barefoot wrote:cyclotaur wrote:The median distance to secondary school in Victoria quoted earlier would be skewed by regional figures and kids crossing several suburbs to attend preferred (by parents) private schools. However, a large majority of urban students would attend a secondary school in their own or the next suburb, well within walking/cycling distance.
That being the case, the median distance is an entirely appropriate measure.
By definition, half the population travels more than the median, half travel less. If the large majority live in the same/next suburb, then one of them will be the median data point.
Median filters out the ultra-long-distance minority that would skew the mean.
Mode only works if you're dealing with a discrete countable measure. Otherwise you have to bin the data. And even then, it's a pretty crap measure.
Yeah, you may be right ... long time since I did any stats !!
But I do get the diff between mean, median and mode and while this isn't data with a discrete countable measure I just distrust broad statements that quote the median in isolation. Sometimes the full spread of data can reveal more, including the possibility (when continuous data is grouped into ranges of say 0-1. 1-2, 2-3 kms etc) that the most common distance range students travel to school (statistical mode?) may be less (by a fair amount) than the median quoted.
Anyway I guess the point I was getting at is that for significant numbers of students ( "...a large majority of urban students...") the max distance to school is well within walking/cycling range.