4 bitesize report summaries on data in governance: June 2021

1. Combining racial groups in data analysis can mask important differences in communities

Source: Urban Institute, read the report here.

The idea: Race is a nebulous concept. When we try to boil it down for data analysis, we can end up ignoring serious differences within groups that share a label. This can have serious consequences for “how government funds are distributed, how services are provided, and how groups are perceived."

Key takeaway: Don’t throw the baby (here, data) out with the bathwater, but be very mindful about the appropriate level of precision needed, the variance within labels, and to scan for the harmful consequences of ignoring subtleties.

Quote: "Within the American Indian or Alaska Native category, poverty rates vary from 5.9 percent for the Aleut (an Indigenous community primarily living in Alaska) to 36.9 percent for those who identify as part of the Sioux Native American tribe."

2. Public sector innovation in Kansas' school system

Source: OECD Observatory on Public Sector Innovation. Read the full report here.

The idea: The Kansas school system created a 'database' of 2,500 stories and made them available to decision-makers and the individuals who shared their narratives. The “community was called upon to spot patterns and make sense of the collection”.

Key takeaway: Bottom-up qualitative insights can be collected at scale — including by citizen journalists — and interpreted to support system-level decision-making.

Quote: "From a complexity perspective, these goals translate to developing a ‘human sensor network,’ embedding citizen feedback loops and sensemaking processes into governance, and complexity-informed intervention via portfolios of safe-to-fail probes."

3. How can we measure productivity in the public sector?

Source: World Bank, read the full report here.

The idea: This article summarises a World Bank report on productivity in the public sector. It's a tough question as public sector work isn't guided by commercial nous, swift return on investments, or easily measurable outputs.

Instead, it recommends:

Complementing traditional ’macro’ measures of public-sector productivity with fine-grained '‘micro’ measures at the individual employee and organization level. Monitoring and reporting output (performance) measures and inputs (costs) separately. Combining multiple measures of productivity tied closely to a conceptual service-delivery chain.

Key takeaway: Their website has a cool feature that makes it possible to Tweet quotable stuff.

Quote: “According to the Worldwide Bureaucracy Indicators, globally the public sector accounts for around 25% of GDP and 38% of formal employment.”

4. What data can’t do

Source: The New Yorker, read the full article here.

The idea: Once a useful number becomes a measure of success, it ceases to be a useful number. Reality is notably unruly and doesn’t like being measured (even if it matters). This observer effect is starting to haunt AI design.

__Key takeaway: __Numbers are better than gut instinct alone — but to be data-driven alone is really quite silly.

Quote: This whole article is gorgeously quotable. Here’s a neat one: “Whenever you try to force the real world to do something that can be counted, unintended consequences abound." – Apolitical Content Team

__We want to hear what you think about this article. Pitch an article to us or send your comments to discussion@apolitical.co