Software Problems That Motivate Web Searches - Finding Source Code on the Web for Remix and Reuse - page 253

13.3.2 Procedure
We shadowed each developer for 1 day of work. We took notes about the activi-
ties developers were performing and time stamped switches between activities. We
paid particular attention to Web searches. After they performed a search, we asked
some questions about the goal of the search, expectations for the search, how the
candidates were evaluated, and the use of the information.
At the end of the day of observations, the researcher conducted a short debriefing
interview. The participant was asked to reflect on the activities performed and the
patterns of searches on the Web.
Because the searches were conducted so quickly, we developed an extension
for the Chrome Web browser to collect data. The extension automatically recorded
searches as well as the results developers visited after each search. Data collected
included the search engine used, terms in the query, and time. For visited results,
the extension recorded: the number of result page, the rank of the visited result, the
time of visit, the title, and the URL. A subset of participants in the US (10) installed
and used the Chrome extension the day they were observed.
We analyzed the data inductively and iteratively [ 9 ]. We used open coding to
identify categories, sometimes revisiting data to apply new categories. We used ax-
ial coding to relate different categories to each other to create descriptions. The
names of participants that we use in this document are pseudonyms to protect their
confidentiality.
13.3.3 Participants
Tab le 13.1 shows a summary of our participants and companies. We observed 12
developers at Novatronic, 11 men and 1 female. Eleven participants completed the
background questionnaire. They were between 23 and 38 years old and had 1-
15 years of programming experience. They used the following programming lan-
guages at work: Java, C/C++, SQL, Visual Basic, and JavaScript. They used these
IDEs: Netbeans, Eclipse, Visual Studio, and Notepad++ (for C). Five participants
indicated that they looked for source code on the Web a couple of times per week,
four almost every day, one several times per day, and one less than once per month.
We observed 11 software developers, all men, from Health Connection. Devel-
opers were between 23 and 36 years old with 2.5-13 years of programming expe-
rience. They used the following programming languages at work: Java, SQL, JSP,
and JavaScript. The IDEs that they used were Netbeans, Intellij IDEA, and Eclipse.
Five participants indicated that they look for source code on the Web almost every
day, three a couple of times per week, and three several times per day.
One participant was from AppFolio. He had 2.5 years of work experience, and
uses TextMate to code in Ruby. He indicated that he looked for source code on the
Web several times per day.