New Article on the Analysis of Crowdsourced Visual Survey Data

New websites such as StreetSeen at The Ohio State University or Place Pulse at the MIT Media Lab use crowdsourcing to analyze urban landscapes by showing visitors a large number of paired Google Street View images, and asking them to pick the image which rates higher on the characteristic of interest (such as beauty, liveliness, walkability, etc). In a Research Note published in the journal Landscape and Urban Planning titled “An Evaluation of the Elo Algorithm for Pairwise Visual Assessment Surveys,” I propose using the Elo algorithm to analyze the resulting data, and use a dataset from a survey conducted in Philadelphia to compare the results of different analysis methods.

The visual characteristics of cities are an important ingredient of urban quality of life, and often determine the success or failure of urban neighborhoods, but they can seem subjective and difficult to measure. The long history of visual survey methods show they can be rigorously analyzed, and these new online survey tools provide researchers and professionals alike new ways to conduct surveys which allow communities to consider visual perceptions of their cities. My paper shows that the most successful forms of crowdsourcing requires not only drawing on previous research but also developing new methods of data analysis.

The link above will provide free access to this article until August 14, 2016, after which it will be available from libraries with a subscription to the journal Landscape and Urban Planning.

Author: Rob Goodspeed