1 Introduction
Have you ever thought about the processes activated in your mind when you are moving toward the source of a sudden incoming sound? Or when focusing on a specific person’s voice in a crowded and noisy room? When these processes are at work, two separate mental systems are activated according to the dual process model of mind (Reference EvansEvans, 2003; Reference De Neys and GlumicicKahneman, 2011). Respectively, Type 1 processes are mostly evolutionarily older and characteristically produce automatic and intuitive responses whereas Type 2 corresponds to analytic and controlled processes which developed later in our evolutionary history and typically require considerable working memory resources (Reference Evans and StanovichEvans & Stanovich, 2013). Differential reliance on these two systems and concomitant cognitive styles may shape our social and political responses.
Researchers have recently argued that religious belief depends more on Type 1 thinking (Reference Gervais and NorenzayanGervais & Norenzayan, 2012; Reference NorenzayanNorenzayan, 2013; Reference Pennycook, Cheyne, Seli, Koehler and FugelsangPennycook, Cheyne, Seli, Koehler & Fugelsang, 2012; Reference Shenhav, Rand and GreeneShenhav, Rand & Greene, 2012; Reference Yilmaz, Karadöller and SofuogluYilmaz, Karadöller & Sofuoglu, 2016; but see Reference Piazza and SousaPiazza & Sousa, 2014). Social conservatism, which is positively correlated with religiosity, can also vary depending on the cognitive or thinking styles. For example, the need for cognitive closure is higher in conservative people than in liberals (Reference KruglanskiKruglanski, 2004). In addition, conservative people exhibit more negativity bias than liberals (Reference Hibbing, Smith and AlfordHibbing, Smith & Alford, 2014). Liberals also show higher levels of integrative complexity — the tendency to consider and link multiple perspectives — than conservatives (Reference TetlockTetlock, 1983). Moreover, one of the core elements underlying conservative ideology — acceptance of hierarchy — is also related to cognitive style. For example, Reference Zitek and TiedensZitek and Tiedens (2012) showed that social hierarchies are remembered and processed more easily and liked better than less hierarchical stimuli, thereby showing the intuitive, low-effort nature of the processing and acceptance of hierarchy. These findings from diverse research streams imply that social conservatism is associated with Type 1 or intuitive thinking, much as religious belief is (see also Reference Pacini and EpsteinPacini & Epstein, 1999; Reference Shook and FazioShook & Fazio, 2009; Reference SidaniusSidanius, 1985; Reference Van Berkel, Crandall, Eidelman and BlancharVan Berkel, Crandall, Eidelman & Blanchar, 2015). Although all of these characteristics differentiating liberals and conservatives are not directly connected to cognitive style in the same way, we suggest that most, if not all, of them are parsimoniously explained by the distinction between intuitive and analytic thinking styles.
In recent years, there has been an increase in the number of studies investigating cognitive style differences between liberals and conservatives (Reference Eidelman, Crandall, Goodman and BlancharEidelman, Crandall, Goodman & Blanchar, 2012; Reference Brandt, Evans and CrawfordBrandt, Evans & Crawford, 2015; Reference Talhelm, Haidt, Oishi, Zhang, Miao and ChenTalhelm et al., 2015). For example, Talhelm et al. (2015) demonstrated that thinking analytically led both American and Chinese participants to favor more liberal social attitudes whereas thinking holistically led them to favor more conservative social attitudes. Some studies have found that liberals show higher scores on the Cognitive Reflection Test (CRT; Reference Deppe, Gonzalez, Neiman, Jacobs, Pahlke, Smith and HibbingDeppe et al., 2015; Reference Pennycook, Cheyne, Seli, Koehler and FugelsangPennycook et al., 2012; Reference Iyer, Koleva, Graham, Ditto and HaidtIyer, Koleva, Graham, Ditto & Haidt, 2012), a test used to measure the dispositional tendency to think analytically (Reference Baron, Scott, Fincher and MetzFrederick, 2005). However, Kahan (2013), using a different sampling procedure that attempted to sample the American population in a more representative way, found that conservatives are in fact slightly (but not significantly) more likely to correct their intuitive responses than liberals on the CRT. There is also hardly any evidence on this issue outside of Western cultures.
We report data collected in Turkey, a non-western and predominantly Muslim country whose political system has more than two major political parties. Turkey presents interesting challenges to political psychologists. For instance, it is difficult to apply the traditional left-right or liberal-conservative distinctions, since European style social democracy is not prevalent (see Onis, Reference Onis2007, 2009, for a detailed discussion). Although few researchers have studied the nature of the left-right distinction in Turkey, a recent set of studies found that participants’ self-placement on the left-right continuum predicts their moral foundations (Yilmaz, Harma, Bahçekapili & Cesur, 2016; Yilmaz, Saribay, Bahçekapili & Harma, invited revision) in parallel with the U.S. findings (Reference Graham, Nosek, Haidt, Iyer, Koleva and DittoGraham et al., 2011).
Here we investigate the relationship between analytic cognitive style (ACS) and political ideology in Turkey, while measuring several other variables. Study 1 examined the relationship between ACS and political orientation in a Turkish sample, as an extension of this topic of investigation to non-Western cultures (see Reference Henrich, Heine and NorenzayanHenrich, Heine & Norenzayan, 2010). In Study 2, we investigated the relationship between ACS and different aspects of political ideology while measuring other potential causal factors such as personality traits, need for cognitive closure, and religiosity. In Study 3, we tried to manipulate ACS in two different samples in order to investigate its causal role on political ideology. In Study 4, we attempted a conceptual replication of Eidelman et al.’s (2012) results to investigate the causal influence of intuitive thinking style on political ideology.
2 Study 1
2.1 Method
Participants
With an estimated correlation coefficient of .15, an 85% power for detecting an effect required a sample of at least 314 participants. We therefore collected data from 356 participants (mean age = 25.83, SD = 9.37, 203 females, 142 males, 11 unreported). Two hundred and thirty-two of them were enrolled in an introductory psychology course at Doğuş University (Istanbul) in return for extra course credit. The remaining sample was non-student, with ages ranging from 25 to 76, collected via snowball sampling. All participants were native Turkish speakers.
Materials and Procedure
All materials in the current set of studies were administered in Turkish. The CRT consists of three different questions designed to measure a dispositional tendency to think analytically (Reference Baron, Scott, Fincher and MetzFrederick, 2005) and is a widely used measure of cognitive style (Reference Toplak, West and StanovichToplak, West & Stanovich, 2011). Each question has one correct, reflective (Type 2) and an incorrect, intuitive/spontaneous answer (Type 1). ACS score was operationalized as the total number of correct answers given to the three CRT questions.
Political orientation was measured by the traditional one item self-placement question from 0 (“left”) to 10 (“right”). Participants were first asked to respond to the CRT, then a demographic form (age, sex, SES, hometown size, ethnicity, religious and political affiliations, preference for political party, and identification with political party were obtained), and finally the political orientation question.
2.2 Results and discussion
As predicted, ACS (here just the 0–3 score on the CRT test) was negatively correlated with political orientation [r (348) = –.163, p = .002], replicating past research (Reference Deppe, Gonzalez, Neiman, Jacobs, Pahlke, Smith and HibbingDeppe et al., 2015; Reference Pennycook, Cheyne, Seli, Koehler and FugelsangPennycook et al., 2012; Reference Iyer, Koleva, Graham, Ditto and HaidtIyer et al., 2012). Table 1 shows the correlations of each of the other variables with ACS and political orientation. It is apparent that, of these, only Age can help to explain the observed negative correlation, because older subjects get lower scores on ACS and they are more conservative; but it does not fully explain the ACS-politics correlation; the relationship is still significant (standardized β = −0.138, p=.013) when ACS is regressed on Age and political orientation.
Table 1: Correlations of each variable with ACS and political orientation (POL, high scores are right wing)
Note: a correlation of 0.095 is significant at p <. 05, 0.134 at p <. 01, two tailed.
In the next study, we used an additional measure of ACS (see Baron, 2015; Reference Adorno, Frenkel-Brunswik, Levinson and SanfordBaron, Scott, Fincher & Metz, 2015, for a justification of using measures in addition to CRT), and differentiated between separate components of political ideology.
3 Study 2
3.1 Method
Participants
With an estimated correlation coefficient of .15, a 95% power for detecting an effect required a sample of at least 476 participants. We collected data from as many participants as possible and exceeded the minimum sample requirement, since participants were readily available. As a result, we collected data from 750 participants (mean age = 20.63, SD = 2.13, 452 females, 256 males, 22 no answer), who were enrolled in an introductory psychology course at Boğaziçi University (Istanbul) and participated in return for extra course credit.Footnote 1 All students were native Turkish speakers.
3.1.1 Measures
Analytic Cognitive Style. In addition to the CRT (see Study 1), we employed three different base-rate conflict (BRC) problems as used by Penncycook et al. (2012). BRC problems consist of a salient stereotype with probabilistic information such as the following:
In a study 1000 people were tested. Among the participants there were 4 kindergarten teachers and 996 executive managers. Lilly is a randomly chosen participant of this study. Lilly is 37 years old. She is married and has 3 kids. Her husband is a veterinarian. She is committed to her family and always watches the daily cartoon shows with her kids.
What is most likely?
a) Lilly is an executive manager (correct answer)
b) Lilly is a kindergarten teacher
In this kind of question, people generally ignore the base-rate probability (99.6 % chance that Lilly is an executive manager) and select the intuitive, wrong answer (Reference De Neys and GlumicicDe Neys & Glumicic, 2008). This mistake can be prevented by thinking analytically.Footnote 2 For the six questions, coefficient α (for the number of correct answers) was .66.
Big Five Personality. Since personality has been found to be related to political orientation (Reference Carney, Jost, Gosling and PotterCarney, Jost, Gosling, & Potter, 2008), we measured it using the Big Five Personality Inventory (BFI; Reference Benet-Martinez and JohnBenet-Martinez & John, 1998). BFI was translated and adapted to Turkish by Sümer, Lajunen, and Özkan (2005). The inventory has five subscales: Neuroticism (α = .80), Extraversion (α = .88), Openness (α = .79), Agreeableness (α = .69), and Conscientiousness (α = .78). Forty-four personality descriptors are presented and the participant is asked to indicate the extent to which each of these is self-descriptive (e.g., “I see myself as a talkative person”), ranging from 1 (strongly disagree) to 5 (strongly agree).
Social/Political Conservatism. We used two different scales to measure social conservatism. The first — Social/Political Conservatism scale — was developed by Reference Olcaysoy and SaribayOlcaysoy and Saribay (2014) by compiling items from the Social Dominance Orientation scale (Reference Pratto, Sidanius, Stallworth and MallePratto, Sidanius, Stallworth & Malle, 1994), Right-Wing Authoritarianism scale (Reference Altemeyer and HunsbergerAltemeyer & Hunsberger, 1992), F-scale (Reference Adorno, Frenkel-Brunswik, Levinson and SanfordAdorno, Frenkel-Brunswik, Levinson & Sanford, 1950), Social and Cultural Attitudes scale (Reference KüçükerKüçüker, 2007), Egalitarianism-Inegalitarianism scale (Reference Kluegel and SmithKluegel & Smith, 1983), and items measuring resistance to change used by Jost et al. (2007). The scale (Appendix B, Table A1) measures the two dimensions of conservatism as theorized by Reference Jost, Glaser, Kruglanski and SullowayJost, Glaser, Kruglanski & Sulloway (2003): resistance to change (two different samples’ Cronbach’s α = 0.80, 0.83) and opposition to equality (α = 0.90, 0.88). The response scale ranged from 1 (totally disagree) to 7 (totally agree). Scores on resistance to change (α = .88 for this study) and opposition to equality (α = .75 for this study) are summed and averaged for each participant in order to form a composite conservatism score in which higher scores indicate more conservative attitudes (one factor solution α = .88 for this study).
The second measure we used — Revised Version of Scale of Social Conservatism — was developed by Henningham (1996) over almost two decades in the Australian culture and adapted to the current American political system by Reference Piazza and LandyPiazza and Landy (2013). In the original scale participants respond dichotomously on each social issue by choosing “opposed to it” or “not opposed to it”. We revised the scale in accordance with Turkish politics removing some items (e.g., “Outlawing the buying and selling of firearms”) and adding new ones, resulting in 15 items (Appendix B, Table A2). In addition, we changed the response format to an 11-point scale ranging from –5 (strongly disagree) to +5 (strongly agree). The items had good reliability (α = .89). Two items with low (below .20) item-total correlations were excluded. Higher scores indicated greater social conservatism.
Personal Conservatism Scale. This scale was also developed by Reference Olcaysoy and SaribayOlcaysoy and Saribay (2014) in order to measure personal (as opposed to political) conservatism on the same two dimensions: resistance to change and opposition to equality. The items (Appendix B, Table A4) were devoid of political content and focused instead on the individual’s personal preferences and lifestyle. It was developed by compiling items from Resistance to Change scale (Reference OregOreg, 2003), Need for Cognitive Closure scale (Reference Kruglanski, Webster and KlemKruglanski, Webster & Klem, 1993) and 14 new items. The response scale ranged from 1 (totally disagree) to 7 (totally agree) and scores on resistance to change (α = .86 for this study) and opposition to equality (α = .80 for this study) were summed and averaged for each participant in order to form a composite personal conservatism score in which higher scores indicate more personal conservatism (one factor solution α = .86 for this study).
Economic Conservatism Scale. In Turkish, to the best of our knowledge, there is no scale measuring economic political attitudes. We composed 16 items suitable for Turkish participants (Appendix B, Table A3). Since these 16 items had good reliability (α = .73), we averaged them into a single score of economic conservatism.
Religiosity. We used the Turkish adaptation of the intuitive religious belief scale (IRS; see Reference Yilmaz and BahçekapiliYilmaz & Bahçekapili, 2015) developed by Reference Gervais and NorenzayanGervais and Norenzayan (2012). It has 5 items with responses given on a scale ranging from 1 (strongly disagree) to 5 (strongly agree) (e.g., “When I am troubled, I feel the need to seek help from God”).
Need for Cognitive Closure. Need for Cognitive Closure scale (NFCC) was developed by Reference Webster and KruglanskiWebster and Kruglanski (1994) and revised by Roets and Van Hiel (2007). Subsequently, Roets and Van Hiel (2011) validated an abridged version of the scale. This scale consists of 15-items and the response scale ranged from 1 (strongly disagree) to 7 (strongly agree). NFCC has five subscales: order, predictability, decisiveness, ambiguity, and closed-mindedness. Every subscale is represented by 3 items (e.g., “I find that a well ordered life with regular hours suits my temperament”). We combined all the subscales into a single need for cognitive closure score (α =.84).
Demographic Questions. Information about participants’ demographic background including age, sex, SES, hometown size, ethnicity, religious and political affiliations, preference for and identification with political party was obtained.
3.1.2 Procedure
At the beginning of the Spring 2015 semester, students taking Introduction to Psychology course at Boğaziçi University received an e-mail invitation to complete an online battery of measures for extra-course credit. Students were given two weeks to complete the battery and they were free to complete it at their own pace but were asked to complete it in only one session. The battery contained all the measures listed above and its completion took approximately 45 minutes. We counterbalanced the order of the analytic thinking and political conservatism measures.
3.2 Results and discussion
Table 2 shows the main correlations of interest. As predicted and consistent with Study 1, ACS is negatively correlated with Political Orientation, Social Conservatism (both forms), Personal Conservatism, and NFCC, but did not significantly correlate with Economic Conservatism. In addition, ACS is negatively correlated with Religiosity, replicating previous research (Reference Gervais and NorenzayanGervais & Norenzayan, 2012; Reference Pennycook, Cheyne, Seli, Koehler and FugelsangPennycook et al., 2012; Reference Shenhav, Rand and GreeneShenhav et al., 2012; Reference Yilmaz, Karadöller and SofuogluYilmaz et al., 2016). Appendix B shows the main correlations by item, along with the text of the items.
Table 2: Correlations of measures with Political Orientation (POL), Personal Conservatism (PER), Economic Conservatism (ECO), Social Conservatism (SC, and SCR, revised), and Analytic Cognitive Style (ACS)
Note: Correlations greater than .140 are significant at p < .0001, .117 at p < .001, .088 at p < .01, and .062 at p < .05, two tailed and uncorrected for multiple tests.
Among the other measures, possible variables that could help to explain the correlation between ACS and social conservatism are those that correlated positively with ACS and negatively with a measure of social conservatism, or the opposite. These include Age, NFCC, Conscientiousness, and especially Religiosity. Note that, unlike Study 1, Age correlates slightly positively with ACS, surely a result of the different sampling procedure. For Political Orientation, the variables that could play a similar role are NFCC, Conscientiousness, Openness to Experience, and Religiosity.
Overall, these results demonstrated that ACS is correlated with social conservatism, personal conservatism, and political orientation, but not economic conservatism. In Study 3 and 4, we aimed to go beyond these promising correlational findings and investigate the causal relation between cognitive style and political ideology.
4 Study 3
In this study, we aimed to investigate the causal role of ACS on political ideology. Before conducting the main study, we conducted two studies to replicate, in a Turkish sample, the effectiveness of analytic primes previously used in some published research (Reference Gervais and NorenzayanGervais & Norenzayan, 2012; Reference Shenhav, Rand and GreeneShenhav et al., 2012) in order to find a way of manipulating ACS.
4.1 Study 3A
We first tried to replicate the thought prime technique, originally used by Shenhav et al. (2012), with a total of 145 (105 females, 39 males, 1 unreported; mean age = 21.45, SD = 1.77) undergraduates in Doğuş University. They participated in this study for extra course credit. They were randomly assigned to the Intuitive-positive condition (n = 39), the Intuitive-negative condition (n = 35), the Analytic-positive condition (n = 37), or the Analytic-negative condition (n = 34) and then they were asked to respond to the standard three-item CRT and BRC (base rate) problems used in Study 2. We computed CRT and BRC scores and combined them into a single ACS score.
In the Intuitive-negative condition, participants were asked to “write a paragraph (approximately 8–10 sentences) describing a time your intuition/first instinct led you in the wrong direction and resulted in a bad outcome” (The italics represent the parts that differed between the four conditions). In the Intuitive-positive condition, participants were asked to “write a paragraph (approximately 8–10 sentences) describing a time your intuition/first instinct led you in the right direction and resulted in a good outcome”. In the Analytic-positive condition, participants were asked to “write a paragraph (approximately 8–10 sentences) describing a time carefully reasoning through a situation led you in the right direction and resulted in a good outcome. Lastly, in the Analytic-negative condition, they were asked to “write a paragraph (approximately 8–10 sentences) describing a time carefully reasoning through a situation led you in the wrong direction and resulted in a bad outcome”.
In contrast to Shenhav et al.’s (2012) results, a one-way ANOVA revealed no significant main effect of thought-priming on ACS, F(3, 123) = 0.18, p= .912, η p² = .004.
4.2 Study 3B
As part of our attempt to discover a suitable procedure that primes analytic thinking, we also tried to replicate the cognitive disfluency paradigm (Reference Gervais and NorenzayanGervais & Norenzayan, 2012) on a Turkish sample. We estimated a medium effect (f) of .3, which required a total sample of 90 with 80% power of detecting any effect. Three participants (one in the Analytic, two in the Neutral condition) were excluded from the analyses because they did not respond to any manipulation check questions. Eighty seven (73 females, 13 males, 1 unreported; mean age = 22.22, SD = 3.67) undergraduates in Doğuş University participated in this study for extra course credit and they were randomly assigned to either the Analytic (n = 44) or the Neutral (n = 43) conditions and then they were asked to respond to the standard three-item CRT used in Study 1. In the Analytic group, participants were given the materials in a difficult to read font to prod analytic thinking whereas in the Neutral condition, the materials were given in a standard font. However, as in Study 3A, the results revealed no significant effect of our manipulation on analytic thinking tendency, F(1, 85) = 1.18, p= .281, η p² = .014.
After these failures of replication (see Deppe et al., 2015 for similar replication failures), we decided to manipulate intuitive, instead of analytic, thinking style in the next study.
5 Study 4
Eidelman et al. (2012) found that people under cognitive load reported more conservative opinions in long-term attitudes. However, Talhelm et al. (2015) primed analytic versus holistic thinking and found that the latter increased conservatism in contextualized, less stable opinions, but unlike Eidelman et al., they did not find an effect of priming on long-term stable opinions. Thus, we decided to do a conceptual replication of Eidelman et al.’s basic results in order to test whether the activation of intuitive thinking will lead people to adopt more conservative political attitudes on contextualized opinions.
5.1 Method
Participants. We estimated a medium effect (f) of .3, which required a total sample of at least 90 with 80% power of detecting any effect. We considered potential attritions and collected data from 104 participants. Participants were selected from those who participated in the online survey (see Study 2). They participated in this study for extra course credit. All students were native Turkish speakers. They were randomly assigned to either the Cognitive load (n = 56) or the No-load (n = 48) condition. However, in the cognitive load condition, we excluded participants if they responded incorrectly in more than half of the 12 different load trials. This resulted in seven participants being excluded from the analyses.
5.1.1 Materials and measures
Manipulation. We used a cognitive load method which includes memorization of random sequences of numbers and letters presented in a random order (e.g., “ig4j6sf”)Footnote 3. In the cognitive load condition, in each of 12 trials, participants saw a sequence of numbers and letters, always consisting of a total of 6 or 7 characters, before engaging in the primary task (e.g., responding to an item of the conservatism scale). They were asked to keep this alphanumeric sequence in memory for an uncertain period of time while continuing to engage in the primary task and to report the sequence back when asked. The first sequence appeared at the beginning of the first questionnaire. The remaining 11 sequences were interspersed among items of the various questionnaires used in this study. For each sequence, memory recall was requested after a few items of the questionnaire that the participant was working on, followed by provision of the next load sequence. In each trial, if an alphanumeric sequence reported by the participant had more than half of the same characters (at least 4 correct characters out of 7) of the original sequence, it was considered a correct recall. In the no load condition, participants did not see any load items nor were they given any of these instructions.
Conservatism Measures. The Scale of Social Conservatism (α for this experiment = .90) and Economic Conservatism Scale (α for this experiment= .74) were used in this study.
Contextualized Political Opinions. In Talhelm et al.’s (2015) study, training people to think analytically or holistically led to a change in a contextualized political opinion (responses to a news article) but did not lead to a change in stable political attitudes such as having the opinion that “flag burning should be illegal.” Thus, we exposed participants to two semi-fabricated news articles that included contextualized political opinions and were also each related to a particular policy (Appendix A). We predicted that the cognitive style manipulation would not influence stable political opinions but would influence political opinions which are actively being processed by the participants.
Collectively, the articles presented two different disputes between rightist and leftist positions in the Turkish political system: One of the disputes was related to a conservative policy and included a conservative anchor (a new internet law which enables government to totally ban internet sites in only 4 hours after a complaint). The other dispute was related to a liberal policy and included a liberal anchor (an argument that the primary function of the prison system is rehabilitation of the prisoners).
Participants were asked to carefully read the article. Once finished, they answered a single question measuring their attitude on the presented issue. The 7-point response scale had conservative (e.g., “I am strongly supportive of the internet law.”) and liberal (e.g., “I am strongly against the internet law.”) anchors at the extremes and a neutral (“I do not have an opinion”) option in the middle. Higher scores in the first article represent higher conservative values whereas higher scores in the latter article represent higher liberal values. We used responses to these two news articles as two subscales: liberalism (score of rehabilitation article) and conservatism (score of internet law article) scores. Additionally, we also treated liberalism score as a reverse-coded item, and combined it with the conservatism score to get a total conservatism score.
Positive and Negative Affect Scale. This scale was developed by Reference Watson, Clark and TellegenWatson, Clark and Tellegen (1988) and was adapted into Turkish by Gençöz (2000). It measures participants’ current moods on two dimensions separately: positive and negative affect.
Design and Procedure. The study was conducted in two sessions. The first session was identical to Study 2. In the second session, at least five weeks later, participants were individually invited to the social psychology lab at Boğaziçi University and were randomly assigned to either the Cognitive-load or the No-load condition. After the manipulation phase, participants were asked to read the two semi-fabricated news articles and indicate their opinions regarding the disputes presented in them (as in Talhelm et al., Reference Talhelm, Haidt, Oishi, Zhang, Miao and Chen2015). Finally, they completed the Social Conservatism and Economic Conservatism scales. We counterbalanced the order of these two scales and we individually randomized the order of items within each scale. We implemented the experiment using Medialab (Reference JarvisJarvis, 2012).
5.2 Results and discussion
Contextualized political opinions. Contrary to our prediction, a one-way ANOVA did not reveal a significant effect of the manipulation on either the liberal-anchored article, or the conservative-anchored one (both Fs < 1).
In the liberal-anchored article (rehabilitation), participants in the load condition (M = 5.63, SD = 0.96; 95% CI [5.35, 5.90]) reported more liberal attitudes than no-load condition (M = 5.44, SD = 1.07; 95% CI [5.13, 5.75]), a non-significant difference, F(1, 94) = 0.82, p = .368, η p²= .009.
In the conservative-anchored article (Internet law), participants in the load condition (M = 2.51, SD = 1.45; 95% CI [2.10, 2.93]) reported less conservative attitudes than no-load condition (M = 2.69, SD = 1.43; 95% CI [2.27, 3.10]), but this difference was also not statistically significant, F(1, 95) = 0.37, p = .545, η p²= .004.
When we controlled for baseline political orientation, gender, SES, age, and the current mood of the participants, the results remained constant (all ps > .05).
When we treated the liberalism score as a reverse-coded conservatism item, and combined them as a total conservatism score, a one-way ANOVA revealed no significant differences among groups, F(1, 94) = 0.86, p = .355, η p²= .009.
Stable political attitudes. As expected, we did not find a significant effect of our manipulation on the social and economic conservatism scales (both ps > 21). The results remained constant when controlling for baseline political orientation, gender, SES, age, and the current mood of the participants (all ps > .05). We also used the pre-experimental Social and Economic Conservatism scores of the participants collected in the online survey and compared them with their post-experimental scores to test whether our manipulation had an effect. To assess differences between pretest and posttest, we subtracted the pretest value from the posttest and computed a single difference score. A one-way ANOVA failed to yield any significant terms (all ps > .17).
Overall, in this study, we predicted that a cognitive load manipulation would activate intuitive thinking, which in turn would lead people to adopt more conservative attitudes on the issues they are actively processing (i.e., the news articles), but would not lead to a change in people’s stable attitudes. In contrast to the previous finding that low-effort thinking enhances political conservatism (Reference Eidelman, Crandall, Goodman and BlancharEidelman et al., 2012), our manipulation did not influence people’s long-term stable political attitudes, nor political opinions that they were actively processing.
One potential limitation of this study concerns the manipulation technique used. We did not do a real manipulation check, therefore we cannot know with certainty whether our cognitive load manipulation really influenced people’s cognitive styles. Moreover, the load conditions may have been too difficult for the participants, because of the length and complexity of the alphanumeric strings, thus we may have encountered a floor effect in this manipulation. The finding that almost no one was able to memorize the load items exactly supports this argument. On the other hand, one alternative interpretation of the current results is that the original findings (i.e., those of Eidelman et al., Reference Eidelman, Crandall, Goodman and Blanchar2012) are spurious because our sample size is higher than 2.5 times (n = 104) the number that Eidelman et al. (Reference Eidelman, Crandall, Goodman and Blanchar2012) used (n = 38, see Reference Simonsohn, Nelson and SimmonsSimonsohn, Nelson & Simmons, 2014). In other words, our results may imply that since political attitudes are formed over a long number of years in a person’s life, they might be less prone to experimental manipulations.
6 General discussion
The present study aimed at clarifying the relation between cognitive style and political ideology, which our data revealed to be more complex than one might initially assume. Study 1 demonstrated a correlation between cognitive style and conservative political ideology, and Study 2 replicated and extended this result to other measures of political ideology, but not economic conservatism. In Study 3, we tried to manipulate ACS to examine its causal role on political ideology, but we could not succeed (our manipulation checks failed). In Study 4, we attempted to direct participants’ cognitive style toward intuitive thinking by a cognitive load manipulation to see whether they would become more conservative as a result, but we found no significant effect.
The original contribution of the current set of studies rests primarily on the following conclusions from our findings: (1) When merely measured, there is some evidence that ACS is associated with political ideology in Turkey. (2) Standard procedures commonly used in the literature to prime analytic thinking may not work as expected (see also Deppe et al., Reference Deppe, Gonzalez, Neiman, Jacobs, Pahlke, Smith and Hibbing2015). (3) Manipulations of intuitive thinking do not seem to affect conservative attitudes. These results are in line with some previous research findings showing the relationship between cognitive style and political ideology (Reference Brandt, Evans and CrawfordBrandt et al., 2015; partially Eidelman et al., 2012; Reference Deppe, Gonzalez, Neiman, Jacobs, Pahlke, Smith and HibbingDeppe et al., 2015; Reference Iyer, Koleva, Graham, Ditto and HaidtIyer et al., 2012; Reference Jost, Glaser, Kruglanski and SullowayJost et al., 2003; Reference Pennycook, Cheyne, Seli, Koehler and FugelsangPennycook et al., 2012; Reference Talhelm, Haidt, Oishi, Zhang, Miao and ChenTalhelm et al., 2015; Reference Van Berkel, Crandall, Eidelman and BlancharVan Berkel et al., 2015; but see Kahan, 2013).
The current research was also motivated by Henrich et al.’s (Reference Henrich, Heine and Norenzayan2010) account of WEIRDness according to which only 15% of people live in Western, educated, industrialized, rich, and democratic societies and the remaining 85% are generally out of the scope of psychological research. Thus, it is important to differentiate the characteristics of the minority (15% of the world population mostly represented by Western people) and the majority (85% of the world population mostly represented by Eastern and Southern people). Talhelm et al. (2015) suggested a sixth characteristic of Westerners: their liberal tendency. On this basis, they offer a new concept: WILDER (Western, industrialized, liberal, democratic, educated, and rich). Perhaps, a seventh letter — “A” for “analytic” — is necessary to add to this list. Like the WEIRD samples that most psychological research employs, our sample was also highly educated (most of them were university students). However, unlike the literature we refer to, our samples were all Turkish and predominantly Muslim. Furthermore, Turkey (compared to the U.S.) is unique in terms of the complexity of its political structure and its multi-party political system (see Onis, 2007, 2009). These facts increase the value of our replications (cf. Henrich et al., Reference Henrich, Heine and Norenzayan2010).
6.1 Conclusion
The current results are in line with the general view of conservatism as motivated social cognition (Reference Jost, Glaser, Kruglanski and SullowayJost et al., 2003). Some core characteristics of conservatism such as preference for hierarchy and uncertainty avoidance are seen to be a product of intuitive thinking, whereas being more tolerant and egalitarian may require more effortful thinking (Reference Van Berkel, Crandall, Eidelman and BlancharVan Berkel et al., 2015). Thus, the main difference between conservative and liberal people may stem from the general cognitive thinking style as characterized in the present research in terms of intuitive and analytic thinking styles.
Our findings also suggest the presence of boundary conditions surrounding the effect of cognitive styles on political orientation. For instance, changing stable political opinions with contextual primes may not be that easy, as shown in the current research and some previous non-significant results investigating the effect of cognitive style on political attitudes (e.g., Deppe et al., Reference Deppe, Gonzalez, Neiman, Jacobs, Pahlke, Smith and Hibbing2015, see also Yilmaz, Reference Yilmaz2015). All in all, the findings are compatible with the view that liberals are dispositionally more analytic and conservatives are more intuitive, but it is experimentally difficult to manipulate these long-term political attitudes. Therefore, in today’s culture wars between leftist (liberal) and rightist (conservative) ideological views, being more tolerant, rather than attempting to persuade the opposite parties using analytic arguments, might be a more effective tool for agreement because convincing people of certain ideologies or manipulating their long-term political attitudes may be more difficult than initially assumed.
Appendix A: News Articles, Study 4
WHAT DOES THE NEW INTERNET LAW BRING?
The Turkish Parliament accepted the legislative proposal for the new regulations concerning internet use. Although the new law received criticisms regarding the limitations it imposes on freedom, the government states that they “are restraining prohibition.”
The new law is based on a previous law numbered 5561 which took effect in 2007 and has resulted in the prohibition of access to many web sites. Its full title is “law on regulation of broadcast via internet and combating crimes committed by means of such publications.” The government claims that the proposal was prepared to regulate issues such as web site prohibition and removal which have posed difficulties in Turkey for years. Main goal of this association is to ban a web site in case of receiving such a directive from the headship of telecommunication. That is, with this new apparent procedure, the government will no longer ban internet sites directly, but a new association of access providers will ban those sites upon the directives of a governmental institution, the headship of telecommunication. When the confidentiality of private life is at issue, the prohibition will be ordered by the headship of telecommunication without waiting for a court decision. How the concept of “private life” should be interpreted is uncertain. A review of internet laws around the world reveals that there is no particular state intervention on the internet, especially by Western governments. This is the most important difference concerning Turkey. With this law, personal information will be kept for two years and it will be legal to carry out address-specific prohibition (i.e., the banning of the address of a single video in a web site). Some political parties and civil society organizations who are in opposition to this new law state that the law has a restrictive flavor regarding all freedoms. While some items of the proposal sound appealing, they complain of potential difficulties that may stem from uncertainties regarding how and by whom the legal status of internet content will be decided. Most generally, they fear that personal rights will be interfered with. Conversely, some other political parties and civil society organizations do not agree that this is a prohibition of the internet. These non-governmental organizations that defend the law even claim that “it is the abolition of the mechanism by which the prohibition of internet is easier”. They argue that the main goal here is to protect personal rights and amendment of grievances. Those organizations mention that speculations such as that the “internet will be banned and censored, the headship of communication will prohibit everything” have nothing to do with reality.
Which side are you on?
-
1. I am strongly against the internet law.
-
2. I am quite against the internet law.
-
3. I am against the internet law.
-
4. I do not have an opinion.
-
5. I am supportive of the internet law.
-
6. I am quite supportive of the internet law.
-
7. I am strongly supportive of the internet law.
NEW REHABILITATION PACKAGE FOR PRISONERS
In a project based on cooperation between the state and universities, a decision has emerged to carry out activities for better adaptation of prisoners to society.
Prof. Dr. Candeğer Yılmaz, rector of Ege University; Durdu Kavak, chief public prosecutor of İzmir; faculty members, and penal institution administrators participated in a ceremony for the signing of a protocol of cooperation between Ege University rectorate and the Penitentiary Campus of İzmir Aliağa, which houses the closed prison of the children and the young and the closed prison for women. Speaking at the ceremony, chief prosecutor Kavak stated that “Problems of prisons are significant. It is a very difficult and laborious task to adapt to society the convict and arrested individuals staying in prisons. Convicts and arrested individuals should be reintegrated to society, taking concrete steps rather than leaving them to ‘serve whatever sentence their crime deserves.’ This is the first project in Turkey designed to overcome this difficulty.”
Rector Prof. Dr. Candeğer Yılmaz spoke as follows: “This is the first time that we will apply the wisdom of the university to arrested and convicted individuals. Therefore, we are very excited for their reintegration to society. We are starting a new chapter with this project. We will transform prisons to educational campuses and improve the potential of prisoners. We will convey to them a new life perspective. With this protocol, we will attempt to ensure cooperation on topics such as general health, oral and dental health, addiction, psychiatry, women’s health, socio-cultural activities, and support for projects.”
Some non-governmental organizations and journalists reacted to this cooperation. They argued that rehabilitation is not the main goal of prisons and that prisons should be based on the idea of excluding from society those who disrupt societal order. Moreover, a non-governmental organization that includes some political parties advocates that this rehabilitation package should not include offenses such as rape and murder. They also emphasize that this protocol creates an image of prisons as pleasant places.
Which side are you on?
-
1. I am strongly against the new rehabilitation package.
-
2. I am quite against the new rehabilitation package.
-
3. I am against the new rehabilitation package.
-
4. I do not have an opinion.
-
5. I am supportive of the new rehabilitation package.
-
6. I am quite supportive of the new rehabilitation package.
-
7. I am strongly supportive of the new rehabilitation package.
Appendix B: Item-by-item correlations, Study 2
Table A1: Correlations of ACS with items of the Social/Political Conservatism Scale. Items marked with “(R)“ are reverse-coded. That is, for all items in these four tables, higher scores indicate higher conservatism.
Table A2: Correlations of ACS with items of the Scale of Social Conservatism.
Table A3: Correlations of ACS with items of the Economic Conservatism Scale.
Table A4: Correlations of ACS with items of the Personal Conservatism Scale.
Open access
