First up apologies if this is a violation of a rule or COC.(adsbygoogle = window.adsbygoogle || []).push({});

Anyway new paper was published a couple of days ago concerning first flowering dates of flowers in the UK.

http://rspb.royalsocietypublishing.org/content/early/2010/04/01/rspb.2010.0291.full" [Broken]

A report on it was published in the http://www.guardian.co.uk/environme...-study-shows?showallcomments=true#comment-51".

And as always a somewhat exuberant discussion broke out. In the end one comentator made this criticism of the paper.

Now I understand that 2 standard deviations are a confidence interval of 95%. I was very wary however of the suggestion that this was all "accounted for solely by chance variation" even at less than 95% confidence surely that is not just chance, just less likely not to be chance? This paper does not demonstrate that first flowering has been occurring on average earlier during the previous 25 years than in any of the 9 previous 25 year periods. The variability in 25-year mean date of first flowering can be accounted for by chance alone. The mean date of first flowering in the most recent 25 year period is within two standard deviations of the mean for the previous 250 years. In other words, exactly where we expect it to be, accounted for solely by chance variation. There is no evidence of any systematic process whatsoever influencing date of first flowering.

I emailed the author and they very kindly responded, although again I lack the skills to fully understand their response. They said I could use the response but not cut and paste it, which I am very happy to comply with. The author infromed me that they do not simply apply an average for the estimated indices but apply a hierarchical Bayesian model that in their words can "estimate parameters with uncertainties in the estimate".

I dont quite follow what that means, I understand the baysian statistics are a means of reducing the randomness in statistics. (My ex does this for a living and I am going have her go through it with me hopefully but thought I would also ask here). The phrase that the author used that has me foxed is this one: "estimate parameters with uncertainties in the estimate". Can someone explain what that means and what the result would be?

As I said apologies if people do not think this is what the forum is for, but I keep seeing things like baysian methods being mentioned all over the place and am never quite sure about it.

**Physics Forums | Science Articles, Homework Help, Discussion**

The friendliest, high quality science and math community on the planet! Everyone who loves science is here!

# Understanding criticisms of a paper

**Physics Forums | Science Articles, Homework Help, Discussion**