Abstract
Demands for transparency in algorithms and their processes increase as the usage of algorithmic support in human decision-making raises. At the same time, algorithm aversion – abandoning algorithmic advice after seeing an algorithm err – persist [1]. The current paper proposes a way to investigate the effect of transparency, i.e., disclosing an algorithm’s certainty about its future performance, on the usage of algorithms even when they err. A respective experimental setting requires varying algorithmic certainty while keeping the algorithm’s error rate constant. However, experiencing discrepancy between the certainty information and the actual performance could distort participants’ behavior. The paper, therefore, proposes a solution to the question: How can a study design prevent a discrepancy between indicated success rate and observable performance?
This poster describes an experimental weight estimation task that allows to measure advice deviation from the recommendation. It introduces a way to choose probability values so that the amount of observed algorithmic errors that occur is equally likely for two different probability conditions. With this design, researchers are able to manipulate the success probability disclosed by an algorithm while providing a sequence of algorithmic advice with a constant number or errors. This provides a way to prevent discrepancy between the indicated certainty and the actual performance in comparable experimental conditions.
The poster describes the process as well as the resulting test material. It furthermore discusses the benefits as well as limitations of the proposed study design.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
References
Burton, J.W., Stein, M.-K., Jensen, T.B.: A systematic review of algorithm aversion in augmented decision making. J. Behav. Decis. Mak. 33, 220–239 (2020). https://doi.org/10.1002/bdm.2155
Logg, J.M., Minson, J.A., Moore, D.A.: Algorithm appreciation: People prefer algorithmic to human judgment. Organ. Behav. Hum. Decis. Process. 151, 90–103 (2019). https://doi.org/10.1016/j.obhdp.2018.12.005
Grove, W.M., Zald, D.H., Lebow, B.S., Snitz, B.E., Nelson, C.: Clinical versus mechanical prediction: a meta-analysis. Psychol. Assess. 12, 19–30 (2000). https://doi.org/10.1037/1040-3590.12.1.19
Dietvorst, B.J., Simmons, J.P., Massey, C.: Algorithm aversion: people erroneously avoid algorithms after seeing them err. J. Exp. Psychol. Gen. 144, 114–126 (2014). https://doi.org/10.1037/xge0000033
Werz, J.M., Borowski, E., Isenhardt, I.: When imprecision improves advice: disclosing algorithmic error probability to increase advice taking from algorithms. In: Stephanidis, C., Antona, M. (eds.) HCII 2020. CCIS, vol. 1224, pp. 504–511. Springer, Cham (2020). https://doi.org/10.1007/978-3-030-50726-8_66
Önkal, D., Goodwin, P., Thomson, M., Gönül, S., Pollock, A.: The relative influence of advice from human experts and statistical methods on forecast adjustments. J. Behav. Decis. Mak. 22, 390–409 (2009). https://doi.org/10.1002/bdm.637
Prahl, A., Swol, L.V.: Understanding algorithm aversion: when is advice from automation discounted? J. Forecast. 36, 691–702 (2017). https://doi.org/10.1002/for.2464
Sniezek, J.A., Henry, R.A.: Accuracy and confidence in group judgment. Organ. Behav. Hum. Decis. Process. 43, 1–28 (1989). https://doi.org/10.1016/0749-5978(89)90055-1
Bonaccio, S., Dalal, R.S.: Advice taking and decision-making: an integrative literature review, and implications for the organizational sciences. Organ. Behav. Hum. Decis. Process. 101, 127–151 (2006). https://doi.org/10.1016/j.obhdp.2006.07.001
Alexander, V., Blinder, C., Zak, P.J.: Why trust an algorithm? performance, cognition, and neurophysiology. Comput. Hum. Behav. 89, 279–288 (2018). https://doi.org/10.1016/j.chb.2018.07.026
Yeomans, M., Shah, A., Mullainathan, S., Kleinberg, J.: Making sense of recommendations. J. Behav. Decis. Mak. 32, 403–414 (2019). https://doi.org/10.1002/bdm.2118
Dzindolet, M.T., Pierce, L.G., Beck, H.P., Dawe, L.A.: The perceived utility of human and automated aids in a visual detection task. Hum Factors. 44, 79–94 (2002). https://doi.org/10.1518/0018720024494856
Madhavan, P., Wiegmann, D.A.: Similarities and differences between human–human and human–automation trust: an integrative review. Theor. Issues Ergon. Sci. 8, 277–301 (2007). https://doi.org/10.1080/14639220500337708
Castelo, N., Bos, M.W., Lehmann, D.R.: Task-dependent algorithm a version. J. Mark. Res. 56, 809–825 (2019). https://doi.org/10.1177/0022243719851788
Longoni, C., Bonezzi, A., Morewedge, C.K.: Resistance to medical artificial intelligence. J Consum Res. 46, 629–650 (2019). https://doi.org/10.1093/jcr/ucz013
Hastie, T., Tibshirani, R., Friedman, J.: The Elements of Statistical Learning: Data Mining, Inference, and Prediction, Second Edition. Springer Science & Business Media (2009)
Goodfellow, I., Bengio, Y., Courville, A.: Deep Learning. MIT Press, Cambridge (2016)
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2021 Springer Nature Switzerland AG
About this paper
Cite this paper
Werz, J.M., Zähl, K., Borowski, E., Isenhardt, I. (2021). Preventing Discrepancies Between Indicated Algorithmic Certainty and Actual Performance: An Experimental Solution. In: Stephanidis, C., Antona, M., Ntoa, S. (eds) HCI International 2021 - Posters. HCII 2021. Communications in Computer and Information Science, vol 1420. Springer, Cham. https://doi.org/10.1007/978-3-030-78642-7_77
Download citation
DOI: https://doi.org/10.1007/978-3-030-78642-7_77
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-030-78641-0
Online ISBN: 978-3-030-78642-7
eBook Packages: Computer ScienceComputer Science (R0)