MathCastle
Statistics & DataSuperchallengeData and Displays

Data and Displays · Superchallenge problem

A data set contains the value kk exactly kk times for each k=1,2,…,20k = 1, 2, \ldots, 20 (so 210210 values). It is drawn as a histogram with bins 0.50.5–5.55.5, 5.55.5–10.510.5, 10.510.5–15.515.5 and 15.515.5–20.520.5. The median is estimated from the histogram by assuming values are spread evenly within each bin (the point where half the area lies to the left). By how much does the true median exceed this estimate?