Seeking Risk or Answering Smart? Experimental Evidence on Framing Effects in Elementary Schools

Losses are painful but how painful are negative outcomes? This paper presents results of a field experiment among 1377 elementary pupils on the effectiveness of loss framing and avoiding negative outcomes on pupils test performance. I compare answers of pupils in a multiple-choice test in math who are endowed with 0 points and earning points is framed as a gain (Control Group) to pupils who are endowed with the maximum number of points but earning points is framed as a loss (Loss Treatment). In a second treatment arm (Negative Treatment) earning points is again framed as a gain but pupils are endowed with a negative amount of points. However, pupils in this treatment could achieve a positive outcome by earning at least half of the points. I find that, on average, the number of correctly solved questions increases significantly in both treatments (Loss and Negative). The multiple-choice testing format allows to identify the underlying channels of improvements. While pupils in the Loss Treatment significantly seem to seek more risk---answer more multiple-choice questions---pupils in the Negative Treatment significantly answer more accurately---increase the share of correctly answered questions. Nevertheless, the Negative Treatment is preferable to the Loss Treatment as both treatments indeed significantly increase performance of high-ability pupils but low-ability pupils significantly perform worse under a loss frame.