3 ms·
That's a poor analysis. What basis do you have for assuming that it's a linear growth over the entire period? As you see, that 1997 data point is very high. Th
by dalke 14y ago
That's a poor analysis. What basis do you have for assuming that it's a linear growth over the entire period?
As you see, that 1997 data point is very high. The Yahoo Finance link I gave notes "The high number of expatriations that year is thought to be a result of Hong Kong reverting back to Chinese sovereignty, according to tax attorney Andrew Mitchel. Chinese citizens are not allowed to have dual citizenship, so many Hong Kong residents renounced their U.S. citizenship prior to Hong Kong joining China in 1997, so they would become fully Chinese citizens." You might be able to verify that a bit by looking to see if there's a higher number of Chinese sounding names for that year.
In any case, remove that first data point and you'll see that the trend is much more pronounced and statistically significant. Because you have so few data points, any analysis will be sensitive to a single large outlier like that one.
You'll should also investigate if there were similar effects elsewhere. Why, for example, was there a seeming slump in 2006-2008?
You'll also need to compare your linear model to other models. The one proposed by many is that the tax and enforcement changes around 2010 caused a big reaction. Can you model that piece-wise?
Also, there are many factors which affect the numbers. The rate of US citizens who renounce citizenship in order to take on high office in another country should be about the same, compared to those leaving for tax reasons, so you should be subtracting some baseline for that.
With this few data points, and with a complicated underlying system, it's basically impossible to make a valid statistical interpretation. That's why these articles end up talking to the people who work with those who renounce their citizenship.
- mikeyouse 14y agoYou might be able to verify that a bit by looking to see if there's a higher number of Chinese sounding names for that year. Once again, why should I have to? These are professional journalists writing articles with massive holes, it's embarrassing. In any case, remove that first data point and you'll see that the trend is much more pronounced and statistically significant. Here's what that looks like: http://imgur.com/a/FBeAv http://imgur.com/a/FBeAv -- Still ugly. The one proposed by many is that the tax and enforcement changes around 2010 caused a big reaction. Can you model that piece-wise? Why should the reader have to perform these exercises? If the author's premise relies on such a situation existing, they should absolutely perform the analysis. The rate of US citizens who renounce citizenship in order to take on high office in another country should be about the same, compared to those leaving for tax reasons, so you should be subtracting some baseline for that. Once again, someone would have to actually do some reporting to determine what that baseline should be. With this few data points, and with a complicated underlying system, it's basically impossible to make a valid statistical interpretation. That's why these articles end up talking to the people who work with those who renounce their citizenship. Precisely my problem with this type of reporting. What do you think a tax lawyer is going to say when you ask him why people are renouncing their citizenship? How does that add any value?
- dalke 14y agoGood luck in your search for a news service with the rigorous quality that you demand.