On The Reasonableness Of Traffic Forecasts

Abstract

It is difficult to approach the topic of this short paper without commenting on overconfidence. Range estimation is popular in many disciplines where professionals, facing uncertainty, estimate possible outcome ranges. Where estimates are biased by overconfidence, narrower ranges result. The psychology literature is rich with examples of individual and professional overconfidence. This form of behavioural bias has been tested by many academic studies and has been documented for – among others – investment bankers, financial analysts and economic forecasters (van der Venter & Michayluk, 2008). Are traffic forecasters any different? A literature review of behavioural research material reveals some consistent findings which could have relevance here: • Overconfidence increases with the prediction difficulty level (traffic forecasting is not easy); • The phenomenon occurs more frequently among males than females (this is a male-dominated profession); • . . . and increases with age and experience (the job titles of the survey respondents – described later – suggest relatively high levels of seniority). Overconfidence is not the central issue here. The paper focuses on a small survey and the survey results. However, reflecting on these results, some comments are made about overconfidence towards the end.