Mean deviation can give us a sense of how much data is dispersed from one of the average measurements (mean,mode,median). Mean deviation depends on the difference between the data and the average measurement.
$MD=1/n∑(x - y)$ where x is the different data values and y is the mean/mode/median.
I don't think it really depends on the average. Rather it is completely controlled by the data differences from the average value.
What I thought at first is maybe we need the $Coefficient$ $of$ $mean$ $deviation$ to compare two or more data lists. But when we measure the mean deviation we measure how much the data is dispersed. More difference means greater mean deviation. I don't think there is any effect on this of what the average value is. So we don't really need to come up with a ratio of mean deviation and the average value to see what's actually going on. Finding the mean deviation should be the end of story. So what's the use of coefficient of MD?