chapter 7 Evidencing Culture for Policy
Well-being data and investment in culture
For now, let’s look at some well-being data to observe the relationship between culture and well-being. To be specific, we are not going to look at the concept of culture as a whole, or, as is normally investigated, the concept of participating in culture (in some way). Instead we are going to look at the money spent on culture. If advocacy for policy spending on culture is based on its positive impact on well-being, this implies that increased investment in culture is assumed to improve well-being. If this is the case, then this should be visible in some data, right? New Labour claimed a 90% increase in expenditure (in real terms) in its so-called cultural manifesto, ‘Creative Britain’1. You would maybe expect to be able to see a relationship between increased investment in cultural infrastructure and improved well-being as a result. You might also expect to see this demonstrated through statistics, whether they come from administrative data or from national-level surveys. Can we see this relationship in the data? How might we check?
We do not necessarily even need to find administrative data to answer the question ‘Did increased spending result in increased well-being?’ We can find sources that tell us about well-being over time and spending over time. The increase in spending is described in a number of other literatures, and specified in some as well, including Hesmondhalgh et al.2:
New Labour increased central government grants to local government from £82 billion in 1999 to £173 billion in 2010 (UK Public Spending website). This enabled local government to invest, particularly in ‘cultural infrastructure’ such as refurbished or completely new galleries and concert halls.
So, this means we could use the numbers published elsewhere, and simply consult well-being data, or literature, to see whether the investment identified by Hesmondhalgh and his co-authors affected well-being. However, the reference we have here indicates a credible data source for data on cultural investment, so we can use data from the UK Public Spending website and the data on well-being that would be most appropriate.
Box 7.2 Primary, Secondary and Tertiary Data
Recall from Chap. 3 that…
Primary data are collected by you or a project you are working
on. In Chap. 3 we used the example of a questionnaire outside a music event in a local park.
Secondary data refer to data collected by someone else or another organisation that is made available at individual level. They will almost always be either anonymous or de-identified.((There are two helpful explanations on how data are anonymous, depersonalised or de-identifed. One is here from the Future of Privacy Forum (2017). A simpler example is available from Understanding Patient Data (n.d.).)) They are usually quantitative data but can be qualitative. In Chap. 3, I discussed reanalysing qualitative data from the Measuring National Well-being debate that was collected by the ONS.
Tertiary data consist of summaries of primary or secondary data, often called headline data. If you go to the ONS’ well-being pages (n.d.), you will find headline statistics, so you do not have to do the maths yourself.
Should you want or need to find data yourself, I am sure the idea of it can feel daunting, and for many reasons. I try to tackle the most obvious ones to me in Box 7.3.
Box 7.3 Concerns with Finding Appropriate Data
1) Where to look?: The UK Public Spending website offers figures for year-on-year spending (tertiary data) that is a good place to start. It can be difficult to have faith in your ability to find the right data, but you can always begin by referring to how someone else has gone about it. In our case, we have started with Hesmondhalgh et al. (2015).
2) Suitability: There are various funding streams that subsidise ‘culture’, so what are you looking for?((For discussion on these various streams, see Hesmondhalgh et al. (2015). For further discussion on how increased National Lottery spend on museums was justifed in terms of increased visitors, see Selwood and Davies (2005). It is worth noting, as well, that fundraising became more professionalised in parallel, with philanthropy and private sources of investment and sponsorship also contributing.)) As you will see in Table 7.2, I chose to use declared total government spend and Grant in Aid to ACE (being one of four arts councils in the UK). That is not to say that this is not complicated, but again, I followed how it was used in the literature and Hesmondhalgh et al. offer detailed descriptions of funding at this time (Hesmondhalgh et al. 2015, pp. 71–75) that can help you decide which is best to use. I used the clearest to me.
3) Availability: The availability of recent historical data that was readily available on websites may have gone through a process of archiving. This changes links and might make it difficult to find the data you have identified as useful from the literature. You can consult the UK government web archive (The National Archives n.d.) if it is government data, or data from a non-departmental government public body like the ONS or ACE. As we have already encountered, back when we were thinking about the role of methodology in data in Chap. 3, there are pros and cons to all data, but administrative data are easy to access and managed by public bodies, with strict guidelines. It is therefore a great place to explore possible relationships and patterns for further research.
4) Assurance: Knowing you have made the right choice can feel impossible. It is not always explicit that many choices are made in even a simple data process, like the one I describe here. The key thing is to know that most choices will have pros and cons and that there are limits to all claims of what can be known with the data and methods used. You just want to be sure to be aware of the limits, and state them when you describe your findings.
I have chosen to consult the ONS for well-being data, as their platform is most familiar to me, and therefore feels easiest to refer to. Going back to the choices we make about which data we choose (Table 3.1), there can be a trade-off between resources (skill, time, money) and robustness. In another situation, you might find other tertiary data more accessible. The data I use here are headline statistics, rather than the whole dataset of every response. Therefore, basic data practices (cleaning and aggregation) have already been done by those who administer the data, for ease of use by the media, government and indeed anyone who is interested. The same is true for the public spending data I have chosen.
As we have previously discovered, Life Satisfaction (LS) is probably the most popular measure of subjective well-being (see Sect. 4.5 for reasons why). While the UK’s Measuring National Well-being (MNW) programme did not officially begin until 2010, the UK had national-level surveys that had a question about life satisfaction for decades. Other national statistics offices, and international statistics bodies, have also administered surveys with life satisfaction questions in. The tertiary data I use here are from the British Household Panel Survey. It followed the same representative sample of individuals—the panel—over a period of years between 1991 and 2009. The same households who took part in BHPS were asked to participate in a larger survey, called Understanding Society.((As an aside while I accessed the headline data from the ONS website, the survey itself is not administered by the ONS, but in fact the Institute for Social and Economic Research (ISER) at the University of Essex. This has no bearing on my use of the data in this instance, but it is important to acknowledge the data source. Also, administration of Understanding Society is slightly more complicated than I explain in-text. Those who administer the survey have to re-sample due to what is known as ‘respondent attrition’ which means that members of a panel who have been recruited fall away over time and are then lost from the sample from whom longitudinal data are being collected. This does not impact how we use the data in this chapter; however, it would be a concern were other types of claims made regarding the longitudinal qualities of the data.)) The same questions are asked of participants in the later survey, so data are available for after 2009.
Table 7.1 demonstrates that using data for satisfaction with life overall, as measured by the BHPS, does not show an increase in life satisfaction over time. While this is a somewhat crude attempt to use data that is readily available, it demonstrates that it can be easy to explore a fundamental question quickly and sensibly. In this case, the question might be: ‘if we know that investment in a particular policy initiative or policy domain has increased substantially over time3, how can headline well-being statistics help us understand the influence of investment on well-being?’ As Table 7.1 shows, the increase in funding is not seen in an increase in LS scores.
There are many limits to what we can know from the data sourced—we know very little of its context in this table, for example, but it tells a clear story. As it was from an ONS summary (for ease), rather than LS data from the UK Data Service, the years represented (2002/2003–2009/ 2010) are those available and only a subset of New Labour’s time in government exactly (1997–2010). This does not mean they are not useful.
Table 7.1 Life satisfaction data 2002/2003–2009/2010
| Q: Satisfaction with life overall | 2002/2003 | 2003/2004 | 2004/2005 | 2005/2006 | 2006/2007 | 2007/2008 | 2008/2009 | 2009/2010 |
|---|---|---|---|---|---|---|---|---|
| Somewhat, mostly or completely satisfied | 77.3 | 78.3 | 77.0 | 74.6 | 76.2 | 77.0 | 78.1 | 77.1 |
Data Source: ONS (2010a)
Table 7.2 Policy spending on the arts and life satisfaction
| Q: Satisfaction with life overall | 2002/2003 | 2003/2004 | 2004/2005 | 2005/2006 | 2006/2007 | 2007/2008 | 2008/2009 | 2009/2010 |
|---|---|---|---|---|---|---|---|---|
| LS | 77.3 | 78.3 | 77.0 | 74.6 | 76.2 | 77.0 | 78.1 | 77.1 |
| Total govt spend (billion) | 1.59 | 1.84 | 1.77 | 1.96 | 1.94 | 2.03 | 1.88 | 1.97 |
| Govt Grant in Aid to ACE (million) | 289.405 | 324.955 | 368.859 | 408.678 | 426.531 | 423.601 | 437.631 | 452.964 |
Data source variable (see endnotes)
The UK government’s changes in funding and policy are unlikely to see an instant impact on a national population’s life satisfaction. There are likely to be lags in effects. However, as noted with the poverty data in Chap. 1, selecting your timeframe can alter the narrative about the effects of government policy, be that life satisfaction or poverty. But we can check. Hesmondhalgh et al. kindly gave us the rest of the data for Grant in Aid to ACE, as follows:
1997–1998, £186.60 million
1998–1999, £189.95 million
1999–2000, £228.25 million
2000–2001, £237.155 million
2001–2002, £251.455 million
Therefore, the increase in Grant in Aid spending was about the same in the five years that we didn’t include, as in the eight years we did, and it increased quite steadily.
What if we want to ask a more complex question, or see if there is any pattern between well-being and funding? In Table 7.1 we were only exploring one dimension of data: life satisfaction over time. Table 7.2 uses the same LS data points over time with some additional rows to report data on arts funding too. This will let us see a relationship between ‘amount of funding’ from one set of data and the level of life satisfaction over time from another set of data. We can then plot these data over time as a line graph that looks like Fig. 7.1. A positive relationship between increase in funding and life satisfaction over time would see the lines on the graph charting a similar course, so to speak.
There is no obvious relationship between policy spend on culture in the data plotted and life satisfaction. Even if we account for the additional five years of data, life satisfaction does not appear to relate to policy spend. Interestingly, LS data from the BHPS from the longer timeframe((Fitzroy and Nolan (2020) was the first article that came up in my search for life satisfaction data over these dates. Their plotting of life satisfaction over the whole period shows it is even more erratic, or, in other words, the line would be even less straight on the graph.)) are even less inclined to show a steady increase than our subset. While the easily available data do not have all of the 13 years in which New Labour were in office, you might expect that 8 years’ data would be enough to find a relationship between policy spending on the arts and life satisfaction, if there is one to find.
So, what about the limits of what we can know about the relationship? Figure 7.1 may only report life satisfaction data, but we know some other things about cultural investment, based on the literature presented so far.
Fig. 7.1 Patterns between arts funding and life satisfaction over time. (Total spend data from UK public spending website https://www.ukpublicspending. co.uk/download_multi_year_1997_2010UKb_17c1li111mcn_F0t8nt. Grant in Aid data via Hesmondhalgh et al. 2015)
For one, we have evidence that policy spend on the arts does not reach everyone equally, because not everyone participates in the culture that this money is spent on. This could mean that the way that culture was funded in this timeframe would, therefore, possibly limit potential increases in life satisfaction overall across a whole population. We need to acknowledge that there is a difference between cultural participation and investment in culture.
Let’s quickly return to what we have already learnt about life satisfaction data as a measure of subjective well-being. Firstly, let us consider the question: ‘How dissatisfied or satisfied are you with your life overall?’ This does not capture all aspects of subjective well-being. In fact, if we think back to Chap. 4 on subjective well-being and Table 4.2, with the ONS4 questions, you will remember that life satisfaction falls under one of the three dimensions of subjective well-being: evaluative. Then, it follows that there may be increases in aspects of well-being that were not captured by responses to this question. Life satisfaction data are therefore useful, and may be the most useful according to some4, but still limited in evaluating overall subjective well-being (if we are to follow the accepted reasoning presented so far).
So, we need to acknowledge that there are many limits to knowing the extent to which policy spending in one area can have a clear relationship with life satisfaction, and what that means for the culture–well-being relationship. There are, in fact, numerous limits to any claim that might be made for causation. The life satisfaction data could also include the effects of countless other things happening at the same time which could be counteracting the effect, if, indeed, it existed. Remember the conditions of a good measure of well-being in Chap. 3? It should
be sensitive to important changes in wellbeing and insensitive to spurious ones. In practice, distinguishing between the two is quite a challenge and often relies on judgement based on a priori expectations.
(Dolan and Metcalfe 2012, 411)
Clearly, the process I have described is not seeking a metric. All I have done here is describe the data easily available to look for a relationship between arts funding and LS. Therefore, no attempts have been made to account for confounders (which we will come to in others’ research later). There are so many variables that might affect life satisfaction in a way that would be captured by life satisfaction data, that it is extremely difficult to pinpoint the impact of one aspect only in this descriptive way. People who analyse data, rather than simply describe it, will use a theory or hypothesis about pathways that shape well-being to help them create models that do this work. We will return to this in Chap. 8.
Life satisfaction is a very influential measure that we have encountered numerous times in this book. We have been measuring it for years, as it was in the first wave of well-being indicators (see Chap. 2). Realising that life satisfaction had not changed as expected with income over the years, resulted in Easterlin’s paradox that was influential in the second wave of well-being as happiness economics5. Life satisfaction is also measured using Big Data technologies (Chap. 5) and is thought to be the measure of subjective well-being that people most readily understand (Chap. 4). Crucially, because questions about satisfaction with life (although worded slightly differently) have appeared in numerous surveys, and for decades, we have a lot of life satisfaction data to make simple comparisons over time, as we have just seen. LS can also be used to show very powerful relationships to outcomes of well-being, such as suicide rate and the familiarity of LS, together with the prevalence of the data, make it useful for simple exercises, as we have attempted here.
We’ve briefly looked at ways that the relationship between different variables (different policy spend data and life satisfaction) can be plotted. This will hopefully make it a bit easier for you to engage with similar representations in future. This section also demonstrated that it is quite easy to play with data that are publicly available. You can download the data into a table, like those featured, and use a simple function in Excel to plot line graphs to look for relationships over time.
Of course, there is another key point to this section, really, and that is to problematise the assumption that the arts and culture are a priority for policy spending if you want to improve well-being6. If you look at historic well-being data that coincide with previous increases in policy spend, you cannot find patterns in the data that prove that this relationship exists. There are many limitations to the claims that can be made with these data. The increase in arts funding coincides with a more general increase in public spending overall, therefore it is hard to disaggregate policy spend from other things that may affect life satisfaction in this time. Another issue is that life satisfaction data only capture one aspect of well-being. I’m sure you have thought of other limits, as well. What is key is that while using data in this way may not prove anything, sometimes exploring data can be good enough reasons to ask questions—remember this is what Easterlin did when he found that life satisfaction did not have the relationship to income that had been long-assumed in the data he had. This is said to have changed well-being research forever—even if people still argue about it. Sometimes data help us question the status quo in productive ways. They are not only there to help certain people answer certain questions.
