After the revelation described in the previous blog post I have been working to implement and test this scoring mechanism.
As described in the previous post this scoring mechanism involves measuring and summing the minimum distances between each drawn point and the answer area or line. Then in the same way measuring and summing the minimum distances between each point of the answer area or line and the drawn area or line.
These two totals are added together and divided by the total number of drawn and answer points to calculate a single average distance. This average distance is a single numeric measure of the accuracy of the drawn area or line compared against the answer. The lower this measure the more accurate the drawn response of the user is. Inversely the higher this measure, the greater the average distance between the points and so the lower the accuracy of the drawn response.
To turn this numeric measure into a percentage score the activity is using upper and lower limits of this measure to define the values of this measure that relate to scores of 0% and 100%. So for example the upper limit may be set as 30 which is the value of this average distance measure which relates to a score of 0%. Values of this measure that exceed this limit will also score 0%.
An example lower limit may be set as 5 which is the value of this average distance measure that relates to a score of 100%. Values of this measure lower than this lower limit will also score 100%.
Values of this average distance measure which fall between these two limits are assigned a percentage score based on a straight line interpolation between these two limits and the respective percentage values (see the graph below).
These limits are set in the activity settings by the activity author. Increasing the upper limit means more inaccurate responses will receive a score. Increasing the lower limit increases the allowable inaccuracy for achieving high scores.
These are three screenshots illustrating attempts of varying accuracy and their related scores:
Very inaccurate – 0%
Medium accuracy – 38%
Very accurate – 100%
To demonstrate this scoring mechanism in action below is a link to a scoring version of the biceps example activity. Have a go at this activity and make multiple attempts to draw the correct area and vary your accuracy to see how your attempts are scored. This activity has the upper and lower limits set at 30 and 5 and so you should see it is possible to get 100% for a very accurate drawing of the area and if you draw an area any distance away from the correct area you will get 0%. Attempts that are close to correct are scored based on their accuracy and in general it seems to give appropriate scores (in my testing anyway!).