while loss and security frames should be most effective for cultural conservative issues.
The Present Research
For the present research, we propose that different message frames should be most effective when addressing an economic or a cultural conservative audience because different motivational bases underlie these two dimensions of conservatism. To be persuasive messages may address outcomes that are motivationally relevant for the targeted individuals (Cesario et al., 2013) or for the targeted attitude object (Zhou & Pham, 2004). Therefore, we propose differential matching effects for gain/loss and for achievement/security argument frames with regard to both economic or cultural conservative individuals and economic or cultural conservative political issues. More specifically, we assume that gain-framed and achievement-related arguments should match with economic conservative issues and individual economic conservatism and that loss-framed and security-related arguments should match with cultural conservative issues and individual cultural conservatism. Arguments with matching frames should be perceived as more persuasive than arguments with mismatching frames.
The argument frames we use in the present research were taken from regulatory focus research. In this branch of research, a number of studies have carefully varied persuasive messages with regard to a gain/loss or an achievement/security framing. These past studies have provided evidence for separate effects of both gain/loss frames (Cesario et al., 2004, Study 2) and achievement/security frames (Quinn & Olson, 2011). However, some results hint at interactive effects of these two types of framing. Both Cesario et al. (2004, Study 1) and Yi and Baumgartner (2009) found that gain-framed messages were more persuasive than loss-framed messages when the message also emphasized achievement-related benefits. Therefore, both types of framing could produce main effects in our studies or they could interact with each other. In case of two main effects and in case of an interaction, the highest persuasiveness should be observed when both argument frames match with issue and/or individual attitude.
Study 1
Study 1 served both as a pretest and as a first test of the proposed differential framing effects for economic and cultural conservative issues. As outlined above, we hypothesize that gain-framed and achievement-related arguments should be perceived as more persuasive for economic conservative issues (Hypothesis 1), whereas loss-framed and security-related arguments should be perceived as more persuasive for cultural conservative issues (Hypothesis 2). Pretesting was necessary to identify political issues that were clearly either economic or cultural conservative and that were sensitive to message framing effects.
Method
Materials and Design
Aiming to test the differential effects for two types of framing in the context of two different types of political issues, we implemented a 2 (type of issue: economic vs. cultural conservative) by 2 (outcome focus: gain vs. loss) by 2 (type of benefit: achievement vs. security) design. Given our interest in finding systematic framing effects across issues while keeping the influence of the specific political issue small, arguments were created for eight different issues assumed to be economic conservative and for eight different issues assumed to be cultural conservative. Issues were taken from the broader areas of 1) education, either highlighting a competitive business-oriented education (economic conservatism) or discipline and traditional values in education (cultural conservatism), 2) capitalist ideology and 3) social order and preventing crime. All the issues employed here were part of the current political debate in Germany. For each issue, four distinctly framed arguments were created. Arguments were about nineteen to twenty-five words long. They were matched in length and great care was taken to create messages that differed only in the focal aspects (see Table 1). Arguments were modeled after those presented by Yi and Baumgartner (2009), who already crossed gain/loss frames and achievement/security frames in their experimental design. Participants were assigned to one out of four counterbalanced questionnaire versions, so that they would read and rate only one argument per issue (see Table 2 for full design).
Table 1
Example Arguments for an Economic Conservative Issue
| Issue: Schools should always compete with other schools
|
|
|---|---|
| Gain end-state | Loss end-state |
| a) Achievement-related benefit or harm | |
| When there is competition between schools, we can achieve that German students improve their position in international school studies. | When there is no competition between schools, we risk that German students will be left behind in international school studies. |
| b) Security-related benefit or harm | |
| When there is competition between schools, we can achieve that German students will have higher security of employment later. | When there is no competition between schools, we risk that German students will have higher unemployment rates later. |
Table 2
Study 1: Full Design.
| Issue | Type of issue | Argument frames and counter-balanced questionnaire versions
|
||||
|---|---|---|---|---|---|---|
| A | B | C | D | |||
| (1) | State-funded elite universities | EC | 1 | 2 | 3 | 4 |
| (2) | Sanctions for disruptive behavior in school | CC | 2 | 3 | 4 | 1 |
| (3) | Corporate funding for schools | EC | 3 | 4 | 1 | 2 |
| (4) | Police checks for internet shops | CC | 4 | 1 | 2 | 3 |
| (5) | Ban for violent video games | CC | 1 | 2 | 3 | 4 |
| (6) | Competition between schools | EC | 2 | 3 | 4 | 1 |
| (7) | Sanctions for staying away from school | CC | 3 | 4 | 1 | 2 |
| (8) | Stricter treatment of unemployed people | EC | 4 | 1 | 2 | 3 |
| (9) | Support programs for the gifted | EC | 1 | 2 | 3 | 4 |
| (10) | School degree required for immigration | CC | 2 | 3 | 4 | 1 |
| (11) | Temporary employment | EC | 3 | 4 | 1 | 2 |
| (12) | Better equipment for police force | CC | 4 | 1 | 2 | 3 |
| (13) | Religious education | CC | 1 | 2 | 3 | 4 |
| (14) | Networking between industry and universities | EC | 2 | 3 | 4 | 1 |
| (15) | Alcohol ban in certain locations | CC | 3 | 4 | 1 | 2 |
| (16) | Extra tuition for long-term students | EC | 4 | 1 | 2 | 3 |
Note. EC = economic conservative, CC = cultural conservative, 1 = gain/achievement, 2 = loss/achievement, 3 = gain/security, 4 = loss/security.
Participants
The sample consisted of 111 employees (63 women) from the University of Tübingen. They were roughly equally distributed across the age categories 20 to 30 years (N = 27), 31 to 40 years (N = 31), 41 to 50 years (N = 25) and 51 to 65 years (N = 21). One participant was younger than 20 years and six participants were older than 65 years. About two thirds of all surveyed employees (N = 70) were members of the academic staff, while one third (N = 40) were members of the administrative staff. One participant did not state their occupation. Subjects were randomly assigned to the four questionnaire versions.
Procedure
The study was conducted online using the platform www.soscisurvey.de. Participants were contacted by email. After answering some demographic questions, participants read a statement referring to the first political issue (“There should be state-funded elite universities in Germany”i), and then answered two items concerning the classification of that statement as an economic or as a cultural conservative issue. The items read as follows: “A statement like this should be supported by people who emphasize the importance of economic strength” and “A statement like this should be supported by people who emphasize the importance of traditional cultural values”. These classification items were answered on 7-point Likert scales ranging from (1) do not agree at all to (7) fully agree. Afterwards, participants read one argument supporting the statement (e.g., the loss/achievement argument in questionnaire version B; “Without these elite universities, we risk that Germany’s young scientists will lag behind in the international comparison”). The dependent variable, perceived argument persuasiveness, was rated on another 7-point Likert scale ranging from (1) not persuasive at all to (7) very persuasive. Overall, each participant provided ratings for sixteen issue statements and sixteen arguments. Issues were presented in a fixed order (see Table 2). As an incentive, participants could take part in a raffle for book vouchers worth 50 Euro.
Results
Descriptive Statistics
Concerning political orientation, our sample showed a mild center-left bias. On a left-right-continuum (with 1 indicating a very left-wing and 7 a very right-wing attitude), 55% of all participants rated their political orientation with 3 or 4 (overall M = 3.21, SD = 1.07). However, across all issues and arguments the average persuasiveness rating of the presented (conservative) arguments was around the scale midpoint (M = 4.09, SD = 0.99).
Classification of Political Issues
A first aim of Study 1 was to identify political issues that could be classified as either economic or cultural conservative. All issues had been conceptualized a priori as either economic conservative or cultural conservative (see Table 2). To confirm that participants would classify issues in this way, ratings for the two classification items were compared using paired-sample t-tests. Fifteen out of sixteen issues were classified in the expected way (all ps < .01). More specifically, for economic conservative issues, the item tapping economic conservatism received significantly higher ratings than the item tapping cultural conservatism, whereas the opposite was found for cultural conservative issues. One issue (“Immigrants should be required to have graduated from school to enter Germany”) had been classified a priori as cultural conservative, but was perceived as an economic conservative issue by participants. This issue was excluded from further analyses.
Framing Effects
Data was analyzed using SPSS Linear Mixed Models. Our experimental factors type of issue, type of benefit and outcome focus and their interactions were included in the model as fixed factors along with the control variables age and sex. Participants and issues were included as random factors. This way, we could control for individual variation and for stimulus variation (i.e., for differences between our specific issues). Intercepts and slopes were specified for both random factors as recommended by Judd, Westfall, and Kenny (2012, 2017). Type of benefit and outcome focus were crossed with both participants and specific issues. Therefore, random slopes were specified for both factors and their interaction for participants and for issues. Type of issue was also crossed with participants but specific issues were nested under type of issue. Therefore, random slopes were specified for type of issue and the factor’s interaction with outcome focus and type of benefit for participants but could not be specified for specific issues. As recommended by Judd et al. (2012) we tested all these random effects and then eliminated the ones with zero variances. Models were then run again to obtain final estimates. None of the random slopes could be calculated with an unstructured covariance matrix; so, we resorted to the more restrictive variance components structure.
After eliminating random slopes with zero variances the model included random error components for the intercept, the effect of type of issue and the type of benefit by outcome focus interaction for participants and random error components for the intercept and type of benefit by outcome focus interaction for issues. This analysis revealed a marginal effect for type of issue, F(1,15.85) = 3.93, p = .065, and a significant main effect of type of benefit, F(1,42.26) = 7.97, p = .007, indicating that arguments for cultural conservative issues (M = 4.46, SE = .27) received slightly higher ratings than arguments for economic conservative issues (M = 3.91, SE = .27) and that achievement-related arguments (M = 4.36, SE = .24) generally received more favorable ratings than security-related arguments (M = 4.02, SE = .24). More important, two significant two-way interactions emerged for both type of issue by type of benefit, F(1,42.65) = 6.88, p = .012, and type of issue by outcome focus, F(1,42.59) = 7.42, p = .009, indicating both experimentally varied framings played out differently for economic versus cultural conservative issues.
Testing the variance components in this mixed model revealed significant variation between participants, that is, the random intercept for participants was significant, B = .71, SE = .14, p < .001, 95% CI [.49, 1.03]. This indicates that participants differed in how they rated argument persuasiveness across different argument frames. The random slope for type of issues for participants was also borderline significant, B = .13, SE = .07, p = .056, 95% CI [.05, .37], suggesting there was some variation between participants in how they rated arguments for the two different types of issues. Variation between specific issues was also observed. Both random intercept and type of benefit by outcome focus interaction were significant for specific issues with B = .23, SE = .10, p = .027, 95% CI [.09, .55] and B = .11, SE = .04, p = .017, 95% CI [.05, .24], respectively. This means that persuasiveness ratings differed considerably between different issues and that the interaction of type of benefit by outcome focus framing affected argument ratings differently for different issues.
To decompose the significant two-way interactions of type of issue and both framings, we ran the same model, without the factor type of issue, separately for economic and for cultural conservative issues. For economic conservative issues significant main effects were observed for type of benefit, F(1,22.86) = 17.99, p < .001, and for outcome focus F(1,22.81) = 5.10, p = .034. In line with expectations, achievement-related arguments (M = 4.24, SE = .28) were rated as more persuasive than security-related arguments (M = 3.59, SE = .28) and gain-framed arguments (M = 4.09, SE = .28) were rated as more persuasive than loss-framed arguments (M = 3.74, SE = .28). As Figure 1 clearly shows these are two additive main effects, that is, when both argument frames matched with type of issue, ratings were most favorable.
Figure 1
Perceived persuasiveness as a function of type of issue, outcome focus and type of benefit (Study 1).
For cultural conservative issues no difference was observed between achievement- (M = 4.48, SE = .29) and security-related arguments (M = 4.45, SE = .29), F(1,19.35) = 0.01, p = .915. Means as depicted in Figure 1 suggested that there might have been a difference between loss arguments (M = 4.61, SE = .29) and gain arguments (M = 4.31, SE = .29). However, this effect failed to reach significance, F(1,19.34) = 2.41, p = .137.
Issue Sensitivity to Framing
Considerable variation between specific issues was observed. Therefore, separate two-way ANOVAs were conducted for all fifteen issues separately to identify those issues where framing effects were observed and those issues where no framing effects could be observed. Significant framing effects were found for five out of eight economic conservative issues. The type of benefit framing yielded significant effects for the issues elite universities (p = .01), corporate funding for schools (p = .001), competition between schools (p = .002), support programs for the gifted (p = .002), and also for networking between industry and universities (p = .05). A significant effect for outcome focus was found for the issue corporate funding for schools (p = .001). Further, there was a marginal effect of outcome focus for the issue networking between industry and universities (p = .09). All of these effects were consistent with expectations.
For cultural conservative issues four significant framing effects were found. Significant effects for outcome focus were observed for the issues ban for violent video games (p = .002) and sanctions for staying away from school (p = .02). An effect for type of benefit was found for the issue of having the police check on internet shops (p = .02) and providing the police with better equipment (p = .03). Three out of four effects were consistent with expectations. However, for the issue of providing the police with better equipment, a reversed effect was found. Here, achievement arguments were seen as more persuasive than security arguments. For the issue of banning alcohol in certain locations in town, there was a marginal interaction between the two framing factors (p = .06). Pairwise comparisons revealed that for this issue the gain/security argument was more persuasive than the gain/achievement argument, while there was no difference between the loss arguments.
Discussion
Study 1 aimed at identifying issues that could be classified as either economic or cultural conservative and that would be sensitive to argument framing. Also, the study presented a first test of the proposed differential framing effects for two types of political issues. Results showed that, first of all, participants classified fifteen out of sixteen issues in line with our a priori assumptions concerning these issues. More important, significant framing effects were found. Two two-way interactions were observed between type of issue and both experimentally varied framings; that is, achievement/security and gain/loss frames played out differently for economic versus cultural conservative issues. For economic conservative issues achievement frames were perceived as more persuasive than security frames and gain frames were also perceived as more persuasive than loss frames (supporting Hypothesis 1). These effects took the form of two additive main effects. Thus, persuasiveness ratings were most favorable when both argument frames matched with type of issue. Loss/security arguments, where both frames were mismatched, were the least persuasive ones. For cultural conservative issues we expected higher ratings for security versus achievement and for loss versus gain frames but neither effect reached significance (not supporting Hypothesis 2).
To this effect, it is important to note that considerable variance was observed for specific issues. Persuasiveness ratings differed between different issues and argument frames also led to different persuasiveness ratings for different issues. Some of the specific issues might have yielded the expected framing effects while others did not. Therefore, framing effects were also tested separately for all specific issues. Five economic conservative and four cultural conservative issues yielded framing effects that were consistent with expectancies. However, three economic conservative issues and two cultural conservative issues did not yield significant framing effects and one cultural conservative issue (providing the police with better equipment) actually yielded an effect that was opposite to the hypothesis. Possibly, the focus on better equipment emphasized the police’s performance, not security matters. These problematic issues were dropped after Study 1. The remaining four cultural conservative issues were all included in the questionnaire of Study 2. Similarly, four of the five remaining economic conservative issues were retained (the least appropriate issue was dropped). Study 2 served two purposes. Specifically, it aimed (1) to replicate results from Study 1 with suitable materials, and (2) to test the hypotheses that the expected framing effects would be moderated by recipient differences in political conservatism.
Study 2
In Study 2, we aimed to replicate results for Hypothesis 1 and possibly find additional support for Hypothesis 2. Further, we hypothesized that for economic conservative issues, gain-framed and achievement-related arguments should be perceived as more persuasive by participants high in economic conservatism as compared to participants low in economic conservatism (Hypothesis 3). For cultural conservative issues, on the other hand, we hypothesized that loss-framed and security-related arguments should be perceived as more persuasive by participants high in cultural conservatism as compared to participants low in cultural conservatism (Hypothesis 4). However, we did not necessarily expect these effects to emerge independently of type of issue. Following our reasoning above arguments are likely to be perceived as most persuasive when argument frames match both type of issue and individual conservatism.
Method
Design and Stimuli
As in Study 1, a 2 (type of issue: economic vs. cultural conservative) by 2 (outcome focus: gain vs. loss) by 2 (type of benefit: achievement vs. security) design was implemented. To achieve a balanced design, four economic conservative and four cultural conservative issues were used that had been identified as suitable in Study 1. Again, four questionnaire versions were created in order to counterbalance the assignment of message frames to issues (see Table 3 for full design). Further, participants’ economic and cultural conservatism was measured.
Table 3
Study 2: Full Design.
| Issue | Type of issue | Argument frames and counter-balanced questionnaire versions
|
||||
|---|---|---|---|---|---|---|
| A | B | C | D | |||
| (1) | Support programs for the gifted | EC | 1 | 2 | 3 | 4 |
| (2) | Ban for violent video games | CC | 2 | 3 | 4 | 1 |
| (3) | Corporate funding for schools | EC | 3 | 4 | 1 | 2 |
| (4) | Sanctions for staying away from school | CC | 4 | 1 | 2 | 3 |
| (5) | Alcohol ban in certain locations | CC | 1 | 2 | 3 | 4 |
| (6) | Competition between schools | EC | 2 | 3 | 4 | 1 |
| (7) | Police checks for internet shops | CC | 3 | 4 | 1 | 2 |
| (8) | Networking between industry and universities | EC | 4 | 1 | 2 | 3 |
Note. EC = economic conservative, CC = cultural conservative, 1 = gain/achievement, 2 = loss/achievement, 3 = gain/security, 4 = loss/security.
Participants
The sample consisted of 234 employees from the University of Tübingen (121 women). One third of participants (N = 71) were 20 to 30 years old, 58 participants were 31 to 40 years old, 39 participants were between 41 and 50 years old, and 27 were between 51 and 65 years old. One participant was younger than 20 years and 12 participants were older than 65 years. 26 participants did not state their age. About two thirds of all surveyed employees (N = 148) were members of the academic staff, while one third (N = 84) were members of the administrative staff. Two participants did not state their occupation. Subjects were randomly assigned to the four questionnaire versions. We made sure that there was no overlap between the samples of Studies 1 and 2.
Procedure
Again, the study was conducted online. In Study 2, participants rated eight arguments for eight different political issues. Issues were presented in the same order as presented in Table 3. Individual economic and cultural conservatism were assessed afterwards. We chose this order to avoid an increased salience of participants’ general political orientation while rating the arguments.
Measures of Cultural and Economic Conservatism
The two types of political conservatism were assessed with the Middendorp scales for cultural and economic conservatism (De Witte, 1990, as cited in Cornelis & Van Hiel, 2006). Middendorp’s economic conservatism scale is concerned with “adherence to capitalist ideology, private initiative, and unrestricted competition” (Cornelis & Van Hiel, 2006, p. 40), in opposition to concerns of equality or worker participation. The scale consists of eighteen items. A sample item reads “Our country can only get ahead if the government gives industry free reign to control its own affairs”. The scale was translated and adjusted to our contemporary German sampleii. The original scale includes six items concerning the role of labor unions. Two of these items were replaced with questions regarding other economic conservative issues (e.g., networking between industry and universities) as not to overemphasize this topic. The scale was reliable with Cronbach’s α = .87.
Middendorp’s cultural conservatism scale, on the other hand, is concerned with traditional versus alternative lifestyles, authoritarian parent-child relationships and traditional gender roles. A sample item reads as follows “Men and women are equally suited to take care of small children” (reverse coded). Again some adjustments were made; for example, the item “Homosexuality is actually a disease that can be cured” was replaced with an item on same sex marriage. The scale consists of eighteen items and was reliable with Cronbach’s α = .81. Similar to previous research (Crowson, 2009; Feldman & Johnston, 2014), economic and cultural conservatism were positively correlated, r = .46, p < .001. Due to some deviation from normality (one-sample Kolmogorov-Smirnov test; D(234) = .08, p = .001) a square root transformation was applied to scores for cultural conservatism.
Results
Descriptive Statistics
Concerning political preferences, the sample again displayed a center-left bias. On a left-right-continuum (with 1 indicating a very left-wing and 7 a very right-wing attitude), about 53% of all participants rated their political orientation with 3 or 4 (overall M = 3.05, SD = 1.06). In line with this, mean scores for economic conservatism (M = 3.00, SD = 0.82) and cultural conservatism (M = 2.48, SD = 0.76) were relatively low. More important, average persuasiveness ratings across issues and argument frames were again close to the scale midpoint (M = 3.51, SD = 0.92).
Type of Issue and Argument Frame
The first aim of Study 2 was to replicate the framing effects found in Study 1. To do this, again we used a linear mixed model and included our experimental factors and their interactions as fixed factors and participants and specific issues as random factors. Once more, we started with a full model, then eliminated random slopes with zero variances. Thus, the final model included random error components for the intercept, the effect of the type of issue by type of benefit interaction for participants and random error components for the intercept and the type of benefit by outcome focus interaction for issues. This analysis again yielded two significant two-way interactions for both type of issue by type of benefit, F(1,22.39) = 11.07, p = .003, and type of issue by outcome focus, F(1,22.17) = 8.33, p = .009. Thus, in Study 2 again both framing factors played out differently for economic versus cultural conservative issues. A significant effect was also observed for the control variable sex, F(1,205.63) = 6.25, p = .013, indicating that women (M = 3.91, SE = .24) generally rated arguments as more persuasive than men (M = 3.60, SE = .25).
Testing the random variance components in the mixed model again revealed significant variation between participants, that is, the random intercept for participants was significant, B = .46, SE = .08, p < .001, 95% CI [.32, .64]. There was also a significant effect for the type of benefit by outcome focus interaction for specific issues, B = .12, SE = .05, p = .017, 95% CI [.05, .28], indicating that again the interaction of type of benefit and outcome focus framing affected persuasiveness ratings differently for different issues.
To decompose the two-way interactions the same model was run again, without the factor type of issue, separately for economic and cultural conservative issues. For economic conservative issues a significant main effect was observed again for type of benefit, F(1,12.25) = 7.49, p = .018, and a borderline significant main effect was observed for outcome focus, F(1,11.72) = 4.63, p = .053, indicating that achievement-related arguments (M = 4.15, SE = .31) were perceived as more persuasive than security-related arguments (M = 3.51, SE = .31) and that gain arguments (M = 4.08, SE = .31) were also perceived as slightly more persuasive than loss arguments (M = 3.58, SE = .31). As Figure 2 shows, these were again two additive main effects, that is, when both argument frames matched with type of issue, ratings were most favorable.
Figure 2
Perceived persuasiveness as a function of type of issue, outcome focus and type of benefit (Study 2).
For cultural conservative issues means as depicted in Figure 2 seemed to suggest an interaction between both experimentally varied framings. However, this interaction did not reach significance, F(1,11.40) = 3.12, p = .104. Instead, significant main effects were observed for both type of benefit, F(1,11.40) = 5.00, p = .046, and outcome focus, F(1,11.40) = 5.22, p = .042. In line with expectations, security arguments (M = 3.85, SE = .31) were perceived as more persuasive than achievement arguments (M = 3.51, SE = .31) and loss arguments (M = 3.85, SE = .31) were perceived as more persuasive than gain arguments (M = 3.50, SE = .31).
Moderating Effects of Economic Conservatism
To test if the just reported framing effects would be moderated by personal economic conservatism, the exact same model was used. Economic conservatism was centered and added to the model along with all interactions of economic conservatism and our experimental factors. Cultural conservatism was also centered and included as a control variable along with age and sex.
This analysis revealed a significant type of issue by type of benefit by economic conservatism interaction, F(1,616.62) = 7.54, p = .006. A significant effect was also observed for cultural conservatism, F(1,205.86) = 16.05, p < .001. The effect on the dependent variable was positive, B = 1.15, SE = .29, p < .001, that is, arguments were rated as more persuasive by participants high in cultural conservatism. The random intercept for participants was significant also in this analysis, B = .37, SE = .07, p < .001, 95% CI [.25, .54]. The variance for the type of benefit by outcome focus effect for specific issues was significant, as well, B = .12, SE = .05, p = .017, 95% CI [.05, .27].
To decompose the significant type of issue by type of benefit by economic conservatism interaction, the same model was run again for high and for low economic conservatism. To do this economic conservatism was re-centered at one standard deviation above and below the mean. The analysis revealed that the type of issue by type of benefit interaction was not significant for participants low in economic conservatism, F(1,38.81) = 2.56, p = .118. However, for participants high in economic conservatism the two-way interaction was significant, F(1,38.38) = 18.37, p < .001; that is, achievement versus security arguments were rated differently for different types of issues by participants high in economic conservatism but not by participants low in economic conservatism. To better understand these two-way interactions, the same models were again run separately for economic and cultural conservative issues.
When economic conservatism was low, the main effect for type of benefit was marginally significant for economic conservative issues, F(1,17.84) = 3.28, p = .087 and non-significant for cultural conservative issues, F(1,31.05) = 0.17, p = .679. When economic conservatism was high the main effect was significant for both economic conservative issues, F(1,17.69) = 10.38, p = .005, and cultural conservative issues, F(1,30.43) = 10.02, p = .004. However, as described above, the main effect played out differently depending on type of issues; that is, for economic conservative issues achievement arguments were rated as more persuasive (M = 4.27, SE = .32) than security arguments (M = 3.44, SE = .32), while for cultural conservative issues security arguments (M = 3.90, SE = .32) were rated as more persuasive than achievement arguments (M = 3.30, SE = .32). This two-way interaction is also clearly visible in the right-hand part of Figure 3.
Figure 3
Perceived persuasiveness as a function of type of issue, type of benefit and economic conservatism (Study 2).
Moderating Effects of Cultural Conservatism
Next, cultural conservatism was centered and added to the model along with all interactions with our experimental factors. Here, centered economic conservatism was included as a control variable along with age and sex. The analysis again revealed a significant main effect of cultural conservatism, F(1,205.86) = 16.06, p < .001. More important, a significant type of issue by type of benefit by outcome focus by cultural conservatism interaction emerged, F(1,821.86) = 7.50, p = .006. In the analysis of random effects again the intercept for participants was significant, B = .36, SE = .07, p < .001, 95% CI [.25, .53]. There was also a marginal effect for the intercept for specific issues, B = .21, SE = .13, p = .097, 95% CI [.06, .68] and a significant type of benefit by outcome focus effect for specific issues, B = .12, SE = .05, p = .019, 95% CI [.05, .27].
To probe the significant four-way interaction, the same model was run again for high and for low cultural conservatism. To do this cultural conservatism was also re-centered at one standard deviation above and below the mean. The three-way type of issue by type of benefit by outcome focus interaction was significant for high cultural conservatism, F(1,39.35) = 5.86, p = .020, but not for low cultural conservatism, F(1,38.95) = 0.07 p = .793. Under low cultural conservatism only the aforementioned two-way interactions for type of issue by type of benefit, (F(1,39.67) = 6.33, p = .016) and for type of issue by outcome focus (F(1,38.83) = 5.54, p = .024) were significant (see Figure 4a). It is noteworthy that under low cultural conservatism the main effect of type of benefit was significant (F(1,17.66) = 5.27, p = .034) and the effect of outcome focus was marginally significant (F(1,17.63) = 3.30, p = .086) for economic conservative issues. For cultural conservative issues neither main effect was significant under low cultural conservatism (F(1,30.31) = 1.77, p = .193 and F(1,30.31) = 2.45, p = .128).
Figure 4a
Perceived persuasiveness as a function of type of issue, type of benefit and outcome focus under low cultural conservatism (Study 2).
To further probe the significant three-way interaction found under high cultural conservatism, the same model was again run separately for economic conservative issues and cultural conservative issues. For economic conservatism, no type of benefit by outcome focus interaction could be observed, F(1,18.04) = 1.53, p = .231. Here, only the aforementioned main effects of type of benefit (F(1,18.04) = 7.74, p = .012) and outcome focus (F(1,18.01) = 4.54, p = .047) were significant (see Figure 4b). For cultural conservative issues both main effects of type of benefit (F(1,31.62) = 5.16, p = .030) and outcome focus (F(1, 31.62) = 4.45, p = .043) were significant, as well. However, these effects were qualified by a significant type of benefit by outcome focus interaction when cultural conservatism was high, F(1,31.62) = 6.28, p = .018. Pairwise comparisons confirmed that loss/security arguments (M = 4.52, SE = 0.36) were rated to be more persuasive than gain/security arguments (M = 3.63, SE = 0.36), F(1,12.03) = 8.63, p = .012. However, there was no significant difference between gain/achievement (M = 3.68, SE = 0.36) and loss/achievement arguments (M = 3.60, SE = 0.36), F(1,12.03) = 0.14, p = .714. Figure 4b clearly shows this interaction.
Figure 4b
Perceived persuasiveness as a function of type of issue, type of benefit and outcome focus under high cultural conservatism (Study 2).
Discussion
A first aim of Study 2 was to replicate results from Study 1. Indeed, the two-way interactions between type of issue and both experimentally varied framings were found again, indicating that achievement versus security and gain versus loss framings were differentially effective for economic and cultural conservative issues. For economic conservative issues the main effects of type of benefit and outcome focus were also replicated (supporting Hypothesis 1), albeit the effect of outcome focus was only borderline significant. For cultural conservative issues no significant effects were found in Study 1, in Study 2 the initial analysis yielded main effects for both types of framing in the expected direction, providing new support for Hypothesis 2. As reported above, considerable issue variance was observed in Study 1. Dropping problematic issues after Study 1 might have led to stronger effects for cultural conservative issues.
A second important aim of Study 2 was to test whether argument frames would also interact with participants’ individual economic or cultural conservatism. In particular, gain-framed and achievement-related arguments were assumed to be more persuasive for participants high in economic conservatism as compared to participants low in economic conservatism (Hypothesis 3). In line with this assumption, the higher perceived persuasiveness of achievement-framed arguments compared to security-framed arguments was found to be more pronounced for participants high (vs. low) in economic conservatism. However, recipients’ economic conservatism did not moderate the effect of the gain versus loss framing and, more important, the higher persuasiveness of security-framed compared to achievement-framed arguments for cultural conservative issues was also more pronounced under higher economic conservatism. This finding might hint to a greater importance of matching argument frames to the type of issue than to individual conservatism.
For cultural conservative issues, loss-framed and security-related arguments were assumed to be more persuasive for participants high in cultural conservatism as compared to participants low in cultural conservatism (Hypothesis 4). Results showed that the type of benefit by outcome focus interaction reached significance for cultural conservative issues when cultural conservatism was high. When rating arguments for cultural conservative issues participants high in cultural conservatism rated loss/security arguments as most persuasive. Persuasiveness ratings for economic conservative issues were also high due to a main effect of cultural conservatism but the two-way interaction was specific to cultural conservative issues.
In summary, support was found for both matching effects between framing and type of issue and between framing and individual conservatism. The match between framing and type of issue might be the more important one. Nevertheless, results were again consistent with the idea that a multiple matching should yield the highest persuasiveness ratings, while a multiple mismatch should yield the lowest ratings. A match between both argument frames and type of issue already led to the highest ratings in the initial analysis and these effects were boosted by matching individual conservatism. The figures presented even hint at a linear trend. However, the unexpected result of economic conservatism also boosting framing effects for cultural conservative issues raises the question, if general conservatism served to boost pre-existing framing effects or if there were specific effects for economic and cultural conservatism. We would like to argue that effects were, at least in part, specific. First of all, different effects were moderated by economic and cultural conservatism. Further, higher cultural conservatism specifically caused a significant two-way interaction to emerge for cultural conservative issues, whereas higher cultural conservatism did not really change framing effects for economic conservative issues.
General Discussion
We started this article with the example of a German politician who apparently framed his political messages with regard to the audience he addressed. Was he right then when he addressed opportunities and potential gains in an interview with a business newspaper and when he addressed (job) security in an interview with a Christian newspaper? Presuming readers of the Handelsblatt score high on economic conservatism, while readers of Christ und Welt score high on cultural conservatism, the answer should be yes. However, taking the type of issue into account should be as important or even more important to create persuasive messages. Assuming that the government’s energy policy is an economic conservative issue, gain- and achievement-framed arguments should be most effective. To the best of our knowledge the present studies are the first ones showing differential framing effects by type of issue in a political context. Thus, the present research has clear practical implications for the field of political communication. There are, however, also important theoretical implications.
The present research can further contribute to our understanding of the motivational underpinnings of two different kinds of conservatism and it also contributes to the discussion on the dimensionality of political conservatism. Results show that different argument frames are most effective for different types of issues and also for different kinds of conservatism. Gain and achievement arguments were more effective for economic conservative issues and loss and security arguments or especially loss/security arguments were more effective for cultural conservative issues and individuals high in cultural conservatism. Therefore, results support the assumption that economic conservatism is motivated by a need for personal gain and accomplishment, while cultural conservatism is motivated by concerns of stability and security and a fear of loss. In consequence, the results also support the assumption that there are at least two psychologically different dimensions of conservatism (Crowson, 2009; Feldman & Johnston, 2014). To the extent that economic and cultural conservatism are motivated by different psychological needs, it should also be justified to differentiate between the two dimensions. Hence, the presented results also contradict the assumption that both economic and cultural conservatism are ultimately motivated by security needs and hence can be summarized into one dimension (Jost et al., 2007). If both economic and cultural conservatism were chiefly motivated by security needs then security arguments should have been the most persuasive arguments for economic conservative issues and for highly economic conservative participants, as well.
An interesting point of discussion lies in the possible relationships between the two dimensions of conservatism and regulatory focus. The arguments used in the present research were modeled after those created by Yi and Baumgartner (2009) in a study on regulatory focus and message framing. Gain and achievement arguments were repeatedly found to be more persuasive for promotion-focused individuals, while loss and security arguments were repeatedly found to be more persuasive for prevention-focused individuals. Consequently, one might argue that economic conservatism should be related to the promotion focus, while cultural conservatism should be related to the prevention focus. Surely, there is some overlap with regard to the motivational bases of the respective constructs. On the other hand, recent studies on regulatory focus and conservatism pointed out different interrelations. Cornwell and Higgins (2013) connected the regulatory foci to Haidt’s binding moral foundations (ingroup/loyalty, authority/respect, purity/sanctity; Haidt & Graham, 2007). The binding moral foundations should correspond to cultural conservatism and Cornwell and Higgins (2013) reported a positive relationship of this dimension and the prevention focus and a negative relationship of this dimension and the promotion focus. This finding is also consistent with the notion that the promotion focus is related to openness and creativity (e.g., Friedman & Förster, 2001). Nevertheless, our findings might point to some interesting future research questions. Research on regulatory focus and political attitudes could surely benefit from differentiating between at least two different ideological dimensions.
Despite the theoretical relevance some limitations to the present studies need to be addressed, as well. For one, our sample displayed a political center-left bias. This should not be too much of a concern, however, because there was variance in both economic and cultural conservatism and mean persuasiveness ratings for our (conservative) arguments were around the scale midpoint. Our sample was altogether more diverse than commonly employed student populations. A more relevant point is the selection of political issues. We employed political issues that referred to actual social issues that are currently discussed in Germany. While these issues could be clearly identified as either economic or cultural conservative, they were not necessarily homogenous and might have differed in more than one characteristic. In consequence, considerable issue variance was observed. Fortunately, this source of variance can be controlled when linear mixed models are used. The problem of issue variance might also be hard to avoid. Creating artificial political issues might not actually be feasible and would decrease ecologic validity.
Further, we selected issues that were clearly either economic or cultural conservative. If unclear issues were used or issues that contained aspects of both economic and cultural conservatism, different effects might emerge. In this case, influence of type of issue might be weaker, whereas the influence of personal conservatism might be stronger. This should be an interesting question for future research, as well. In the present research we saw that individuals high in cultural conservatism also gave rather high ratings to arguments supporting economic conservative issues. It would be interesting to see conservatives’ reactions to arguments for culturally liberal or economic egalitarian issues when argument frames matched those issues or when argument frames matched recipient motivations (i.e., when security arguments were used to argue for social justice). Research on ideology is usually focused on conservatism but future research could indeed try to frame arguments for culturally liberal or economic egalitarian issues, as well. Drawing on Schwartz’ model of basic human values economic egalitarianism should be motivated by self-transcendence values. Thus, arguments supporting social justice issues could address benevolence and universalism. Likewise arguments supporting issues of political tolerance could address openness to change values. Further, future research could also fruitfully analyze the arguments presented by policy makers in actual political debates. Do politicians intuitively use the effects demonstrated here? Many politicians might not be aware of the theoretical background presented here but they might know their targeted audiences quite well. Content analysis of arguments brought forward for different issues by different political parties or facing different audiences should be worthwhile.
In summary, the present research can contribute to our understanding of political communication and future research could further experiment with different combinations of matching and mismatching argument frames to find out more about how persuasive arguments can be framed for different political audiences.