Taiwan take-home pay falls 1.36% in Q1 2012

[GuyInTaiwan]

archylgp: Yeah, if the SD is okay, then the mean would probably be okay. I'd put a couple of caveats on that though. Firstly, maybe a standard distribution wouldn't be preferable (I really don't know). Secondly, you'd also want to look at whether the SD had changed significantly over time if only because that could still have social ramifications even if it looked okay (e.g. previously, the entire population was scewed to the right and was no becoming distributed like a normal bell curve).

[Feiren]

I just took them from the official stat site:

http://www.dgbas.gov.tw/ct.asp?xItem=31380&ctNode=5624

archylgp: The mean is a poor measure of anything, for the reasons HH2 has mentioned/alluded to. It's too affected by outliers. In theory, the mean could actually have increased, but the median could have decreased, due to someone who was very rich making a lot more money. For example, consider the following two examples:

3, 5, 7 (mean = 5)
2, 4, 12 (mean = 6)

You need to look at the overall distribution. I wouldn't even look at the median. I'd probably look at something like the inter-quartile range.

I know the difference (typed the wrong word.). You're right; it comes down to distribution. If the standard deviation is acceptable,then averages should be fine, correct?

It seems the numbers in the papers a few days ago were wrong, anyways, if the numbers provided by Feiren are correct. Feiren, what method was used to get those numbers?