Page 121 -

5.2 Benefits and drawbacks of surveys 107

method compared to asking users to recall how many times they completed a task.
In that type of situation, a combination of computer-collected data and a user survey
might make the most sense (see the sidebar on photo tagging and sharing). It is also
possible that the individuals who you are most interested in studying may come from
a culture that is more oriented toward oral (spoken) approaches to communication
than written approaches. If that is the case, then interviews or ethnography might be
a more appropriate research method than a survey.

RESEARCHING PHOTO TAGGING AND SHARING BEHAVIORS

Two separate studies researching photo tagging and photo sharing behaviors
illustrate how a combination of a survey and computer-collected data could be
performed in a research study.
Nov et al. (2008) were interested in learning more about tagging behavior
on Flickr (a website on which people can post pictures and notes [“tags”] about
those pictures). The researchers were aware that it would not make sense to ask
users how many tags they had created in a certain time period, as their responses
were likely to be only a guess or an estimate, not an accurate count. However,
there were a number of research questions that could best be investigated using a
survey, so a combination of a survey and data logging, was used.
The researchers contacted a random sample of Flickr users who had
posted at least five unique tags on pictures, in English (although this might
have limited the sample to certain nationalities, the researchers wanted
to make sure that the respondents understood the survey questions). A
random sample of 1373 users was selected and e-mailed with an invitation
to participate in a survey. At the end of the survey, respondents were asked
to authorize the researchers to access data about tagging from their Flickr
account (if the user gives permission, Flickr allows access to data from a
user's account). Once the respondents filled out the survey and authorized
access to their account data, the researchers were able to collect data on
the number of tags. There were 237 valid survey responses and the average
respondent had used 370 tags.
In a separate study related to Flickr usage published 8 years later, Kairam
et al. (2016) were interested in studying how users chose to share photos.
The researchers recruited participants who were already active Flickr users,
to respond to a survey, requesting permission to access their photos as a part
of the study (and the participants were paid a few dollars through Amazon's
Mechanical Turk). There were 96 respondents to the survey. As a part of the
research, 20 photos were selected from each participant's account, stratified
across the five possible privacy settings for each picture. Of those 20 photos,
10 photos were randomly selected to be presented to the participants during
the survey, with questions related to the privacy and content of the picture.

(Continued)