E1n1verse – WoW, Learning, and Teaching by Michelle A. Hoyle

When I last saw J, my supervisor, we were disagreeing about how to do the motivational essay coding for my first World of Warcraft survey.. My plan was to go through the essays first to come up with some themes. Then Basil and I would independently code them for theme. My reasoning was I wanted the coding to be free from subjective bias. If two of us agreed independently, then that would be better than just my assessment of the data. J. thought it was unlikely Basil and I would agree, so she set me the “Great Date Night Experiment.” In this experiment, Basil, my partner, and I would sit down on “date night” and test out my theory on a small scale. Basil would read one essay and summarize the main themes or ideas he thought were represented in the essay. I would independently do the same. Then I would report back to J.

Read the rest of this entry »

General

I know people are very curious about the results of my recent E-Mail in the Cloud: An Open University Survey. Time is a bit short for me, so I decided to write up this quick and dirty post outlining the key result. An analysis of the comments people left about why they made the choice they did will be covered in a later posting, as those comments proved to be extremely interesting.

In a more formal report, the order of detail presented would be different. I’ve started with the results first, as that’s likely to be of interest to most people, and then discussed the methodology, survey deployment, and motivation.

Read the rest of this entry »

I’ve been thinking about the design of the study I want to do on motivation in World of Warcraft. My immediate approach, similar to introductory programming students, was to jump right into the meat of it and start writing survey questions instead of planning. In order to get the data you need in the study, you need to know what questions you want answered. You need to plan. Without knowing that, how can you write survey questions to elicit those answers? So what is it I want to know?

Read the rest of this entry »

I had a good meeting with Thufir on May 14th, lasting almost the full allotted hour. This was because I’ve recently had a breakthrough with my MATLAB analysis and can quantitatively evaluate the similarity between different people or different algorithms with my multi-dimensional scaling (MDS) diagrams. I took some output to the meeting which compared my half-baked algorithm against the cosine normalization version. Both use hypernyms, but how they weigh the hypernyms is different. My automated analysis algorithm also produces an MDS cluster diagram as output for each of the data files provided (see anal1ahyper and anal2ahyper).

Anal1a, in terms of clumping, doesn’t look very good, at least not anymore. That was not previously the case, but I had revised my algorithm to make it symmetrical as per the insructions of a computing statistician here at the University of Sussex. He claimed that the Procrustes Rotation needed symmetric data and my nonsymmetric data, where Doc1 vs Doc2 didn’t have the same similarity as Doc2 vs Doc1, was not going to work. That change has, I believe, altered the efficacy of the algorithm and things are no longer clumped together as promisingly as they were previously. The clumps should be a two- or three-letter short code followed by a digit. Therefore, ac1 and ac2 belong together. Pl1, pl2, and pl3 belong together, and so on. The clumping is significantly better in the already symmetric cosine normalization algorithm (anal2a). The two speech processing documents are clumped together (sp1 and sp2), all of the Power PC and G4 documents are together (pp1, pp2, g4c), and the three Pine Lake tornado stories are clumped far away from everything else (which is all computer-related) and together on their own. Excellent clumping, in fact. So the hypernym hypothesis looks like, on these short documents, it is working well with cosine normalization.

Here’s the final bit of loveliness: comparing one MDS cluster diagram against another. MDS output is mapped to the vector space independently. That is, the same data will produce the same visualization or mapping, but different data is mapped to a different vector space, so you cannot just compare one MDS matrix to another directly. That is where Procrustes Rotation comes in. It applies a series of intelligent matrix transformations, trying to map the second vector matrix onto the source vector matrix. As a side benefit, essential in my case, it always provides a fitness measure to tell you how close the two were. on a scale of 0 to 1. So these two, as you can see (see above image), even after the transformations, were not that close together. As it happens, though, this is not particularly useful information to know. I am currently more interested in assessing how close the two algorithms are to human classifiers.

This recent success gave us plenty to discuss, particularly with respect to metric and non-metric data. The MDS community calls source data metric when the similarity or dissimilarity data is symmetric. That is, the value at row 2, column 1 is the same as the value at row 1, column 2. Classical multi-dimensional scaling (MDS) is designed to only work with metric data. SPSS includes the ALSCAL and PROXSCAL MDS algorithms which can work with non-metric data, but MATLAB’s classical MDS does not because it treats things as Eucledean distances–another reason why I had to alter the Anal1a algorithm. The primary reason I now had metric data for everything, however, was because the computing statistician had told me I needed it for the Procrustes. Hawever, as we were examining my output, it occurred to me that Procrustes did not really care if the data was symmetric, so long as the dimensions of the data were the same (the same number of rows and columns). Which leads us to question whether the application of the method is statistically sensible or not. To that end, I need to track down a new computing statistician and perhaps a mathematician and discuss the process with them. My original computing statistician has retired.

Earlier I said that comparing one machine to another, to see how they fit is not useful information, but what would be interesting is to prepare a matrix of all the possible combinations of human judgements, cosine normalization, and weird formula:

cosine   wrd form.   human
cosine (anal2a)		x
weird formula (anal1a)           x
human                                        x

So that is my task for my next meeting (on the 16th of June). Before then, I need to figure out how to get MATLAB to take multiple tables as data. In SPSS, I could paste in several tables (representing all of the people’s individual data, for example) and it would work with that. That is necessary in order to aggregate the peopel to do the comparison. Onward ho, then! Progress at last!

I have to confess to feeling a bit stupid. I have been struggling with MATLAB for weeks now, trying to get it to read in my data files so I can automate my analyses. My data is in a tab-delimited file and looks something like:

Doc1	Doc2	Doc3	Doc4
Doc1	100	76	18	91
Doc2	76	100	22	35
Doc3	18	22	100	65
Doc4	91	34	65	100

This is not too dissimilar from the labelled diagram, part of the MATLAB documentation on data importing. Except that, if you look at the table below it, which describes which functions to use, they don’t have a function with a similar example to their labelled diagram. Early on I thought I should be able to use dlmread, which allows you specify rows/columns for starting points or a range. My idea was just to have a range which excluded the non-numeric troublesome labels. No matter what I did, though, I could not get it to work. It was frustrating, because I could paste the data into the Import Wizard and that could handle the data fine. I wrote people, I researched on the web, and I tried all sorts of things.

Eventually, I came full-circle back to dlmread and experimented by making a small data file with unrelated data in it. That worked fine. So I then copied half of one of my data tables into the test file and tried that. That also worked fine. I copied the whole data table into the test file and used dlmread on it. It worked fine! What was the difference between the two identical data files other than their filenames? When I uncovered the answer to that, I kicked myself. My data files were generated years ago and stored on my Mac OS 9-based laptop. My laptop and the data have since migrated to Apple’s swoopy BSD-based UNIX goodness and that’s the environment that MATLAB runs under. So… Have you guessed the problem? Yes, it was linefeeds! The data files had original Mac linefeeds and MATLAB wanted UNIX linefeeds. D’oh! It just goes to reaffirm that the things you don’t see can really hurt you.

Read the rest of this entry »

By the end of April, I seemed to be stymied in my quest for an affordable copy of MATLAB that I could run on my local laptop. The university maintains a license pool for the base software plus toolkits. I do have access to run a copy from the license server when I’m on campus. It actually worked off of campus too, when I’d been told it shouldn’t, but that turned out to be a mistake. When I reported it, the firewall was closed to the outside world for requests for the license server. That’s where honesty gets you: no MATLAB accessibility from anywhere with an Internet connection.
I was hoping to snag one of the concurrent licenses for my permanent use and offered even to buy an additional one for that purpose as that would be cheaper. I was told that I couldn’t have one and I should investigate the student version of the software. Unfortunately, the student version is only available to students in taught courses, not Ph.D. research students, so that wasn’t any good. I mentioned to Thufir that I’d been turned down, reportedly by the head of software/hardware procurement within our department. Thufir promised to see what he could do. Then, last week, I received an e-mail last week from the lab manager, the person in charge of the procurement. He offered, if somebody would pay for it, to install an academic version of the software on my university-owned equipment for only £525 (~930 US/780 €). That was just for the base software and not also for the toolkit I need.

Read the rest of this entry »

I need some help in using MATLAB and MDS, so I looked to Google to find resources. There seem to be more MDS resources than when I last looked quite some time ago. I found a useful page with links and pointers to MDS-related resources at http://www.granular.com/MDS/. From there, I obtained most of the resources for a pyschology course organized around MDS taught by one of the MDS’s primary researchers Forrest Young. I downloaded all the notes in PDF format and stored them away to browse through. Young is the same researcher responsible for developing the ViSta software (Visual Statistics System), which looks a lot like that Canadian object-oriented, icon-based programming language. I remember looking at ViSta before, but I don’t think it supported doing things like MDS and it hasn’t been recently updated for anything other than Windows.

Read the rest of this entry »

Played around a little more with MATAB, although I didn’t really get very far. I was trying to figure out how to use dlmread to import my data files properly into MATLAB automatically. It didn’t really seem to want to fly somehow.
I spent some time trying to figure out how to automate my MATLAB work. I couldn’t figure out how to get it to import the tab-delimited data files that my programs had produced using the dlmread command which should put it into a matrix. I can do it via the clipboard but not anyway else at the moment. Reviewing various MATLAB tutorials on the web, looking for hints, I realized that I need to review my understanding and knowledge of matrices again, so I’ve added this to my list of things to be done.
Assuming that my methodology is correct, which it might not be, I can now map one MDS cluttering onto another using the Procrustes in MATLAB. I need to figure out now how to get a single measure of how different the two are from that. I’m at least further along than before, so that’s promising.

Read the rest of this entry »