Discard the irrelevant: Statistics don’t bleed, but our students do.

I have written an new article for the LLAS blog (in a personal capacity).

Some rise by sin and others by virtue fall. William Shakespeare, Measure for Measure

Statistics are everywhere in education. We have the National Student Survey (NSS), the First Destinations Survey, newspaper league tables, and the Times World University rankings among others. Universities are now required to publish ‘Key Information Sets’ (KIS) from 2012. The KIS has data from the NSS (the higher the agreeing percentages the better), the cost of university accommodation (presumably the lower the better), fees (the lower the better), graduate employment rates (the higher the better), percentage of assessment which is written exams (depends on the student) and number of ‘contact’ hours (again, depends on the student). In short if it can be measured the data is out there. And if it can’t be measured, we’ll find a way to measure it anyway, (research impact anyone?). Add to all this the information that students get from visit days, Facebook, twitter, the online student forums, friends and the phrase ‘information overload’ comes to mind. In his report Dimensions of Quality Graham Gibbs warns us about that immeasurable factor, reputation, which can override any real measure of quality. I suspect that all this information only serves to make reputation all the more important.

Get accessing to UK postal code data for use in MapWindow.

The relationship between postal codes and latitude and longitude has only recently come into the public domain (see www.freethepostcode.org). This has been something of problem for users of GIS software.

Here are two ways to plot UK postcode data on a map in MapWindow.

1. Method 1 is probably the easiest.
3. Generally seems to work quite well, but was hanging when coping with larger amounts of data. The main problem is that it is limited to 2000 postcodes a day. A bit of a problem if you have a very large dataset.

Method 2: A work around using a free utility called batchgeo.com www.batchgeo.com

1. Paste your data from excel into the area indicated on their website.
2. (Optional) Validate data—good idea to check that the part it thinks is your postcode data really is.
3. Press ‘Map Now!’
5. Scroll to the bottom of the page and click on save as google earth .kml file. NB: It took my ages to find this bit!
6. Save the .kml file to your hard drive.
7. Map Window does not read .kml or. kmz, so you will need to convert your .kml file to a shapefile (.shp).  You can do this online at http://www.zonums.com/online/kml2shp.php, but I’m sure that there are other converters.
8. Save you .shp (and the partner files it generates to your hard drive.
9. Go back to MapWindow and Add layer selecting the shapefile you have just created.

I’m sure that someone has been able to find another way of doing all, but these are the ways I managed to get it done.

The yellow symbols on the map above are the locations of the Links into Languages lead universities. Click to see it more closely.

Teaching quantitative methods

Part of my job in teaching the the LLAS research methods workshops is to teach the sections on questionnaire design and quantitative methods. As the workshop is taught over just two days it is difficult to know what exactly to teach and how to present it. If I was teaching such a course in a regular classroom setting over the course of a semester I would probably do a short lecture followed by a two hour 'lab' exercise.

I feel I have quite a strange relationship with quantitative methods. I wasn't great at maths at school, but statistics were to become an inevitable part of my A-levels subjects (business studies in particular) and my undergraduate and master's studies in geography. Few of my fellow students seemed to enjoy these courses, but I quite enjoyed them. I even gained something of reputation for “liking stats”. If my memory serves me correctly my multiple regression model of UK population change was awarded a higher mark than anything else I studied on my master's course.

I remember Tony Moyes, the geography lecturer who taught statistics at Aberystwyth saying that the stats course may turn out to be the most useful thing we ever learn in a geography degree (I'm sure he said it something like that anyway!). Despite using qualitative methods in my PhD thesis, it has turned out to be a very useful skill, especially working a humanities department where few colleagues have experience of questionnaire design or statistical analysis packages. My teacher at the University of Bristol was the late Les Hepple. It was only in a tribute to him that I learnt that Les's A-level background was actually in arts type subjects. However, he not only become an expert user of statistics, but he also contributed to the theory of spatial statistics.

In all honesty, I still find stats quite hard and like everything else in life there is always more to learn. However, I have managed to find some good resources from the US which have given me lots of ideas about the best way to teach quantitative methods. If you would allow me to generalise the broader curriculum in the US means that teaching non-specialists is a normal part of the teacher's job. A couple of years ago whilst browsing my (Canadian) sister-in-law's bookshelves I found an old edition of David Moore's Statistics: Concept and Controversies which struck me as a book which was written for the exact audience I intended it for. His verbal reasoning approach to statistics seemed to be perfect for the humanities audience. Today I came across the lectures of Dan Judge on YouTube. So far I have only watched the first three parts of lecture 1, but I love the approach he takes. It's a lecture by a guy writing on a white board but he makes it so engaging using language to which his audience can relate. I will definitely be recommending these to our workshops participants.