Home / posts / Blog / Can we generate value profiles from policy documents?

Can we generate value profiles from policy documents?

Oct 11, 2023 | Blog

By Richard Gunton (Oct 2023)

What kinds of value do policies seek to deliver?  A trivial answer would be “value for money”, but this doesn’t say anything about the kinds of value that we seek to maximise with respect to expenditure.  For a financial investment strategy, the answer is simply “return on investment”, meaning maximum monetary value at a future point in time.  But in real economics – or the real world more generally – there are no such easy answers.


In some cases, the remit or mission of an organisation constrains the kind of value that must be sought.  The main aim of the Red Cross is to save lives in crises; the Department for Education must address learners’ achievement.  But this still leaves much room for different kinds of goodness, and in any case, it’s important to take account of side-effects and unintended consequences.  In order to compare policies within and across sectors of government, business and civil society, it would be useful to have a way to create value profiles: standard summaries of the kinds of goodness that an initiative aims to promote.  This is one goal of the next phase of development of the Pluralistic Evaluation Framework (PEF), which I’ve written about here before.


The PEF is based on eleven categories of value, starting with biotic value (e.g. health and fitness), then sensitive or psychological value (safety, comfort), then continuing through analytical value (consistency, diversity), formative value (heritage, novelty, freedom), lingual value (informativeness, clarity), social value (friendliness, appropriateness), economic value (productivity, efficiency), aesthetic value (beauty, harmony), jural value (fairness, equitability), ethical value (care, cherishing) and finally ultimate value (worthiness, sacredness).  Each of these categories covers a range of meanings, but together they provide a refraction of “goodness” into distinct categories.  In workshops, we have found this approach quite robust, at least when sufficient context is given.  For example, consider the goods of “resilience”, “biodiversity” and “equality”.  “Resilience” needs further context for us to understand why it is good: a resilient pathogen might not be good at all, but psychological or economic resilience generally are.  Biodiversity tends to mean whatever ecological phenomena we want to conserve, but probing might reveal sensitive, analytical or aesthetic kinds of value.  And while equality needs some context to understand its goodness (e.g. social, economic or legal equality?), it generally implies a jural good.  Consonances can be seen with Maslow’s Hierarchy of Needs, Stephen Kellert’s modes of biophilic value and Philip Phenix’s Realms of Meaning (see here for a systematic comparison).


What happens when we pass a policy document through these categories to try and obtain a value profile?  At present, this has to be done using an intuitive approach, in the absence of an accepted thesaurus of value using these categories.  However, gathering and inspecting value terms from workshops with policymakers and academics as preparation for our journal article (with support from CECAN) has led to a partial thesaurus, enough to provide a starting point.  I looked at the UK Government’s “25-year plan” for improving the environment, and compared the text of the forewords and executive summary against the main body text.  The results are shown in the chart below.

Although these are preliminary results, they indicate the idea – and show one striking pattern.  Economic forms of value are most prominent, but far more so in the summary (the forewords and executive summary taken together) than the main body of the document.  After the summary material, other forms of value become more visible, with references to social and ultimate goodness making an appearance.  Of course, the summary material had far fewer words: 1,872 compared to the 42,236 words of the main body, and this gives a statistical reason for the omission of some value categories.  But the shift in emphasis on economic goods is striking.


The next task is to develop a more objective way of detecting value-laden words and assigning them to the value categories.  This should be possible using recent advances in language models – as I plan to do with colleagues in a forthcoming project.  Profiles have been explored for people’s held values, but as Andrew Darnton says, focusing on forms of value (goodness) rather than people’s values is much more conducive to collaboration.  If you’d like to get involved or keep an eye on progress, why not join the PEF LinkedIn group?


Share This