Adhering to initial judgment: How power distance belief increases preference consistency

ABSTRACT This paper demonstrates that individuals with high (vs. low) power distance belief (PDB), who tend to support inequality in society, are reluctant to change their initial judgments when receiving preference inconsistent (vs. consistent) information, thereby showing a greater preference consistency effect through three studies. This effect is driven by high (vs. low) PDB individuals’ greater resistance to change. Specifically, high (vs. low) PDB individuals are less likely to change their initial judgment because they are more reluctant to make changes in their lives, which leads to a greater preference consistency effect. Furthermore, we demonstrate one important boundary condition: a positive affective state. Key implications are also discussed.


Introduction
Culture has been extensively examined as one of the most fundamental influences on consumer attitude and behavior (Shavitt et al., 2009). Consumers react differently to the same stimulus depending on the culture in which they have learned certain associations, knowledge, beliefs, and so on, and abundant evidence suggests that culture (either in countries as a proxy or chronically measured) influences consumer attitudes in different domains and situations. For example, consumers in the United States showed a more favorable attitude to an advertisement with individualistic (vs. collectivistic) appeals while consumers in Korea showed the opposite (Han & Shavitt, 1994). Moreover, Chinese (vs. American) consumers take a greater risk in the financial area but not in the academic or medical areas because they have more financial assistance and help from others when they are in need (Hsee & Weber, 1999). Furthermore, consumers in the United States tend to adopt an attenuation strategy, whereas those in China tend to adopt an additive strategy when they face incongruity between source and attribute information (Aaker & Sengupta, 2000). While prior studies focused more on how to form consumers' initial attitudes, the current research focuses on how to change their initial attitudes. Specifically, does culture impact consumers' tendency to maintain or change their initial preferences/judgments when they encounter inconsistent information? If so, what specific cultural variable is driving this effect? This topic is especially important due to a wealth of product information from a wide range of sources nowadays (Mansoor, 2018).
What happens when a consumer encounters conflicting information from different sources? For example, suppose a person reads that a digital camera, Brand A, is superior in terms of its attributes to its competitor, Brand B, on a blog written by a technology expert. Based on this, he or she has a favorable attitude toward Brand A. However, after browsing some more, the person encounters the opposite information on a recently released consumer report that states that Brand B has better attributes than Brand A. Is the person likely to change his or her initial judgment after encountering subsequent inconsistent information or stick with the initial judgment and neglect the subsequent information?
Previous research on preference (in)consistency has demonstrated that individuals tend to favor preference consistent information (Ditto & Lopez, 1992;Jain, 2003;Jain & Maheswaran, 2000;Kunda, 1990). For example, individuals are less likely to examine preference consistent information critically compared to preference inconsistent information (Ditto & Lopez, 1992). Similarly, individuals tend to create more counterarguments for preference inconsistent (vs. consistent) information (Jain & Maheswaran, 2000). In this stream of research, the preference (in)consistency effect has also been studied from a cultural perspective (see Aggarwal et al., 2013), demonstrating that Westerners (vs. Easterners) show greater discomfort and strong motivation to resolve the difference between inconsistent information. Specifically, Aggarwal et al. (2013) demonstrate that students from a Canadian university (i.e., Westerners) are more likely to change their purchase intention when provided with preference inconsistent information compared to students from a Korean university (i.e., Easterners) because Westerners try to reduce cognitive discomfort by generating more counterarguments. Importantly, they used two different countries to represent Westerners and Easterners. While typifying an important step toward understanding the preference (in)consistency effect from a cultural perspective, this research did not thoroughly detangle different aspects of culture (e.g., individualism-collectivism, power distance, masculinity, etc., as suggested by Hofstede, 2001) because they used the country as the independent variable rather than focusing on specific cultural variables, such as individualism, power distance, and uncertainty avoidance. In other words, we were unable to determine whether the effect was driven by different levels of individualism-collectivism or power distance, and so on, thereby limiting our understanding of the exact driver for the effect and its mechanism. As cultures are combinations of different components, this limitation calls for a systematic examination of certain cultural dimensions on the preference (in)consistency effect and the understanding of the exact underlying mechanism behind the effect.
To meet this need, this article focuses on the role of one specific cultural aspect, power distance belief (PDB) -the extent to which people endorse and expect inequality in a particular society (Hofstede, 2001;Oyserman, 2006) -on individuals' tendency to change their initial judgments and preferences after receiving preference inconsistent information. Specifically, we predict that high PDB individuals' greater resistance to change -that is, the extent to which people resist or avoid making changes (Oreg, 2003) leads them to stick with their initial judgments even after encountering inconsistent and contradictory information. This is because individuals with greater resistance to change (i.e., high PDB individuals) regard changes as negative, thereby showing a greater tendency to maintain the status quo. This suggests that high (vs. low) PDB individuals are reluctant to change their initial preferences and judgments.
This research makes both theoretical and managerial contributions. Theoretically, this research contributes to the preference (in)consistency literature by identifying a new antecedent for the preference (in)consistency effect: PDB. As PDB is an important component of culture, this is expected to enhance our understanding of the role of specific cultural aspects in the preference (in)consistency literature. It also reveals that the impact of PDB on preference (in)consistency stems from individuals' resistance to change. This suggests that various consumer behaviors, such as brand switching, willingness to adopt new technology, or endowment effect, can differ by consumers' level of PDB. This research also has important managerial implications. For example, managers should pay greater attention to what information their target customers are likely to encounter first, especially if their target customers are high in PDB, as their initial judgments are less likely to change with subsequent information. Therefore, marketers should ensure that positive information is delivered to their target customers first, and this is especially important when their target customers are high in PDB.

Preference inconsistent information and preference (in)consistency
The previous literature has demonstrated how motivational factors such as self-serving bias (Ditto & Lopez, 1992), motivation to be consistent with their goal (Ahluwalia et al., 2000), and the difference between desired and actual confidence levels (Jain, 2003) make individuals favor their preexisting beliefs, suggesting that these factors lower the tendency to change individuals' initial judgments when they encounter inconsistent information (Ditto & Lopez, 1992;Jain, 2003;Jain & Maheswaran, 2000;Kunda, 1990;Maheswaran & Chaiken, 1991;Ahluwalia et al., 2000). For example, individuals tend to process preference inconsistent information more critically (Jain, 2003) and require more information if the subsequent information is not consistent with their initial judgment (Ditto & Lopez, 1992), thus showing a preference for consistent information (i.e., defensive or motivated information processing).
However, there are several circumstances under which individuals are more likely to change their initial judgment. For example, individuals with low commitment levels are more likely to be influenced by subsequent negative information (Ahluwalia et al., 2000), and individuals with weak preferences are more likely to change their preferences after encountering strong preference inconsistent arguments (Jain, 2003). Moreover, individuals who feel shame or gratitude are more likely to change their initial preferences (Agrawal et al., 2013). While previous research mainly focuses on the moderating role of individuals' attitudes toward brands or products on the preference (in)consistency effect, we attempt to understand the role of individuals' fundamental characteristics on preference (in)consistency, thereby expanding the scope of research. Specifically, we propose a novel factor that impacts individuals' motivation to maintain or change their initial judgments based on their societal beliefs in this research, that is, power distance belief. While previous research demonstrated the differences between Westerners and Easterners in the preference (in)consistency effect (Aggarwal et al., 2013), neither specific cultural variables nor specific mechanisms were empirically tested. We attempt to fill this gap by focusing on the role of power distance belief. Next, we elaborate on how power distance belief increases individuals' preference consistency.

Power distance belief, resistance to change, and preference (in)consistency
People demonstrate different characteristics depending on their endorsement of inequality in society, and this phenomenon has been termed 'power distance belief (PBD)' (Hofstede, 2001;Oyserman, 2006). For example, individuals with high PDB (i.e., people who support inequality in society) have greater self-regulation (Zhang et al., 2010) but are less likely to donate money or time (Winterich & Zhang, 2014) or support user-designed products (Paharia & Swaminathan, 2019) compared to individuals with low PDB (i.e., people who support equality in society). These examples demonstrate how societal beliefs impact various individuals' judgments and behaviors.
In this study, we predict that high PDB individuals have greater resistance to change -the tendency to resist or avoid making changes (Oreg, 2003). Resistance to change is a multidimensional construct that includes routine seeking, emotional reaction, short-term focus, and cognitive rigidity (Oreg, 2003;Oreg et al., 2008). By definition, individuals in high PDB cultures or organizations have a rightful place where they belong to, which increases the stability in their lives (De Mooij & Hofstede, 2011). Moreover, high PDB individuals desire to have more structure in their lives (i.e., need for structure; Lalwani & Forcum, 2016), arrive at a firm answer quickly (i.e., need for closure; Lee et al., 2020), and expect clearer descriptions of behaviors (i.e., desire for predictability; Biggart & Hamilton, 1984;Lee et al., 2020;Lalwani & Forcum, 2016), all of which suggest that high PDB individuals expect more stability rather than unexpected life changes (i.e., greater resistance to change). Indeed, high PDB individuals perceive changes as more threatening because they have fewer experiences with changes due to the rigid societal structure (Erez & Gati, 2004) and they place greater value on conformity and tradition (Spencer-Oatey, 1997; Triandis & Gelfand, 1998). Therefore, executives with higher PDB levels tend to stick with the status quo (Geletkanycz, 1997), thus showing a greater preference for the current situation. Moreover, high PDB countries show a lower adoption level of new technologies (Erumban & De Jong, 2006) and national level innovation score (i.e., global innovation index; Rinne et al., 2012), thereby supporting the positive link between PDB and resistance to change.
Furthermore, individuals with greater resistance to change prefer sticking with familiar things rather than trying new ones (Oreg, 2003). Indeed, conservatives (vs. liberals) who seek stability and tradition (vs. changes) are more likely to be nostalgic about the past (Lammers & Baldwin, 2018) as well as less likely to favor new products (Khan et al., 2013) and adopt changes in the workplaces (Fay & Frese, 2000). Based on this, we further propose that high PDB individuals who devalue change in their lives show a greater preference consistency when they encounter contrasting information that challenges their preexisting judgments. This is because they tend to consider changes as unfavorable, owing to their greater resistance to change, thus leading to a greater willingness to stick to their initial judgments (i.e., showing a greater preference consistency effect). However, the indirect effect of PDB on preference consistency through resistance to change is salient when individuals are provided with inconsistent information. When provided with consistent information, both high and low PDB individuals are less likely to change their initial judgments. Therefore, resistance to change should not play a critical role in this situation. Hence, we propose the following hypotheses: H1: Individuals with high (vs. low) PDB are less likely to change their initial preferences when provided with preference inconsistent (vs. consistent) information.

H2:
The impact of PDB on preference change is driven by different levels of resistance to change.

The moderating effect of positive affect
In general, marketers provide and emphasize the positive aspects of their brands in the early stage. However, consumers are more likely to encounter negative (i.e., inconsistent with their initial positive attitude) information from different sources as brands or products are more accessible to other consumers. To increase the practical implication of the findings of this paper, we explored an important boundary condition whereby marketers could prevent their low PDB consumers from changing their initial (i.e., positive) attitude in response to the subsequent preference inconsistent (i.e., negative) information.
Research has revealed that a positive affective state increases individuals' tendency to prefer the status quo (Yen & Chuang 2008) because they want to maintain their current positive state (Isen & Simmonds 1978). Similarly, we also predict that individuals in a positive affect state are less likely to change their initial judgment even when they encounter inconsistent information subsequently. However, this effect should be stronger among low PDB individuals who tend to change their initial judgment significantly compared to high PDB individuals. Therefore, affect manipulation should be more salient among low PDB individuals. Hence, we expect the interactive effect between the affect condition and PDB. This is stated formally as follows (see Figure 1 for the conceptual framework): H3: Low (not high) PDB individuals are less likely to change their initial preference when receiving subsequent preference inconsistent information when they are in the positive affective state compared to the control condition.

Study 1
Study 1 tests whether high (vs. low) PDB individuals are more reluctant to change their preferences when subsequently provided with preference inconsistent (vs. consistent) information (H1). The study used a 2 (preference information: consistent vs. inconsistent condition; between subjects) x PDB (continuous) design.

Participants and PDB
A total of 105 undergraduate students (49.5% female, M age = 20.76) at a Midwestern university in the United States participated in exchange for partial class credit. PDB was measured using a 5-item, 9-point scale (α = .81) developed by Yoo et al. (2011). Higher scores indicate greater endorsement of inequality. See Table 1 for the items.

Preference change
We followed the overall procedure described by Agrawal et al. (2013; also based on the procedure description by Jain, 2003;Jain & Maheswaran, 2000). The participants were made to believe that they were participating in two separate marketing surveys. In the first survey, they were given information about two new digital cameras (Brands A and B) provided by a technology researcher, stating that Brand A was superior in five (three important and two less important attributes) out of eight product attributes (four important and four less important attributes that were pre-tested in Agrawal et al., 2013). Next, the participants evaluated both Brands A and B using a 4-item, 9-point scale: ('bad' = (1) -'good' = (9); 'undesirable' = (1) -'desirable' = (9); 'useless' = (1) -'useful' = (9); 'unfavorable' = (1) -'favorable' = (9)). Ostensibly, in a second unrelated survey, the participants were given the evaluations of actual users of the digital camera Brands A and B, conducted by a marketing research company. A total of 52 students were assigned to the preference consistent information condition, and 53 students were assigned to the preference inconsistent information condition. In the preference consistent (inconsistent) information condition, the majority of the consumers felt that Brand A (Brand B) was better in most attributes. The participants evaluated both Brands A and B again using the same measures.
As a manipulation check for information consistency, we also measured the extent to which participants felt that the information from the second product survey 'disagreed' = (1) -'agreed' = (9) with their initial preference.

Manipulation check
The participants in the preference consistent (vs. inconsistent) information condition indicated that the second product survey agreed with their earlier preference more (M consistent = 6.15 vs. M inconsistent = 3.85, t(103) = 5.58, p < .001), suggesting that the manipulation was successful.

Preference change
We predicted that participants with high (vs. low) PDB would be less likely to change their initial judgments when given preference inconsistent (vs. consistent) information (H1). The data supported this hypothesis. Following Agrawal et al. (2013), we created a dependent variable by subtracting the second evaluation from the initial evaluation of Brand A and used this as the preference change. A general linear model (GLM) with preference change as the dependent variable and PDB, preference information condition (0 = consistent condition, 1 = inconsistent condition), and their interaction as the independent variables revealed significant main effects for PDB (F(1, 101) = 10.71, p < .005), the preference information condition (F(1, 101) = 39.02, p < .001), and, importantly, a significant interaction between the two (F(1, 101) = 7.47, p < .01).
Furthermore, the results of the Johnson-Neyman technique (i.e., floodlight analysis; Spiller et al., 2013) indicated a significant positive effect for the inconsistent (vs. consistent) information condition on preference change for those whose PDB score was less than 5.52 (b JN = 1.30, SE = .66, p = .05). This indicates that low PDB individuals show a greater tendency to change their preferences when provided inconsistent information compared to when they are provided consistent information. However, for those whose PDB score was higher than 5.52, the effect of the preference information condition on their preference change was not significant, suggesting that preference change was not significantly different when they were provided with preference inconsistent information compared to consistent information. See Figure 2 for details.

Ancillary analyses
In the preference inconsistent information condition, the effect of PDB was significantly negative (β = −.68, t(51) = −3.04, p < .005), suggesting that low (vs. high) PDB individuals were more likely to change their initial judgment when provided with preference inconsistent information. However, in the preference consistent condition, the effect of PDB was not significant (β = −.06, t(50) = −.75, p > .45), suggesting that low and high PDB individuals did not differ in their preference change when they were provided with preference consistent information.

Discussion
Study 1 shows that high (vs. low) PDB individuals are more reluctant to change their initial preferences when provided with preference inconsistent information compared to when given preference consistent information. Thus, H1 was supported. Study 1 demonstrates that there is no effect of PDB on preference change when participants are given consistent information. Thus, we focused on preference inconsistent information conditions in subsequent studies. Furthermore, we examined the underlying mechanism for this effect, testing the mediating role of resistance to change (H2) in the next study.

Study 2
Study 2 tested the mediating role of resistance to change for the effect of PDB on preference change. Specifically, we predicted that individuals with high (vs. low) PDB are more reluctant to change their initial judgments because they are less open to making changes in general (H2).

Method
A total of 102 members of a TurkPrime online panel (57.8% female, M age = 43.12) participated in this study in return for a financial reward, and PDB was measured with a 3-item, 7-point scale (α = .98) used by Zhang et al. (2010). See Table 1 for the items. Furthermore, we measured preference change using the same procedure as in Study 1 and focused on the preference inconsistent condition. The participants' resistance to change was also measured via a 17-item, 6-point scale (α = .92) used by Oreg (2003). Higher scores indicate greater resistance to changes. See Table 1 for the items.

Results
We predicted that the negative relationship between PDB and preference change would be mediated by resistance to change. Supporting this, PDB significantly positively predicted resistance to change (r = .25, p < .025) and negatively predicted preference change (r = −.31, p < .005). Furthermore, resistance to change was significantly negatively associated with preference change (r = −.24, p < .025). Similarly, regression analyses revealed that PDB significantly negatively predicted preference change (β = −.32, SE = .10, t(100) = −3.27, p < .005). Next, a bootstrapping procedure (10,000 iterations; Model 4, Hayes, 2012) suggested that the indirect effect of resistance to change on the link between PDB and preference change was significant (β = −.05, SE = .03, 90% CI [−.1058, −.0012]), indicating that the effect of PDB on preference change was mediated by the resistance to change. These results support our H2.

Discussion
Study 2 replicated the findings of Study 1 and also tested the mediating role of resistance to change on the effect, thus supporting both H1 and H2. However, one possible confounding effect in Studies 1 and 2 is that we used two different information sources. Specifically, the first piece of information was given by a technology researcher and the second one was given by actual consumers. Therefore, it may be possible that high PDB individuals stick with the first information (i.e., they are less likely to change their initial attitude) due to higher perceived expertise from the first information source (Winterich et al., 2018). Therefore, we controlled for this confounding effect in Study 3. Moreover, we tested an important boundary condition that provides practical implications in the next study.

Study 3
We test H3 in Study 3 by employing a 2 (affect condition: positive vs. control; between subjects) x PDB (continuous) design.

Method
A total of 166 members of a Prolific online panel (61.4% female, M age = 37.53) participated in this study in return for a financial reward. The participants were assigned to either a positive (N = 81) or control (N = 85) affective state. The participants in the positive affective state condition were asked to write about an event that made them happy, while those in the control condition were asked to write about their typical day. As a manipulation check, we measured how happy/enjoyable it was to be in the situation they wrote about (on a 7-point scale; r = .95, p < .001). Thereafter, we used the same procedure as in Study 2 to measure preference change. In addition, we measured perceived expertise for two information sources with a 5-item, 7-point scale (for a technology researcher α = .95; for actual consumers α = .91) used by Netemeyer and Bearden (1992). PDB, our independent variable, was also measured with a 3-item, 7-point scale (α = .96) used by Zhang et al. (2010). See Table 1 for the items.

Affect manipulation check
Participants in the positive affective (control) condition scored higher on the manipulation check items (M positive = 6.67 vs. M control = 4.52, t(164) = -12.69, p < .001), thus confirming the effectiveness of the affect manipulation.

Preference change
We conducted a GLM with preference change as the dependent variable and PDB, affect condition (0 = control, 1 = positive), and the interaction between these two as independent variables. We also included perceived expertise for the information sources (one for a technology researcher and the other for actual users) as covariates. The results revealed the significant main effects of PDB (F(1, 160) = 5.53, p < .05) and the affect condition (F(1, 160) = 5.48, p < .05), and, importantly, a marginally significant interaction between the two (F(1, 160) = 3.46, p = .065). Two covariates were not significant in the model (p > .10). To examine this interaction in detail, we conducted a floodlight analysis (Spiller et al., 2013) with the same variables included and found a significant negative effect of affect condition on preference change for participants whose PDB score was less than 1.82 (b JN = −.68, SE = .34, p = .05), indicating that low PDB individuals were less likely to change their initial evaluation after receiving inconsistent information when they were in a positive affective state compared to the control condition. However, this pattern was not found among individuals whose PDB scores were higher than 1.82. Therefore, the data supported H3. See Figure 3 for a visualization of these results.

Ancillary analyses
The effect of PDB was significantly negative (β = −.50, t(83) = −3.28, p < .005) in the control condition, suggesting high (vs. low) PDB individuals were less likely to change their initial preference when receiving inconsistent information. This result replicates the previous studies. However, the effect of PDB was not significant (β = −.08, t(79) = −.37, p > .70) in the positive affective condition.

Discussion
Study 3 reveals one important boundary condition in which low (not high) PDB individuals are less likely to change their initial judgment when they encounter inconsistent information: a positive affective state. This finding provides important insights to marketers who wish to maintain their target customers' positive brand judgment even when their customers are exposed to inconsistent (i.e., negative) information from different sources other than marketers' controlled information sources. Specifically, marketers can heighten their target customers' positive affective state using various environmental stimuli and make them less susceptible to subsequent negative information about their product. In addition, we demonstrated that different levels of perceived expertise toward information sources were unlikely to be responsible for our effects.

General discussion
This study demonstrates that high (vs. low) PDB individuals are more reluctant to change their initial preferences after encountering preference inconsistent (vs. consistent) information (Study 1). This is because high PDB individuals are more reluctant to make changes in their lives, leading to a greater preference consistency effect (Study 2). Furthermore, we demonstrate that a positive affective state moderates this effect (Study 3).

Theoretical and managerial contributions
While previous studies on preference (in)consistency focused on why and how individuals favor their earlier preferences (Ditto & Lopez, 1992;Jain, 2003;Jain & Maheswaran, 2000;Kunda, 1990), this research expands this scope by introducing a novel factor for the preference (in)consistency effect: PDB. This is an important step from a cultural perspective because it is one of the first attempts to examine the role of culture on the preference (in)consistency effect (for an exception, see Aggarwal et al., 2013). As Aggarwal et al. (2013) compared students from a Canadian university (Westerners) with students from a Korean university (Easterners), we were unable to reveal which cultural dimension leads to the effect because Canadians and Koreans differ in various cultural dimensions, such as individualism, power distance, masculinity, and uncertainty avoidance. However, this study focused on a specific cultural component, PDB, and revealed the exact underlying mechanism for this effect: the tendency to avoid changes. Our findings suggest that Aggarwal et al.'s (2013) effect may be attributed to different levels of PDB in the two countries. In fact, Korea (scores 60 on the Power Distance Index; Hofstede, 2015) is high in PDB compared to Canada (scores 39 on the Power Distance Index; Hofstede, 2015). Therefore, our paper reconciles the previous research by identifying the exact driver for the effect. Moreover, based on this finding, we can also predict various other behaviors exhibited by high PDB individuals that are possibly related to the tendency to avoid changes. For example, high PDB individuals may be more reluctant to switch brands or products, even if they are not satisfied with them. Alternatively, they may be less likely to adopt new products compared to low PDB individuals. These and other predictions open up new paths for future PDB research.
This research also provides important insights to marketers in terms of how to effectively allocate their limited resources. With an abundance of available information, consumers often encounter inconsistent information which runs counter to their initial judgments or preferences. Based on this research, marketers should allocate their valuable resources into making positive (vs. negative) information accessible first to their customers at the initial stage, when customers' initial attitude toward their brands or products is formed. This is because consumers are generally less likely to be influenced by subsequent negative information if they have already formed a favorable attitude toward the brands or products. Notably, this strategy is more effective among high PDB customers, who tend to stick to their initial preferences. Conversely, low PDB consumers are more likely to be influenced by subsequent information, which may suggest higher vulnerability to subsequent negative information. Based on this, marketers can develop more effective marketing strategies by understanding their target customers' PDB levels. Furthermore, based on the findings of Study 3, marketers can induce positive affect among their target customers so that they are not influenced by the subsequent negative information; this strategy should be more effective among low PDB individuals.

Limitations and future research
It may be possible that product type can moderate this effect. For example, high PDB individuals have a greater preference for status-related products (Kim & Zhang, 2014), which suggests that high PDB individuals may be more influenced by subsequent inconsistent information if the products are related to status. Future research can examine different moderating effects for the relationship between PDB and the preference (in) consistency effect. Another limitation is that it is possible that the proposed effect is driven by other alternatives (e.g., risk aversion, prevention vs. promotion focus, etc.) that are not tested in this research. Future research should thoroughly test other alternatives to increase confidence in the underlying mechanism of this effect. Finally, we measured participants' preferences in this research. However, if future work can validate this effect by measuring participants' actual choice behaviors, it would have more managerial implications.

Data availability statement
The data that support the findings of this study are available from the corresponding author, HL, upon reasonable request.