Using the median isn't strange at all for this kind of calculation.
The average (and by extension the total) wealth has gone down by a much smaller percentage since 2007. The fluctuations just happened to have hit the people around the middle particularly hard. The people at the bottom haven't seen their net wealth decline at all, because they never had any. The people at the top have seen their net wealth decline substantially, but since it was largely in other asset classes less so [percentage-wise] than the people right in the middle, whose net wealth was dominated by home equity.
Bottom line: recessions suck. They happen every decade or so. Plan with this in mind.
Wealth increases from the bottom 1% to the top 1% at an ever accelerating clip. If you take the median, you have absolutely no clue what the health of the groups above or below look like.
For example: if the median wealth is $77,000 --- you can't use that to then calculate how much actual wealth is in the middle 80% of the country (not even close in fact), because that $77k figure is radically off base for the massive skew that the top 10% in that 80% group is going to contain. But you have no means to derive how much wealth you're talking about being contained by that 10% group in the top of the 80% bracket (no means to figure out the skew at the top or bottom).
If you had 1 million data points, and your median is $77k, what does that tell you about the total wealth in the 1 million? It tells you almost nothing. What's the increasing skew like for the top 10,000 in that million? Because that probably contains 30% of all the wealth in the million set.
Calculating wealth by a median, in a nation of 300 million people, only barely kinda-sorta works if you have a very large number of data points, otherwise you get a completely useless data point. Also, medians don't work very well due to population bunches when it comes to wealth, it's not smooth. So if your middle numerical just misses a wealth spike a few points up above the median how useful is your information? If it just misses a big wealth plunge a few points down, what does it tell you about the health of the people in the bottom 40%?
The median is a much more useful statistic when dealing with this kind of distribution, as it applies to a much larger population than the average, which is useless for ~80% of the population as an indicator.