Abstract
Every investment decision by organizations is an implicit expression of a prediction on the future state of companies, demand trends, future challenges, or politics. Usually, decision makers can chose between different options to invest and also tend to diversify investments in order to reduce risk. Research on cognitive processing, however, showed that unconscious heuristics triggered by “heuristic cues” influence such decisions in an irrational manner. Partition dependence, the partition of the state space, is such a heuristic cue that leads to an irrational bias in the assessment of probabilities towards a uniform distribution. This work demonstrates that this bias is also persistent in group-based forecasting. We explain and predict the occurrence of the partition dependence bias by applying the Dual-Process Theory of cognitive processing. The results of our three consecutive experiments suggest that besides task complexity, individual expertise, and motivation are moderators for the occurrence of partition dependence. We contribute to the research of biases in forecasting by applying approaches of the Dual-process Theories to explain the occurrence of partition dependence for the first time and shed light on future pathways in this research field.
Similar content being viewed by others
Notes
Different Dual-Process Theories usually also use different wording for the systems. Sometimes they also use slightly different understandings of the interactions of the two, or more, types of processing.
For example, the favourite-longshot bias is attributed too low liquidity that leads to the situation that in the border areas orders do not get matched, which can again be explained using the prospect theory (Snowberg and Wolfers 2010). Other explanations argue with the composition of the field of participants (Feess et al. 2014; Restocchi et al. 2018), the motivation of the participants, and subsequently incentives (Servan-Schreiber et al. 2004).
The probability that the realization of the event will fall into this interval.
In the unlikely event that two or more participants would have entered the experiment at the same time, participants would have seen each other trading. This, however, did not happen in current case.
Participants were assigned to the treatments using round-robin according to the time of first entry into the experiment. Due to a higher rate of drop-outs in the market treatments, the poll treatment was closed and the rest of the participants were only distributed among the market groups.
See Table 7 for the complete list of items.
Boerseninitiative e.V. Karlsruhe
Significance codes for all analyses: 0.000 *** 0.005 ** 0.01 * 0.05 . 0.1 n.s. 1.
A one-tailed test was applied, as the theory behind partition dependence already induces the direction of the effect.
“Rational” and “attributional thinking” in the wording of Brakel et al. (2002).
In the lmsr2 treatment the estimation of a trader was influenced by the estimation of the previous trader. This effect is reduced in the lmsr3 setting, as the first numbers shown to the trader always equals the uniform distribution.
Own translation.
The REI-10 measures the two reflective constructs “need for cognition” and “faith in intuition” by five items on a 5 point Likert-scale from completely false to completely true. The five items were selected from a longer version of the test (REI-42), selected by the highest factor loading each.
See Table 8 for the complete list of items (in German).
See Table 9 for the complete list of items (in German).
See Table 10 for the complete list of items (in German).
In the German election system usually no party is able to receive more than 50% of the votes. Hence, several parties have to form coalitions in order to gain a majority in the parliament to be able to govern.
A two-sample test for “equalitiy of proportions” (prop.test()) shows no significant inequality (\(p=.250\)) regarding the gender.
We extensively discussed to utilize incentive compatible fees, as this is usually the case in experimental economics. However, there are several reasons, why we decided to use a flat fee instead: (1) We need to incite the true assessment of the likelihood for intervals. If one assesses 80% to the correct interval, another 60%, should then both receive the “full” payment? If yes, how much need they to assess to an interval to receive the payment in order to qualify? This would be therefore not incentive compatible. If one would pay 80% of the full payment if the correct interval is assesses 80%, it would be game theoretically optimal to assess 100% the the interval that is perceived to be the most likely one. Therefore, this would be also not incentive compatible. (2) We aimed an similar incentive structure for the poll and the lmsr treatment. In the lmsr treatment it would be possible to pay out based on their outcome after the stocks paid out. However, this is not possible for the poll treatment. This would necessarily result in different incentive structures. (3) We aimed to perform the experiment on Amazon Mechanical Turk. To ensure that the exact time of completion of the survey does not have an advantage, the event of forecasting needs to be sufficiently in the future. We decided for approx. 90 days. However, it is not possible (or reasonable) to hold back the payment for the crowd workers for such a long period. Therefore, a flat fee, that is high enough to incite a truthful participation is most suitable in our context.
The experiment was conducted in english, therefore we used the original item formulations, as can be found in the corresponding papers.
This needs to be provided in order that the participants can successfully close the task on Amazon Mechanical Turk and redeem their payoff.
As several workers just started the task in order reserve it for a later completion, which resulted in a “long-tail” at the right end, the median provides a more realistic picture of the completion time than the average.
A two-sample test for “equalitiy of proportions” (prop.test()) shows no significant inequality (p=.380) regarding the gender.
References
Bardolet D, Fox CR, Lovallo D (2011) Naïve diversification and partition dependence in capital allocation decisions: field and experimental evidence. Strateg Manag J 32:1465–1483. https://doi.org/10.1002/smj.966
Berg JE, Rietz TA (2018) Longshots, overconfidence and efficiency on the Iowa electronic market. Int J Forecast 35(1):271–287
Brakel LAW, Shevrin H, Villa KK (2002) The priority of primary process categorizing: experimental evidence supporting a psychoanalytic developmental hypothesis. J Am Psychoanal Assoc 50(2):483–505
Brünken R, Plass JL, Leutner D (2003) Direct measurement of cognitive load in multimedia learning. Educ Psychol 38(1):53–61. https://doi.org/10.1207/S15326985EP3801_7
Campbell SD, Sharpe SA et al (2009) Anchoring bias in consensus forecasts and its effect on market prices. J Financ Quant Anal 44(2):369
Chen O, Castro-Alonso JC, Paas F, Sweller J (2017) Extending cognitive load theory to incorporate working memory resource depletion: evidence from the spacing effect. Educ Psychol Rev 30(2):483–501. https://doi.org/10.1007/s10648-017-9426-2
Chen W, Li X, Zeng DD (2015) Simple is beautiful: toward light prediction markets. Intell Syst IEEE 30(3):76–80. https://doi.org/10.1109/MIS.2015.54
Cipriano MC, Gruca TS (2014) The power of priors: how confirmation bias impacts market prices. J Predict Mark 8(3):34–56
Cowgill B, Wolfers J, Zitzewitz E (2009) Using prediction markets to track information flows: evidence from Google. In: Das S, Ostrovsky M, Pennock D, Szymanksi B (eds) 1st international conference on auctions, market mechanisms and their applications 2009. Springer, Berlin, p 3. https://doi.org/10.1007/978-3-642-03821-1_2
Dohmen T, Falk A, Huffman D, Sunde U, Schupp J, Wagner GG (2011) Individual risk attitudes: measurement, determinants, and behavioral consequences. J Eur Econ Assoc 9(3):522–550. https://doi.org/10.1111/j.1542-4774.2011.01015.x
Epstein S, Pacini R, Denes-Raj V, Heier H (1996) Individual differences in intuitive-experiential and analytical-rational thinking styles. J Personal Soc Psychol 72(2):390–405. https://doi.org/10.1037/0022-3514.71.2.390
Evans JSBT (2012) Dual process theories of deductive reasoning: facts and fallacies. In: Holyoak KJ, Morrison RG (eds) The oxford handbook of thinking and reasoning, OUP USA, 2013, chap 8, pp 155–133 https://doi.org/10.1037/0033-295X.90.4.293
Feess E, Müller H, Schumacher C (2014) The favorite-longshot bias and the impact of experience. Bus Res 7(2):217–234. https://doi.org/10.1007/s40685-014-0013-9
Fischhoff B, Slovic P, Lichtenstein S (1978) Fault trees: sensitivity of estimated failure probabilities to problem representation. J Exp Psychol Hum Percept Perform 4(2):330–344. https://doi.org/10.1037/0096-1523.4.2.330
Fox CR, Clemen RT (2005) Subjective probability assessment in decision analysis: partition dependence and bias toward the ignorance prior. Manag Sci 51(9):1417–1432. https://doi.org/10.1287/mnsc.1050.0409
Fox CR, Bardolet D, Lieb D (2005) Partition dependence in decision analysis, resource allocation, and consumer choice. In: Zwick R, Rapoport A (eds) Experimental business research. Kluwer, Dordrecht, pp 338–360
Gómez-Chacón IM, García-Madruga JA, Vila JÓ, Elosúa MR, Rodríguez R (2014) The dual processes hypothesis in mathematics performance: beliefs, cognitive reflection, working memory and reasoning. Learn Individ Differ 29:67–73. https://doi.org/10.1016/j.lindif.2013.10.001
Harvey N (2007) Use of heuristics: insights from forecasting research. Think Reason 13(1):5–24
Jones RC (2014) Making better (investment) decisions. J Portf Manag 40(2):128–143
Jung D, Dorner V (2018) Decision inertia and arousal: using neurois to analyze bio-physiological correlates of decision inertia in a dual-choice paradigm. In: Davis FD, Riedl R, vom Brocke J, Léger PM, Randolph AB (eds) Information systems and neuroscience. Springer, Cham, pp 159–166. https://doi.org/10.1007/978-3-319-67431-5_18
Kahneman D (2012) Thinking, fast and slow. Penguin, London
Kloker S, Straub T, Weinhardt C (2017) Designing a crowd forecasting tool to combine prediction markets and real-time Delphi. In: Maedche A, vom Brocke J, Hevner A (eds) Designing the digital transformation. DESRIST 2017. Lecture Notes in Computer Science, 10243rd edn. Springer, Cham, pp 468–473. https://doi.org/10.1007/978-3-319-59144-5_33
Luckner S, Weinhardt C (2008) Arbitrage opportunities and market-making traders in prediction markets. In: Proceedings of the 10th IEEE conference on E-commerce technology and the 5th IEEE conference on enterprise computing, E-Commerce and E-Services, 2008. IEEE, pp 53–59
Meservy TO, Jensen ML, Fadel KJ (2013) Evaluation of competing candidate solutions in electronic networks of practice. Inf Syst Res 25(1):15–34. https://doi.org/10.1287/isre.2013.0502
Meub L, Proeger T (2016) Can anchoring explain biased forecasts? Experimental evidence. J Behav Exp Finance 12:1–13. https://doi.org/10.1016/j.jbef.2016.08.001
Page L (2012) ‘It ain’t over till it’s over.’Yogi Berra bias on prediction markets. Appl Econ 44(1):81–92
Pouget S, Sauvagnat J, Villeneuve S (2017) A mind is a terrible thing to change: confirmatory bias in financial markets. Rev Financ Stud 30(6):2066–2109. https://doi.org/10.1093/rfs/hhw100
Prokesch T, von der Gracht HA, Wohlenberg H (2015) Integrating prediction market and Delphi methodology into a foresight support system—insights from an online game. Technol Forecast Soc Change 97:47–64. https://doi.org/10.1016/j.techfore.2014.02.021
Qiu L, Cheng HK, Pu J (2017) Hidden profiles in corporate prediction markets: the impact of public information precision and social interactions. MIS Q 41(4):1249–1273. https://doi.org/10.25300/MISQ/2017/41.4.11
Reichelson S, Zax A, Bass I, Patalano AL, Barth HC (2017) Partition dependence in consumer choice: perceptual groupings do not reliably shape decisions. Psychon Bull Rev. https://doi.org/10.3758/s13423-017-1326-4
Restocchi V, Mcgroarty F, Gerding E, Johnson JEV (2018) It takes all sorts: a heterogeneous agent explanation for prediction market mispricing. Eur J Op Res. https://doi.org/10.1016/j.ejor.2018.04.011
Rothschild D, Wolfers J (2013) Forecasting elections: voter intentions versus expectations. https://doi.org/10.2139/ssrn.1884644
Samson A, Voyer BG (2012) Two minds, three ways: dual system and dual process models in consumer psychology. AMS Rev 2(2–4):48–71
Servan-Schreiber E, Wolfers J, Pennock DM, Galebach B (2004) Prediction markets: does money matter? Electron Mark 14(3):243–251. https://doi.org/10.1080/1019678042000245254
Snowberg E, Wolfers J (2010) Explaining the favorite-longshot bias: is it risk-love or misperceptions? J Polit Econ 118(4):723–746. https://doi.org/10.1086/655844
Sonnemann U, Camerer CF, Fox CR, Langer T (2011) Partition dependence in prediction markets: field and lab evidence
Sonnemann U, Camerer CF, Fox CR, Langer T (2013) How psychological framing affects economic market prices in the lab and field. Proc Natl Acad Sci 110(29):11779–11784
Spears B, LaComb C (2009) Examining trader behavior in idea markets: an implementation of GE’s imagination markets. J Predict Mark 3(1):17–39
Speirs-Bridge A, Fidler F, McBride M, Flander L, Cumming G, Burgman M (2010) Reducing overconfidence in the interval judgments of experts. Risk Anal 30(3):512–523. https://doi.org/10.1111/j.1539-6924.2009.01337.x
Stastny BJ, Lehner PE (2018) Comparative evaluation of the forecast accuracy of analysis reports and a prediction market. Judgm Decis Mak 13(2):202–211
Tannenbaum D, Doctor JN, Persell SD, Friedberg MW, Meeker D, Friesema EM, Goldstein NJ, Linder JA, Fox CR (2015) Nudging physician prescription decisions by partitioning the order set: results of a vignette-based study. J Gen Intern Med 30(3):298–304. https://doi.org/10.1007/s11606-014-3051-2
Teschner F (2012) Forecasting economic indices: design, performance, and learning in prediction markets. Karlsruhe Institute of Technology (KIT). http://digbib.ubka.uni-karlsruhe.de/volltexte/1000029512
Tetlock PE, Gardner D (2015) Superforecasting: the art and science of prediction, 1st edn. Crown, New York
Trumbo CW (2002) Information processing and risk perception: an adaptation of the heuristic-systematic model. J Commun 52(2):367–382. https://doi.org/10.1111/j.1460-2466.2002.tb02550.x
Tversky A, Kahneman D (1974) Judgment under uncertainty: heuristics and biases. Science 185(4157):1124–1131
Tversky A, Kahneman D (1983) Extensional versus intuitive reasoning: the conjunction fallacy in probability judgment. Psychol Rev 90(4):293–315. https://doi.org/10.1037/0033-295X.90.4.293
Vakkari P (1999) Task complexity, problem structure and information actions: integrating studies on information seeking and retrieval. Inf Process Manag 35(6):819–837. https://doi.org/10.1016/S0306-4573(99)00028-X
Watts S (2015) Application of dual-process theory to information systems: current and future research directions. Found Trends® in Inf Syst 1(2):69–162. https://doi.org/10.1561/2900000004
Watts S, Shankaranarayanan G, Even A (2009) Data quality assessment in context: a cognitive perspective. Decis Support Syst 48(1):202–211. https://doi.org/10.1016/j.dss.2009.07.012
Winkler J, Moser R (2016) Biases in future-oriented Delphi studies: a cognitive perspective. Technol Forecast Soc Change 105:63–76. https://doi.org/10.1016/j.techfore.2016.01.021
Woodland LM, Woodland BM (2011) The reverse favorite longshot bias in the national hockey league: do bettors still score on longshots? J Sports Econ 12(1):106–117. https://doi.org/10.1177/1527002510368792
Zhang Y, Fang Y, Wei KK, He W (2013) Cognitive elaboration during wiki use in project teams: an empirical study. Decis Support Syst 55(3):792–801. https://doi.org/10.1016/j.dss.2013.03.004
Acknowledgements
We thank the Deutsche Forschungsgemeinschaft (DFG) for generously funding this research (WE 1436/12-1), as well as Saskia Bluhm, Marie Reger, and Maximilian Schaedlich for their help in collecting the underlying data.
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
About this article
Cite this article
Kloker, S., Straub, T. & Weinhardt, C. Moderators for Partition Dependence in Prediction Markets. Group Decis Negot 28, 723–756 (2019). https://doi.org/10.1007/s10726-019-09622-9
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10726-019-09622-9