“Other charities and leaders in the sector are slowly warming up to the potential of mining big data and analysing it to identify social trends and solve social problems of the day” (Social Sector Mining Data to Solve Problems, Janice Tai).
That charities and leaders in the social service sector “are slowly warming up to the potential of mining big data and analysing it to identify social trends and solve social problems of the day” (ST, Jul. 24) is encouraging, and in particular needs analysis (understanding, for instance, the profile of the beneficiaries who require assistance) and programme evaluation (ascertaining the extent to which a project has been effective or successful, against benchmarks) should take centre stage. In fact, the intermediate steps preceding big data and analytics – that is, the collection, management, and usage of data in general – are equally critical too. And in this vein there are at least three ways through which such progress can be furthered.
First, greater programme experimentation and evaluation – of services delivered by the non-profit organisations (NPOs) – should be encouraged. In other words, through pilots with smaller groups at the start, they should design their own programmes with research-supported hypothesis on the potential benefits. Statistically, randomised controlled trials (RCTs) provide these organisations with data to make causal conclusions, that their programmes or services have resulted (or not) in stipulated outcomes. Few NPOs at the moment, if any, embark on these experiments or even quasi-experiments, and as a result most only have output or observational data showing correlations, not causal effects.
Related to this is the second proposal to diversify data collection methods and data sources. Research within the social service sector in Singapore rarely go beyond perception studies, which are oftentimes plagued by problems of representativeness and survey or self-response biases. In the report, the National Council of Social Service said its dashboard project will appraise the demand for social services by “quality of life measures [and] survey of public attitudes towards certain social issues” – again, premised upon perceptions, not experimental data – yet remained vague about the “other needs assessment”. Big data and analytics start with the data itself and its quality, and therefore methods for its collection cannot go unnoticed.
And finally, the government potentially houses much data that NPOs will find useful, and it is not clear the extent or the frequency to which such data is shared. The report cited the official data site data.gov.sg, yet the data is often not granular enough, and data that is directly relevant to the work of NPOs appear hard to come by. The Charity Portal, hosted by the Ministry of Culture, Community, and Youth, provides useful financial reports and governance details of all registered charities in Singapore, but the data is neither collated nor available in machine-readable format. Of course, sector collaboration and a greater willingness of organisations to share data will be progressively productive, though the government can provide that head-start.