The Price of Tea in China

Local data on home sale median prices always gets our attention, doesn’t it?

I would like to convince you that not only should you ignore it, but you should be especially wary about any conclusions that come along with it. 

Median is an often-misunderstood calculation. Not to be confused with average, median is the single data point which is found in the exact middle of a set of data ranked in descending order. I know 'median' has that smart and sophisticated sound to it.

The problem occurs when you try to use median values to study a very small pool of data, and then try to make conclusions from it.

This is especially bad when the items being studied are not homogeneous, like the homes and properties in our area with significant differences in square footage, acreage, location, age, and condition. Here is a step-by-step example that clearly reveals the problem:

OK let's see, looks like 11 homes sold in Lake Bluff for this current period. We rank the homes in descending sales price and home No. 6 becomes the median for this period, right?

You realize that we completely ignore the data from the other 10 transactions. Then let's look back at the year-ago period, we have nine homes that sold, so sort them and the sales price of home No. 5 is that median price and we throw out the data for the other eight transactions.

You can see where this is going. Our 'analysis' boils down to simply comparing the selling price of these two transactions. Then we somehow make observations as to why the median sales price changed and what this might mean about where we are headed? Come on, are you kiding me?

You just cannot pretend that whatever conclusions that follow were derived from or are supported by this data. Not only is the sales mix within these small samples of data a major problem, but if the timing of a single closing in either period was altered, or if the period used was widened or narrowed the slightest bit, we would be left with quite a change as an entirely different pair of homes would go head-to-head as the median prices being compared.

Median price comparisons can be quite relevant when the number of transactions being analyzed goes way up. In large markets they can be used to describe what is happening to actual prices. Too bad we just never have enough data in one place and at one time within our small local housing markets to use median values to reliably show us anything.

On May 4, I and about 90 others attended "Food for Thought, North Shore Real Estate Trends" at the Gorton Community Center. While the vast majority of the information exchanged was interesting and informative, there were a few moments that relied on median price comparisons using a small number of transactions.

A couple of slides required going to a larger current period than other periods due to sparse data. Some other slides broke data down into smaller groupings for clarity. Geez, as predictions were offered based on this, it was time to stop taking notes. 

So if you wonder why past articles or posts make the same mistake, whether knowingly or not, quickly realize it so you won't be swayed by unsupported conclusions.

Oh, and what does all of this have to do with the price of tea in China? Well changes to the price of tea in China, from one period to another, represents an excellent example of where one can use median prices to analyze what's happening in a relevant way...