The National Association of Realtors Uses It. Big Media tends to prefer median sales price over average sales price. [What exactly is the ‘median’ price? [OCR]](http://www.ocregister.com/ocregister/money/homepage/article_1071944.php) Although it is the primary numeric representation of the residential housing market, it is surprisingly misunderstood. Other statistics such as _[Average or Mean](http://www.answers.com/topic/mean?method=22),_ or to a much lesser degree (statistical Siberia), _[Mode](http://www.answers.com/topic/mode-statistics?method=6)_, don’t get the same prominence in national statistics.

[Median _defined_: ](http://www.answers.com/median&r=67) Relating to, located in, or extending toward the middle.

The idea that the middle number is used puts many at ease because the outliers, in other words, the extreme data points like very high and very low sales prices, get excluded. Half of the numbers are above the median and half are below.

When its used in the context of real estate, however, the biggest concern is the mix of sales that make up the data set and how it is used. In the OCR article, the northern and southern portions of the county have a different median sales price largely to the differing mix of condos, smaller and older housing within the mix of sales.

In other words,

>a lower median sales price in one neighborhood doesn’t always mean (no pun intended) that the value of a specific house is worth less than the value of a similar house in another neighborhood.

I often see median sales price defined (such as the OCR article) within the context of the article yet average sales price seldom is. It seems a bit ironic that in light of the outpouring of statistics on the housing market and the fact that median sales price dominates statistics in the housing sector, _many still don’t understand it._ Maybe its not a great indicator after all.

For the next housing boom, since many of us are confused, lets use the statistic: [mode](http://math.about.com/library/weekly/aa020502a.htm). Some people are able to [find their perfect number [Seth Godin]](http://sethgodin.typepad.com/seths_blog/2006/03/the_perfect_num.html), so finding a perfect statistic is not too much to ask.


2 Comments

  1. VDH March 28, 2006 at 9:46 am

    Full text here because the link did not work:

    Yeah, median can be problematic.
    If median is the exact middle then with the following example:
    200K, 250K, 285K, 1MM, 1.1MM, 25M
    The median value is 642.5K (you take the mean of the even pair in the middle). If someone was given that median number how would they ever get to the actual numbers I listed? They would never get close. The median of 600K, 642.5K, 650K is 642.5K. Same result, totally different numbers. So not only is the exact middle problematic if you have extremely low and extremely high numbers (quite possible in NYC) but it is also not very useful as a means of comparing different numbers from different periods because the same type of spread of numbers may not be applicable at those different periods. Just look at my two examples to see that.

  2. Jonathan J. Miller March 29, 2006 at 3:23 pm

    Yes – this is a brilliant example of too much comfort in the median sales price. Cutting out the extremes doesn’t always make the result more reliable.

Comments are closed.