DataONE:Notebook/Data Citation and Sharing Policy/2010/07/26

Project name

Entry title

To do this morning:

This afternoon:

Analysis Planning:

The significant categories from what I gathered seem to be society affiliation, publisher, data requested or required, instructions for citing data.
We can show an upward trend in Impact Factor with journals that request / require data sharing....but not when we look at just journals that require sharing... this seems problematic. *Nic Weber 07:41, 26 July 2010 (EDT)

Nic Weber 15:34, 26 July 2010 (EDT): Heather and I decided on the following categories for our Journal Analysis

Publishers: Divided into a binary of LargePub (Taylor and Francis, Elsevier, Wiley and Springer) and Other (The rest)
ISI Categories: Divided into the three categories we used for this journal collection- Evolutionary Biology, Ecology and Environmental Sciences
Subscription Model: Divided into a Some OA (meaning Open Access or Mixed model) or Subscription ( requiring a subscription for access)
Society Affiliation: Divided into a Has Affiliation or Does not have affiliation

With a dependent variable of our Journal had policy required / requested data sharing

First Pass at Code is now on gist

Nic Weber 16:42, 26 July 2010 (EDT):Running this code I found the following sig variables:

Impact Factor to have a p .000464 and conf int of 0.30950297 1.342204

Affiliation to have p .0005 and conf int of 0.66235193 - 2.307410

Evolutionary Biology (as an ISI category) to have p 0.033748 and conf int of 0.09942984 - 2.600871

Nic Weber 17:57, 26 July 2010 (EDT): Above is not accurate...Everything above should have been debugged... new code is up

Impact Factor found to have a p .0004 and conf int of 0.46 - 1.58

Affiliation found to have p .0003 and conf int of 0.74 - 2.4

EvoBio is no longer Statistically Sig (doh!)

Nic Weber 18:14, 26 July 2010 (EDT): I added a column to my spreadsheet for a publisher code

This code includes five variables: Taylor, Wiley, Elsevier, Springer and Other (catch all for 125 various publishers)

This presents a statistically significant variable for all other publishers:

Publisher Other found to have p- .032 and conf int of .23 - 3.11