Internet has become a new form of mass media since its commercialization in early 1990¡¦s. While the transaction platform moves from bricks-and-motar to Internet, potential factors that influence consumers¡¦ purchase decisions changed. Because they cannot touch the product and interact with sales person, Internet buyers can only make decisions based on information presented on web pages. Under this circumstance, how the presentation of information such as advertisement and product description influence consumers¡¦ buying decision is an important issue. When the information is presented in different ways, people might make biased decisions. This study conducts four laboratory experiments which aim to investigate two decision biases in e-commerce context: framing effect and ancoring effect. The first two experiments focus on the framing effect and the last two experiments focus on the anchoring effect. Framing effect refers to the situation in which people¡¦s buying intention is influenced by different framing messages. Anchoring effect center on the situation in which people¡¦s price estimates are influenced by different anchor points presented in web pages. Three different kinds of framing messages which are formed by combining the attribute framing, goal framing and risky choice framing message and positive and negative presentation are considered in the first two experiments. Moreover, the subjects were assigned into two groups in different level of intrinsic self-relevence to understand whether it plays the moderating role in framing effect. In anchoring effect, in addition to the influence of high and low anchor points on subjects¡¦ price estimates, we also consider the moderating role of the operation of anchor points (one-way/two way), the reinforcement of anchor points (normal/intensified), and the relevancy between anchor and target (relevant/unrelevant). The results indicated that attribute framing effect is stable and is not influenced by subjects¡¦ level of intrinsic self-relevance, whereas the occurrence of goal framing effect and risky choice framing effect depends on the participants¡¦ level of intrinsic self-relevance. For subjects low in intrinsic self-relevance are more influenced by framing message and thus results in different buying intention or choices than those high in intrinsic self-relevance. This study also test and verify the robustness of anchoring effect. Estimaes made by participants in high and low anchor conditions is significantly different no matter the anchor is manipulated in one-way or two-way. In addition, the result of anchoring experiment supports the argument that the relevancy between anchor and target is important for the occurrence of anchoring effect. The moderating effect of anchor reinceforcement depends on the anchor was operated in one-way or two way condition. Anchoring effect is stable despite that the anchor is manipulated in normal or intensified condition when the anchor is manipulated in two-way. On the other hand, when the anchoring effect is manipulated in one-way condition, the anchor reinceforcement plays the role the moderator. Anchoring effect can be observed only when the anchor point is reinforced by appearing for three times. This study serves as a foundation for future study in e-commerce area. The procedures and experimental designs in this study can be either replicated or modified with a different sample to gather further evidence for the results discovered. Further, it can benefit practitioners in improving the design of e-commerce interfaces in real world applications.

No description available.

Aim: When service failures occur, companies must use service recovery actions to recover customer satisfaction. The service recovery paradox refers to the paradox which occurs when a customer is more satisfied after successful service recovery, than they would have been if no failure had occurred. The purpose of this thesis is to examine the effectiveness of using the anchoring effect to lower the threshold value for service recovery paradox occurrence in a specific service failure scenario in the hotel industry. I.e., using the anchoring effect, this thesis aims to reduce monetary compensation required for the service recovery paradox to occur in the specific scenario. Method: 225 respondents were given a survey based on a scenario of a service failure in the hotel industry. The method used was a between-subject experimental design survey, manipulating anchor size, with 5 conditions; SEK 200, SEK 450, SEK 700, SEK 950, and no anchor, to see if the different conditions affected requested compensation whilst customer satisfaction remained constant. Gender and age of the participants was also measured to see if the requested compensation rates depended on these variables. Results: Results showed that it was possible to reduce monetary compensationrequired to reach the SRP by using anchoring values. The control group, which was given the scenario without any anchors, rendered a higher requested compensation than all groups which were given anchors. Furthermore, results showed that the mean compensation required to reach the SRP for the scenario decreased as the anchor values decreased, which indicates that the lower the anchor is set the lower the claim for requested monetary compensation becomes. Finally, the results indicated that age was a relevant variable whilst gender was not a relevant variable regarding compensation requirements. Contribution: Reaching the SRP can sometimes be expensive, especially when it comes to major service failures such as double-booked hotel rooms as is analysed in this thesis. The main contribution of this thesis is to provide an instrument for managers with which they might reach the SRP at lower costs. The results from this research can be directly implemented by managers in similar situations in the hotel industry and may also be used as a foundation for reaching the SRP in other industries as well as other scenarios. Suggestions for future research: As this survey only analyses Swedish participants, results may not be directly applicable to hotels in other cultures. Future research should, thus, focus on examining whether results vary depending on geographic and cultural factors. Other aspects that should be further investigated are what causes the differences in answers between different age groups and if gender displays any differences in other cultures, as it does not seem relevant with Swedish participants. Furthermore, the same surveys, with adjustments, could be used to conduct similar research on other scenarios and in other industries. Finally, researching whether anchoring levels below the lowest anchor used in this thesis, SEK 200, could further lower the required compensation would provide a more complete framework for this research.

Syfte: Denna uppsats behandlar en undermedveten kognitiv bias vilken benämns som ”anchoring effect” eller förankringseffekten. Effekten uppenbaras då människor tenderar att lägga för mycket tillit till den första informationen som görs tillgänglig vid olika typer av beslutsfattande. Teoriramen för detta forskningsområde är tämligen utbredd med drygt 40 år av studier som på senare tid börjat undersöka förankringseffektens påverkan vid förhandlingar. Dessutom finns utbredda konstateranden för att betydande kunskap inte lindrar effekten i någon större omfattning. Däremot föreligger bristande forskningsslutsatser kring hur kunskapsvarians vid förhandlingar  påverkar förankringseffekten. Till följd av detta avser studien att undersöka nedanstående syfte: Syftet med denna studie är att undersöka hur förankringseffekten påverkar utfallet i en förhandlingssituation, när kunskapsvarians råder mellan parterna gällande det aktuella förhandlingsområdet. Metod: I studien genomfördes ett experiment med totalt 44 deltaganden. Experimenten utgjordes av prisförhandlingar gällande en fiktiv bostad, där varje enskild deltagare fick genomgå två förhandlingar vardera. Den första förhandlingen avsåg en lägenhetsförsäljning och den andra en villaförsäljning där parterna agerade säljare respektive köpare. I experimentgruppen förelåg det kunskapsvarians då tredjeårsstudenter från fastighetsmäklarprogrammet mötte studenter med annan utbildningsbakgrund. I kontrollgruppen ställdes motsatsvis deltagare från samma utbildning mot varandra för att skapa mindre skillnader i kunskap beträffande det aktuella förhandlingsområdet. Resultat & slutsats: Resultatet i denna studie tyder på att deltagare vars kunskap stod dem till förfogande, alstrat förmånligare överenskommelser i jämförelse med deltagare med låg kunskapsnivå. Detta trots att deltagarna vars kunskapsnivå var låg, erhållit fördelen av förankringseffekten då de fick lägga det första budet. Resultatet indikerar därmed att förankringseffekten kan lindras till följd av kunskapsvarians vid förhandlingar. Förslag till vidare forskning: Vidare forskning bör utgå från liknande förhandlingsexperiment där forskaren i första hand eftersträvar att generera större kunskapsskillnader mellan parterna i förhandlingen. En större omfattning av denna studie torde således resultera i ökade statistiska klarheter vilket torde vara gynnsamt för det aktuella forskningsområdet. Uppsatsens bidrag: Studiens bidrag är att forskningsresultaten tyder på en lindring av förankringseffekten vid kunskapsvarians inom förhandlingar. Detta till skillnad från tidigare studier där olika kunskapnivåer inte visats ha någon större betydelse. I och med att ingen tidigare studie undersökt detta förhållande har denna studie lyckats identifiera ett tydligt forskningsgap som bidragits till. / Aim: This paper is about a subconscious cognitive bias referred to as "Anchoring Effect". The effect is revealed by the fact that people tend to put too much trust in the first information that is made available in different types of decision-making situations. The theory framework for this research area is fairly widespread with over 40 years of studies, and lately the research has begun examining the anchoring effect in different types of negotiation dyads. In addition, there are widespread findings that significant knowledge does not mitigate the effect in any notable degree. However, there is a lack of research findings regarding how differences in knowledge within negotiations might affect the anchoring effect. Therefore, this study intends to investigate the following: The purpose of this study is to investigate how the anchoring effect is affecting the outcome of a negotiation, when the parties have different levels of knowledge regarding the negotiated area. Method: This study has conducted an experiment with a total of 44 participants. The experiments have involved simulated price negotiations regarding a condominium and a residence property. Each participant performed two negotiations each, one for respective dwelling place. In the experimental group, there was a difference in knowledge when third year students from a real estate brokering program negotiated with students from other programs. In contradistinction to the experimental group, the control group included students with similar education background in order to create minor knowledge differences within the negotiated area. Result & Conclusion: The result of this study shows undeniably that the experiment participants with greater knowledge have generated more favourable agreements, compared to the participants with lower relevant knowledge. Even though the participants with lower knowledge had the advantage of presenting the initial offer in the experimental group. Thus, a mitigation of the anchoring effect has been identified as a result of differences in knowledge within the negotiations. Further research: Further research should be based on similar negotiation experiments with focus on creating greater differences in knowledge between the participants. This in combination with a larger replica of our study should enable increased statistical clarities with fruitful outcomes in this research field. Contribution of the thesis: The theoretical contribution of our study is primarily the fact that the anchoring effect tends to be mitigated by variance in knowledge within negotiations. Considering that no previous studies have examined this before, we argue that a clear research gap have been identified and that our findings has contributed to the theoretical framework.

Fastighetsmarknaden i Sverige har under en lång tid bestått av en uppåtstigande marknad med låg ränta, denna trend bröts under 2023 där räntorna började höjas och marknaden började vända. Vid en tid där priset på bostäder inte självklart stiger blir det intressant att undersöka vilka faktorer som påverkar priset på bostäder. Priset på en bostad är en komplex uppsättning av flera faktorer såsom skick, läge och marknad. Inom beteendeekonomin har det uppkommit fler idéer om vad som kan påverka priset på en bostad. Heuristik, eller mentala genvägar, har också en effekt vid värdering och köp av bostad. En typ av heuristik är förankringseffekten som denna studie hanterar. Studien undersöker om det finns en ankare i form av utgångspris som påverkar slutpriset och om styrräntan har en roll i hur stark förankringseffekten är. Analysen har gjorts genom en kvantitativ metod med en statistisk analys där tester genomförts för att se skillnaden i utgångspris och slutpris under år 2021 och år 2023 där skillnaden är styrräntan. Datan om utgångspris och slutpris har samlats in via hemsidan Hemnet och det geografiska området är Slottstaden i Malmö. Slutsatsen är att det finns en signifikant skillnad mellan tidsperioderna som kan förklaras av förankringseffekten där styrräntan påverkar dess inverkan. / The real estate market in Sweden has for a long time consisted of a rising market with low interest rates, this trend was broken in 2023 where interest rates began to rise and the market began to turn. At a time when the price of housing does not obviously rise, it becomes interesting to investigate which factors affect the price of housing. The price of a home is a complex set of several factors such as condition, location and market. Within behavioral economics, more ideas have emerged about what can affect the price of a home. Heuristics, or mental shortcuts, also have an effect when evaluating and buying a home. One type of heuristic is the anchoring effect that this study addresses. The study examines whether there is an anchor in the form of starting price that affects the final price and whether the policy rate has a role in how strong the anchoring effect is. The analysis has been done through a quantitative method with a statistical analysis where you test the difference in starting price and closing price in the year 2021 and the year 2023 where the difference is the policy rate. Data on starting price and closing price have been collected via the website Hemnet and the geographical area is the Slottsstaden in Malmö. The conclusion is that there is a significant difference between the time periods which can be explained by the anchoring effect where the policy rate affects its impact.

Trust is essential for individuals. We want others to keep their promises and meet our expectations (Dan der Cruijsen et al., 2020, p.680-681). This also applies to banks. For many individuals, trusting their bank is essential to ensure financial security. The interest rate increases between 2022 and 2023, and the increased bank profits, as a result, caught the attention of many individuals. Could two events like these possibly disrupt individuals' trust toward banks? As a result of the high inflation in Sweden in recent years, interest rates increased significantly between 2022 and 2023 which shocked many individuals. Furthermore, banks' profits increased because of the increased interest rates. While banks are profit-making companies, they need to be cautious not to lose the trust of their customers.  This study is based on theories fundamental to answering the study's research questions and purpose. The purpose is to provide an understanding of the factors that affect individuals trust toward their main bank. In addition, the study analyzes factors that may influence the disruption of trust toward banks. The focus will be on whether the interest rate increases and banks' increased profits between 2022 and 2023 have disrupted individuals' trust. The theories used to respond to the purpose are the anchoring effect, point of reference, financial literacy, asymmetric information, and overconfidence.  A quantitative method was adapted for this study to collect data through a questionnaire. By building the questionnaire on appropriate theories, we could examine how our results were consistent with previous research. Later, a linear regression analysis was conducted in Stata based on our collected data. Correlation, multicollinearity, and heteroscedasticity were tested to obtain proper values.  The results from this study show several factors that affect individuals' trust toward banks. Also, it indicates that the sudden interest rate increases and the increased bank profits between 2022 and 2023 disrupt individuals' trust toward banks. Some factors that affect the level of trust are whether the individual has experience of incorrect financial advice, lives in a small city, and feels that they do not receive enough information from the bank. These factors can be linked to previous research on asymmetric information. Other factors that affect trust are how easily individuals generally trust other people and whether individuals visit a bank office for banking affairs. These factors can be linked to previous research on point of reference. Furthermore, individuals having an elementary school degree as the highest education level and unemployed individuals are factors affecting trust that can be linked to financial literacy. These mentioned factors are only a few, the remaining ones are to be found in the result.  This study provides a better insight into trust and trust disruption because of the two events, increased interest rates, and increased bank profits. The results are beneficial both for individuals and banks.

信用評等機構對於企業與市場投資者而言有著重要且無可取代的功能，其所提供之信用評等資訊應當是許多市場投資者所仰賴的重要決策依據，但近年來，卻有許多外界聲浪質疑信評機構評等之準確性，本論文之研究目的即是希望從錨定效應此一行為偏誤之觀點切入，探討國內信評機構在對企業評等時，是否會受到錨定效應影響，導致評等調整不正確或是評等落後其他財務指標等現象發生。 研究結果顯示，國內信評機構對受評企業之過往財務資訊存在錨定現象，但不至於大幅影響整體違約風險之準確性，且信評機構對受評企業財務之惡化較為敏感。另一方面，本研究也發現，信評機構對於非上市櫃公司、金融業以及初次評等等級在「twAA」以上企業之評等，存在較為明顯之錨定現象。 / Credit Rating Agencies (CRAs) play a major role in the financial market. Credit rating purport to provide investors with valuable information they need to make decisions about investing, but the accuracy of the rating itself has been called into question by many investors in recent years. The purpose of this study is to examine the anchoring effect on CRAs while the rating is being given. The results indicate that domestic CRAs tend to be anchored on the past financial information of the issuers. But the impacts are very slightly. Besides this, CRAs seem relatively sensitive to the financial deterioration. Moreover, the anchoring effect are much more significant when the debt issuers are private firms, financial institutions or the companies with greater or equal to twAA initial credit rating.

住宅市場是典型的不完全訊息市場，每個市場參與者並不知道潛在交易對象的所在位置、偏好，與保留價格。不完全訊息意涵著，交易者必須透過搜尋才能找到交易對象，因而必須支付搜尋成本，也會形成搜尋市場。不論是住宅交易的買方或者賣方，都可以選擇是自行搜尋交易對象，或者透過仲介業者來協助交易的達成。然而，仲介業者對於住宅搜尋市場之影響，目前仍存在著許多問題是尚待釐清的，而拍賣市場在台灣所扮演的重要性愈來愈高，其市場機制的價格效果也是值得關注的問題： 一、跨區購屋、定錨行為與仲介服務效果 買賣房屋幾乎必然會有議價過程，雙方的議價能力除了受到市場條件的影響外，賣方對於本身所蓋或者所擁有的房屋及附近地區市場等資訊都較買方為多，因此賣方處於較有利的地位。因此，本研究的第一個研究問題即是：就購屋者彼此之間，在地購屋者是否比跨區購屋者具有訊息優勢？參考價格偏誤是否存在？具有訊息優勢的仲介服務能否改善購屋者的搜尋成本與參考價格偏誤？本研究實證結果顯示，基於搜尋成本較高的原因，跨鄉鎮市區的跨區購屋者相對於未跨區者需要多支付3.8%的價格貼水。其次，參考價格愈高的地區，其購屋者會因為定錨效果或參考點偏誤而多支付1%的價格貼水，此外，高價格分配信念的購屋者，平均會支付4.9%的價格貼水。第三，地價上漲率較高地區的購屋者，會誤用自身地區的外推性預期，以為遷入地區也有同等的價格增值空間，而多支付約11.4%的價格貼水。最後，購屋者若尋求仲介服務亦能改善其出價能力，約可降低2.9%的價格貼水，然而，仲介服務在改善搜尋成本與定錨的效果方面則不顯著。 二、仲介服務對於價格分散之影響 本研究利用搜尋成本與價格分散的觀點，檢視具有訊息優勢的仲介服務業者是否真能提高住宅市場的價格搜尋效率。以台北地區的住宅市場資料，指出仲介服務的存在的確可以提高購屋者的搜尋能力。價格分散的估計與檢定結果則顯示：第一，購屋者成交價價格分散小於訂價價格分散；第二，透過仲介服務搜尋者訂價價格分散未顯著異於自行搜尋者的訂價價格分散，可是透過仲介服務搜尋者的成交價價格分散則顯著小於自行搜尋者的成交價價格分散，同時透過仲介服務的價格收斂比率較高。此乃表示，仲介服務業者並未運用其訊息優勢協助賣方進行較有效率的訂價，但是能有效地協助買賣雙方透過配對與議價活動，大幅地降低成交價的價格分散程度。此外，進一步比較國內相關研究結果，目前台北市的住宅市場訊息效率已較過去有顯著的改善，特別是透過仲介服務的改善效果更為明顯。 三、不對稱的仲介服務價格效果 過去有關仲介服務對於交易價格影響的實證結果卻出現許多分歧而不一致的現象，本研究認為過去相關文獻的差異，可能源自以普通最小平方迴歸的方式來估計仲介服務的價格效果時，會忽略住宅價格條件分配的差異。以分量迴歸估計後發現，仲介服務係數在各價格分量呈現很大的差異且顯著，仲介服務的價格效果，在0.10分量約有4.4%的溢價，而 0.75分量以上則約有-5.6%的折價現象。因此，本研究嘗試以高低價格分量的不對稱訂價策略，作為仲介服務價格效果不一致的現象的檢視觀點，並得到實證上的支持。 四、搜尋與拍賣市場機制選擇及拍賣市場績效之再檢視 拍賣市場為購屋者的重要次級市場之一，因此拍賣市場的績效就顯得愈來愈重要。本研究考慮了購屋者的搜尋成本對於市場機制自我選擇偏誤的影響，重新檢視拍賣市場的績效。本研究實證結果顯示，在未考慮自我選擇偏誤下，拍賣市場機制的估計係數為-22.6%，且達1%統計顯著水準。但是，在控制買方與物件的自我選擇偏誤後，我國拍賣市場與搜尋市場間並無顯著的價格差異存在，因而本研究對於過去國內相關文獻認為拍賣市場一定比搜尋市場折價較多的說法，提出了相當的質疑。惟此三年間我國北部地區拍賣市場的拍定率從10%快速成長到30%，而市場條件的快速變化，很可能會造成較大的拍賣價格變異。因此在後續研究上，可以比較拍定率差異較大的時間進行比較研究，以了解本研究結果之穩定度。 / This dissertation employs search theory and behavior theory to study four relative essays. The first essay is to test three questions using a unique data base in the viewpoint of search cost and Anchoring behavior: First, is there anchoring effect or reference price bias on home-purchasing behavior? Second, is there any extrapolative expectation effect of reference price change on homebuyers? Third, can homebuyers reduce price premium from their high search cost or perceived bias? Those answers can help us understand if we can get alternative interpretation to housing price dispersion and if government should provide housing information service. Empirically, we find that out-of-town/district buyers pay a statistically significant price premium in the Taipei area. We also find some evidence consistent with the price premium being driven by high search costs, anchoring effect and extrapolative expectation from heuristics. Finally, homebuyers can lower price premium through real estate brokers in the market. The second essay is to examine the efficiency of housing and brokerage markets in view of price dispersion. We find brokerage service enhance the search ability of homebuyers. We use listing price prices as the prices before search and the actual transaction prices as the price after search, and we also separate the sample into search by homebuyers and search by broker. We find that search by broker decreases the price dispersion compared to search by homebuyers. The third essay is try to explain a number of past and recent studies provided conflicting empirical answers to the effect of real estate brokerage service on housing price. We employ quantile regression to capture the behavior at each quantile of conditional house price distribution and to test the asymmetric effects of brokerage service. An important findings of this paper is that the price effects of real estate brokerage service are significant heterogeneous across the conditional price distribution. The contribution of this paper to the prior literature is to provide empirical evidence by showing that broker might have a positive, negative, or zero impact on the housing prices. The final essay is to discuss the decision making behavior of housing markets mechanism choice. Real estate auction market has been one of main market mechanisms of home purchase. Therefore, the performance of real estate auctions is a very important issue. This article reviews the price premium or discount of real estate auctions by correcting the self-selectivity bias in a view of homebuyers’ search cost. The empirical result shows that the availability of an auction as an alternative has the result of high search cost buyers attending auctions. Next, after accounting for the endogenous nature of this choice and controlling for property and buyer characteristics, prices of properties sold at auction were not lower than those of comparable properties sold in a search market. It is questionable to say the performance of real estate auctions is inferior to that of negotiated sales at Taiwan.

為了處理在願付價格的研究中，極端受訪者對於估計結果所造成的誤差。本文沿用Tsai(2005)所建議採用的三要素混合模型，將受訪者區分為價格再高都願意支付、願意支付合理價格以及價格再低都不願意支付等三種類型。在評估願付價格時，以加速失敗模型(accelerated failure time model，簡稱AFT model)針對願意支付合理價格的受訪者進行估計，並且在考慮不同起價點可能會造成不同程度的起價點偏誤(starting point bias)或是定錨效果(anchoring effect)的情形下，提出一個起價點偏誤調整模型來做探討。我們並以CVDFACTS中的高血壓之願付價格資料進行實證分析。分析結果發現，教育程度越高的男性對於能降低高血壓病患罹患心臟血管相關疾病之新藥願意付較高的金額。此外我們也發現在此筆資料中，不同起價點確實會造成不同程度的偏誤，經由偏誤調整後會得到較高的願付金額。 / A study of willingness-to-pay often suffers from the bias introduced by extreme respondents who are willing to or not willing to pay any price. To overcome the problem, a three-component model proposed by Tsai (2005) is adopted. Under such a circumstance, respondents are classified into three categories, i.e. respondents who are willing to pay any price, unwilling to pay any price, or willing to pay a reasonable price. The willingness-to-pay for those subjects who are willing to pay a reasonable price is again modeled by an accelerated failure time model (AFT model). In this study, we, however, propose an unified model that allows us to look into the issue related to starting point bias and anchoring effect, simultaneously. Willingness-to-pay for cardiovascular disease treatment from a longitudinal follow-up survey- CVDFACTS, is investigated using the new model. Through the use of the model, we are able to detect the effects of starting point biases, and make a proper adjustment accordingly. Our analysis indicates that male respondents with higher education level have an inclination to pay higher price for the new treatment. Besides, we also discover that starting point bias does exist in this dataset.

De par leur nature scientifique, les sciences économiques visent, entre autre, à observer, qualifier, ainsi que quantifier des phénomènes économiques afin de pouvoir en dégager diverses prévisions. Ce mémoire se penche sur ces prévisions et, plus particulièrement, sur les facteurs pouvant biaiser les prévisionnistes au niveau comportemental en référant à l’effet d’ancrage, un biais propre à l’économie comportementale – une sous-discipline des sciences économiques. Il sera donc question de comprendre, par une analyse selon la discipline que représente l’économie comportementale, ce qui peut les affecter, avec un accent mis sur l’effet d’ancrage plus précisément. L’idée générale de ce dernier est qu’un agent peut être biaisé inconsciemment par la simple connaissance d’une valeur précédente lorsqu’il est demandé de faire une estimation ultérieure. De cette façon, une analyse des salaires des joueurs de la Ligne Nationale de Hockey (NHL) selon leurs performances passées et leurs caractéristiques personnelles, de 2007 à 2016, a été réalisée dans ce travail afin d’en dégager de possibles effets d’ancrage. Il est alors possible de constater que les directeurs généraux des équipes de la ligue agissent généralement de façon sensible et rationnelle lorsque vient le temps d’octroyer des contrats à des joueurs mais, néanmoins, une anomalie persiste lorsqu’on porte attention au rang auquel un joueur a été repêché. Dans un tel contexte, il semble pertinent de se référer à l’économie comportementale afin d’expliquer pourquoi le rang au repêchage reste une variable significative huit ans après l’entrée d’un joueur dans la NHL et qu’elle se comporte à l’inverse de ce que prévoit la théorie à ce sujet. / Economic analysis, by its nature, involves observing, qualifying and quantifying economic data with the ultimate goal of making forecasts. In this masters thesis, I am interested in factors that could bias a forecaster's behavior – with special focus on phenomena, like the anchoring effect, that have been proposed in behavioral economics. At a fundamental level, the anchoring effect states that an agent's ability to accurately forecast may be affected by placing unwarranted emphasis on certain economic variables. To study this effect, I analyze how the salaries of National Hockey League (NHL) players are determined by the players' characteristics and past performance. From the results, it would appear that NHL general managers are generally sensible and rational when it comes to using historical data to make decisions about player salaries. However, there is a persistent anomaly regarding the draft position of a player. Although one would not expect the draft position to be very important after eight years of experience in the NHL, the analysis shows that it is remains a significant determinant of player salary. Behavioral economics and more specifically, the anchoring effect, helps explain why this might be so.