2 ms·

> "found a correlation of 0.04 – which means that there was effectively no relationship at all." This is scary: Independence => 0 correlation but ~(0 correl

• simonster 13y ago
  "effectively no relationship" and "statistically independent" are not necessarily equivalent claims. Assuming Nutt used a rank correlation statistic (Pearson correlation is not well-suited to ordinal data) then zero correlation most likely implies no linear relationship between ranks. That's not equivalent to independence, but I think it's accurate to characterize it as "effectively no relationship" in the context of a discussion about drug laws. OTOH, a scatterplot would be more diagnostic than a statistic here.