Visual preference survey comments analysis tool, 2/16/2019

Posted on February 16, 2019

At the 2/13/2019 Planning Commission (PC) meeting, it was clear that the PC had expected Town staff to analyze, or at least categorize, the comments that were part of the Town’s Visual Preference survey.  E.g., how many people mentioned parking as an issue, that sort of thing.  That appeared to be part of the original plan.  But Town staff dismissed the idea.  Just won’t do it, plain and simple.

So I will.

Let me put aside all the things I have said are wrong about this visual preference survey approach, and put in the time to rectify this.  Why am I bothering?   The Town won’t do a simple straightforward survey similar to what I did, to get a proper estimate of citizen sentiment about MAC.  Barring that, these comments are the best source we have for a broad-based sampling of opinion, about MAC, among Vienna citizens.  It’s wasteful not to analyze them.  Particularly as it only takes a couple of hours to set this up so that anyone can ask simple questions about those comments.

The Excel workbook linked below is my comment analysis tool.  It does the simple-minded thing that you would expect.  Type in TRAFFIC, and it will count the number of comments that mention traffic.  Sort the results, and you have all the TRAFFIC comments in one place.  A second sheet lets you type in several related words, so you can count and find comments with (e.g.) TRAFFIC or CONGESTION or CARS or VEHICLES.

Not rocket science.  And not ideal.  But, surely good enough to start pulling some interesting information out of the otherwise wasted data.

Separately, as long as I was going to the trouble to do this, I read all the comments and made my subjective judgment as to whether the commenter was pro-MAC, anti-MAC, or offered no clear opinion.  That’s the final sheet in the workbook.

So, just to give the flavor of this, I tried a few words that I thought might be fairly common.  These are the percent of comments with these words mentioned:

To me, this is how the data should be presented.  But you may, if you wish, think of this as just a very ugly word cloud.  I can’t really get information out of something like this:

This is similar to the key results of my random-sample survey.  People cared about traffic, and green space.  They didn’t much care about infrastructure and taxes.

Of course, there will be some mis-categorization when flagging by a single word.  And if you want to flag all comments with synonyms of a word (bike or bicycle or cycling or bicycling) I have set that a separate sheet within the workbook for that.

Other concepts can’t readily be captured by simple word searches.  Of particular interest was, of course, how many people were for MAC, and how many were against.  For that estimate, I read every comment and made a judgment.  I may have done this in the context of knowing too much, in the sense that if somebody said they’d never accept buildings over three stories on Maple, then, as currently configured, they are anti-MAC.  But, if anything, I bent over backwards to flag pro-MAC, and to avoid flagging anti-MAC comments.  The vast majority of comments were neither clearly pro- nor clearly anti-, but were comments about what they would like to see, or like to see changed, in the MAC zone.

The main point is that Town really should not use the high participation rate as any indication of support for MAC.  If anything, it demonstrates how motivated people are to express an opinion against MAC development.  Of those who commented, and expressed an opinion clearly enough that I thought I could differentiate, anti-MAC sentiment outweighed pro-MAC sentiment by 2.5 to 1.

This is actually considerably more anti-MAC than I found in my random-sample survey, and suggests a first general caution about these results:  this is not a random sample by any means.  The entire visual preference survey was a non-random sample, and the comments consist of a further non-random sample with in that.  (Only about 40% of respondents offered a comment).  It’s possible that anger is the stronger motivator here, and that commentors are skewed toward anti-MAC sentiment, compared to a true cross-section of Vienna citizen opinion.  But that’s just a guess.  If the Town won’t do a true random-sample survey and ask people, we’ll never know.

The second general caution is that these people were, by and large, responding to the survey.  So Town Staff set the agenda, and items that were mentioned in the survey (e.g., sidwalks) will of course be heavily mentioned in the comments.  One-third of the survey was more-or-less about sidewalks, so … no big surprise, a lot of the comments were about sidewalks.  It is simply incorrect to interpret this as showing that Vienna citizens are (or are not) obsessed about sidewalks.  By its very nature, this type of survey can’t show that. When it comes to words that were integral to the survey — such as sidewalk — you’d be crazily incorrect if you interpreted the frequency-of-mention data as any indication of true level of interest of Vienna citizens.

So, caveat emptor.  Just be aware that if the survey focused on it, it will appear disproportionately in the comments.

The Excel spreadsheet should download if you click the link below.  Be aware that Excel will search for the exact set of characters that you enter.  So a simple search for “TREE” will find every instances of TREE or TREES, but also every instance of STREET.  And spaces are characters:  Searching for TREE_ (where _ represents a space) will count TREE, but not TREE., TREES, STREET.  You may need to solve a little logic problem to get the exact search you want.

Comment analysis tool v1, 2-16-2019