WHEN people in London posted on Twitter about going to church in the spring of 2014, they probably didn’t expect their 140-character missive to receive a great deal of attention, let alone detailed analysis concerning its content, message, and location.
They could not have known that researchers at the University of Turku, Finland, would be painstakingly conducting such a study, and eventually concluding that their approach should be taken up by others for clues about churchgoing.
The results were published this month in Exploring the Use of Twitter Data to Better Understand Church Attendance, a doctoral thesis by Anthony-Paul Cooper. Currently co-director of the Centre for Church Growth Research at the University of Durham, Dr Cooper writes of being conscious of “fundamental shortcomings in the available data on church attendance”, which are “incomplete, inaccurate and contradictory”. He was curious whether Twitter data could help to address this.
His study entailed analysing posts containing the word “church” written on Sundays over a ten-week period (beginning on Easter Day 2014), across all London boroughs. (Among the limitations acknowledged is that “it is not evident how truly random the resultant sample of tweets actually is.”)
Only tweets containing geodata (i.e. the location where the tweet was posted) were retained for analysis, giving a sample of 1004 tweets, posted by 723 unique users. These were coded according to topic, including church attendance and discussion about Bible, belief, or theology, and given a “sentiment rating” — indicating whether it was positive or negative. The number of posts about Hillsong at the time resulted in this topic becoming a coding category in itself.
The “mean average tweet net sentiment score” for each London borough was plotted against existing data on whether church attendance in the borough was growing, declining, or stagnant (based on Peter Brierley’s 2012 London Church Census). The study concludes that there is “a statistically significant relationship between the sentiment of church-related tweets and the presence of church growth in the borough from which the tweets were posted”. The tweets “tended to be modestly positive and negative”, with the mean average net sentiment score in all but two London boroughs positive.
Dr Cooper also looked at whether geolocated Twitter data could be used to identify church locations and “uncover previously hidden churches”. In total, it proved possible to identify the location of 42 churches from the data set, nine of which were determined not to have been included in Mr Brierley’s census. This demonstrates, the study argues, that “it is possible to use freely available Twitter data to improve the sampling base of church attendance studies, to improve the accuracy and quality of church attendance data going forward.”
Among the recommendations that Dr Cooper offers is that church leaders should “study and explore social-media discourse and content related to their churches. . . Where churches are seeing attendance decline or stagnation, there might be potential for those church leaders to understand more about their congregations (or potential congregants) and get a sense of some of the attitudes potentially driving that observation, thus permitting the opportunity to act in response.”
Dr Cooper will take up a role as a postdoctoral researcher at the University of Turku next year. Asked this week whether he thought that leaders might feel a degree of trepidation in undertaking such a search, he suggested that this might increase as “controversial” issues became more prominent in the public square. His thesis suggests that it might be valuable to repeat the study “to investigate whether the types of topics discussed in church-related tweets change as church thinking of issues of sexuality and gender progress and receive increased public and media attention”.
Read the study here: https://www.utupub.fi/bitstream/handle/10024/152714/Annales%20F%206%20Cooper%20DISS%20Electronic.pdf?sequence=1&isAllowed=y
