|
|
|
|
|
|
by m_dupont
1042 days ago
|
|
|
Yeah the title here really overstates the strength of the conclusions. Didn't read every table in the paper but they clearly state that a linear fit was parsimonious with the data, and adding quadratic and exponential terms only modestly improved the performance but neither even had a p-value of less than 0.05 (I am aware that overreliance on p-values is a bit of an issue, but nonetheless). Its also easy to see from the graph in the paper that the linear fit goes through all the error bars. * edit / Disclaimer: only skim-read the article, I should be working right now |
|
|