Thursdays are slow for chat?
I've been doing a lot of number crunching for the chat reference service here at Baruch College and for the five other colleges in CUNY that share a QuestionPoint subscription with us (Borough of Manhattan Community College, Brooklyn College, CUNY Graduate Center, Hunter College, and John Jay College). I was just looking over numbers for Baruch's service and discovered that Thursdays, of all days, are the slowest days for our service.
Here's the data with the day of the week, followed by the number of chat sessions requested on just the Baruch College service, the percent of all requests for the month (215 for all of November), and the rank of that day of the week:
Tuesday: 40 chats, 19%, 2
Wednesday: 28 chats, 13%, 5
Thursday: 18 chats, 8%, 7
Friday: 25 chats, 12%, 6
Saturday: 42 chats, 20%, 1
Sunday: 30 chats, 14%, 4
I realize that at least one Thursday was Thanksgiving, but before doing the analysis I didn't expect that Thursday would be our slowest day of the week. Our reference desk is busiest on Tuesdays and Thursdays, so I figured that would hold true for chat reference. I would pull a larger data set to see if this finding holds true across a longer time span if it wasn't so difficult to do this kind of analysis. The QuestionPoint reporting tools don't record what day of the week a chat took place on. I had to export the daily chat session activity into a spreadsheet and manually type in the day of the week for all 215 sessions. (I assume that those with fancier spreadsheet/database skills could make this work of assigning days to dates automated in some way.)
CORRECTION (20 January 2009): Whoops! I recently realized that I didn't use an equal number of days of the week when running my report. The month I analyzed (November 2008) has four Mondays, Tuesdays, Wednesdays, Thursdays, and Fridays and five Saturdays and Sundays. By taking the average number of chats per weekday, I made the mistake of taking the average of five Saturdays and five Sundays. If I drop the numbers for one Saturday and one Sunday, then it turns out that Tuesday is the busiest day of week, not Saturday, which makes a bit more sense to me. Here is the corrected breakdown:
Monday: 32 chats, 15% of total, 3rd busiest day
Tuesday: 40 chats, 19% of total, 1st busiest day
Wednesday: 28 chats, 28, 13% of total, 4th busiest day
Thursday: 18 chats, 9% of total, 7th busiest day
Friday: 25 chats, 12% of total, 6th busiest day
Saturday: 39 chats, 19% of total, 2nd busiest day
Sunday: 26 chats, 13% of total, 5th busiest day
CORRECTION (20 January 2009): Whoops! I recently realized that I didn't use an equal number of days of the week when running my report. The month I analyzed (November 2008) has four Mondays, Tuesdays, Wednesdays, Thursdays, and Fridays and five Saturdays and Sundays. By taking the average number of chats per weekday, I made the mistake of taking the average of five Saturdays and five Sundays. If I drop the numbers for one Saturday and one Sunday, then it turns out that Tuesday is the busiest day of week, not Saturday, which makes a bit more sense to me. Here is the corrected breakdown:
Monday: 32 chats, 15% of total, 3rd busiest day
Tuesday: 40 chats, 19% of total, 1st busiest day
Wednesday: 28 chats, 28, 13% of total, 4th busiest day
Thursday: 18 chats, 9% of total, 7th busiest day
Friday: 25 chats, 12% of total, 6th busiest day
Saturday: 39 chats, 19% of total, 2nd busiest day
Sunday: 26 chats, 13% of total, 5th busiest day
posted by Stephen Francoeur @ 4:18 PM
5 Comments
Links to this post
Flickr
Myspace
Facebook
Linkedin
Twitter
YouTube
Last.fm
Del.icio.us
Technorati
MyBlogLog
5 Comments:
Here at Texas A&M, Thursdays are our slowest days as well. (We have two librarians on at night, 6-10 Su-W, but only 1 on Thursday). We average a 15.8 chat questions per Thursday, over the last two years, compared with 20.7 om Mondays -- our busiest day. (This is single campus VR only.)
The WEEKDAY function will convert a date to a number (1 = Sunday). E.g., =WEEKDAY(A1) where A1 is a date.
The TEXT function will convert to the name of the day, either three letters ("Sun","Mon", ...) or the full name. E.g., =TEXT(A1,"ddd") will give you the three letter name, =TEXT(A1,"dddd") the full name, where A1 is *either* a date or a day number (1 to 7).
Once you have converted dates into either number or name, you can use COUNTIF to count the days one at a time, or HISTOGRAM to generate a chart in one step.
Stephen,
Just a hunch, but if Tue/Thu are busy in person maybe that is a when students are on campus for classes... so they don't need the chat as much.
-George
Graeme, thanks for the tip. I'll have to give that a try. One complication, though: the cells with the dates in them also include the time that the chat began. I think I'll have to do a global edit first to strip out the time stamp that is wedged in the date cells.
George, that's not a bad theory. Tuesdays, though, are actually the 2nd most popular day. Also, Tuesdays and Thursdays both have a unique 2 hour break between 12:25 pm and 2:30 pm when there are no classes scheduled at all (Thursdays is "club day" here on campus, which allows all the various student groups a chance to meet when everyone is free). I haven't run the report lately to verify this, but my sense is that at least 30 to 40 percent of all chat sessions take place here on campus (many from within the library itself). Still, more investigation is warranted given your intriguing suggestion.
Post a Comment
Links to this post:
Create a Link
<< Home