For new authors:
free, easy and fast
For registered authors
Publish your own papers with us - it's easy!Read more
Academic Paper, 2013, 52 Pages
II List of Graphs
1 Overconfidence: The Harmful Optimism
1.1 Background and Motivation
1.2 Definition: Overconfidence
1.2.1 Unrealistic Optimism
1.2.2 Better-Than-Average Effect
1.2.3 Illusion of Control
1.2.4 Illusion of Knowledge
1.2.5 Self-serving bias
3 The German Education System
3.1 Federal Differences
3.2 Influences of the Bologna Process
4 Review of Literature
4.1 Confounders of Overconfidence
4.1.3 Mental Health
5 Empirical Research
5.2 Process of Primary Data Collection
This study was not simply conducted to obtain an academic degree. It was motivated by the same curiosity that makes children wonder why the earth is moving or why some stars shine brighter than others. I had some unanswered questions and the only way to solve this was to create a questionnaire and conduct field research. However, just like in many other situations in life, I could not have done it on my own. Fortunately, I have met many constructive and wise people on my way who have been willing to help me out.
The names of those who influenced this book the most are unfamiliar to me, yet they deserve special recognition. I would like to thank the hundreds of students, who participated in this study. All of this would not have been possible without their contributions. The individual who supported me the most throughout the empirical research and the writing process is Jennifer Pfister. I thank her for pre-testing the questionnaire and listening to my findings so uncomplainingly. I feel privileged to have had many enthusiastic and kind teachers who did not only share their knowledge with me but also dedicated their precious time to collect data for this book. Among those I am particularly grateful to Edgar Kempter, who is an inspiring pedagogue and a true supporter, to Florian Parzefall, a classicist in German literature, who awakened my interest in writing, to Adrian Pickel, who has the unique ability to spread happiness and confidence to those who listen to him, to Sabine Ruetz, whose articulacy and in-depth knowledge in journalism constantly animated me, and to Stefan Neuhauser for reminding me of the importance of history. Special thanks go to my dear cousin Tobias Henniger, who had the resilience to help me with this research while also taking care of his young family and his students. I am indebted to my good friend Martin Deininger for his erudition and his efforts to gather data from his degree program. I would like to thank Joanie Andelin for proofreading this book even though she had little time and for her kindness to forgive me my mistakes. Finally, I am deeply grateful to Claudia Bauer and Manfred Nacke for their contributions to this study. They knew me so little and helped me so much – this is true magnanimousness.
Figure 1: Five Factors of Overconfidence
Figure 2: German Education System
Figure 3: Bavarian Education System
Figure 4: Development of Erasmus Students per Year
illustration not visible in this excerpt
Degree program: International Business and Management FT-10
Overconfidence – A Matter of Education?
Student: Dominik Piehlmaier
Disclaimer: In the interest of a reader-friendly flow of text, gender-neutral terminology has been used wherever possible. If a non-neutral form appears, it automatically refers to both genders.
This work examines the relationship between education and excessive confidence in situations of uncertainty. For this purpose, a questionnaire with 10 pseudo general knowledge questions was designed, whereby their degree of difficulty exceeds the knowledge of an average student by far. It was investigated whether subjects (N = 535) would acknowledge this condition and its associated nescience. If that is the case, they will answer the 10 questions within an extremely wide confidence interval in order to meet the predefined 90% accuracy requirement. The focus of investigation was in Southern Germany, as the school system regularly receives top marks in national educational rankings. The data analysis resulted in the stochastic proof that there are significant differences between the various educational institutions in accuracy and overconfidence.
In addition to the empirical study the paper defines the distortion of judgment and identifies its relevant factors. It gives a detailed explanation of the German education system and states the criticism of the concept of overconfidence. The paper concludes with a recommendation for action and ventures a look ahead.
Kufstein, July 10, 2012
Studiengang: Internationale Wirtschaft und Management VZ-10
Ist Vertrauens-Hypertrophie bildungsabhängig?
Student: Dominik Piehlmaier
Hinweis: Die in dieser Arbeit gewählte männliche Form bezieht sich gleichermaßen auf weibliche Personen. Eine Doppelbezeichnung wurde ausschließlich aufgrund einfacherer Lesbarkeit nicht gewählt.
Diese Arbeit untersucht den Zusammenhang zwischen Bildung und übermäßigem Selbstvertrauen in Situationen von Ungewissheit. Dazu wurde ein Fragebogen mit 10 Quasi-Allgemeinwissensfragen entwickelt, wobei deren Schwierigkeitsgrad das Wissen eines durchschnittlichen Schülers bei Weitem übersteigt. Es wurde untersucht, ob die Probanden (N=535) diesen Zustand und ihre damit verbundene Unwissenheit anerkennen und zur Beantwortung der 10 Fragen, innerhalb eines vorgegebenen 90% Konfidenzintervalls, jenen Vertrauensbereich möglichst weit wählen, um die Anforderungen zu erfüllen. Der Fokus der Untersuchung lag in Süddeutschland, da das dortige Schulsystem regelmäßig Bestnoten im nationalen Bildungsvergleich erhält. Die Datenauswertung resultiert in der Erbringung des stochastischen Beweises, dass es zwischen den Bildungseinrichtungen einen signifikanten Unterschied beim Schätzverhalten und der damit verbundenen Vertrauens-Hypertrophie gibt.
Neben der empirischen Untersuchung definiert die Arbeit den Begriff des übermäßigen Selbstvertrauens und benennt die relevanten Einflussfaktoren. Das deutsche Bildungssystem wird dabei ebenso beleuchtet, wie die Kritik am Konzept der Heuristik. Die Abhandlung endet mit einer Handlungsempfehlung und einem Ausblick in die Zukunft.
Kufstein, 10. Juli 2012
The ability to expect a rosy future and solidly trust in our skills empower us to create monuments of human capability and push mankind toward unexpected technological findings. Marie Curie and her husband Pierre found the fission products radium and polonium and developed the first methods to isolate them in order to use both for further research (Nobel Lectures, 1967). The Chinese Ming Empire finished the Great Wall of China, the biggest man-made structure ever built, in 1620 (UNESCO World Heritage Centre, 2011). Every year millions of people start their own business all around the world; others get married and start a family. Every person who makes such a decision does it because of their faith in their own actions. It is, however, uncertain whether Marie Curie would have continued her intensive research with radium if she had known that radiation causes severe illnesses. It can be seen as certain that the Ming Emperors would not have reconstructed the Great Wall if they had been informed that it would not protect them from being replaced by the Manchurian Qing Dynasty.
This is not only true for historical events, but also for modern decision making. If US Americans decide to get married, most do not take into account that they could be the one out of three couples that divorce (U.S. Census Bureau, 2012). In 1993, when the U.S. divorce rate was close to 50%, a study tested whether people are aware of these numbers before they tied the knot. Although the subjects knew the facts, not one of them thought that this would happen to their relationship (Baker and Emery, 1993). This was unexplainably optimistic and on average not true for almost every second couple. Startup owners tend to share the same view of their business ideas. In a well-known survey, entrepreneurs rated their expected business success. 81% of the 2,994 participating founders thought that their chances of success were 70% or higher. Every third person stated that his chances were not less than 100% (Cooper et al., 1988). These expectations seem unrealistically high compared to research showing a different picture. 34% of all U.S. firms do not get through the first year, and half of the companies go out of business before the end of the second year. 60% do not survive the third year (Wiklund, 2006). A HBS article paints an even worse scenario: 30 to 40% of investors lose most or all of their assets; 70 to 80% of all startups do not achieve the expected return on investment and 90 to 95 of them fall short of meeting their declared goal. In other words “failure is the norm” (Nobel, 2011).
This might be one of the best examples of the discrepancy between unrealistic optimism and reality. The too rosy estimation of their own abilities and the business environment foils many startup founders. At the same time, some companies survive and contribute to society for exactly the same reasons. In that sense, overconfidence is often but not always harmful. There are other examples of this psychological phenomenon showing its destructive side. Dozens of nuclear power plants (NPPs) were built in earthquake risk areas especially in Japan, India and the State of California. In the aftermath of the 2004 Tsunami which destroyed huge coastal areas in Indonesia, Sri Lanka, India and Thailand, the Japanese Nuclear and Industrial Safety Agency (NISA) launched a study to assess the risk of a tsunami for Japanese NPPs. There was a special interest in those plants which face the Pacific without any protective islands off the coast. Two years later the results suggested that “'there is a possibility that power equipment could lose functions if a 14-meter-high tsunami hits the Fukushima plant, with seawater flowing inside the (reactor) turbine buildings” (Kyodo News, 2012). NISA officials saw no need of action “because a total power failure was not seen as an imminent threat” (Sunaoshi et al., 2012). On March 11, 2011 Japan was hit by a magnitude 9 earthquake, which was followed by several tsunamis with a maximum height of 38.9 meters (IAEA Expert Mission, 2011). 20,896 lives were lost (USGS, 2012). The combined catastrophe led to a long-lasting, complete power loss at the Fukushima Daiichi nuclear station which resulted in a meltdown at three of six reactor units as well as a hydrogen explosion at reactor nos. 1, 3 and 4 (Masaya, 2012). This was the second INES 7 accident in history.
The Fukushima meltdown is one of many disastrous events that can be attributed to unexplainably optimistic forecasts. In April 2010, the petrol company BP had no effective method to stop the oil leak caused by the explosion of their offshore drilling platform Deepwater Horizon in the Gulf of Mexico. There was no worst-case scenario plan because it was considered costly and unlikely to happen (Steinberg, 2011). Similar behavior can be seen on a number of other occasions such as the Bhopal disaster in 1984. The public seems to be aware of the human misjudgment but fails to prevent it. One reason might be that, although scientists have contributed a lot to understanding the nature of overconfidence, there are still some “blank spots on the map”. The interdisciplinary field of behavioral economics is younger than the established neo-classical approach and requires further long-term investigation to understand all aspects and conditions of irrationality in economical behavior. This research examines whether there is a statistically provable connection between the level of overconfidence and the educational achievement in a certain school system. Due to the given framework of this paper, it was not possible to extend the research to several international systems. Nevertheless, the outcome should help prevent the harmful aspects of overconfidence. It aims to support the development of a suitable strategy against future misjudgments.
Before continuing with an explanation of the research methodology, a detailed definition of overconfidence is needed. Due to the importance of this word, the terminology discussion is part of the first chapter.
1.2 Definition: Overconfidence
There are several definitions of overconfidence in the standard literature. They all follow the basic idea of irrational behavior and its roots in psychology. With even the first research in this field, scholars have been able to prove irrationality in decision making. In an experiment the test subjects were given two scenarios (A and B) each with two choices:
(A) Would you prefer $ 100 today or $ 110 tomorrow?
(B) Would you prefer $ 100 thirty days from now or $ 110 thirty-one days from now?
Many subjects gave different answers for A and B (Diamond and Vartiainen, 2007). This shows an inconsistent and irrational decision-making process and conflicts with the neo-classical view and the idea of rational decision makers, who should always behave in their best interest given all relevant information and their own preferences. In that experiment, all subjects should have chosen the same answer for A and B. A completely rational person would have calculated the present value (PV) for both scenarios at the exact same discount rate.
As mentioned before, overconfidence is one factor that can explain such irrational behavior. The following terminology is designed to establish a suitable hypothesis as well as to support the empirical research. The phenomenon is commonly referred to as the overestimation of one’s own abilities, knowledge, and future prospects, but it does not occur as one single bias (Barber and Odean, 2001). Numerous psychological mechanisms lead to overconfidence. They will be defined separately and will be discussed in a final overview.
Too rosy expectations are closely linked to other relevant biases in this chapter. The effect can be described as an overestimation of the likelihood of a desirable event to happen or the prediction of a future event to be more positive than it will be (Müller, 2007). Unrealistic optimism can be both a cause and a consequence of overconfidence. If an investor is too optimistic about the future price increase of a certain stock, he will not be able to consider that a decrease in value is equally possible. His bright expectation convinces him that his actions will lead to further gains. In that case, unrealistic optimism serves as a heuristic and causes overconfidence. If, however, a couple is absolutely sure about their ability to maintain a stable marriage, their overconfidence and unrealistic optimism will lead them to disregard the chance of being one out of three couples that divorce.
The BTAE is a major component of overconfidence. In many cases, people use their own characteristics as benchmark for others. This is either caused by a lack of alternatives or by an undefined peer group. When subjects have to rank their performance, they frequently state it is above average compared to their peers (Guenther, 2009). In a well-known study Svenson could prove the BTAE by asking subjects to assess their driving skills. 77% of the participating Swedish drivers felt that they drive safer than average (Benoit and Dubra, 2009). Although this would be desirable, it is impossible. The majority cannot be above average. This bias is closely connected to the representativeness heuristic. People have a systematic misconception about probabilities and distribution (Kahneman et al., 1982).
The flattering self-evaluation leads to overconfidence, because decisions are made under the impression of an unrealistic assessment of one’s own capabilities. A startup owner who thinks that his management skills are above average could ignore important business actions of his competitors which might have a negative impact on his own company.
The concept of the IOC describes the human belief of being able to have an influence on random future events. When IOC is present, people are more likely to take risks. In an often-cited experiment subjects “participated in a lottery where they had choice or no choice of a familiar or unfamiliar lottery ticket”. Then they had an opportunity to exchange their ticket with a ticket from a lottery with better odds, namely, a lottery with a smaller amount of tickets to be drawn from (Harvey et al., 1978). Clearly all subjects should have taken advantage of this opportunity to increase their chance of winning, but it turned out that precisely those with a chosen or familiar ticket rejected the offer. IOC made them believe they would be better off with their first choice.
Furthermore, IOC is present if the outcome not only depends on luck but also involves a certain skill level (Grömminger, 2011). Poker and roulette are examples of skill and luck. It has been shown that poker players systematically underestimate the role of luck in their game (Shead et al., 2008). Extrinsic incentives such as money or other rewards fuel the IOC. In their absence, only subjects with a high desire of control acted according to previous tests with money incentives (Burger and Schnerring, 1982). The illusion of having control over uncertain events leads to overconfident decisions. While a rational thought process would include the possibility of a negative or positive outcome, an IOC-biased decision making process underestimates the chance of an undesired result or even completely ignores it. A good example is a private investor who mistakenly believes that his own selecting of stocks will automatically lead to higher future profit than stocks selected by anybody else. In this case, the IOC is responsible for his overconfidence.
There is a common belief about the connection between the amount of information available and the quality of decisions made. The more input a person gets, the better his judgment will be (Schwartz, 2005). Scientists found that in many cases exactly the reverse is true. In an experiment subjects were asked to predict the results of randomly chosen National Basketball Association (NBA) games of one season. They received statistics about the teams’ records and halftime scores. In addition, half of all participants were told the team names (eg. Dallas Mavericks vs. Chicago Bulls) while the others did not know these names (Team A vs. Team B). Basketball fans insist that knowledge of team names will increase accuracy of their prediction because they are able to connect the information with other significant facts about each team (e.g. injuries). The study showed that the opposite happened. Participants with the additional cues relied less on statistical odds. They bet on more familiar teams and earned on average less than those without the team names (Hall et al., 2007). This phenomenon is called illusion of knowledge.
The core message is that additional information does not ultimately lead to higher accuracy in decision making. Even though our cognitive system needs a certain amount of knowledge to make a decision, more knowledge goes along with a higher confidence level but not better judgment. This is called the “less-is-more effect”. Similar findings were made when the available knowledge exceeded the required amount of information (Iyengar and Lepper, 2000). This can be referred to as the “more-is-less effect” (Hall et al., 2007). This is the reason why experts are affected by the overconfidence bias. Their specific knowledge increases their confidence level but not their accuracy.
In several experiments aiming to test the illusion of knowledge, participants showed clear signs of overconfidence in their decision-making progress. They were more likely to take risks and ignore significant statistical cues. A doctor who is convinced that his in-depth medical knowledge will automatically help him to find an appropriate cure for a certain illness might oversee important aspects which occur less frequently. In such cases, the illusion of knowledge is a major driving force for overconfidence.
The SSB is a special cognitive process that serves as an internal protection mechanism of a person’s self-esteem. The validity of harmful influences resulting from negative feedback are changed or rejected. The bias puts the focus on strengths and achievements and suppresses failures and repercussions. It “internalizes success and externalizes failure” (Sherrill, 2007). In laboratory tests participants had to complete certain tasks either alone, in pairs or groups. Some individuals, pairs or groups received positive feedback after they had finished their tasks, while the work of others was assessed as a failure. In the end, subjects had to evaluate the outcome of their efforts. Those told that they had failed attributed performance to external factors such as misfortune, degree of difficulty or the interference of others. Participants who received positive feedback stated that the outcome was mainly caused by internal factors such as ability or intelligence (Campbell and Sedikides, 1999).
Researchers connect the SSB with unrealistic optimism. People usually expect positive future events. They think they will “pass tests, get good jobs, and have long-lasting relationships rather than fail, get fired, or divorce” (Sherrill, 2007). The frequent illusion of success causes overconfidence. They do not expect a negative outcome because, in their imagination, it has hardly ever happened before. This qualifies the SSB as the most important reason for overconfidence. An employee who is frequently late to work, often sick and works inaccurately will not blame himself for his non-performance because his cognitive self-serving process suggests that all that happened is due to external factors.
As stated at the beginning of 1.2, overconfidence is described as the combination of overestimation of one’s own abilities, knowledge, and future prospects as well as underestimation or ignorance of possible failure. The heuristic is influenced or intensified by several cognitive or perceptual processes – the five most important are explained above. The effects of overconfidence on behavior and decision making are varied, and a decrease in risk aversion and wrong calibration are just two examples.
Table 1 summarizes the five most important processes that are closely connected with overconfidence.
illustration not visible in this excerpt
Figure 1: Five Factors of Overconfidence
(Complied by the author)
At the end of this chapter the structure of the research paper is described. The following section, chapter 2, describes the workflow of the centerpiece of this study. A clear overview of how the data was collected and assessed is given. A detailed explanation of the school system where the tests took place follows. Chapter 3 describes the test environment and its limitation and covers the differences in the German education system in various states and the influence of the Bologna Process. Chapter 4 is a review of the literature characterizing overconfidence and explaining criticism of the concept. The fifth chapter begins with a research hypothesis and introduces the empirical part of this paper. The process of the data collection, the findings of the study and the interpretation are also included in this section. Chapter 6 includes the conclusion and discussion of the impact and limitation of the results.
As mentioned in chapter 1, this study aims to test whether there is a statistically provable difference between the level of overconfidence and a person’s education. Therefore, a questionnaire was designed to test students’ miscalibration in several steps. The questionnaire includes ten questions which should be answered first directly and then within a 90% confidence interval (CI). An example was given to explain the procedure. In addition, it was remarked that each question had to be answered without the assistance of other participants or technical devices such as smartphones or laptop computers. Finally, subjects were prompted to state their age, gender, current level of education, and their aspired degree. These data were required to group all subjects correctly in order to test if there is a difference.
In the first step, the questionnaire was handed out to teachers from all basic school types in Southern Germany. The teachers were then advised to give the questionnaire to their ninth graders as well as to observe the experiment. The ninth grade is the last year of compulsory school in Germany and therefore the only choice for a study seeking results from all levels of education. The PISA test, conducted by the OECD, surveys the same group of students and shows that they are able to understand the questions. As there were no extrinsic incentives for participation, the questionnaire was designed to have an intrinsic interest. The teachers received the answers and sources of all ten questions and were advised to tell their students the correct answers after all of the questionnaires had been collected. This way the subjects received immediate feedback. The pretest showed that the subjects had the impression of participating in an interesting exercise rather than in a psychological study. It is important to mention that neither the questionnaire nor the answer sheet gave any indication of the aim or the topic of the study. All participating students thought it was about their ability to estimate the correct answer. However, this is partly true because their metaknowledge was subliminally tested. Earlier research showed that participants will behave differently if they are aware that the aim of the study is to test their confidence level (Mayrhofer, 2006).
In the second step, the same questionnaire was given to several teachers from a vocational school as well as to teachers from a specialized secondary school (SSS). They were advised to give the survey to students from classes with different educational backgrounds. All students had either a General Certificate of Secondary Education (The German Mittlere Reife; ISCED 2) or a General Certificate for University Entrance (The German Abitur, Austrian Matura; ISCED 3A). This step was to determine if the results obtained from the ninth graders are consistent with the result from older students with the same educational background and first work experience.
In the third and last step, the same questionnaire was handed out to undergraduates in an electrical engineering degree program at a technical university. This last test was carried out to determine whether the trend of step one and two continues. It is certain that all students have a university entrance certificate.
Finally, every questionnaire received a unique number and all data were collected in SPSS®. The scanned questionnaires, data sets and calculations are available upon request. This guarantees complete transparency and eliminates a need for recalculation. The procedure of data analysis is explained in chapter 5. The next section will cover the educational level aspect of this thesis.
According to Article 26 of the 1948 Universal Declaration of Human Rights education is a fundamental right (Spring, 2000). It has been described as “a process of teaching, training and learning, especially in schools or colleges, to improve knowledge and develop skills” (Oxford Advanced Learner’s Dictionary of Current English, 2005). Germany is a federal republic and education is a responsibility of its federal states. Therefore, there are 16 slightly different forms of one basic system. The following figure explains the conceptual framework.
illustration not visible in this excerpt
Figure 2: German Education System
(Complied by the author)
All of the slightly different school forms have the three certificates of secondary education in common. After the ninth grade, there is a possibility of receiving the Qualifying Lower Secondary General Education Certificate (the German Qualifizierte Hauptschulabschluss; ISCED 2). One year later, students have a chance of acquiring the GCSE and, two years after that, the GCUE. Since 1997, the UNESCO has provided a standardized concept for these educational achievements to ease the interpretation and significance of a local certificate on a global scale (UNESCO, 1997). The acquisition of the German certificates is subject to obtaining positive results in the final exams. These tests are different in every state and the equivalency is not guaranteed. The annual school report shows that students in the south of the republic score higher than their counterparts in the north. In particular Bavaria has a competitive school system which contributes to its being the ranking leader among the 16 states (Bavarian State Government, 2003). This is the reason why the field study was conducted in Southern Germany, and also why it is important to mention the federal differences between the basic German school system and the Bavarian concept.
The fundamental distinction is the separation of the various secondary schools. There is no comprehensive school in Bavaria. This, at first view, might curb social mobility but there are educational paths which connect each school type and allow mobility. It is possible, although not easy, for a student of a lower secondary school (e.g. Hauptschule) to receive the GCUE in order to be able to study at a university. Pupils holding a GCSE have numerous ways to obtain the entrance certificate. Immediately following the tenth grade, they can complete the last two years at the highest secondary school (Gymnasium). Alternatively, they may also continue their education in social studies, technology or business at the SSS (Fachoberschule). Another possibility is to continue at an upper vocational school (Berufsoberschule) which requires a completed apprenticeship and a vocational certificate. This is a typical combination in the German dual education system, which connects professional training with vocational learning at a vocational school (Berufsschule). Figure 3 explains the various paths and possibilities in contrast to the basic system which was charted on page 11.
illustration not visible in this excerpt
Figure 3: Bavarian Education System
(Complied by the author based on Bavarian Ministry of Education)
As mentioned before, the Bavarian system is competitive and the annual failure rate is rather high. On average, 6.1% of all Bavarian Gymansium students fail to make the grade, about one-third of them change to a lower secondary school (Hackl, 2004). Those who obtain the entrance certificate have basically two options. They can study in Germany at a university or a UAS or study abroad. Most students stay in Europe and benefit from the Bologna Process and its European Credit Transfer and Accumulation System. The next subitem will explain the reform and its impact on this research.
On 19 June 1999, the European Ministers of Education published a joint declaration to implement the “European Higher Education Area” (Campbell and Van der Wende, 2000). The official announcement is also called the Bologna Declaration and started the corresponding process. Its predefined aims were mainly to introduce bachelor, master, and doctorate studies instead of many different national university degrees. In addition, the purpose of the reforms was to provide a “quality assurance” and to guarantee “recognition of qualifications and periods of study” (European Commission, 2011). The official launch of the Higher Education Area was more than 10 years later on March 11/12 2010 in Budapest and Vienna. The ministers and the commission expected the reform to be an incentive to increase student mobility. It was one attempt to decrease the long-term European unemployment rate by taking first steps towards a common European labor market. The idea was to implement policies which would lead to higher student mobility resulting in an increase in labor mobility in the future.
In 2009 the Eurobarometer showed that, although one-third of all students planned to study abroad, the majority did not want to participate in a higher education exchange program. 41% never planned to study in a foreign country and 11% gave up their plan to go abroad (The Gallup Organization, 2009). Nevertheless, a combination of the Erasmus Student Mobility Program and the Bologna Process shows a clear, upward trend in student mobility:
illustration not visible in this excerpt
Figure 4: Development of Erasmus Students per Year
(Source: Erasmus - Facts, Figures & Trends)
As shown in Figure 4, there were 231,410 students studying in another European country in 2010-11. This represents an annual increase of 8.5% compared to the previous year (European Commission, 2012).
This is an important consideration since the research includes university students. Degree programs are not as homogeneous as school classes. There are students from various countries with a different educational and cultural background. As stated above, the Bologna Process abets the student mobility in Europe and, therefore increases the chance of heterogeneous subjects in tertiary education. The next chapter reviews the literature and explains why previous studies in this field suggest that this might lead to variance inflation in this particular subgroup.
Overconfidence is a distortion of judgment which has been tested in various experiments although the knowledge of its presence is not new. In the 19th century, the economist and philosopher John Stuart Mill wrote “that any opinion of which they [mankind] feel very certain may be one of the examples of the error to which they acknowledge themselves to be liable” (Russo and Schoemaker, 1992).
Many different explanations and concepts of overconfidence can be found in the relevant literature. Camerer and Lovallo (1999) describe it as the overestimation of one’s own relative abilities compared to a peer group. The authors state that whenever people have to assess their position in a distribution they are likely to rate it above average especially regarding positive attributes such as income prospects or longevity. This, however, is impossible in a symmetrically distributed trait. Their definition focuses on the BTAE and includes the aspect of an unreasonably optimistic view on future events. Brenner et al. (2004) say that most studies examine overconfidence in a situation of uncertainty in which a judgment is required. In those cases, irrational behavior manifests itself in non-regressive predictions and overly-narrow confidence intervals. This is an important perception because the student questionnaire on page A tests whether it is also true if subjects do not have to think about their degree of confidence for each individual question, but rather the width of their interval on pseudo common knowledge questions. The study tested students’ metaknowledge and the possible differences between the various school types.
Studies which focus on irrational behavior on financial markets state that “human beings are overconfident about their abilities, their knowledge, and their future prospects” (Barber and Odean, 2001). Biased investors sell and buy more frequently, they apply risky stock-picking methods and follow guru advices (Shiller, 2000). These traders invest more time and money to gather information about their stocks and the market (Barber and Odean, 2001). They influenced by the illusions of knowledge and control because they think that additional information helps them to increase their profits. Similarly, they believe that the stocks they buy will do better than average. Financial market participants tend to disinvest in stocks which experience further price increases after they sold them. At the same time, they keep those that constantly decline because they do not want to realize their losses (Shefrin, 2002).
Daniel et al. (2001) suggest that individuals are especially confident about private information. They overestimate its precision and overreact accordingly. Overconfidence could not be detected in cases where the trader did not have such information. Overestimation of one’s own abilities and positive future events seems to be a common problem in almost all professions. Previous experiments testing physicians and nurses, investment bankers, engineers, entrepreneurs, lawyers, negotiators, and managers showed that the heuristic influences their decision-making process (Barber and Odean, 2001). In his often-cited research Oskamp (1965) gave 32 psychologists and psychology students a case study based on real medical records. The case was divided in 4 sections and every subject was asked to answer 25 specific questions about the case after reading each section. They also had to provide their confidence level on how certain they were that the given response was correct. The surprising results showed that the average final accuracy was less than 28% with no significant changes over the four stages. However, the confidence rose significantly from 33% at Stage 1 to 53% at the end of Stage 4. Though the results cannot be generalized, the study is still a good example of overconfidence and illusion of knowledge among experts.
It is due to the complexity of the human cognitive system that a bias does not occur at the same intensity in every single person. There are various determinants that can influence overconfidence. This section contains the five most important factors and describes them as confounders to the heuristic. They can have a positive and negative impact on a person’s confidence level. Furthermore, they bias the outcome of a statistical calculation in one particular direction. For each subitem there will be an explanation of how to avoid a misinterpretation of the final results caused by the factor itself.
The overestimation of one’s own abilities is part of human nature but, on average, men are more affected than women. Barber and Odean (2001) showed that male investors trade more than their female counterparts and by doing so, they adversely affect their performance. Barber and Odean stated that a rational investor only trades if the expected return exceeds the costs of the transaction, such as taxes and commission. An overconfident trader overestimates the accuracy of his information and thereby the expected return. The authors said that people may even trade when the true expected net gain does not result in breaking even. However, the gender difference is not universal; it depends on the task. Lundeberg et al. (1994) said that the gap between the male and female confidence level is bigger in typically masculine areas. The finance industry is one area where men are disproportionally represented. The specific domains are also known for their lack of clear and immediate feedback, which therefore also increases overconfidence. Whenever unambiguous feedback is available without delay, the gender differences disappear (Barber and Odean, 2001).
This research contains no masculine tasks. The pseudo common knowledge questions do not favor a specific gender. They are equally hard to answer for both male and female students. The analysis contains different groups, also for men and women, and all subjects received immediate feedback after the test. The detection of a gender difference is discussed in chapter 5.
Common sense says that older people are better in judging their estimations. They have a sound metaknowledge after a lifetime of decision making in various forms and situations. This social group knows what it does not know and therefore knows the limits of its knowledge. Kovalchik et al. (2005) followed this argumentation. In their study, they formed two groups with 51 junior college students and 50 highly educated, neurologically healthy seniors aged from 70 to 95. The authors tested their confidence level by giving each subject a questionnaire with 20 trivia questions. All questions had two possible answers. Participants were instructed to select an answer and then provide a confidence assessment of their choice. The lowest possible assessment was 50% rising in increments of tens up to 100. The older group did slightly better on the test and showed a lower level of overconfidence. This suggests that age has a positive impact on self-evaluation.
Hershey and Wilson (1997) tested 28 undergraduate students and 32 university alumni with an average age of 71.1 years. About half of each group’s participants were trained, the other half untrained. The scholars created subgroups and examined their behavior on six questions concerning financial planning for retirement. After completing these problems, subjects had to rate the quality of their answers on a Likert-scale ranging from 1 (very poor solution) to 7 (very good solution). The findings revealed that “the absolute magnitude of errors made by older participants were equivalent to those made by younger ones”. A clear connection between age and overconfidence was not visible. Crawford and Stankov (1996) drew a reversed picture. They stated that “older subjects showed a consistent tendency towards greater overconfidence compared to younger subjects”.
This shows that, so far, scholars have not been able to prove a significant correlation between age and confidence level which might be representative for the whole population. The connection remains task-dependent and inconsistent. Due to these aspects and the fact that the age difference among the tested students is by far not as large as in the above mentioned research, age difference is considered as a minor determinant in this paper
A sane neurological and cognitive system is another important aspect of overconfidence. It can be seen as a prerequisite because overestimation of one’s own abilities protects our self-esteem from threat and injury. Campbell and Sedikides (1999) tell the story of a psychology student called Lence and his reaction to self-threat. Lence got a bad grade on his midterm exam. Instead of admitting his failure he blamed external factors. His instructor graded harshly and he couldn’t sleep the night before the test. Although the student realized he got a bad mark, his ego stayed untouched. He rejected the validity of the negative outcome and denied his share of the responsibility for it.
These situations are familiar to most human beings. Sherrill (2007) states “that a number of cognitive, motivational, and psychological factors combine to create” this protection mechanism, but it does not work for everyone. Individuals who suffer from depression or chronically low self-esteem tend to internalize failure and externalize success. They overestimate their weaknesses and underrate their abilities. The SSB has no or only little effect on this group. This leads to a process of adjustments in their confidence level because their ego cannot be protected from negative feedback. Calibration studies result in a phenomenon called underconfidence. Their accuracy, which is usually the percentage of right answers, exceeds the stated confidence level. While mentally healthy people rarely decrease their overconfidence the reverse is true for depressed individuals.
Mental health plays an important role in analyzing the heuristic. It is possible that the data includes students suffering from a mental illness. Underconfidence occurs only in rare situations or under certain conditions. Every case of accuracy > 90% confidence, which means 10 out of 10 right answers, will be assessed in order to determine if mental health could be the cause.
The distortion of judgment is not exclusively a mental aspect. It can also be provoked by biochemical processes. Russo and Schoemaker (1992) suggest that adrenalin and endorphins, which are responsible for strong emotions, may lead to overconfidence. The decision-making process under the influence of euphoria can be biased. Overconfident actions are more likely to happen after an event of personal or professional success. The consumption of alcohol and drugs can have the same effect. The authors provide an example of the Ford Motor Company and their experience with introducing a suggestion system which led to a tremendous euphoria within their workforce. The top management had to impose a cooling-off period before the implementation of any given suggestion. The risk of rashly made decisions was simply too high.
Although it became clear that euphoria can be a confounder of overconfidence, it is not expected to influence the result of this research. There is no evidence that the questionnaire, with its 10 pseudo common knowledge questions, provokes strong biochemical processes. During the observed pre-test, there were no signs of euphoria among participants.
The link between education and overestimation of one’s own capabilities is the quintessence of this research. Yates et at. (1996) argued that there are cultural differences in overconfidence. Asian students tend to be more affected than their American counterparts. The authors saw a connection to the different school systems and their applied teaching method. Ex-cathedra teaching that focuses on rote learning fuels overconfidence. These are widely-used teaching methods in many traditional Asian school systems and may therefore explain the differences. One reason is that many of these countries have languages with thousands of characters which have to be internalized by all students. Li et al. (2006) substantiated the previous findings. They tested 316 university students from Singapore and 340 from Mainland China. All students were ethnical Chinese and had the same cultural background. The striking difference was the education system. Singapore adopted a more westernized model, whereas China retained its traditional method of learning. All subjects had to answer the same peer-comparison problem. The result confirmed that Chinese students exhibit a higher degree of overconfidence than their Singaporean counterparts.
These findings stress the importance of the Bologna Process to this paper. It is not the cultural difference that influences a person’s confidence level, but his educational background. The reforms incentivized student mobility and created more heterogeneous degree programs. Figure 4 on page 14 shows the steady increase of exchange students. This development is the reason why it is more likely that the study contains university students from various school systems. The single choice question about the highest reached level of education is crucial to address this specific confounder. Participants’ answers were used to create subgroups in order to minimize the harmful effect of a heterogeneous group on the standard deviation.
 In fact the cost of maintenance and defense for the Wall were one reason for the decline of the Ming Dynasty.
 The first mayor accident (INES 7) happened in 1986 in Chernobyl, Ukraine. INES 7 is the highest category of nuclear disasters with severe impact on people and environment (IAEA, 2010).
 For further information about the toxic contamination of the Indian city of Bhopal caused by a chemical explosion see: (Bryan, 2003)
 These standards follow the model of “Homo economicus”. For a short overview and criticism see: (Krugman, 2007)
 The PV shows the value of future inflows at the time of investment. For detailed information see: (Tietze, 2011)
 Other concepts and criticism are presented in chapter 4.
 In behavioral science a heuristic is a short decision-making process that excludes some relevant information and/or stochastic facts.
 Researchers found out that the phenomenon increases on questions regarding moral or subjectively construed characteristics.
 For example in (Bank and Kottke, 2005) or (Hall et al., 2007)
 Wrong calibration can be seen in the choice of an overly-narrow confidence interval if the question cannot be answered directly. This effect is important as it will be tested in the student survey to determine the overconfidence level and eventual differences among several school types. Further explanation can be found in chapter 2 and 5.
 The original questionnaire can be found in the appendix on page B. It was modified to a smaller scale. The actual font size was 12 and can be seen on the research disc attached to this paper. The translated version is on page A. The English questionnaire was not used for research purposes and is only provided for reader’s convenience.
 For more information about PISA see: (OECD, 2012)
 The original answer sheet can be found in the appendix on page D. It was modified to a smaller scale. The English translation is on page C.
 Chapter 5 describes the pretest and its impact on the survey.
 While common knowledge comprehends facts and details, metaknowledge represents the human understanding of nature and its limits. It helps a person to know what he does not know. If a man calls a plumber after a green breakage, he admits that his knowledge of plumbing is not enough to fix the problem. In that case his metaknowledge helped him to realize that the task would exceed his common knowledge.
 All school types which have been used for the survey will be explained in chapter 3.
 The International Standard Classification of Education (ISCED) is part of the next chapter.
 Contact: stud.Dominik.Piehlmaier@fh-kufstein.ac.at
 There is one exception: Students from the highest secondary school (German Gynasium) receive the GCSE automatically after they have passed the tenth grade.
 The Erasmus Program started in the academic year of 1987-88. It was designed to ease the bureaucratic procedure of an exchange semester and to give those students a financial support. Especially the bureaucratic aspect was later replaced by the ECTS in the Bologna Process.
 A list of participating schools and institutes can be found in the appendix on page F.
 The average ninth grader does not know the length of Greenland’s coast and the students are not expected to. They were asked to estimate the answers. Most questions cannot be seen as common but rather as specific. That is why it is called pseudo common knowledge questions. Further details will follow in chapter 5.
 See subitem 1.2.3 and 1.2.4 on pages 5, 6.
 Subitem 4.2 gives further details on other conditions of underconfidence.
 It is important to underline that the English version of the questionnaire was not used at any time for the data ascertainment. Subjects could only choose between German (or Austrian) certificates. This was done on purpose to detect statistical outliers more easily.
Master's Thesis, 91 Pages
Textbook, 290 Pages