Sampling -

Want the trend of how it's going. not exact number.  comparing ARL stats year to year would give us a trend. 
Compare last year's arl sample, with this year's arl. Compare ARL sample trend to yearlong trend? Mat can do this.

 What's required for ARL & MIT Libraries:

  • Steve: We need to be able to measure how many questions were asked in general categories; how they're being asked and where.  At desk, away from, in person, on phone, e-mail.  Would be nice to get handle on complexity issue, maybe by taking time.  Uses them to inform general strategy and direction, but doesn't use them to make in day-to-day decisions.
  • Question 31 on page 4 of ARL statistics submission form asks for "number of reference transactions" and "Is the reference transactions figure based on sampling? Yes or no?"
  • The instructionsfor that question (page 6) define "reference transaction" and it's supposed to exclude simple directional questions (questions we define as "other").  "Sampling based on a typical week may be used to extrapolate TO A FULL YEAR for Question 31."
  • Saw this report, which confirms that many ARL libraries are having a hard time with this: http://www.arl.org/bm~doc/spec268web.pd

Lisa had a spreadsheet trying to correlate door counts with ref stats. It doesn't seem to have a pattern, but there are only 5 years to look for a trend. We may want to include ARL stats with this to look for a trend.

Maybe take stats only for ARL weeks and then doing other smaller statistics-taking during the course of the year.
ARL requires reference stats, but we are not required to record directional or "other" stats. Maybe we need to get away from recording other stats.

Zoho Creator - There is a cost associated with it for institutions. Free 1 edit, 5 users. Subscription fee $175/mo for 50 users (including 1 admin). Developer class exists for pay version to allow others to see back end. Simple, flexible. It is very easy to set up and use.
Reporting structure, doesn't seem to batch, need to go get the stats. Requires you to log in.

  • Remlee played with Excel example described here: http://www.bibliotechweb.com/archives/2005/10/27/reference-statistics/, but doesn't recommend.
    • Pros:
      • simple!!!
      • takes stats by hour
      • can run reports easily
      • Does create some pretty cool graphs on the fly.
    • Cons:
      • Can keep only at one desk, and would have to combine individual stats at end of month anyway.
      • Can't delete a stat easily, and you can't go back and enter stats for a previous time period.
      • No gratification - people can't see how many stats have been entered so far in month.
      • Just another window on your browser to open.

AltaRama - likely very expensive. Could do a trial as part of the show and tell. Send Steve a note asking if there is any budget for reference statistics? Highly unlikely, but Remlee can ask. Would this work for a sampling method? It is unlikely to be worth it for four weeks a year. Maybe if we also used it for capturing other information in other studies during the year.

Heather to look at DeskTracker. The problem was that the reports that it generated were not the ones we needed. Not as flexible as we needed it to be.

WOREP - (Lisa) $1.25, definitely a sampling option, because it is in an in depth form.

Take to RISG:
What is the value of collecting reference statistics?

Show the tools, talk about sampling and other conclusions we made from this project.

  • No labels