The data is from 125 individuals, each listed on a separate row with a unique ID for each individual.
Each participant viewed each victim gender and blood loss amount combination three times, with different videos. The data in this dataset is averaged across those three videos.
Participants took part in a controlled experiment in which they viewed a series of 78 five-second film clips featuring a person with a simulated bleeding. They were asked to, as quickly as possible, classify the video as showing a life-threatening or not life-threatening bleeding using a keyboard response. After each video, the participant was asked to estimate how large the bleeding was, classify the severity of the injury, and, if they classified the video as showing a life-threatening bleeding, to estimate how minutes it would take for the victim to die from their bleeding. The participants also completed a basic demographics questionnaire at the end of the experiment. The entire experiment took between 40 to 60 minutes to complete.
The participants were students at a large south-eastern university in the USA. Participants with prior medical training or stop the bleed education were excluded. Thus, all participants were medical laypeople without prior experience.
The variables varied in the videos were victim sex (male or female), blood volume (ml of blood on the floor), and rate of blood flow (in ml per minute). Further, two video sets were created, one with a top-view (camera placed above the victim) or front view (camera placed facing the victim from the front). In each video, the victim was dressed in blue, hydrophobic scrubs and were seated against a white wall. The simulated wound was not visible. The actors were positioned such that the blood flowed down their thigh and pooled between their legs. The same male and female patient actor were used for all videos. The flow rates used were 80, 200, and 400 ml/minute. The blood volumes used were 0, 50, 100, 150, 200, 300, 400, 500, 700, 900, 1100, and 1900 ml. The combination of three flow rates, 13 blood volumes, and two genders meant that there was 78 videos in total.
For the current dataset, the data has been collapsed across flow rate, and the volume 0 has been excluded, meaning that there are 24 combinations (2 genders x 12 volumes). The dataset includes the response time for the initial classification, the classification response, the volume estimate, and the volume error (calculated as the difference between the true amount and the estimated amount of blood loss).
Ten participants are missing data for the variables in Volume Estimate and Volume Error and 16 participants in Response Time and Classification.