Included here is the data from the Johns Hopkins article search
comparison study of September 2012.  As reported in an article still
in progress, likely to be published in the Code4Lib Journal 
(http://journal.code4lib.org) in 2013, with the tentative title 
"A Comparison of Article Search API's via blinded experiment and developer
review," by Jonathan Rochkind.

We are making this data available for purposes of complete transparency of
our research process and analysis, and in case anyone would like to analyze our
results on their own.  

The data here is a dump of the database collected by the bento_battle tool used
as our survey instrument, the exact version of the tool used is as tagged at
https://github.com/jrochkind/bento_battle/tree/jhu-study-10Sep2012 and
https://github.com/jrochkind/bento_battle/tree/jhu-study-17Sep2012

There are two files included here:

1.  complete_database_dump.sql.zip : Complete dump of bento_battle database,
using `mysqldump`. Can be used to restore a mysql database with identical
contents to ours when we concluded the study. There are three tables included,
`selections`, `timing`, and `errors`. 

2. selections.csv.zip :  The `selections` table (also included in complete dump
above), included in comma-seperated value (CSV) format, for potentially easier
use of the key table involved in our results. The first row of the CSV is column
headers.