A Framing-Effect-Based Analysis of Income Inequality Perceptions

Abstract

Keywords: Behavioral economics; framing effect; perception of income inequality

1. Introduction

Income inequality is a significant problem, affecting almost all countries to varying degrees. Addressing this issue is challenging because, in addition to income inequality itself being negative, perceptions of it vary widely and are perhaps an even greater problem. The perception of income inequality is a highly diverse and intriguing phenomenon. This is because, even though data on income inequality is readily available, its interpretation varies considerably. For example, individuals with high incomes may not react to income inequality in the same way as those with low incomes. In this context, according to a study conducted by Horowitz et al. (2020), 52% of low-income individuals think that income inequality in the US as the most important problem that politicians should solve, while only 36% of high-income individuals share this view. However, 61% of people in the US believe that income inequality is high. What is interesting here is that while 61% of people think income inequality is high, 39% do not. Perceptually, these differences are also likely to make it difficult to resolve income inequality. Our perceptions can indeed change or be distorted easily, and this situation is particularly studied by behavioral economists. According to this school of thought, psychological factors cause various biases in decision-making processes, and these factors lead to perceptual differences that influence our decisions or perceptions. The framing effect is one such bias. According to this effect, the reality we perceive can differ when the same situation is presented in different ways. The first section of this study examines the behavioral economics school, which rise to prominence in the 1980s with Daniel Kahneman and Amos Tversky, and the framing effect, one of the psychological biases that cause bounded rationality in this school. The second section of the study addresses the issue of income inequality. The third section of the study presents a literature review on the perception of income inequality, and the final section presents an empirical study on this perception.

Effect

Traditional economics' “homoeconomicus” assumption posits that individuals are rational in all circumstances and make choices that maximize their utility. In contrast, behavioral economics assumes that individuals may not always behave rationally, even if they have the capacity to do so. This assumption of behavioral economics is called “bounded rationality” (Mullainathan & Thaler, 2000). When behavioral economics was not yet popular, its relationship with traditional economics was not very good. This was because traditional economic actors accepted economics as a mathematical field and had adopted a dismissive view of behavioral economics for a long time (Rehman, 2016). Keynes's 1936 book began to change the relationship between traditional economics and behavioral economics. Traditional economics actors sought to develop economics by examining Keynes's views within a mathematical framework, while behavioral economics actors sought to address the shortcomings of the behavioral economics school and develop it further. Towards the end of the 1970s, two important figures in behavioral economics, Daniel Kahneman and Amos Tversky, published Prospect Theory, ensuring that behavioral economics gained more ground in economics. The theory essentially determined that individuals' preferences change depending on whether they are in a gain or loss situation, with individuals generally being riskaverse in gain situations and generally taking risks in loss situations (Akerlof, 2002). In contrast to traditional economics’ rigid assumptions about preferences, behavioral economics offers a more flexible perspective and provides some critical examples challenging the assumption of rational human behavior. For example, consider a fair coin with equal and equal probabilities of landing heads or tails. Suppose this coin is tossed 50 times and lands heads 50 times. Some people may see a higher probability of heads coming up on the 51st toss because it came up tails 50 times, or they may think that since it came up tails every time, it will now come up heads

Yıldız Social Science Review, Vol. 11, No. 2, pp. 106–115, 2025

next time. However, no matter how many times the coin is tossed, the probability of heads and tails coming up each time is equal and independent of previous events. Rational individuals assume that the probabilities are equal, but many people cannot behave rationally due to previous events and may make flawed choices (The University of Chicago, 2024). The basic logic of bounded rationality, one of the fundamental assumptions of behavioral economics, is loss aversion. According to loss aversion, individuals generally prefer guaranteed choices in positive situations and take risks in negative scenarios. The worse the negative scenario, the greater the degree of risk-taking. Furthermore, individuals react more strongly to loss situations than to gain situations. In this regard, Kahneman and Tversky (1979) conducted a simple experiment in their studies to explain how rational behavior can easily be disrupted. In the experiment, participants were asked to make choices in the following scenario:

1. Which one would you prefer?

a. Definitely earning $240 (84%) b. A 25% chance of earning $1,000 and a 75% chance of losing nothing (16%)

2. Which one would you prefer?

c. Definitely lose $750 (13%) d. Lose $1,000 with a 75% probability and lose nothing with a 25% probability (87%) The first question is a typical gain scenario, and participants tended to choose option a, the more guaranteed choice. The second question is a typical loss scenario, and a large proportion of participants chose option d, i.e., taking the risk. Participants were then asked to make a choice in the following scenario:

3. Which one would you prefer?

e. A 25% chance of gaining $240 or a 75% chance of losing $760 (0%) f. A 25% chance of gaining $250 or a 75% chance of losing $750 (100%) In this question, all participants chose option f. It can be said that everyone acted rationally only in terms of this question. However, when all three questions are considered, it can be seen that the vast majority of participants did not act rationally. This is because option e in the third question was created by combining options a in the first question and d in the second question, while option f in the third question was created by combining options b in the first question and c in the second question. Therefore, in order to behave rationally, participants who strongly preferred options a and d in the first and second questions should have chosen option e in the third question. However, no participant chose this option and instead chose the other option, failing to behave rationally. This example is a simple illustration of how rational behavior is more difficult than assumed and how easily rationality can be disrupted. The main focus of behavioral economics on behaviors

that undermine rationality is psychology, and these behaviors are referred to in the literature as “bias.” Some of these effects are loss aversion, mental accounting, confirmation bias, sunk cost fallacy, availability effect, context effect, and framing effect. In addition to these effects, other biases that have emerged in recent years include the fundamental ratio fallacy, action bias, majority effect, Dunning-Kruger effect, and IKEA effect. Among these effects, the framing effect is particularly popular and significant (Azzopardi, 2021). According to the framing effect, decision-making can vary depending on how a situation is presented. For example, let's assume that a person named John goes to a pharmacy and wants to buy an antibacterial cleaning wipe. There are two brands of the product John wants to buy, and the price and number of wipes in these brands are the same. The only difference between the two brands is the information on the front about the percentage of germs. These two brands are “Bleachox,” which claims to kill 95% of germs, and “Bleach-it,” which claims that only 5% of germs survive. When John compares these two brands of wipes, he is not very interested in buying the “Bleach-it” brand, which indicates the probability of germs surviving. As a result, John purchased the “Bleachox” brand wipe. While everything seemed normal here, there was actually a problem. This is because both brands were equally effective in combating germs. The Bleachox brand killed 95% of germs while allowing 5% to survive. Similarly, the Bleach-in brand killed 95% of microbes while allowing 5% to survive. Therefore, both brands were saying the same thing. However, John fell victim to the framing effect and interpreted the information labels on the brands as if they were different, thinking he had made a better choice (The Decision Lab, 2024). The most famous experiment conducted on the framing effect is the one Kahneman and Tversky (1981) called the Asian Experiment. The experiment was conducted on a total of 307 students (first group 152, second group 155) at the University of British Columbia and Stanford University. In the experiment, it was expected that a virus that had emerged in Asia would also affect people in the United States and cause the deaths of approximately 600 people. It was not entirely possible to prevent the spread of this virus, so some of the participants were asked to choose one of the following treatment methods: • If Treatment A is applied, 200 people will definitely be saved. (72%) • If Treatment B is applied, there is a 1/3 chance that everyone will be saved and a 2/3 chance that no one will be saved. (28%) The choice scenario presented to the first group of participants was framed positively. This is evident from the use of the term “saved.” In positive scenarios, individuals are likely to gravitate toward guaranteed choices, so 72% of participants in this scenario chose Treatment A, which

Yıldız Social Science Review, Vol. 11, No. 2, pp. 106–115, 2025

was a more guaranteed choice than the other option. The second group of participants was also asked to choose one of the following treatment methods: • If Treatment C is applied, 400 people will definitely lose their lives. (22%) • If Treatment D is applied, there is a 1/3 chance that no one will lose their life, while there is a 2/3 chance that everyone will lose their life. (78%) The choice scenario presented to the second group of participants was framed negatively. This is evident from the phrase “lose their lives.” Since individuals are likely to make risky choices in negative scenarios, 78% of the participants in this scenario chose Treatment D, which was a riskier choice than the other option. The framing effect is an effect that indicates that decision-making can change when the same situation is presented in different ways. In the Asian Experiment, Kahneman and Tversky also presented the same situation to participants in different ways. So, if the virus is expected to kill 600 people, in the scenario presented to the first group of participants, if you apply Treatment A, 200 people will be saved, but at the same time, 400 people will also lose their lives (because the virus is expected to affect 600 people for sure). Similarly, if Treatment C is applied, 400 people will definitely lose their lives, but 200 people will also be saved (because the virus is expected to affect 600 people for certain). In other words, Treatment A and Treatment D describe the same situation. In this context, it can be said that Treatment B and Treatment C also mean the same thing. This is because the 1/3 probability of everyone being saved expressed in Treatment B also means the 1/3 probability of no one losing their lives expressed in Treatment C. Similarly, a 2/3 probability of no one being saved is equivalent to a 2/3 probability of everyone losing their lives. As a result, participants fell prey to the framing effect and made irrational choices.

tries and within income groups. In particular, while people in the upper income group became wealthier, those in the lower and middle income groups were unable to increase their incomes to the desired level. There are many methods for measuring income inequality. These methods include the Gini coefficient method, the Atkinson index method, the coefficient of variation method, the decile ratio method, the generalized entropy index method, the Kakwani index method, the total income ratio method, the Robin Hood index method, and the Sen poverty index method. The most popular of these methods is the Gini coefficient method. This coefficient takes values between 0 and 1, and it can be understood that as the coefficient approaches 0, income inequality decreases, and as it approaches 1, income inequality increases (De Maio, 2007). For example, in 2023, the countries with the lowest Gini coefficients were Slovakia (0.24), Slovenia (0.24), and Belarus (0.24), while the countries with the highest Gini coefficients were Colombia (0.55), Brazil (0.52) and Zambia (0.51) (Our World in Data, 2024). The Gini coefficient is actually an extension of the Lorenz curve, and details about the Lorenz curve can be found in the Figure 1 below. The Lorenz curve represents the share of income received by a specific percentage of the population in a country. In a situation of perfect equality (where the Gini coefficient equals zero), the Lorenz curve has a 45-degree slope, meaning that, for example, each 20% segment of the population receives 20% of the income. As inequality increases (the Gini coefficient rises), the Lorenz curve begins to shift toward the lower right. In this case, part of society begins to receive a smaller share of income; for example, the 20% segment receives 5%. In this case, inequality increases (De Maio, 2007). These examples are supported by statistical data, and these details can be found below. Income is distributed so unevenly that this situation is more clearly demonstrated in the World Inequality Report published in 2022. For example, 50% of the world's adult

3. The Problem Of Income Inequality

Income inequality refers to the unfair distribution of income among the population (OECD, 2024). Looking at the historical development of income inequality, between 1945 and 1970 (after World War II), countries grew economically and began to increase their incomes. During this period, the income gap between the lower and middle income classes and the upper income class, while significant, did not change much. Therefore, income differences between countries did not attract much attention. However, starting in the 1970s, countries' economic growth performance weakened, and various unemployment problems began to emerge (Sherman, Trisi, & Cureton, 2024). So much so that while the world economy's growth rate was close to 5% at the end of the 1980s, it declined to around 1.5% at the beginning of the 1990s (World Bank, 2024). This led to a marked increase in income disparities both between coun-

Figure 2. Distribution of the world's adult population by income groups (left) and the shares of income and wealth received by these income groups (right) (2021). population (2.5 billion people) is in the low-income group, 40% (2 billion people) is in the middle-income group, and 10% (517 million people) is in the high-income group. Of those in the upper income group, 10% (51 million people) are in the highest income group. The real problem here is the share of income and wealth that these income groups receive (Fig. 2). For example, while the 2.5 billion people in the low-income group received 8.5% of income in 2021, the 51 million people in the top-income group received 19% of income in 2021. On average, a person in the lower income group earned €2,800 in 2021, while a person in the top income group earned an average of €321,600 in 2021. The difference is approximately 115 times. The imbalance in income is even worse in terms of wealth. At this point, it is known that in 2021, the lower income group received only 2% of the world's wealth, while the top income group received 38% of the world's wealth. On average, a person in the lower income group received €2,900 of the wealth in 2021, while a person in the top income group received €2,755,200. The difference is 950 times, which is a much worse situation in terms of income (Chancel, Piketty, Saez, & Zucman, 2022).

4. Literature Review

Some studies in the literature on the perception of income inequality have reported that people may have different perceptions of income inequality. For example, according to Knell & Stix (2020), individuals reach conclusions about income inequality not based on general data, but only by looking at reference groups. However, when it comes to income inequality, low-income (high-income) individuals tend to perceive themselves as worse off (better off) than they actually are. This result was also obtained in the study by Faggian et al. (2023), which also revealed that people living in the same region have similar perceptions of income inequality. Among studies examining perceptions of income inequality specifically in the US, Xu & Garand (2020) found that, regarding income inequality in the US over the past 20 years, people living in states with higher income in-

Yıldız Social Science Review, Vol. 11, No. 2, pp. 106–115, 2025

equality perceived a greater increase than people living in states with lower income inequality. Chambers et al. (2014) found that people living in the US perceive income inequality as a bigger problem than it actually is and underestimate their own average income. Furthermore, liberal politicians exaggerate income inequality more than conservative politicians. Another study examining perceptions of income inequality in the context of electoral processes, Engelhardt & Wagener (2014), found that perceptions of income inequality are more distorted than normal during political election cycles and that people vote based on perceptions rather than data. According to Hauser & Norton (2017), who examined the variability of perceptions of income inequality based on reference groups and social status, people tend to underestimate income inequality in their own country and make incorrect inferences about income inequality based on information obtained from their immediate environment. (2019) found that people's perceptions of their status within society also influence their conclusions about income inequality. In terms of this study, apart from the studies mentioned in the literature, the study by Amiel & Cowell (1992) stands out on the issue of income inequality. In this study, an experiment was conducted on the perception of income inequality, and it was found that people have variable perceptions. In the experimental procedure of the study, participants were asked numerical and verbal questions about income inequality. The numerical and verbal questions asked had the same meaning. Therefore, participants were expected to answer the questions with the same meaning in a similar way. However, as the complexity level of the questions increased, the answers given to the numerical and verbal questions differed significantly. In subsequent years, two similar studies were conducted: Ballano & Ruiz-Castillo in 1993 and Jancewicz in 2016. These two studies yielded results similar to those obtained in the study by Amiel & Cowell.

5. Materials And Methods

This study was conducted to reveal perceptions of income inequality. In this context, the study first conducted by Amiel & Cowell in 1992 was used as a basis. In that study, a survey was conducted to determine perceptions of income inequality, and it was found that participants' perceptions of income inequality varied under different scenarios. This study differs from other studies in several ways. For example, in other studies, participants were labeled as supporters of certain axioms found in the literature on income inequality after giving their answers, but this study does not include such details. Furthermore, while in other studies the questions “Doubling Income,” “Adding a Fixed Amount,” and “Reducing a Fixed Amount” were all included in the “Income Transformation Scenario,” in this study only the question “Doubling Income” is included in the “Income Transformation Scenario.” and the “Fixed Amount Addi-

Yıldız Social Science Review, Vol. 11, No. 2, pp. 106–115, 2025

tion” question is presented as a separate scenario. The “Fixed Amount Reduction” question is not included in this study in order to reduce the number of similar questions. In addition, the options presented to participants in the verbal questions in Amiel & Cowell's “Income Transformation Scenario,” “Transfer Principle Scenario,” and “Population Replication Scenario” have been simplified in this study and presented as “Decreases,” “Increases,” and “Remains Unchanged.” Again, unlike this study, Amiel & Cowell's study asked participants if they wanted to change their preferences immediately after each verbal question, but this option is not offered in this study. While Amiel & Cowell and Jancewicz's studies depict individuals' incomes as 5, 8, and 10, this study changes individuals' incomes to 2, 5, and 8. In the question posed in the “Fixed Amount Addition Scenario,” 6 units are added to individuals' incomes in this study, whereas 5 units are added in other studies, and this amount is the same as in Ballano & Ruiz-Castillo's study. One important point is that the “Unequal Enrichment Scenario” in Amiel & Cowell's study, as Jancewicz also mentions, is a scenario that is quite difficult for participants to understand and answer. Therefore, this scenario will not be included in the analysis section of this study. Finally, the main objective of this study is to explain the perception of income inequality using the link between behavioral economics and the framing effect. Other studies also present a similar analysis, but they do not include the link between behavioral economics and the framing effect. The study employed an online survey method, and 101 academics working at a foundation university were selected as the experimental group. In the survey application procedure, participants were first asked 9 numerical questions, followed by 6 verbal questions. Some of these numerical questions and each of the verbal questions have the same meaning, and participants will be tested to see if they gave the same answer to these questions in the context of the framing effect. For this purpose, frequency tests were performed using the SPSS 29.0.2.0 program to analyze the data in the study. Although this study has certain limitations, participants who wish to participate in the survey are required to share certain demographic information (such as age, gender, or income status). Those who do not wish to share this information cannot participate in the survey. However, compared to other studies, the number of participants is lower. The following scenarios are listed in order: “Income Transformation Scenario,” “Fixed Amount Addition Scenario,” “Transfer Principle Scenario,” “Population Replication Scenario,” and “Decomposability Scenario.” Since a

different method was followed from other studies in terms of scenarios, a comparative analysis with other studies was performed for the three scenarios other than the “Income Transformation Scenario” and the “Fixed Amount Addition Scenario.” Since the main objective of this study is to measure the perception of income inequality through the framing effect, it is irrelevant whether the questions in all scenarios are answered correctly. What matters is whether the participants give the same answer to the numerical and verbal questions (two numerical questions in the case of the Decomposability Scenario) within the relevant scenario. As mentioned earlier in this study, the framing effect is an effect where the same situation is presented differently, and preferences may change accordingly. Therefore, when examining the scenarios, it is important whether participants are influenced by the framing effect rather than whether they give the correct answer.

6. Result And Discussion

This study is similar to those conducted by Amiel & Cowell in 1992, followed by Ballano & Ruiz-Castillo in 1993 and Jancewicz in 2016. The most fundamental difference between this study and others is that it examines the perception of income inequality through behavioral economics and the framing effect. The framing effect is an effect whereby decisions can differ depending on how the same situation is presented. This study also tested the perception of income inequality by utilizing this effect. To this end, in the first stage of the survey, designed as an experiment, participants were asked numerical questions related to income inequality, while in the second stage, they were asked verbal questions with the same meaning as the numerical questions. The basic logic is this: The answers given to numerical questions should match the answers given to verbal questions. If this does not happen, we can say that participants have been influenced by the framing effect and their perception of income inequality has been distorted. The five basic scenarios in Amiel & Cowell's study (Income Transformation, Transfer Principle, Population Replication, Decomposability, Unequal Enrichment) were presented in this study in a slightly different form. The reason for this is the considerable complexity of the Unequal Enrichment scenario in particular and the similarity of question types within the Income Transformation scenario. The complexity of the Unequal Enrichment scenario is also mentioned in Jancewicz's study.

7. Conclusion

In terms of the results of this study, it was found that as the complexity of the questions increased, participants were more susceptible to the framing effect and their perception of income inequality was distorted. This result is broadly consistent with other studies. It was found that in simpler scenarios (e.g., comparing two numerical questions in the decomposability scenario), participants' perception levels were better and they were less susceptible to the framing effect. One of the biggest problems in the perception of income inequality is how income inequality is communicated to people. It is important that data and facts that can be perceived as complex are communicated in the simplest and most under-

Ethics: There are no ethical issues with the publication of this manuscript. Peer-review: Externally peer-reviewed. Conflict of Interest: The authors declare that there is no conflict of interest. Authorship Contributions: The authors contributed to the study equally. Financial Disclosure: The authors declared that this study has received no financial support.

Share and Cite