False Consensus Basic Algorithm - Term Paper Example

TRUTH-TELLING IN ONLINE FORUMS Introduction Consumers purchase products and services based on different criteria. Most of the time, their preferenceis influenced by various factors. They consider several aspects that they perceive as related t the product or service in question. In this regard, consumers employ biases and heuristics that help them come up with purchasing decisions. As such, purchasing decisions appear to be more subjective rather than objective. There is no set of procedures that consumers follow when deciding on what to buy. Furthermore, each individual may employ a distinct process of determining his preference. In any case, purchasing decisions are clearly made through subjective judgment. Subjective judgment, particularly those concerning elements viral to purchasing decisions are integral elements of marketing. It is important for firms to identify the prevailing trends and preferences in the market. It is in this regard that market research comes in play. Market research is the primary venue by which firms determine the consumer tastes, likes, and beliefs. All these factors influence purchasing decisions. Knowledge of such elements are important in order for enterprises to come up with products that are successful in the consumer market. However, since market research usually deals with the subjective judgment of consumers, it is quite difficult to assess the validity or truthfulness of the information gathered through the various market research tools. The quality of data that a market research gathers is highly dependent on the truthfulness of answers that the respondents give. The most common problem though, is that it is quite difficult to ascertain how truthful the respondents are in giving their opinions or preferences. More importantly, it is difficult to ascertain that the results of the survey reflect the true preference of the respondents. Thus, it is important for market researchers to identify a method by which they can determine the most truthful choice of consumers. This choice is usually the correct purchasing decision and as such, it is the one that market researchers must be most concerned about. This paper will attempt to present a method of determining the correct purchasing decision of consumers. The method to be presented is based on the theorem known as the false consensus effect. Moreover, this discussion involves ideas regarding one popular form of market research today, online forums and surveys. Basically, this paper will attempt to detail how an online forum may be conducted so as to be effective in determining the correct purchasing decisions of consumers. The False Consensus Effect: A Background Various psychologists have come up with numerous studies on the false consensus effect. In doing so, they have given several interpretations of the definition of such concept. However, the most prevalent definition associated with the false consensus effect is that used by Mullen et al (1985): False consensus refers to an egocentric bias that occurs when people estimate consensus for their own behaviors. Specifically, the false consensus hypothesis holds that people who engage in a given behavior will estimate that behavior to be more common than it is estimated to be by people who engage in alternative behaviors. Basically, the false consensus effect is the tendency for individuals to put too much weight on their own decisions and overestimate the degree to which others agree with them. This happens when an individual readily guesses that his own preferences, opinions, and beliefs are more prevalent in the general public than they really are. According to the Wikipedia Encyclopedia: This bias is commonly present in a group setting where one thinks the collective opinion of their own group matches that of the larger population. Since the members of a group reach a consensus and rarely encounter those who dispute it, they tend to believe that everybody thinks the same way. (2006) Moreover, Gilovich (1990) describes the false consensus effect as, “…A tendency for peoples own habits, values, and behavioral responses to bias their estimates of the commonness of the habits, values, and actions of the general population.” The term was first coined by Ross, et al (1977). The said authors came up with an experiment wherein they asked a group of students to asked to walk around campus wearing a sandwich board saying “Repent.” They then asked those who agreed to the task for their estimate of how many other students would agree to the task as well. Also, they asked those who did not agree to complete the task how many they believe share their disagreement. The researchers found that those who agreed gave higher estimates than the actual frequency. Specifically, they found that those who agreed estimated that 63.5% of their peers would do so, while those who refused expected only 23.3% to agree. (Ross et al, 1977) In short, the false consensus effect materializes when people choose option A from a set of options {A, B}, and expect that there choice has a higher frequency than the alternative option. (Engelmann & Strobel, 2004) The false consensus effect downgrades the popular belief that the consensus is the basic criterion for truth. It decreases the value that most people give to consensus decisions or opinions. Instead, it identifies the truth based on individual preferences in relation to estimated and actual frequencies of the individual’s decision or choice. As previously mentioned, the false consensus effect has been studied by various social psychologists and researchers involved in different fields. In the following section, some of such studies will be discussed in order to provide more information on the false consensus effect and how it is realized in certain situations. False Consensus Effect: A Literature Review Scientists and researchers have made several attempts to prove or disprove the false consensus theorem. Also, psychologists have tried to explain the theorem and how it is related to various elements in human cognition. Lastly, others have attempted to make improvements on the principles of the theorem. One of the researches conducted that aimed to apply the false consensus effect in everyday circumstances is that conducted by Monin and Norton (2003). Their study involved the analysis of the behavior of people during times of drastic change and how various social projection mechanisms affect their actions. One of the mechanisms that they studied was the false consensus effect. Basically, the researchers studied a total of 415 individuals during and after a shower ban. According to the authors the participants in the study, …Demonstrated multifaceted social projection and the tendency to draw personality inferences from simple behavior in a time of drastic consensus change. Bathers thought showering was more prevalent than did non-bathers (false consensus) and respondents consistently underestimated the prevalence of the desirable and common behavior—be it not showering during the shower ban or showering after the ban (uniqueness bias). Participants thought that bathers and non-bathers during the ban differed greatly in their general concern for the community, but self-reports demonstrated that this gap was illusory (false polarization). Finally, bathers thought other bathers cared less than they did, whereas non-bathers thought other non-bathers cared more than they did (pluralistic ignorance). The study captures the many biases at work in social perception in a time of social change. (Norton & Monin, 2003) Basically, the researchers studied how biases, such as the false consensus effect, affect an individual’s behavior when social change occurs. The results of the research indicated a clear false consensus effect. They found that participants who took showers gave higher estimates of those who they believe showered as well as compared to the same estimates of those that did not. The results of their study are highlighted by the graph shown below: More importantly, the researchers found that during the time of crisis, in this case the shower ban, false consensus was more evident in the fact that both bathers and non-bathers believed that they belonged to the majority. Their study demonstrated one scenario in which the false consensus effect is clearly observed. The false consensus effect has likewise been reviewed in relation to political opinions. There have been several studies regarding elections and political polls wherein the false consensus effect has been observed. In a study conducted by Brown (1982), the choices of 179 psychology students regarding their preferred candidate in the 1980 U.S. presidential elections were analyzed. The students were asked for their choice between the candidates, Anderson, Carter, or Reagan. Also, they were asked to estimate the percentage of students in the class believed to prefer each candidate. The author found that supporters of all three candidates estimated significantly higher support for their own candidate compared to the predictions of the rest of the class. In a similar study, Morwitz and Pluzinsky (1996) discovered a strong false consensus bias for the 1992 US presidential election and the 1993 New York mayoral election. Lastly, the same situation was observed in an empirical study of the 1992 Constitutional Referendum in Canada. The authors found that those who signified their intention to vote “Yes” predicted a significantly higher proportion of “Yes" votes than those planning to vote “No.” Moreover, the converse scenario was likewise observed. Not all studies on the false consensus effect deal with applying to real-life polls and surveys. Some attempt to explain how the said social concept is affected by other factors. In a study conducted by Gilovich (2004), he aimed to determine whether the false consensus effect is partly due to peoples failure to recognize that their choices are not solely a function of the “objective” response alternatives, but of their subjective construal of those alternatives. Simply, Gilovich attempted to relate the false consensus effect to the differential construals that individuals form. The author conducted 4 separate studies. The said study was described as follows: Study 1 provided initial support for the importance of differential construal in peoples consensus estimates by showing that larger false consensus effects tend to be obtained on items that permit the most latitude for subjective construal. Study 2 replicated this effect experimentally by asking Ss either a general or specific version of the same question. Larger false consensus effects were obtained on the general version that offered more latitude for construal. Studies 3 & 4 provided further support by showing that (a) Ss who made different choices tended to interpret the response alternatives in ways that reflected the choices they made and (b) subjects who were led to construe the alternatives in the same way tended to make the same choices. (Gilovich, 1990) The study was able to prove the significance of the differential construal in relation to the false consensus effect. In situations where an individual has to make a choice, he must first interpret the meaning embedded in each of the alternatives presented. His interpretation of the alternatives will influence his decision. More importantly, without the awareness that other people may have varied interpretations of the alternative, the person will most likely be led to overestimate the frequency of his choice. In this way, the construals he came up with influence his estimates of the number of other people who would choose similarly. Basically, most studies agree that the false consensus effect may be observed. However, there are those who believe that the use of the word “false” is in appropriate. According to Dawes (1989), it is perfectly rational to use the information about one’s own decision in the same way as the information about any other randomly selected sample of size one. The effect is only false if too much weight is assigned to one’s own decision. It is in this regard that a new definition of a truly false consensus effect was developed. According to Engelmann and Strobel, “A (truly) false consensus effect is considered to be present if people, when forming expectations concerning other people’s decisions, weight their own decision more heavily than that of a randomly selected person from the same population.” This definition will figure significantly in the manner by which the hypothesis of this discussion will be proved. The following section will state the hypothesis as well as the evidence to support the claim. Hypothesis The hypothesis for this study is stated as follows: Using the principles of the false consensus effect, an online forum can help identify the correct purchasing decisions of consumers. The use of the false consensus effect for the online forum is hinged on Bayesian reasoning. Specifically, it is based on the Bayesian argument which states that, “One should expect that others will underestimate the true frequency of one’s own opinion or personal characteristic.” (Prelec, 2004). Such implication is derived from the more popular principle in Bayesian reasoning which states that: The highest predictions of the frequency of a given opinion or characteristic in the population should come from individuals who hold that opinion or characteristic, because holding the opinion constitutes a valid and favorable signal about its general popularity. (Prelec, 2004) Primarily the use of the false consensus effect for the online forum will employ the method developed by Drazen Prelec (2004) known as the Bayesian Truth Serum. Prelec (2004) describes this as a, “Method of eliciting subjective information, designed for situations where objective truth is intrinsically or practically unknowable.” Since online forums, specifically those designed to gather data for market research, deal with subjective information, (i.e consumer preferences and opinions), Prelec’s system is appropriate for the given situation. The Bayesian truth serum is basically made up of an information-scoring system that induces truthful answers from a sample of rational (i.e., Bayesian) expected value– maximizing respondents. Such method veers away form the notion of the consensus as the only basis for truth. Instead, the BTS method, as described by its developer, …Assigns high scores to answers that are more common than collectively predicted, with predictions drawn from the same population that generates the answers. Such responses are surprisingly common, and the associated numerical index is called an information score. This adjustment in the target criterion removes the bias inherent in consensus-based methods and levels the playing field between typical and unusual opinions. (Prelec, 2004) In the BTS method, each of the subjects being studied is asked for his personal answer. Also, he is asked to provide an estimate or a prediction of the empirical distribution of answers. Simply put, he is asked to give his estimate of how many share his opinion and how many contradict it. Predictions in the Bayesian Truth Serum are given scores for accuracy. This means that they are scored based on how close or far they are from the actual empirical frequencies that the researchers gather. According to Prelec (2004), The personal answers, which are the main object of interest, are scored for being surprisingly common. An answer endorsed by 10% of the population against a predicted frequency of 5% would be surprisingly common and would receive a high information score; if predictions averaged 25%, it would be a surprisingly uncommon answer, and hence receive a low score. Prelec provided mathematical proof to support his method. In his system, the population endorsement frequency is denoted by , and which denotes the geometric average of predicted frequencies. These variables are computed using the following equations: The information score for each answer is the log-ratio of the actual to the predicted endorsement frequencies. To obtain such score for an answer k, the following equation may be used: This equation may be better understood in the following form: The answer that has the highest information score is the most truth-telling answer. To better illustrate the argument presented by the BTS system, a diagram is provided below: Figure 1 illustrates the rationale for the surprisingly common criterion. The scenario it uses is one wherein respondents are asked to state their preference between red wine and white as well as their prediction of how many others share their preference. The black box enclosure refers to a group which is made up of individuals who prefer one of the two choices, Red and White. The small red dot indicates a Red-endorsing individual. Her estimate of how many in the population share her choice is indicated by the largest oval with a red boundary. This may be seen as the numerator in the equation of the information score. On the other hand, the denominator in the equation is a mixture of the estimates of the two groups, those who prefer white and those who prefer red. The average of mixture is denoted by the dashed oval. In a way, this represents the geometric mean of the predicted frequencies. This likewise represents the best estimate for the population that prefers Red. Because the expected collectively estimated share for Red falls somewhere between the two, indicated by the dashed oval, the preference for red is expected to be surprisingly common. The argument is equally valid for a white-preferring individual, who expects that the preference for white will be collectively underestimated. Using the same scenario, Table 1 demonstrates how the BTS scoring system works. Using the said figures, the information scores for the two choices Red and White were computed. The equation stated previously was used. It was determined that the information score for the answer Red was 0.21. On the other hand, the information score for White was computed to be -0.16. Thus, it has been made evident that the most truthful choice in this case is Red. The discussion above focused on dichotomous questions. However, the BTS may likewise be used for multichotomous questions. Although the process involved is more complex, the results will be based on the same principles as with dichotomous questions. The most truthful answers are given the highest information score. This is basically what the proposed online survey aims to determine. The Bayesian Truth Serum has provided proof of how the false consensus effect maybe used to determine the correct purchasing decisions. Through the information scoring system, the most truthful preference may now be ascertained. Such scoring system will be used in the online forum to be designed. The computed scores for the answers to the surveys will serve as the basis upon which the decision of which is the correct purchasing decisions. Application in Surveys Having proven that the principles of the false consensus effect help identify the most truth-telling response or answer, it can now be applied to surveys. However, in order to apply the said principles, the surveys must be structured in a certain way. Basically, the survey’s questions must have two possible answers. The respondent is to choose one of the two. Moreover, each question must be followed by another question wherein the respondent is asked to give a prediction of the frequency of those that share his answer. In each question in the survey, the respondent will be asked to supply an estimate. An example of such survey questionnaire is shown below: Such type of survey may be applied to several scenarios. These scenarios must involve the use of an individual’s subjective judgment. For instance, the survey may be used to study the sexual activities among teenagers. Often times, teenagers argue that they are sexually active because most of their peers are. Also, they always say that, “everyone does it so why shouldn’t I?” This belief may be tested by the survey detailed above. The researchers may ask a teenage respondent the following set of questions: Do you engage in sexual acts? Have you had sexual intercourse? Have you had multiple sexual partners? Do you believe that you are too young to be sexually active? The “yes” or “no” answers to the questions must be accompanied by estimates as to the frequency of those who share his/her opinion. Basically, each of the said questions must be followed by the following question: What percentage of teenagers do you think shares your answer? With the collected actual answers and the predictions and estimates, the researcher may then employ the Bayesian Truth Serum scoring system in order to determine which answer is the most truthful. Another scenario wherein the survey may be used is to determine consumer preferences for certain products. If a market researcher for a software firm wants to determine what the most truth-telling preference prevailing in the market, he may employ the survey discussed. He may ask questions pertaining to the respondents’ views regarding his firm’s products, as well as their views regarding the product of their rival. As with the previous example, these questions must be accompanied by questions that ask for an estimate. From the answers of the respondents, and with the use of the BTS scoring system, the market researcher may be able to identify the true preference that prevails in the consumer market. The survey may likewise be utilized as a market research tool in order to help the firm develop products that are well-suited for the market. The survey can help identify the fads and trends in the consumer market by asking questions that pertain to the product the firm plans to develop. A sample set of questions for this particular task may be as follows: Quality or Quantity? Style or Functionality? Trendy or Classic? The respondents’ answers to such questions will help the market researcher identify the consumers’ preferences. More importantly, the accompanying estimates will allow the research to use the BTS information scoring system so as to determine the most truthful answers. There are other instances wherein the survey may be applied. In some cases, the said scenarios have already been studied by other scientists. However, since the Bayesian Truth Serum information-scoring system is quite a new discovery, it can be applied to verify the results of the surveys that have been conducted previously. Some of such surveys were discussed in the earlier section of this discussion. These include political opinion surveys, individual decisions in times of drastic social change, and other commonly studied subjects as well. Since the purpose of this paper was to detail the use of the false consensus theorem in the creation of an online forum for market research, it is vital to discuss the manner by which the said forum will be designed. The following section will provide an overview with regards to the algorithm and programming language to be used in developing the online forum and survey. Design Overview for Online Forum An online forum that aims to apply the principles discussed in this paper must be structured in such a way that information about the preferences of the respondent as well as their estimates regarding the consensus of the group is gathered. These two elements are vital to achieving the goal of the online forum. A flowchart, shown below, presents a simplified algorithm that the online forum will follow in order to determine the most truth-telling response. It must be noted that for the online forum being designed, the questions to be asked must have only two choices. More possible answers require a more complex analysis. Basically, it begins with one question with two possible answers. Once an individual chooses an answer, he is then asked to provide a prediction as to how many other people share his answer. The next step would be to determine the actual frequencies for each answer. After which, the information score for each of the two answers is calculated using the equation discussed previously. Lastly, the two information scores will be compared and the one with the higher score is determined to be the correct one. Having identified the process that the online forum will follow, it is likewise important to provide details with regards to the other important elements necessary in designing the online forum’s website. Since the goal of the forum is to gather information regarding consumer purchasing decisions, and the primary method by which it does so is through online surveys, the online forum is basically a site-centric research. The survey will consist of 10 questions with two possible choices or options for each. Each option will have its corresponding tick box. Moreover, there will be a space wherein the respondent will place his prediction. Once he has done answering all the questions, he will have to click on the “submit” button in order to transmit the data which will be used later on. The construction of the online forum surveys is not that difficult. In essence even the simple HTML codes may be used. These are especially valuable in designing the webpage itself. From the colors, shapes, forms, frames, and textual characters, all these can be designed using HTML codes. Since participation in the survey is voluntary, the webpage must be designed in such a way that will keep the respondents focused. It must be attractive and entertained so as to keep the respondents interested. Once the respondents lose interest, the online survey will inevitably fail. As such, HTML codes will be used to add vibrancy to the webpage. It must be ensured that the webpage contains enough graphics to keep the viewers’ eyes set on the web page. The concentration of the respondent is vital to the information gathering process. When the respondent does not concentrate, he may be led to give false answers. Also, to add more creativity to the webpage, songs and videos may be placed especially if these are related to the product or service for which the market research is being conducted for. For the survey questionnaire itself HTML codes will be used in combination with JavaScript commands. Alert boxes will be used to remind the respondents of certain things. For instance, if the respondent fails to answer all questions or if he fails to provide an estimate, an alert box will appear to remind him to do so. Also, the two choices will be denoted by check boxes which will be developed using JavaScript. Check boxes are one of the elements of forms in the said programming language. Since JavaScript relies on variables, the following variables will be assigned for each question: Q1: A1 and B1 Q2: A2 and B2 The same variables are attached for all 8 other questions with just the subscripts changing to correspond to the question number. Basically, each of the choices are variables themselves. Having assigned the proper variables, and the values of which are derived from user inputs, JavaScript will again be used to calculate the information score of each option. Specifically, the logarithmic function in the said programming language will be utilized. The last step would be to determine the answer for each question with the highest information score. The if—then statement can be used in comparing the information scores for the two choices. The development of the online forum requires the use of forms, images, and buttons. All these may be constructed using the HTML codes and the JavaScript programming language. Another alternative to the programming design explained above will be to use software that was specifically developed to create online surveys and questionnaires. Nowadays, there is an ample number of such type of software that an individual can choose from. Each of these has features that make them distinct from the rest. It is but a matter of analysis in order to determine which software fits the needs of the researcher. Summary The discussion was able to show how the false consensus effect may be used to determine the correct consumer purchasing decisions. Also, it has shown how such concept may be applied to marketing research specifically to online surveys and forums. Furthermore, various ideas have been given with regards to how the said theorem may be used for information gathering. Primarily, the Bayesian Truth Serum was the core principle in the discussion. Its method of information scoring was utilized in the design of the online forums in order to identify the correct purchasing decisions. Lastly, it has been shown that the most informative answer is usually the one that is correct and which receives a low estimated frequency. Thus, the hypothesis stated earlier has been proven. Reference: Baker, L., Koestner, R., Worren, N.M., Losier, G.F., Vallerand R.J. (1995). “False Consensus Effects for the 1992 Canadian Referendum.” Canadian Journal of Behavioural Science 27, 2-2. Brown, C.E.(1982). “A False Consensus Bias in 1980 Presidential Preferences.” The Journal of Social Psychology 118, 137-138. Dawes, R. M., (1989). “Statistical Criteria for Establishing a Truly False Consensus Effect”; Journal of Experimental Social Psychology 25, 1-17. Engelmann, D & Strobel, M. (2004). The false consensus effect: Deconstruction and reconstruction of an anomaly. Gilovich, T. (1990). “Differential construal and the false consensus effect.” Journal of Personality and Social Psychology, Vol 59(4), pp623-634. Monin, B. & Norton, M. (2003). “Perceptions of a Fluid Consensus: Uniqueness Bias, False Consensus, False Polarization, and Pluralistic Ignorance in a Water Conservation Crisis.” Personality and Social Psychology Bulletin, Vol. 29, No. 5, pp 559-567. Morwitz, V.G., Pluzinsky, C. (1996). “Do Polls Reflect Opinions or Do Opinions Reflect Polls?” Journal of Consumer Research 23, 53-67 Mullen, B., Atkins, J. L., Champion, D. S., Edwards, C., Hardy, D. Story, J. E., and Venderklok, M., 1985. “The False Consensus Effect: A Meta—Analysis of 115 Hypothesis Tests”; Journal of Experimental Social Psychology, 21, 263-83. Prelec, D. (2004). “A Bayesian Truth Serum for Subjective Data.” Science, vol. 306. Retrieved 12 May 2006 from: http://www.sciencemag.org. Ross, L., Greene, D., and House, P., 1977. “The ‘False Consensus Effect’: An Egocentric Bias in Social Perception and Attribution Processes”; Journal of Experimental Social Psychology 13, 279-301. Read More