JAN Forum: your views and letters


Response to: Elisions in the field of caring by V. Deary, I.J. Deary, H. McKenna, T. McCance and R. Watson (2002) Journal Of Advanced Nursing 39, 96–102.

I am, to begin with, enormously flattered by the attention my paper (Paley 2001) has received from the massed ranks of ‘caring’ researchers (Deary et al. 2002). ‘Elisions in the field of caring’, a reply to my ‘An archaeology of caring knowledge’ (ACK), is a series of clever ripostes and devastating arguments. Unfortunately, the ripostes are to points I never made, and the arguments are against views I don't hold (more precisely, views I quite clearly don't hold). Deary et al. have created a mirage, and attacked it with commendable gusto. It is a highly entertaining performance – rather like watching a man wrestle himself – so it seems almost churlish to point out that they are tilting at windmills. However, at the risk of being a kill-joy, I would like to draw attention, briefly, to the way in which the authors misrepresent my position, and to the endearingly Pythonesque silliness of one of the central ideas in their critique.

Speaking and doing

Let me try to make one of the main arguments in ACK as clear and succinct as possible.

  • 1There is a distinction between the study of X and the study of what-people-say-about-X, which I will abbreviate as WPS(x).
  • 2According to Foucault, Renaissance thought did not distinguish between knowledge of X and knowledge of WPS(x). This is illustrated (note that word) by Aldrovandi's treatize on snakes ( Foucault 1970 ).
  • 3After about 1630, again according to Foucault, the distinction between X and WPS(x) became absolutely critical. What we now call the scientific revolution turned on it, precisely because science is (largely) the study of X, not the tabulation of WPS(x).
  • 4In the nursing literature, where X = caring, no distinction is made between the study of X and the tabulation of WPS(x). So there is a fairly direct parallel between the nursing literature (in this respect) and Renaissance, pre-scientific, thinking.
  • 5A research-and-theory literature that cannot distinguish between X and WPS(x) will be plethoric but poverty-stricken, for the reasons given by Foucault. The literature on caring is (in my view) both.
  • 6Where X = caring, the distinction between WPS(x) and X is equivalent to the distinction between what people say about caring and what people do when they are caring . It is, in effect, a version of the familiar distinction between attitudes and behaviour.

Anyone who has a problem with this – anyone who cannot see the difference between talking about it and doing it – should stop here.

We can now proceed to what Deary et al. make of this.

‘Paley…[suggests] that research into the phenomenon [of caring] should cease.’ Quite the contrary. I think it would be rather nice if research on caring got started. What we have currently is not research on caring – that is, X. It is research on what people say about caring – that is, WPS(x). Unfortunately, many researchers seem not to recognize the difference – and it looks as though Deary et al. are among them. Otherwise, when I suggest that research on what people say about caring is poverty-stricken, they would not imagine that I was saying research on caring should cease.

Of course, Deary et al. do announce that ‘there is no confusion concerning the things said about and the things done in the name of caring in the minds of the present authors’. If only. In the next few pages, they manage to get the simple distinction between ‘what she said’ and ‘what she did’ mixed up with the distinction between phenomena and noumena in Kant. The noumenon – the thing-in-itself, the Ding-an-sich– is completely unknowable. It cannot be talked about, observed, experimented on. It is, so to speak, ‘behind’ the world of phenomena, the world we can see, touch and theorize. Still, according to Deary et al.‘what she did’ is a noumenon. Beyond our ken. So let's be clear. In Kantian terms the ‘what she did’ and the ‘what she said’ are both phenomena. They can both be seen, heard and gossiped about; and my point, to labour it, is that there should be more observation of ‘what she did’, and less gossip about ‘what she said’.

However: ‘If he has any suggestions accessing [sic] the Ding-an-sich, then he really should let us know as quickly as possible, since the scientific and philosophical communities have been waiting for some 350 years’. The sublime daftness of this is engaging. Of course I have no such suggestions. I think we should be interested in the ‘what-she-did’ of caring. Kant has nothing to do with it – except in the minds of Deary et al., who think that the ‘what-she-did’ is a Ding-an-sich which cannot be ‘accessed’. Most of us spend quite a bit of time watching, and discussing, what others do. Deary et al. believe, on philosophical grounds, that this is impossible. Heaven knows what they find to talk about.

Basically, Deary et al. have invoked Kant in order to explain why they cannot do research on caring-as-acting (rather than caring-as-spoken-about). It is a complete misunderstanding of Kant and a pretty lame excuse. In the same way, I would have thought that the Sokal parody (Sokal & Bricmont 1998) would – by now – be discouraging people from invoking Bohr (or Heisenberg) in their attempts to prop up social constructivism. In the quoted sentence, Bohr is referring to the Copenhagen interpretation of quantum mechanics. He is not proposing that physicists go round asking people what ‘reality’ means to them.

Associated words, attributes and factor analysis

An inability to distinguish ‘doing it’ and ‘talking about it’ continues in the discussion of factor analysis, complicated by the invention of various things I'm supposed to have said, and an unfortunate confusion of ‘arbitrary’ with ‘random’. Again, let me try to present a summary of the relevant bit of argument in ACK.

  • 1The bulk of the research-on-caring literature presents us, first, with long lists of words associated (by either researchers or respondents) with ‘caring’. These lists are ‘potentially endless’ because they are thesaurus-like in their construction, and they can quickly be reproduced by turning to Roget.
  • 2It presents us, second, with lists of ‘attributes’. The attribute lists are much shorter than the lists of associated words, and sometimes represent a ‘condensed’ version of the latter. They are, as I said in ACK, ‘superordinate, “second-order” lists, frequently derived from the first–order lists of association’.
  • 3The sense in which attributes are ‘derived from’ the associated-word lists varies quite a bit. The associated words are grouped together on the basis of what I called ‘resemblances’…and these are very often arbitrary. Despite the huffing and puffing of Deary et al. ‘arbitrary’ is not the same as ‘random’. ‘Arbitrary’ means roughly ‘dependent, in an unexplained way, on discretion’; randomness is, of course, a statistical concept. Roach's five ‘C's of caring, for example, are alliterative. This is quite arbitrary, but clearly not random, or even capricious.
  • 4The lists, of both types, are taken to represent a description of ‘what caring is’. To this extent, nursing is engaged in an endless definitional project – endless, because there is always a new word to be added, always a new permutation of attributes waiting to be formulated. That is pointless enough, but what makes it far worse is that the project is a vicarious one. The ‘underlying structure of caring’, supposing there could ever be a consensus about what that is, will not be the structure of any actual process, an in-the-world activity, but a condensed version of what thousands of people associate with a mere word.

If anyone seriously thinks that this collective, global twist on the Super Match Game in Blankety Blank is likely to be useful to health care professionals, they can spare themselves the burden of proceeding any further.

Now let us consider the use of factor analysis, specifically, as a way of deriving attributes. If Deary et al. were to read ACK a bit more carefully than they have done, they would notice that I do not describe factor analysis as ‘arbitrary’. Nowhere in the paper did I object to the technique of factor analysis, as such; and nowhere did I suggest that it is anything other than an entirely appropriate way of reducing a list of items to a smaller set of attributes or factors, identifiable with ‘structure’. In particular, I did not claim that ‘factor analysis leads to the proliferation of lists when in fact it accomplishes the reverse’. I was very careful to point out, instead, that the most obvious difference between quantitative studies and qualitative studies of caring ‘is the way in which the former generate attributes from associations’; and I have already observed that attribution lists are much shorter than word association lists. So I do not need any lectures on a ‘procedure that reduces additive lists into fewer categories’.

My problem is not with factor analysis, as a method, but with the use to which (in this case) it is being put. In other words – and I apologise for the constant, grating, repetition – the attempt to identify the ‘underlying structure’, not of caring, but of ‘what people say about caring’. Deary et al. do not see the difference between ‘caring’ and ‘what people say about caring’… so they accuse me of ‘naive realism’ (and if this means that I do see the difference, they are right).

The diabetes example illustrates all this rather nicely, and I am grateful to Deary et al. for suggesting it. In this case, factor analysis was used to discriminate between two distinct clusters of symptoms, which were later confirmed to be associated with two distinct generating mechanisms. Fine. Unfortunately, this example has everything which current ‘caring’ research does not have. To see this, imagine what a comparable study of ‘caring’ would look like. Factor analysis (or some other appropriate technique) would be employed to distinguish between two distinct types of ‘caring attitude’, two distinct clusters of ‘things said about caring’. Subsequently, it would be shown that these clusters were associated with behavioural differences (or, conceivably, physical differences) between nurses. Now, a result like this could be genuinely interesting. It would mean that, finally, an empirical link had been made between ‘what she said’ and ‘what she did’– and that different types of ‘what she said’ had been associated with different types of ‘what she did’. This, in turn, might lead to research designed to assess the impact, if any, that ‘caring’ has on patients.

I'm not against this, I'm all for it. What I am against – what I think completely pointless – is research intended to identify ‘the elements of ‘caring’ in nursing' (Lea et al. 1998). It's pointless because (i) it is research limited to things said, ignoring things done; (ii) worse, it represents ‘things said’ and ‘things done’ as the same; (iii) it makes no attempt to study the link between ‘caring’ and the rest of the world, and in particular patient outcomes. Its one purpose is to ‘describe’ caring as a ‘phenomenon’– where the phenomenon is understood exclusively in terms of what people say about it. Moreover, unlike the diabetes study, where one symptom cluster was differentiated from another, this style of research treats the ‘phenomenon’ as an undifferentiated something-or-other, whose characteristics and attributes can progressively be identified, then added to the list. Finally, and as I keep saying, the interest which this genre takes in what people do, and what real consequences their actions have, is virtually non-existent. The reason for this, at least according to Deary et al. is that you can only ‘access’ what-people-say, not what-people-do. Which is strange, because they obviously believe that you can ‘access’ generating mechanisms (as in the diabetes study). Why should ‘what-she-did’ be an unknowable Ding-an-sich, if ‘generating-mechanisms’ turn out not to be? Where are those naive realists when you really need them? But perhaps that's what you get when you write by committee.

Last word

It worries me that Deary et al. think that some of what I say is ‘abusive’. ‘Tedious parade’ I will own up to (though it is an accurate reflection of how I felt after reading 150 research reports on this subject). But as for their other examples, I am at a loss. ‘There is nothing empirical that can be inconsistent with it.’ What, pray, is abusive about that? It is the conclusion of an argument which Deary et al. clearly disagree with (as, obviously, they are entitled to do). But does the fact of their disagreement make this conclusion an instrument of abuse? If not, what other feature of the sentence do they find offensive? And I would ask the same question about their other examples.

Finally, I could perhaps say this. ‘I must apologise for the fact that I don’t do much empirical research these days – something which Deary et al. seem to find worthy of comment. Sadly, I have to make do with my ability to read. It is a skill which, in the case of Deary et al. seems to need a bit of polishing.' I could, but I won't. That would be abusive.