Understanding Well-being Data

chapter 8 Talking different languages of value

Museums and Happiness and other relationships

The research reported in ‘Museums and Happiness’ was actually looking for more than one relationship. The equation I cited earlier in that presentation in the Manchester hotel was one of two presented in the text. The report states:

We look at the impact on wellbeing and health of participating in and being audience to the arts and of being involved with museums and compare these impacts to other activities such as participation in sport.

(Fujiwara 2013, 7)

This means, that as well as the ‘general happiness’ question, used in the TPS questionnaire, the researchers were also able to use other general questions on health.((TPS also now asks more specific, subjective questions about whether people put an increase in activities down to improved or worsened health. This question is only in the longitudinal version of the survey which has been going on since 2012. It has small differences to the version of the survey used in ‘Museums and Happiness’.))

They also used income. We’ll come back to this. But for now, we know that there are some culture variables (participating in and being an audience to), some other activities, including sport. For ease, we are going to call all of these ‘participation variables’. You can find these in Table 8.1. The participation variables are the independent variables (or, you might find it easier to think of them as the predictor variables). The two dependent (or outcome) variables are health and subjective well-being.

Table 8.1 Participation variables modelled in ‘Museums and Happiness’

Museum variableswhether participants visit museums in their free time
whether they volunteer in museums
a measure of the number of hours spent in museums per year
the number of museum visits per year
The non-museum variables
whether participants had done sport or other physical activity in the last four weeks
whether participants had (in the last year) participated in each of ballet, dance, singing, playing music, painting and drawing, photography or crafts
whether participants had (in the last year) attended exhibitions (also referred to as ‘audience to arts’), opera, concerts and live music, ballet and dance
Adapted from Fujiwara (2013)

The same process was used to calculate the relationship between visiting museums and happiness and ‘has done sport or physical activity in the last four weeks’ and happiness. This is a fairly simple process for someone who knows what they are doing, as they can run the same model multiple times, swapping out one participation variable for another. You might then do the same thing again with the outcome variable as health, going through the process of swapping the predictor participation variables.((Perhaps confusingly, the calculation is slightly different for health in ‘Museums and Happiness’ (it does not include income), but you could swap the outcome variables in principle.)) The takeaway point is really, that we are going to proceed by talking about the processes involved in calculating the relationship between museums and happiness, for ease of understanding, but really there are multiple museum and non-museum ‘participation’ variables used to calculate different associations with health and happiness in this research.

To go about achieving the aims of this research: looking at the ‘impact on wellbeing and health of participating in and being audience to the arts and of being involved with museums’1, a well-being valuation approach was used.((The report tells us that a similar approach was used in the CASE (Culture and Sport Evidence) programme (DCMS 2010), but with different data. The CASE programme used the BHPS study to value: sport, going to the cinema and going to concerts (as the variables available). It also used data from life satisfaction questions to measure well-being, rather than ‘happiness’ as with our case study here.)) This approach aims to estimate ‘monetary values by looking at how a good or service impacts on a person’s well-being and finding the monetary equivalent of this impact’1. In order for us to engage with this process of valuation, it may be helpful to get into a mindset in which we think of participating in an activity (let’s stick with museums for ease) as a ‘good’ or a service. By goods here we mean the same as ‘trading in goods’: that is, this experience has a market value; this experience can be valued in this way. That is, people can choose to spend time or money on attending museums, as opposed to on another thing, like the cinema or rock-climbing. This is a slightly different mindset, perhaps, than the idea of culture as a social good.

You have maybe spent most of this book thinking of well-being as a social good, without thinking about a social good as having a market value. In many ways, instinctively they feel opposite, as often actions to maximise something’s financial value, feel at odds with a social good (we discussed this in Chap. 2 in thinking about McDonald’s and the rainforest). But theoretically, all things which are good can become ‘goods’. In this mindset, culture is not just a qualitative, incommensurable (has no common measure) experience. It is not only a way of experiencing fulfilment and happiness, but people can choose to consume culture, and it is something that makes them feel satisfied. This means it has utility (because it makes them happy).

Getting into this mindset helps us ‘talk the talk’ of valuation and imagine how culture may be quantified (in theory). When you think about it, we all have limited time to do anything, whether that is watching Netflix, going to the gym, playing video games, blowing dandelions or going to museums. Different ways of spending time might be associated with different value, but because we don’t have unlimited time, we have to prioritise. The relationship between value, museums and experiential benefit is there; it is just not always readily visible to us, or something we think about.

So, if someone wants to estimate museums’ impact on well-being, then they might say that they hypothesise that attending museums has a positive association to well-being, but we know more about the ways different types of well-being have a relationship with money. The amount of research on the relationship between income and different forms of subjective well-being far exceeds that on participation and well-being. As we discovered in Chap. 4, the relationship between income and happiness (the Easterlin paradox) is even described as the very turning point in well-being research. So, using income enables us to

  1. begin to understand the relationship between museums and happiness, and
  2. express this relationship in financial terms.

Of course, we have many prior estimates of the culture−well-being-money relationship to work with. This is not one undisputed value. For example, there have been thousands of studies on the relationship between income and well-being. This inevitably means that there are different approaches with different results. So, a decision has to be made by the researcher about the most suitable way to estimate the relationships they are interested in, in the specific context in which they are working. This refinement of which variables to use is standard practice, so long as the decisions made are subsequently clearly outlined and are justified and the limitations to research and the caveats to claims acknowledged and discussed.

In a valuation approach like Willingness to Pay (or another of the stated preference techniques we have previously covered in Chaps. 2 and 3), the data used are from people’s responses to questions which asked them for their preferences or what they value. The questions ask people to state the value themselves for a good or service. In the simplest of terms in this example, this would be: ‘how much would you be willing to pay to attend museums?’ There are noted cons to asking people to attribute value themselves that are acknowledged in the report.

Page 28 of the report explains that a study in Bolton in 2005 found that people were willing to pay £33 a year for museums in Bolton. The reason this is so low, in comparison to the £3200 per year in the Museum and Happiness findings, is explained as follows. It is unlikely that people will state a high value for a currently publicly available service in case they may get asked to pay for it in the future2. This is called strategic bias. However, there is not one way that strategic bias might affect the valuation. This argument works just as well as saying that some people will overinflate their willingness to pay for a museum, knowing that the more they say it is worth, the more attractive it is to fund, and the less likely they will have to pay for it, of course. We might guess that some people would be very likely to apply a high number to their willingness to pay, by virtue of working in the cultural sector. It is not possible to be sure which way strategic bias will go in this context or indeed the motivation.

There are other issues with ‘willingness to pay’ and other contingent valuation methods.((A really clear discussion of the limits of contingent valuation methods can be found in Throsby (2001, Chapter 5).)) They have limits in part because of the hypothetical nature of what you are often asking people. For example, ‘existence value’ is worth thinking about (and is, again, noted in the report). It is hard to imagine how much something like a museum or library is worth to you, as they exist and have value just in people knowing they are there, and some people want them to be there in case they—or others—want it (called option value). There is also the knowledge that they will be there for future generations. This is not the same as using these services, or being prepared to pay for them. When people are threatened with the removal of museums or libraries they do not use, they see a hypothetical value in them. Or, the theory goes that there is a value in knowing they exist at all.

The TPS data used in the ‘Museums and Happiness’ study did not contain people’s own valuations. This means that it was not possible to have ‘preference satisfaction’ measures in the valuation model. Instead, it used a well-being valuation approach. The report explains that this over-comes the biases in people’s own evaluations by estimating for them. The ‘Museums and Happiness’ report states that ‘two very distinct measures of wellbeing are used’ in the Bolton Study on the one hand and ‘Museums and Happiness’ on the other2. The report continues: ‘there is no philosophical or theoretical reason why values from these methods should converge in anyway [sic]’2. This means that even though these two pieces of research are both using eco- nomic approaches to value museums and well-being, the findings should not be expected to be similar. When you think back to Chap. 7, and the importance of how culture and well-being are operationalised, versus the headline findings from reviews of evidence, you might think to yourself that this does not bode well for arguments on how much we can know the relationship between culture and well-being, if we cannot expect studies to have more similar results than £33 and £3200 as answer to the question ‘what is the value of museums to people in terms of well-being?’

Let’s return to the well-being valuation approach used here and how it can know the value of something to people without asking them. It requires a dataset to include a measure of well-being, a measure of the good we are interested in valuing [museums] and other determinants of (things we know are associated with) well-being, such as income. The logic is that say we imagine a unit of happiness as an ‘HAP’, and we know that £1 neatly equals exactly 2 HAPs (how convenient), economic approaches can use what we know about this relationship and apply it to understand others. The technique runs on the following rationale:

so, 1: if ‘museums have a relationship with well-being that we need a value for’

and, 2: ‘money has a relationship with well-being that we have a value for’ then, 3: ‘how much money makes you as happy as a given unit of museums’ is essentially the question.

Box 8.4 Coefficients

What they are estimating here are the coefficients behind particular types of participation. When you look up the meaning of ‘coefficient’, you are likely to see something like this: ‘a numerical or constant quantity placed before and multiplying the variable in an algebraic expression’.

It’s probably important to bear in mind that the amount of museum participation isn’t how many times someone goes, how long they are there or how many people are inside a given museum. Museum participation is a variable in and of itself that will represent whatever people answered in response to the survey question, and/ or how those data have been coded.

The coefficient is basically: if you increase a unit in your independent variable, how large an increase do you get in your dependent variable?

In this example, the variable ‘museum participation’ means ‘visits to a museum a certain number of times a year’, so if you increase the museum participation, how much increase in the happiness variable is there?

The variable might have values between 0 and 1 (which means the unit increase is ‘goes from not doing it to doing it’, and it might be continuous in hours, or could be another type of proportional increase). Say it is one of the questions from TPS in Box 8.1.

During the last 12 months, have you attended a museum or gallery at least once?

1. Yes 2. No -1. Don’t know

This is either 0 or 1, for yes or no. In this instance, if people take the don’t know option, they won’t be included. The coefficient is how much of the variable ‘attended museum or gallery’ there is. So, the coefficient behind ‘goes to museums or not’ might be large, but that’s because it only goes from 0 to 1.

However, if you have ‘number of hours spent in a museum’, with values lying between 0 and 1, you would expect a much smaller coefficient because the max number of hours is much more than 1.

Remember that using income is (1) a way into understanding the relationship between museums and happiness, and (2) a way to express this relationship in financial terms.

Box 8.5 Imagining Units of Happiness, Museums and Money

Say a HAP = 1 unit of happiness.
You went to the British Museum yesterday, and your HAPs

increased by 8.
The day before yesterday, someone gave you £1, and your HAPs

increase by 2.
This suggests that going to the British Museum is equivalent to

getting £4 increase in income. Or if you were to stop going to the British Museum, but were to get £4, you would stay equally happy.

Once you have established this relationship, you can equate museum visitation happiness to happiness from getting more income. This is one way of valuing museums for their relationship to happiness.

In a previous study, the researcher found that ‘when using lottery wins as an instrument for income… the size of the impact of income on happiness increases more than ten-fold’3. The reason why lottery wins are thought to be a good indicator for income is they are from outside of a person’s day-to-day life. Theoretically, this makes it easier to determine the impact of the money on someone’s happiness. You might find yourself asking ‘well, how can you know how much of the happiness from the lottery win is from the increase in wealth, and how much of the happiness is from the joy of winning?’ There is even a whole body of research that argues that winning the lottery doesn’t impact on happiness at all.((The first and most famous of these studies is Brickman et al. (1978). However, as with other previous examples of wealth and happiness, the evidence is not universal.)) However, the rationale in ‘Museums and Happiness’ is that it is suitable ‘to get a good estimate of the causal effect of income’3.

The report also explains prior studies ‘derive implausible large value estimates for non-market goods’ because of this discrepancy in income3. Notably, a CASE21 study using an income compensation approach found that going to the cinema once a year had a value of £9000 per household per year4. The report states that

Since there is no suitable instrument for income in the Taking Part data we also estimate values using an income coefficient that has been multiplied by 8 (which is in the scale between 2 to 10, which is the level of bias found in the studies above, but weighted more towards 10 since the analysis of happiness data using the BHPS suggests that the true impact of income on happiness may be more than ten times larger than the OLS coefficient).

(Fujiwara 2013, 26)

In simplest terms, the idea is that those previous studies are able to ‘instrument for income’—which means that they can isolate the benefits of money from the benefits of being a high earner. It is important to remember that as a higher earner, you are unlikely to have data collected on how long you went to the loo; maybe you have a nicer office and get to expense your coffee, perhaps even someone else goes and gets you nice coffee from your vendor of choice? In the same way it is difficult to disaggregate the joy of winning from the impact of money, it is difficult to account for all the ways that being a higher earner may improve your life outside of money alone.

Returning to these previous studies, they found the discrepancy between income and lottery wins. There is therefore a number for that that can be plugged into the valuation. The report explains that this means it is therefore plausible to use people’s income, as declared in the TPS, and multiply the coefficient by 8, based on the fact that the estimates between studies that ‘instrument for income’ and those that don’t tend to differ by around this much in previous studies. This is accounted for in the report, like this:

This is part of the reason why Wellbeing Valuation studies that do not instrument for income derive implausible large value estimates for non-market goods.

(Fujiwara 2013, 26)

The report accounts for the robustness of this approach, like this:

The wellbeing valuation techniques used here are in line with welfare economic theory on valuation (which underlies all cost-benefit analysis and SROI techniques), but we should note that these values should not be seen as amounts that people would actually be willing to pay per year for these activities. This would only be the case if people satisfy their preferences solely on the basis of what makes them happy, but other factors may impact on people’s preferences and market decisions. These values should be seen as the equivalent amount of money required to create the same impact on people’s happiness and they are useful as they show us the magnitude of importance of museums and the arts to people.

(Fujiwara 2013, 33)

In other words, these valuation principles are considered robust, but these values are not what people would actually be prepared to pay. Instead how much extra money would keep someone at their current levels of happiness if they had to stop participating. In a section called ‘Key Findings’, the valuations are presented as follows:

Using ticks for bullet points may feel an unusual way of presenting findings, especially when there are so many caveats to these estimates, particularly whether people actually do value museums like this, or not. More importantly, these statements imply that people value participation in these monetary terms. Again, it is not that people do not—either consciously, or unconsciously, but the presentation might be confusing to the report’s audience. These are in fact what the report calls ‘the compensating surplus’ for these activities. In other words, according to these calculations, this is the amount of money people would in theory give up in order to undertake the activity. In other words, the finding is that on average, people who go to museums are as happy as people who don’t go to museums but are paid £3200 a year more.

This can be difficult to understand when you are reading the key findings from a report as a non-expert, and especially difficult when they are presented out of context, like in a daily national newspaper. We will now look at how findings can appear in different contexts in ways that can be distracting.

  1. Fujiwara 2013, 7 [] []
  2. Fujiwara 2013, 28 [] [] []
  3. Fujiwara 2013, 26 [] [] []
  4. Matrix Knowledge Group 2010a, b []
  5. Fujiwara 2013, 8 []