using social media research methods to identify hidden...
TRANSCRIPT
Using social media research methods to identify hidden
churches
Anthony-Paul Cooper
CofE Faith in Research Conference 2014
4th June 2014
Overview
• The research described in this presentation is part of the London Church Attendance Baseline study.
• End result of the study will be a tagged map of all known churches in London, from a range of sources (This data will also be available in list format!).
• For the sake of this presentation, we’re just interested in how social media research methods can be used to identify previously undocumented churches.
• Important note: This presentation features snapshots of real data (captured on 20th April 2014). As such, despite efforts to redact offensive language, some offensive content/themes may still remain.
Twitter: What will be done?
• Use the Twitter API to pull all Tweets created within London which contain the word “church” every Sunday (0001 – 2359) for a 6 month period between April and September 2014.
• Where this methodology identifies churches not already known about, add to the online map which will form the new baseline.
Twitter: How will this be done?
• An API search term will be used.
• This term takes a starting point within London (latitude: 51.5117, longitude: 0.1275) and then gathers all Tweets containing the word church from within 60Km of this starting point.
• This will generate some results from outside the M25 area (as 60Km is a deliberately generous radius), but this will become apparent when we come to add results to online map.
Twitter: What does this search area look like?
• The starting point and 60Km radius gives the following coverage area:
Twitter: What will the search results look like?
• Results are initially output into a large .txt file:
Twitter: What will the search results look like?
• These can easily be moved to a .xls spreadsheet. From here the data can be cleaned. Headings can be added, formatting can be tidied and any data from days not being considered can be discarded:
Twitter: What will the search results look like?
• Many results will contain no geo data, so these data can also be discarded:
Twitter: How can any sense be made of the data?
• The data we are interested in is:
- Posted on the date in question (i.e. a Sunday between 0001 and 2359)- Contains the word “Church”- Contains geo data
• Making sense of this subset of data requires manual coding. I have chosen to code as follows:
- Rose: Tweets with no reason to believe posted from a church- Tan: Tweets which may or may not have been posted from a church- Light Green: Tweets believed to be posted from a church
Twitter: How can any sense be made of the data?
Twitter: How can any sense be made of the data?
• The Tweets which were posted within a church can be gathered together, and the geo data for these Tweets can be cleaned, ready for input onto the online map.
Twitter: How useful is the data?Some coded as Tweets with no reason to believe posted from a church:
• “I miss going to church u know”• “Church in the AM.”
Some coded as Tweets which may or may not have been posted from a church:
• “i love church ”❤• “We decided to go to a less modern church for Easter. While I would have loved to go to #Hillsong…”
Some coded as Tweets believed to be posted from a church:
• “Easter Sunday Service! @ Hillsong Church London”• “I'm at Our Lady and St Joseph's Catholic Church”
Geo Mapping (During last 3 months of Tweet gathering)
• Create a version of the online map which includes only churches we already know of, and allow users to tag their own churches.
The Future: Longitudinal Study using Twitter Mini-Reports
• Commence longitudinal attendance reporting over a 12 month period, with willing volunteer churches from the new baseline.
• Use Twitter as the submission tool for this data (e.g. “Westminster Abbey. Adults = 500, Children = 180 #LondonChurchStudy”)
Any Questions?
For any questions following the conference, please email: [email protected]