Wireless usage in Austin public libraries

Preview of a paper I’m writing for my GIS class.

This map shows wireless sessions in proportion to “wireful” Internet terminal sessions at Austin Public Library branches. The regions on the map relate to the spatial analysis part of the paper: I’ve aggregated census block groups to the nearest library so I can estimate income and demographic characteristics for each branch.

Austin wireless map

My hypothesis is that people in the more affluent parts of town are better prepared to take advantage of the availability of public wireless in Austin. Libraries are a good place to study that because they’re open to everyone and because I could get wireful session counts with which to normalize the wireless counts.

I think this map roughly supports my hypothesis, although interestingly Oak Springs and Ruiz show wireless use more like westside libraries. Can anyone think of anything unusual about those branches to explain their high wireless rates? ACC and/or Riverside student apartments might do it for Ruiz, I suppose.

Meanwhile at Yarborough the wireless count is a full 22% of the wireful count! The next highest is barely over half that, at Old Quarry. Yarborough is my neighborhood branch and there’s nothing obvious to explain it. I’m seriously wondering whether someone in the adjacent apartment complex has a Pringles antenna. :-)

4 thoughts on “Wireless usage in Austin public libraries

  1. I took Martha Fuentes’ suggestion and dropped in today at three libraries that represent outliers in my data. There were friendly staffers at each branch who were happy to answer my questions.

    At Ruiz the answer was unequivocal: they said they get a lot of traffic from ACC students with laptops. The Oak Springs people didn’t know why their wireless usage was high but they were happy to hear it.

    At Yarborough they said they believed they get a lot of laptop users because they provide a place for them: big tables with electrical power where you can really spread out and get some work done. I can attest to that myself. I sometimes bring my laptop along when take my daughters to the library. That works well at Yarborough, but when I tried it at St. John’s the only place I could find an outlet required me to sit on the floor in a corner of the stacks.

  2. Based on my experiences at North Village, I am guessing that the real correlation is with the number of power outlets available per branch. The higher the outlets, the higher the number of wireless sessions.

    But you don’t need a GIS class for that, so perhaps better to just move along.

    Do you have any data about the difference in demographics between laptop users and the general population?

  3. David, the Yarborough people would certainly agree with you about the outlets.

    No, I don’t have any direct demographic info on wireless users. The one actual GIS-dependent part of this study, the aggregation of census block groups to the nearest library, is an attempt to redress the absence of demographic data about library patrons in general. Understandably, APL does not ask questions about ethnicity or income when people check out books or sign onto the Internet. The only demographic statistics they could provide me were estimates based on census tract data. I tried to improve on that by using census block groups, which are of higher granularity than census tracts — although of course any analysis based on people’s home addresses fails to take into account transportation patterns or people’s library use near work or school.

    I supplemented the pretty maps with some statistical analysis which showed a significant but not overwhelming correlation between wireless use and the demographics of the surrounding neighborhoods.

    It’s not impossible that outlets would still be a mediating factor: maybe library staff in white, English-speaking, highly educated neighborhoods are more likely to believe it’s important to provide outlets. But I’ve got no evidence that that’s what’s going on.

    Another, perhaps more important factor is whether a library gets a large influx of schoolkids in the afternoons: not only would that affect whether the library is a quiet place for laptop users to work, as the Oak Springs staff suggested, but it also could increase the wireful session counts (the denominator) and thereby artificially depress my normalized wireless counts.

    My guess is that several things are going on at once. St. John, for what it’s worth, would be negatively impacted by all the factors I’ve identified — it’s short on outlets, in a poor mostly nonwhite part of town, and adjacent to an elementary school.

  4. I really wonder how significantly houses or other buildings adjacent to these libraries are utilizing their wifi. I know I sometimes show signal from the library next door in the school where I work. Maybe testing for insufficient wifi signal strength in front of their neighbors’ buildings would rule out all but the most determined (ie pringles antennae builders), whereas high signal in several overlapping buildings would possibly show greater unreliability. Then again, perhaps that’s part of the access you’re trying to measure? You using ArcView? That’s what I used for GIS at UCSC in 2000.

Leave a Reply