Understanding Data and Data Analysis

Data is never raw: every figure you are handed has been shaped by human choices about what to ask, collect, classify and read. Drawing on Lisa Gitelman's argument that raw data is an oxymoron and Cheryl Strauss Einhorn's three data types and their biases, this module follows the six-step data journey, three lenses for reading data and five questions to ask of any number before you act on it.

Too often an organisation reaches for a survey before anyone has agreed what decision the data should inform. This module sets out six steps from decision to method, Joseph McGrath's three-horned dilemma of reach, objectivity and detail, and two Welsh cases: keeping a census in 2031 and two official sources that disagree on Welsh speakers. It closes with five tests for good collection, including John Creswell's mixed methods.

A number is only as good as the way it was made, and most quality is decided long before a figure reaches a board paper. This module applies William Edwards Deming's standard for usable data, sets out the major types of data and the sums each allows, and uses the Labour Force Survey case to show how falling response erodes trust. It also explains the Code of Practice for Statistics and its two labels.

Most misreading of data is not deliberate; it comes from ordinary habits such as trusting an average or reacting to the latest month. This module looks at what averages hide, Walter Shewhart and W. Edwards Deming on common and special cause variation, four reasons things move together, four ways a sample misleads, five chart traps and Gerd Gigerenzer's natural frequencies, ending with a short check before acting on evidence.

Capable, well-intentioned people often read the same figure in quite different ways. This module draws on Daniel Kahneman, Olivier Sibony and Cass Sunstein's work on noise, Dan Kahan's motivated numeracy study, Kahneman's fast and slow thinking and framing effects to explain why, and sets out five lenses every reader brings. It offers six steps of decision hygiene so that a team can read data well together.

Numbers do not speak for themselves; people speak for them, and meaning is settled between people through definitions, categories and stories. This module draws on Berger and Luckmann, Theodore Porter, Espeland and Stevens, Bowker and Star, Lisa Gitelman and Karl Weick, and uses the contrasting ways England and Wales count rough sleeping to show how much choice sits inside an apparently simple figure, and how to make meaning well.

A formula looks neutral, yet every part of it was chosen: its proxies, weights and thresholds. This module uses the Resource Allocation Working Party formula for NHS funding, the NICE cost per QALY threshold and its April 2026 change, Daniel Yankelovich's account of the McNamara fallacy and the 2020 A Level grading algorithm to show why you should use the formula and own the decision.

Every measure changes the behaviour it set out to see. This module draws on Robert Merton's causes of unintended consequences, Goodhart's and Campbell's laws, and Gwyn Bevan and Christopher Hood's three kinds of gaming, then examines the four-hour A&E target in England and a Welsh home care example from Cartrefi Cymru. It closes with the fixes that fail archetype and a process for designing measures for learning rather than control.

Data describes the past, yet leaders must decide about futures that cannot simply be forecast. This module sets sensemaking beside the conventional gather, analyse, decide model, drawing on Karl Weick's seven properties of sensemaking and Dave Snowden's Cynefin framework to show what data can do in each domain. It then uses Jonathan Rosenhead's trumpet of uncertainty and robustness analysis to show why keeping good options open has value.

Every figure a public service leader sees was collected by someone, about someone, to answer someone's question. This module looks at power, voice and fairness in data, using five questions about power, the seven principles of Data Feminism from Catherine D'Ignazio and Lauren Klein, and the FAIR and CARE principles. It closes with how lived experience and participatory practice, including Camden Council's resident Data Charter, change what data means.

The same data is read differently by different people, and the meaning a group settles on depends on how it reads the data together. This module sets out practical methods for collective sensemaking: the Urban Institute's data walk, SenseMaker from Dave Snowden, the Delphi method, data parties and Peter Checkland's Soft Systems Methodology, with a six-step cycle for designing a session and a guide to choosing between methods.

Stories, interviews, complaints and survey comments are often the richest evidence an organisation holds, yet few leaders are helped to judge them. This module shows how to weigh narrative evidence rigorously, using Lincoln and Guba's trustworthiness criteria, Most Significant Change from Rick Davies and Jess Dart, Denzin's forms of triangulation and John Creswell's mixed methods designs, illustrated by the invented Tir Afon Housing case.

How do you tell whether something worked when many services and forces act on the same outcome at once? This module sets out Pawson and Tilley's realist evaluation, John Mayne's contribution analysis, the Human Learning Systems approach from Toby Lowe and colleagues, and Michael Quinn Patton's developmental evaluation, with the Plymouth Alliance for Complex Needs as an example. It helps you make, and challenge, honest claims about impact.

Most performance data reaches boards as a dashboard, and a dashboard is for decisions, not display. This module covers lag and lead indicators, Mark Friedman's Results-Based Accountability and its three questions, Audit Wales's review of how outcome information reaches Welsh council leaders, five features of a good dashboard, choosing honest charts and Cole Nussbaumer Knaflic's Storytelling with Data, closing with five questions any board can ask.

Many numbers that shape your decisions come from official statisticians, regulators and national surveys rather than your own organisation. This module maps who produces public data in Wales and the UK, from the ONS and StatsWales to Data Cymru, traces the move towards measuring well-being, reads the Wellbeing of Wales report, and shows why measures such as Tenant Satisfaction Measures differ across the border.

Data is made by people, so its quality depends on how people feel about producing, sharing and hearing it. This module covers an Avanti view of data maturity, five features of a healthy data culture, Donella Meadows on information flows, the Government Digital Service's Data and AI Ethics Framework, and how bias scales inside automated systems, including the DWP's own fairness analysis of its Universal Credit advances model.

Capable, well-intentioned people can look at the same figures and see very different things. This capstone examines four real cases: the Challenger disaster, read through Edward Tufte, the engineers' reply and Diane Vaughan's normalisation of deviance; the Post Office Horizon scandal; the Mid Staffordshire mortality data; and the Covid-19 R number. It draws the course together in six habits of a data-wise leader.