Thought for the day: the statistical significance of poverty

from Merijn Knibbe

A press release of the USA census bureau of September 16th states the next ‘key findings’:

“The nation’s official poverty rate in 2009 was 14.3 percent, up from 13.2 percent in 2008 — the second statistically significant annual increase in the poverty rate since 2004. There were 43.6 million people in poverty in 2009, up from 39.8 million in 2008 — the third consecutive annual increase. Meanwhile, the number of people without health insurance coverage rose from 46.3 million in 2008 to 50.7 million in 2009, while the percentage increased from 15.4 percent to 16.7 percent over the same period.” 

Two remarks:

1. read this report!

2. ‘the second statistically significant annual increase’ – this is probably the case. However. When A is larger than B but not statistically significant different and  B is larger than C but not statistically significant different, A can still be statistically significant larger then C. This effect probably explains the contradiction between two statistically significant poverty rate increases and three consecutive annual number of people increases.

http://www.census.gov/prod/2010pubs/p60-238.pdf