[Ciimplementer] Please share tips and best practices!

Paul Schilling paul at schillingconsulting.com
Fri Jan 24 14:44:20 UTC 2014

Our recommendations/best practices have been very similar to this.  Start by uploading just a single course so you can really dig in to the results of the CI report without being overwhelmed.  If you see problems review your setup and determine the cause.  Then carry that over to the other courses since they likely have similar issues.  Once you are good with that course try a few more or a course from a different student level.  Repeating that process makes dealing with problems much more manageable than getting a 1000 page report (with less than helpful lists of program objectives and PCRS as I am guessing most schools are finding).

I also agree with Sascha's last point.  Take on what you can get done this year.  If that is just getting school objectives in and linked to PCRS get that done.  Then look at projects that can be done in a reasonable amount of time.  For example M1 course objectives, or maybe fall semester of M1.  Don't focus so much on the 12-13 data that it takes away from your ability to make the 13-14 year's data better.

On Jan 24, 2014, at 7:14 AM, Cohen, Sascha <Sascha.Cohen at ucsf.edu> wrote:

> I would concur with all the previous notes, and add that checking and validating the output for the report in the initial tables --- including titling, date ranges, cumulative hours, etc --- is key.
> Not only iterative submission in chunks, but also frequent and repeated submission! I've been urging our folks here to model, submit, review, and refactor, and then repeat, as many times as possible until they have as close to what they want in the report as is feasible. This helps not only to provide training, a better understanding of the process overall, but also to highlight the strengths and weaknesses present in the current systems.
> I would also add that while precision is always important, at this stage of the game none of us should let the perfect become the enemy of the good --- this becomes important when modeling issues such as clerkship presentations and other variable and/or complex sequence blocks which may wind up being imprecise in this first iteration. We (UCSF, and Ilios) are taking the approach of finding the most effective balance between data input, modeling, and presented output, and working with that balance. I expect that next year we will have significantly more robust tools with which to model and validate (and adapt) our data.

Paul Schilling
Schilling Consulting LLC
paul at schillingconsulting.com
Phone: 1.608.628.6313
Fax: 1.888.696.8405

More information about the Ciimplementer mailing list