Original article

Abductive reasoning and qualitative research nup_532 244..256

Martin Lipscomb RN PhD

Senior Lecturer, Faculty of Health and Life Sciences, University of the West of England, Gloucester, UK

Abstract Abduction, deduction and induction describe forms of reasoning.

Deduction and induction are discussed in the nursing literature.
However, abduction has been largely neglected by nurse scholars. In this
paper it is proposed that abduction may play a part in qualitative data
analysis specifically, in the identification of themes, codes, and catego-
ries. Abduction is not, in research, restricted to or associated with any
particular methodology. Nevertheless, situating abduction in qualitative
research facilitates the identification of three interlinked issues. First, it is
suggested that abductively derived claims require support from deduc-
tive and inductively sourced evidence if they are to hold and, yet, in
qualitative research this is clearly problematic. Second, difficulties in
choosing between alternative plausible hypotheses (i.e. concerning
theme, code, and category description) are explored through an exami-
nation of the generality problem. Third, the role of background and
auxiliary theories in adjudicating between hypothesis options is dis-
cussed. It is argued that if qualitative researchers utilize abductive infer-
ence in the manner suggested, then the peculiarly fallible nature of
abduction must be acknowledged and, in consequence, the action
guiding potential of qualitative research findings is compromised.

Keywords: nursing research, qualitative research, applied research,

explanation, abductive reasoning.

the creative, imaginative or insightful moment in

Introduction
which understanding is grasped or is thought to be
Abduction is defined more fully within the body of grasped. Described in this way abduction is both an
this essay; however, put simply it can be envisaged as everyday event and, it will be argued, a problematic
one. Thus as a commonplace happening I see my
Correspondence: Dr Martin Lipscomb, Senior Lecturer, Faculty wifes car in the drive and I abduct that she has
of Health and Social Care, University of the West of England, arrived home before me. This form of reasoning often
Gloucester Centre Hartpury Campus, Gloucester GL19 3BE, generates correct conclusions. However, as Peirce
UK. Tel.: +44 077 328 5624; e-mail: martin.lipscomb@uwe.ac.uk (1998a [1903]) notes, while the abductive suggestion

244 2012 Blackwell Publishing Ltd

Nursing Philosophy (2012), 13, pp. 244256
 Abductive Reasoning 245

comes to us like a flash. It is an act of . . . extremely nursing literature, then a very large number of sources
fallible insight (p. 227) and, e.g. it might also be that would have been located. Nonetheless, the small
having been given a lift into work today, the presence number of hits generated from CINAHL might indi-
of a car in the drive does not signal my wifes return. cate that nurse researchers and scholars are relatively
Here it is proposed that, while abductive reasoning indifferent towards the concept and this indifference
informs nursing and research practice, interesting appears to be replicated in nursing research text-
questions are raised for qualitative researchers books. Thus, reference lists in Bowling (2009),
and qualitative research consumers (readers) when Parahoo (2006), Polit & Beck (2006), and Cormack
abduction is considered. This paper does not suggest (2000) identify entries for induction and deduction
that abduction is an inappropriate or unsuitable form but not abduction. Davies (2007) does note abduc-
of inference to deploy in qualitative or other forms of tion. However, the topic merits only a few lines in a
investigation (it may be indispensable). Nor is it pro- glossary at the close of that book and it is not possible
posed that alternative forms of logic or inference are to capture the terms meaning from the description
absent from the interpretive process. Rather, it is sug- given. Therefore, while the non-nursing literature on
gested that if abduction is granted a role in qualitative abduction may be being accessed this is not evidenced
data analysis then the peculiarly weak nature of this and it is therefore concluded that, despite this
form of inference must be recognized. This is not an reviews admittedly partial and limited character,
attack upon qualitative research; it is an exploration nurse researcher/scholars have not, to date, demon-
of the problems inherent in belief formation and strated in their writings a substantive interest in this
belief justification when, in qualitative studies, abduc- topic.
tion occurs.

Does this lack of interest matter?

Abduction in nursing texts
Lawson (2010) states that how scientists reason and
Abductive inference plays a part in logic, reasoning, argue . . . has long been viewed as a critical compo-
and belief formation. It is, however, a form of infer- nent of scientific literacy (p. 336). Nurses interest
ence that is underexplored; it is largely ignored, by themselves in a wide range of human activities and
nurses. this interest involves or includes using science to
An all-text e-search conducted on 17 November understand action/behaviour. Scientific literacy
2011 of the database Cumulative Index to Nursing entails awareness of or engagement with the pro-
and Allied Health Literature (CINAHL) located 89 cesses of belief formation and belief justification.
hits for the search phrase inductive reasoning, 87 hits Abductive inference is a part of reasoning and abduc-
for the search phrase deductive reasoning and just 5 tion thus underpins aspects of belief formation/
hits for the phrase abductive reasoning.Two of the five justification and, in consequence, action/behaviour.
located papers on abductive reasoning (Richardson & Examining the inferential forms that logic or rea-
Kramer, 2006; Haig, 2008) were in journals that do not soning takes therefore has real import. Indeed, just as
specifically target nursing or health and, for that unpicking the nature of argument potentially allows
reason, they may be unfamiliar to nurses (i.e. Journal the robustness of argument to be enhanced (by per-
of Theory Construction and Testing and Qualitative mitting weak elements to be strengthened or elimi-
Research). Of the remaining three papers, two are by nated) so acknowledging the role of abductive
Rholm (2010a, 2010b) and recent publication prob- reasoning in nursing practice and research can benefit
ably means that they have not yet been absorbed by practice and research insofar as this acknowledgment
the profession; the third, by Gordon et al. (2005), mis- enables the beliefs that we act upon to be informed by
construes abduction (see below). more transparent and defendable logic and reasoning.
Papers discussing abduction will have been missed This paper offers both a supportive codicil and
by this search and, had this review gone beyond the challenge to authors such as Rholm (2010a) who

2012 Blackwell Publishing Ltd

Nursing Philosophy (2012), 13, pp. 244256
246 Martin Lipscomb

have sought to bring the concept of abduction to a As Rholm (2010a) notes modern usage of the
nursing audience. It is supportive insofar as it recog- concept is closely associated with the work of Charles
nizes that benefit should accrue from acknowledging Sanders Peirce (18391914). Peirces understanding
the role played by abductive reasoning in belief for- of abduction and its place in scientific reasoning
mation and belief justification. It challenges by varied over the course of his intellectual career. Here,
emphasizing that the case for abduction must not be the concept is presented as it appears in Deduction,
overstated and, perhaps, Rholms (2010a) recent Induction, and Hypothesis (Peirce, 1992 [1878]
paper (in this journal) underestimates the problems henceforth DIH) though numerous other writings
accompanying use of this form of inference. Here it is including, e.g. Pragmatism as the Logic of Abduction
suggested that, even more than Rholm (2010a) (1998a [1903]), Sundry Logical Conceptions (1998b
allows, abductive inference must be employed judi- [1903]), and The Nature of Meaning (1998c [1903])
ciously for, while abduction can play a role in the develop abduction further.
discovery of processes of [nursing] theory develop- In DIH Peirce somewhat misleadingly, to modern
ment (p. 268) this form of reasoning is especially readers at least, uses the term hypothesis rather than
vulnerable to challenge and, in consequence, claims abduction (Hacking, 2007) and, in this form, hypoth-
resting upon it must be handled, if not with extreme esis is not to be confused with hypothetico-deductive
caution, then certainly warily. Specifically, not only are reasoning (Haig, 2005) or deductive-nomological
the conclusions of abductive reasoning open to cri- explanation (Hempel, 1966). Rather, hypothesis
tique, the relationship between abduction and other making or abduction, as it came to be called, is intro-
forms of inference (and accompanying evidence) in duced by Peirce in the following manner:
belief formation and justification is contested and this
Suppose I enter a room and there find a number of bags,
contestation must be acknowledged.
containing different kinds of beans. On the table there is a
Developing this line, it is argued that, if abduction
handful of white beans; and, after some searching, I find one
describes the inferential process whereby qualitative
of the bags contains white beans only. I at once infer as a
researchers identify the themes, codes and categories
probability, or as a fair guess, that this handful was taken out
in analysis that thereafter become findings, then
of that bag. This sort of inference is called making an hypoth-
these findings are necessarily insecure or fragile. It is
esis. It is the inference of a case from a rule and result.
not suggested that abduction is the only process
Peirce emphasis in original (DIH, p. 188)
involved in qualitative data analysis and, importantly,
abductions part in this process is asserted rather than Peirce delineates analytic or deductive inference
demonstrated. Nonetheless, if abduction is granted from synthetic induction and hypothesis formation or
even a potential role in analysis then this cannot but abduction and DIH (p. 188) famously distinguishes
problematize the action guiding nature of qualitative between them thus:
research. Of course, the assumption that qualitative Deduction
research has, for clinical nurses, an action guiding Rule. All the beans from this bag are white.
Case. These beans are from this bag.
function is controversial. This issue is addressed
\ Result. These beans are white.
briefly below and it is critiqued, along with insight, in Induction
Lipscomb (2012). Case. These beans are from this bag.
Result. These beans are white.
\ Rule. All the beans from this bag are white.
Hypothesis
Rule. All the beans from this bag are white.
Abductive reasoning Result. These beans are white.
\ Case. These beans are from this bag.
Since nursing publications effectively ignore abduc-
tion it is necessary in a paper such as this to be espe- For Peirce, a hypothesis is formed where we find
cially clear about the meaning of the descriptor at its some very curious circumstance, which would be
core. What then is abduction? explained by the supposition that it was a case of a

2012 Blackwell Publishing Ltd

Nursing Philosophy (2012), 13, pp. 244256
 Abductive Reasoning 247

certain general rule, and thereupon adopt that suppo- formation to generate hypotheses similar to the exist-
sition (DIH, p. 189). Significantly, supposition is not ing ones (Thagard, 1993, p. 54).Abduction by analogy
the same as claiming that, with any degree of cer- may be contrasted with existential abduction (though
tainty, the proposed hypothesis might be true . . . the two forms overlap in use) wherein hitherto
[and instead a weaker claim is made, namely that] the unknown or unexperienced objects or relations are
hypothesis is initially plausible and worthy of further hypothesized. Both forms of abduction are, I suggest,
pursuit (Haig, 2005, p. 377). The reduced nature of evident during data analysis in grounded theory
this claim is important. Gordon et al. (2005, p. 709) (though grounded theorists may object) and, it will be
approvingly cite Josephson (1997) in claiming that claimed, many other forms of qualitative thematic
appropriately derived abductive hypotheses are analysis.
probably correct. This is a mistake and instead such Although the description of abductive reasoning
hypotheses are, at best, only possibly correct. That given by Peirce is, on the face of it, convincing, two
aside, abduction or fair guess making in the presence points require attention. First, despite the innate sen-
of curious or novel circumstances occurs to every sibleness of Peirces bean-based vignette and regard-
person everyday in innumerable contexts. It is neither less of whether analogous or existential abduction is
an esoteric nor an exotic activity and, rather, abduc- considered, making an abductive hypothesis,of a case
tion is a commonplace occurrence that generates, from a rule and result, is fraught with difficulty and in
albeit tentatively, ideas about the world. consequence those engaging in this process must
For example, in a clinical context a nurse notices that temper assertions based on such inference. Second,
a patient appears flushed and drowsy. The patient is the relationship between the various forms of reason-
having or has recently completed a blood transfusion. ing (abduction, deduction, and induction) is, in appli-
The nurse, invoking background theories (derived cation, equivocal or opaque and the extent to which
from personal experience and training) reasonably deductive and/or inductive inferential forms must
surmises or abducts that the patient might be experi- support abductive reasoning is unclear. Put another
encing an unwanted reaction to the transfusion and way, ought deductive and/or inductive logics and the
she acts to gain evidence to support or refute her evidence (broadly defined) associated with them
informed guess (e.g. by talking with the patient and by always be sought to support initial abductive guesses
taking a set of observations). Oftentimes, as evidence or can abduction stand on its own? Must, as in the
emerges, an initial hypothesis is shown to be in error. above example, nurses always act to buttress or
Thus, the patient was not in fact flushed. She was scupper abductive hunches by seeking additional con-
instead wearing too much rouge blusher and, given a firmatory or discrediting information or can justified
hectic day, she was simply having a nap. On the other beliefs about the world be grounded simply on abduc-
hand it is also the case that fair guesses, fair insofar as tive reasoning? As will be shown, this question has
they are grounded on reasonable if disputable assump- special significance in qualitative research particu-
tions, are frequently upheld and proved right.Thus the larly when more than one explanatory option is
patient was experiencing an unwanted reaction. available.
While the focus of this paper is on abduction in
qualitative research rather than abduction in practice
this example demonstrates the generative nature of
Choosing between alternative
abductive reasoning since generative methods reason
explanations
from warranted premises to an acceptance of the
knowledge claims in question (Haig, 2005, p. 383). It The need to temper assertions based on abductive
also describes an instance of abduction by analogy. inference becomes apparent as soon as we probe
Analogical abduction transposes or relates past expe- Peirces exemplar more closely. And, although the
rience (here of kindred unwanted reactions) to the following argument may at first appear digressionary,
new situation and it uses past cases of hypothesis pushing the description of abduction contained

2012 Blackwell Publishing Ltd

Nursing Philosophy (2012), 13, pp. 244256
248 Martin Lipscomb

within the bean exemplar exposes an important assigned to a class of bags (e.g. they come from the
conundrum in qualitative data analysis. class of bags with all or mostly white beans and no or
Peirce presents an uncomplicated scenario wherein few black beans). This again seems straightforward
one handily placed bag contains white beans only. albeit that, first, the point at which too many black
However, what if two white-bean-only bags are avail- beans appear in a legitimate option bag remains open
able? Or, what if, in addition to the two bags contain- and, second, the example might be said to illustrate
ing only white beans there is third bag on the table how different inferential forms closely interact in
within which, among the white beans, one black bean explanation. However, for nurses and especially
sits? It might be argued that the presence of a single qualitative nurse researchers while the importance
black bean does not significantly influence the statis- of abductive reasoning should (as per Rholm, 2010a)
tical likelihood that any particular handful of beans be recognized, in application the existence of alterna-
would be other than white were they to escape and tive hypotheses or hypothesis choices severely under-
be found and, in consequence, the found beans could cuts the amount or degree of confidence that can be
have come from any one of the bags (assuming no invested in any particular choice. Yet before examin-
other option). ing this statement further it is perhaps useful in the
Yet if the handful of found white beans might now first instance to position or relate Peirces ideas to the
come from any of the three bags i.e. there exist context in which they were made.
multiple legitimate explanations or explanatory
options must abduction cease or, more pressingly,
Abduction contextualized
how are beliefs formed and justified when compound
options or choices are available to the hypothesizer? Despite wide ranging interests in theory and logic,
Further, we might ask, for how long would the bag Peirce was a practicing physical scientist (from 1859
containing non-white beans remain a legitimate to 1891 he worked on the US Coastguards Geodetic
option if increasing numbers of black beans were sub- Survey measuring fluctuations in the intensity of the
stituted for white beans? Thus, if in a bag of mainly earths gravitational field, Burch, 2011) and, maybe, it
white beans (no = 99) one black bean is not likely to is in consequence easier to situate his ideas on infer-
appear in any randomly selected handful, what can ence and scientific method within a natural rather
reasonably be assumed about beans taken from a bag than social science or humanities frame.
where white beans are substituted by two, three or n Thus, abduction, as insightful or creative perception
black beans? This problem can be variously inter- is, in Peirces developed theory of scientific enquiry
preted (and its resolution clearly depends, in part, merely the first of three stages namely: abduction or
upon the number of found white beans). Here the hypothesis development, deduction or prediction of
point being made is simply that, again, assuming the consequences, and induction or experimental testing
found beans come from only one of the three bags, of hypotheses (Anderson, 2009, p. 155). That is, first,
and at a stage before which too many black beans reflection or considered but nonetheless speculative
appear in one of the bags, the found white beans could conjecture upon a problem or observation generates
abductively be assigned to any bag. And yet, as the a tentative abductive hypothesis. Second, this hypoth-
number of legitimate options increases, the danger of esis is translated into a statement capable of deduc-
ascription error rises. tive testing and, importantly, event observations that
It will shortly be argued that nurse researchers face will (or will not) support this hypothesis should be
similar dilemmas when analysing qualitative data other than those that generated the hypothesis under
(when more than one interpretive option is always investigation. Third, additional support for the
available). Nevertheless, in this instance, a sensible hypothesis should be sought by inductive testing.
abductive reasoner will refrain from making a Here event-objects are manipulated in ways that can
hypothesis about any one bag (i.e. they came from only or with a high degree of probability happen if the
that bag) and instead the found beans might be framing hypothesis is correct.

2012 Blackwell Publishing Ltd

Nursing Philosophy (2012), 13, pp. 244256
 Abductive Reasoning 249

Lawson (2002, 2010) summarizes this sequence as behaviour is frequently patterned (demi-regularities,
if-then-therefore and, in application and significance, Wilson, 2005), causality, as commonly understood,
the relationship between the various forms of infer- cannot be established in the non-natural sciences as it
ence is both important and contested. Howson (2000), can be, or it is claimed it can be, in the natural sciences
though he does not necessarily agree with the sugges- (Bhaskar, 1997, 1998). Moreover, whenever people
tion, notes that abduction has been presented as a rather than things are discussed, normative assump-
fundamental or synthesizing concept, as the inferen- tions inevitably interpenetrate or coexistent along-
tial method of science (p. 112, italicization in original) side factual claims.
and, in this vein, Sther (1998) positions abduction as Values are abstracted from Peirces exemplars
retroduction (though this term is disputed) as a means since, excepting a desire to maintain logical coherence
of overcoming the problematic of dualist deductive (bar valorizing logic), we do not care about the
and inductive inference (see also Downward & outcome or conclusions of bean-based reasoning.
Mearman, 2007). Thus although different theorists Nurse researchers do, however, care about the con-
interpret the concept differently abduction has been clusions of their arguments/beliefs because they apply
presented as if it were primus inter pares as the form to people and, moreover, normative assumptions
or method of reasoning that generates the insightful (possibly unarticulated) are often conflated in analy-
idea that other inferential forms then hone and test. sis with facts. This may or may not be considered a
However, whether or not abduction holds some problem. However, it is here proposed that reflec-
special place in scientific activity, social scientists and tively derived explanations or arguments grounded in
humanities scholars might regard deductive experi- or on abductive reasoning are vulnerable to critique
ment and inductive testing (the second and third of since, returning to the problem of multiple legitimate
Peirces stages) as inappropriate or unavailable or, at explanation (i.e. where the found beans could have
a minimum, challenging in their disciplines. And, like- come from more than one bag) nurse researchers may
wise, many nurses will, depending upon their inter- fail to notice that the existence of close alternative
ests, also encounter difficulty with Peirces developed explanations potentially undercuts belief justification
theory of scientific method. and, it is here asserted, this problem is exacerbated
where strongly held values foreclose upon or limit the
examination of other explanations. Further, while
Abduction and nursing
all observation is theory laden, even hypotheses
Nurses are not often troubled by bean allocation grounded upon uncontested factual claims are in
dilemmas and they rarely operate as practitioners or the social/human world frequently unavailable for
researchers in circumstances that easily lend them- testing.
selves to formal experimental deductive and induc-
tive testing. More realistically clinical nurses work in
The generality problem
extremely dynamic environments and, in these cir-
cumstances, belief formation and belief justification Problems in choosing between alternative explana-
becomes, relative to bean exemplars and simplifying tory options are not confined to arguments built upon
natural scientific assumptions, exponentially more abductive inference. However, for Peirce, hypothesis
complex. generation is tied to the affective or sensuous
With regard to research, belief formation outside element of thought (DIH, p. 199, emphasis in origi-
of the natural sciences (i.e. branches of sociology and nal) and, likewise, Haig (2005) recognizes that abduc-
psychology together with the humanities) is problem- tion entails both psychological and logical processes.
atic not least because both society and its members Thus, while belief formation inescapably involves
are open insofar as they have sui generis powers that some element of personal or subjective understand-
preclude closed or laboratory-like natural science ing, abductive reasoning is particularly dependent
experimentation and, therefore, although human upon the idiosyncratic reflective abilities of abductors

2012 Blackwell Publishing Ltd

Nursing Philosophy (2012), 13, pp. 244256
250 Martin Lipscomb

and this is troubling for; adapting Bishop (2010), first, thought that, over the long term, scientists might
let it be assumed that every plausible abductive move towards justifiable truth claims even if, at any
hypothesis is reflectively derived where reflection particular moment, all claims are necessarily fallible
signifies that a belief is justified on the basis of and, it could be argued, analogous reasoning might be
. . . [personal] knowledge . . . arrived at . . . as a result deployed against strong versions of the generality
of a highly (but not perfectly) reliable way of reason- problem (i.e. knowledge may be subject over time to
ing (p. 286). Then, second, in such instances belief some form of Darwinian evolutionary fitness test
justification necessarily rests on procedural or reliabi- though Howson, 2000, again problematizes this).
list criteria i.e. the justificatory status of a belief is Nonetheless, focusing upon reflectively derived belief
merely a measure of the reliability of the process that forming processes allows this interpretation of Bishop
produced it or, put another way, reliabilism asserts (2010) to highlight important truisms namely, alter-
that a belief is justified to the extent that it is acquired native hypotheses can always be reflectively abducted
by reliable processes or methods (Haig, 2005, p. 383). from available evidence and alternative evidence can
Third, since the number of variants in belief-forming always be sought to support any particular hypothesis.
process-types (Bishop, 2010, p. 285) is in all instances For example, we might imagine transcribing dia-
unlimited or, if this is disputed, at least extremely logue capable of being interpreted as anxiety or
large (recognizing that hypotheses vary in plausibility aggression. But which interpretation is correct or,
Boutilier & Becher, 1995) it is, fourth, inevitable that more accurately, which interpretation best explains
a range of close or plausible and distant or implau- what was said? In my own work I have spent long
sible alternative explanatory hypotheses are available periods of time reworking theme, code, and category
at each stage in the reasoning process. Given this, fifth, descriptions around the concept (eventually labelled)
on strictly procedural or reliabilist grounds, there is professionalism. I finally chose descriptors that, in
no clear way, no non-self referential way, of choosing my opinion, felt accurate. But I was aware that
on principled grounds between the validity of alter- earlier iterations, earlier interpretations, were also
native plausible hypotheses and, sixth, an abductive compatible with the data.
claim reflectively arrived at can only be, in the last The generality problem is not, to restate, a problem
instance, asserted rather than demonstrated. only of abductive reasoning. However, when the gen-
erality problem is encountered: Since there are true
unjustified beliefs and false justified beliefs the
Bishops challenge and
process relevant for determining the justificatory
qualitative research
status of the belief cannot be the token that produced
The generality problem, as it is known, applies when- it (Bishop, 2010, p. 286). This lends credence to the
ever a beliefs justificatory status is, or is taken to be, idea that abductively derived hypotheses require
a function of the reliability of its production (Bishop, support from deductive and inductively gained logics/
2010, p. 287) and, if this is accepted, then reflectively evidence if they are to hold and, as noted, for Peirce
derived beliefs cannot be accepted on procedural or the second (deductive) stage in scientific method
reliabilist criteria because, as Bishop (2010) notes, specifies that data other than that which led to
there is no principled means of specifying why any hypothesis formation be used to test that hypothesis.
belief forming process is necessarily correct in a par- Perhaps all this says no more than that abductive
ticular instance. This is a strong claim and it can be inference is defeated by superior (i.e. better supported
contested. Justification need not be ultimate or deci- or more explanatory) accounts and that, to have cred-
sive and, in many instances, sufficiently probable ibility, abductively derived conclusions need to be sus-
(Boutilier & Becher, 1995, p. 44) is perhaps good tained by deductive and inductively derived evidence.
enough (though note, Howson, 2000, dismisses this However, this leaves open the question of how we can
sort of argument an argument he associates with pin down what better supported or more explana-
naturalized epistemology, p. 113). Further, Peirce tory mean. These sorts of question raise larger episte-

2012 Blackwell Publishing Ltd

Nursing Philosophy (2012), 13, pp. 244256
 Abductive Reasoning 251

mological issues than a single paper can grapple with. important claim. It implies that data are not mined
Yet while evidence counts, evidence or facts cannot (i.e. understanding is not extracted if this simply
to restate be identified outside of theory and, as the means making overt what exists but is hidden) and,
generality problem makes clear, multiple plausible instead, qualitative data (broadly defined) provide
theories or hypotheses are always available to give researchers with raw material from which interpretive
form to evidence. (Hayek wryly notes that:empirical constructions are made. This, it must be noted, is not
studies taken alone are seldom decisive in determining ontologically implicative. It does not force upon
the facts of the matter in Caldwell, 2005, p. 380.) researchers idealist or postmodern perspectives and
Thus, the force of Bishops (2010) challenge remains the notion of construction presented above need not
and is particularly felt when belief rests on abductive alarm realists. The quotation does, however, acknowl-
assumptions that are in a social/humanities (nursing) edge that researchers work imaginatively to create
context, in practice and possibly in principle, necessar- meaning from data and it is proposed that the logic
ily subjective and formed in relation to experience that informing this process involves abduction.
is unrepeatable, unquantifiable and untestable and However, if qualitative researchers do employ
where, often, facts and values elide (i.e. qualitative abductive reasoning when analysing and interpreting
research). data then, given the aforementioned generality
problem, what can we say about belief formation and
belief justification in qualitative studies? As described
Abductive inference in
the generality problem insists that where a beliefs
qualitative analysis
justificatory status is a function of the reliability of its
Richardson & Kramer (2006) argue that abductive production, and where belief formation rests on
inference describes, in grounded theory, the process process-types that cannot on procedural or reliabilist
of associating data with ideas (p. 500). Abduction or criteria distinguish on principled grounds between
hypothesis formation occurs when researchers, during unlimited or extremely large ranges of potential or
data analysis, shape or insightfully generate beliefs possibly available hypotheses, then we might conclude
about the meaning or significance of their data on that findings based upon abductive hypotheses are, to
non-deductive and non-inductive grounds (Boutilier put it crudely, asserted and not demonstrated.They are
& Becher, 1995). Here it is suggested that abductive not defeasible (Koons, 2011). Moreover, this issue
inference does not simply underpin the creation or problematizing the plausibility of researcher analysis
development of findings in specific versions of and the beliefs derived from analysis applies not only
grounded theory but, rather, abduction may describe in regard to a studys main conclusions, it also applies
the inferential form that underpins many types of at each intermediate step in the production of conclu-
thematic qualitative analysis. Specifically, when quali- sions where these steps rest upon or involve abductive
tative researchers, during analysis, develop themes, inference (i.e. potentially at every stage in theme, code,
codes, and categories that structure data they are, in and category development).
part at least, insightfully abducting or, to invoke For UK nurses who must deliver care based on the
Peirce, they are making fair guesses about the best available evidence (Nursing and Midwifery
meaning of data. Council, 2008, p. 7) this is problematic. Nurses as
This assertion might be challenged. However, let us members of a practice-based discipline do not read or
accept it as a possibility. What then is the status of consume research simply because it is interesting.
data in qualitative analysis? To support their position Rather, research is accessed because it can explain,
Richardson & Kramer (2006) approvingly cite Coffey describe, enlighten, aid understanding or otherwise
& Atkinson (1996) who note that: Our important offer instruction about aspects of practice. Reading
ideas are not in the data, and however hard we research reports can alter belief and in so doing, insofar
work, we will not find those ideas simply by scrutiniz- as belief informs behaviour, research is action guiding.
ing our data ever more obsessively (p. 155). This is an Evidence is not of course synonymous with research.

2012 Blackwell Publishing Ltd

Nursing Philosophy (2012), 13, pp. 244256
252 Martin Lipscomb

However, evidence is most clearly action guiding when, Alternatively, it may be that researchers do under-
emerging from research, it is part of a credible and take the steps suggested here (i.e. they consider alter-
logically cohesive argument that in important (though native options) but, because of constraints imposed
difficult to specify) respects demonstrates rather than by journal word allowances, these details become lost
merely asserts conclusions and, if qualitative research in the writing-up process. Moreover, researchers do,
cannot do this (because the logic of its argument incor- of course, recognize that more than one interpreta-
porates abductive reasoning), then perhaps its findings tion of data is possible and, when teams analyse data,
are of indeterminate use value. various consensus forming strategies are described in
Qualitative researchers would presumably object recognition of this fact (or rather they are stated, the
to this last statement. However, to refute the claim detailed process of consensus formation is generally
that qualitative research is of indeterminate use brushed over). However, in the absence of evidence
value when the logic of its construction rests upon to the contrary we might conclude that these steps are
abductive inference, it is necessary to specify how the not taken and, since qualitative nurse researchers do
problems thrown up by abduction (e.g. the generality not appear to take this matter seriously it is reason-
problem) are being addressed and, as far as I can see, able to suppose that its significance the threat to
this is not being done. Thus, despite having reviewed a belief justification, the threat to the use value of such
reasonably large albeit unquantified number of quali- work is unrealized.
tative nursing research reports, I am unable to locate A retort to this argument is that, to paraphrase
instances of researchers acknowledging or substan- comments made during the writing of this paper,
tively engaging with the possibility that alternative were you me, had you been there, had you experienced
explanations, alternative hypotheses specifically that situation, you would know that this interpretation
here of theme, code or category descriptions could is accurate. This defence has at least two components.
be offered of their data. That is, and mindful of the First, the researcher making the statement clearly felt
problem described earlier when multiple legitimate that experience counts. However, as Hempel (1945)
options were presented in relation to bean allocation, recognized, having a sense of evidence, or a feeling
I cannot find instances of researchers stating the of plausibility (p. 8) regards evidence is an inad-
range of possible interpretations of data that they equate justification for any hypothesis involving, as it
considered (e.g. on these grounds interpretations a, does, a confusion of logical and psychological consid-
b and c appeared plausible as code descriptions) erations (p. 8). Second, it might be argued that the
and I cannot find instances of how any particular statement implicitly allows that abductive data inter-
choice was justified (i.e. against these named criteria pretation or hypothesis making can be justified or
interpretations a, b and c were considered and supported by either formally articulated and identi-
option b was chosen as being the best code fied background or auxiliary theories or, less formally,
description because etcetera). by a priori assumptions about the world (subjective
Perhaps this is to demand too much or, perhaps, this contextualized understandings).
is to demand something foolish? Theme, code, and Re-emphasizing a point made earlier, Evers & Wu
category descriptors are rarely if ever formulated as (2007) highlight the role of background theory [and
propositional statements and the logic by which they one might add here values] in helping to adjudicate
are derived a logic that may be based on abduction the matter of best explanation. Strictly speaking, for
is almost never described and it is possibly only in inferential purposes hypotheses never occur in isola-
this form that they could be defended in the manner tion but rather as embedded in some theoretical
outlined above. Further, presenting this type of detail/ context (p. 207). This is important. Abduction is asso-
explanation runs counter to current research conven- ciated with inference to best explanation and it might
tions that favour the advancement of unambiguous be argued that explanatory power or goodness is
singular truths or research outcomes (reviewer linked in some way to the fittingness of explanation
suggestion). to existing theories/ideas about the world. In this vein

2012 Blackwell Publishing Ltd

Nursing Philosophy (2012), 13, pp. 244256
 Abductive Reasoning 253

critical realists permit conclusions to be drawn confirming or at least contrasting theories have been
because they best explain available evidence and, considered and on good grounds rejected and, in a list
from this perspective, abduction is associated with the that could be extended, the possibility that other
number and sophistication of available background interpretations cannot be equally supported by the
theories. same or other theories should be considered.
More accurately critical realists substitute the Pragmatically (using the term in a colloquial sense),
terms retroduction and retrodiction for abduction it may be reasonable to assume that straightforward
and, although the following definitions are not or non-contentious hypotheses do not require such
applied consistently, retroduction and retrodiction detailed justification. Yet, not only are the descriptors
describe forms of abduction wherein differing quan- straightforward and non-contentious question
tities of background theory are held or applied in begging, it is also of note that within the nursing lit-
relation to study objects. Thus, retroduction occurs erature this level of engagement is hardly ever
when knowledge of mechanism function is relatively demonstrated and, therefore, the idea that (almost
impoverished i.e. when we believe we know little regardless of quantity or objectivity) background
about what is happening in relation to the study theories usefully offer justificatory support to abduc-
object (few background theories are available) and tively inferred hypotheses remains underdetermined.
retrodiction occurs when knowledge of mechanism It could be the case that they can. However, it is not
function is relatively advanced i.e. when we believe shown and, insofar as all theory is epistemically
we have good reason to think we know what is going underdetermined, it may not be showable (Stanford,
on or happening in relation to the study object (many 2011).
background theories are available) (Professor S. Another response to the challenge of justifying
Fleetwood, personal communication). However, as qualitatively derived belief based on abductive infer-
Howson (2000) observes: What is plausible and what ence might be that, where researchers interview
is not is likely to be a highly context-dependent affair, different participants consecutively (or the same par-
if not an outright subjective one, unless it is tied to ticipant on consecutive occasions) then an abductive
some objective standard (p. 114). Objective standards hypothesis formed in relation to one interview could
regarding beliefs about background or auxiliary theo- be checked or tested against the results of subse-
ries are thin on the ground in much qualitative work quent interviews. Temporally later interviews might
and the plausibility of abductively derived convictions then be seen as new data (or later comments in a
therefore remains problematic. And so, on balance, single interview might substantiate earlier abductive
and to repeat points made previously, the question hypotheses though this raises its own problems)
how are we to know what the best explanation is? and, following Peirce, it could be argued that this
remains undecided. meets the need for supportive data to be other than
Nevertheless, let us allow that highlighting the sig- that which generated the initial hypothesis.
nificance or role of background theory in adjudicating However, first, most qualitative research is small in
between discrepant explanatory hypotheses may scale and qualitative researchers often both collect
prove useful if the presence or application of such and analyse data. Full or detailed analysis may occur
theories supports particular belief forming processes. after all interviews are concluded and, when this
While it is attractive to imagine that this might occur, occurs, researchers cannot read or interpret a tran-
it is more difficult to demonstrate the link in concrete script unaware of the content of other interviews. The
form. Thus, merely because theory x exists and, in researchers encompassing or totalizing perspective
relation to the data being analysed, theory x appears may be portrayed as a strength in qualitative work.
to support interpretation y we cannot conclude that Yet because every act of interpretation is performed
y is in fact an adequate interpretation or hypothesis. in full knowledge of all data the possibility that
Before this occurs it must be established that x does themes, codes, and categories can be abducted from a
indeed support y. It must be shown that other dis- transcript as if that data were a separate and distinct

2012 Blackwell Publishing Ltd

Nursing Philosophy (2012), 13, pp. 244256
254 Martin Lipscomb

entity must be discounted. Second, if, somewhat or even most instances our suppositions about the
oddly, full and final analysis did occur at the close of world prove to be correct or, at least, correct enough.
each interview (i.e. before the next interview was con- Generally, though reason may not authorise such
ducted) then this is not disclosed and I am unable to suppositions, they prove sufficient for action.
locate examples of reports in the nursing literature Abduction in research is, however, another thing.
that engage, in this way, with the problem of alterna- Research processes include or require the demonstra-
tive or multiple theme, coding or category options. tion of logical competence and if this competence is
For example, I cannot see statements such as: when not evident then findings are not carried and they
interpreting interview 1 three code descriptors a, must be rejected.
b and c appeared plausible. In interview 2 each of A critique of my own work, where interview tran-
these code descriptors was checked against responses script analysis incorporates abductive reasoning, led
in the following manner etcetera. Thus, as with back- me to this subject. And concern with the validity of
ground theories, although this approach may be the interpretative process in that work directed me to
viable, its effectiveness or utility in meeting what is Peirce. Though I would prefer it otherwise it is diffi-
here termed the generality problem remains to be cult not to conclude that abduction involves vis--vis
shown. belief formation and belief justification, a perilous
step (DIH, p. 192) and, cognizant of the especially
fallible, provisional and contested nature of abductive
Concluding comments
inference Peirce noted that:
For Audi (2011), the uncritical tend to believe too
As a general rule, hypothesis [abduction] is a weak kind of
much that is unsubstantiated; the overcritical tend to
argument. It often inclines our judgement so slightly toward
believe too little that is true (p. 5). Rholm (2010a) is
its conclusion that we cannot say that we believe the latter to
not uncritical; however, if Rholm (2010a) underplays
be true; we only surmise that it may be so. (DIH, p. 189)
the problems inherent in abductive reasoning the
danger in this essay is that its difficulties are over- No research or evidence, no truth directive claim, is
blown. Focusing on the nature of belief formation and ever infallible or immune to refutation/revision.
belief justification might imply or suggest that mental However, surmising that something may be so is not
processes and the conduct that these processes allow the same as demonstrating that it is so (or is likely to
can and should always be logical, rational, reasoned, be so) and, thus, if qualitative researchers do indeed
and justifiable. This is not my position. use abductive inferential forms when analysing data
Peirce (1998a) claimed that: The elements of every then, because they can only surmise on the basis of
concept enter into logical thought at the gate of per- such reasoning, the practical or real world use value of
ception and make their exit at the gate of purposive qualitative studies for nurses is diminished as the
action; and whatever cannot show its passports at process of belief formation cannot sustain or justify
both those gates is to be arrested as unauthorized by substantive action guiding claims.
reason (p. 241). That said a good deal of our knowl- It might at this point be objected that qualitative
edge is possessed and used by us without ever being studies are only action guiding insofar as they gener-
articulated . . . [and whilst] tacit beliefs contain tacit ate insight or new understanding in the mind of the
error as well as tacit knowledge (Gray, 2007, p. 40); reader. However, first, it is a moot point whether
nevertheless: The appropriateness of our conduct is research consumers (readers) can obtain genuine
not necessarily dependent on our knowing why it is insight or understanding on the basis of mere suppo-
so. Such understanding is one way of making our sition and, second, researchers probably want to claim
conduct appropriate, but it is only one way (Hayek, that their findings are more than just plausible
2006, p. 58). Thus, despite the problems outlined guesses. Thus, following Hayek (2006), while in every-
above, in everyday life, perfectly sensible abductive day life it is not necessarily the case that we ought to
inferences are made by every one of us and in many believe only what has been demonstrated to be true

2012 Blackwell Publishing Ltd

Nursing Philosophy (2012), 13, pp. 244256
 Abductive Reasoning 255

(p. 57); nonetheless, research findings, cannot be

References
accepted they cannot be action guiding, in the
absence of adequate demonstration. Anderson D.R. (2009) Peirce and Cartesian Rationalism. In:
To address the inherent weakness of abductively A Companion to Pragmatism: Blackwell Companions to
Philosophy (eds J.R. Shook & J. Margolis), pp. 154165.
derived claims in qualitative work it might be pro-
Wiley-Blackwell, Blackwell Publishing Ltd, Oxford.
posed that, following Peirces three stage scientific Audi R. (2011) The ethics of belief and the morality of
method, if the findings of qualitative researchers are action: intellectual responsibility and rational disagree-
to inform nursing practice in a meaningful way then ment. Philosophy, 86(335), 529.
abductive supposition must be supported by deduc- Bhaskar R. (1997) A Realist Theory of Science, 2nd edn.
Verso, London.
tive and inductively sourced evidence which, impor-
Bhaskar R. (1998) The Possibility of Naturalism: A Philo-
tantly, differs from that which generated the initial sophical Critique of the Contemporary Human Sciences,
hypothesis (this would engage with the generality 3rd edn. Critical Realism Interventions Series. Routledge,
problem).Alternatively, this might seem controversial London.
or unnecessary insofar as qualitative research does Bishop M.A. (2010) Why the generality problem is every-
not make generalizing claims and the requirement is, bodys problem. Philosophical Studies, 151, 285298.
Boutilier C. & Becher V. (1995) Abduction as belief revi-
moreover, clearly problematic. Further, deduction
sion. Artificial Intelligence, 77, 4394.
and induction are, as inferential forms, themselves Bowling A. (2009) Research Methods in Health: Investigating
subject to critique and many aspects or elements of Health and Health Services, 3rd edn. Open University
the social world i.e. common nursing interests do Press, McGraw-Hill Education, Maidenhead.
not, as previously stated, lend themselves to the types Burch R. (2011) Charles Sanders Peirce. The Stanford Ency-
clopedia of Philosophy, Fall 2010 edn. (ed. E.N. Zalta).
of formal testing or experiment associated with these
Available at: http://plato.stanford.edu/archives/fall2010/
kinds of reasoning. Nevertheless, it is not perhaps entries/peirce/ [accessed 26 January 2011].
entirely unreasonable to expect that qualitative Caldwell B. (2005) Hayeks Challenge: An Intellectual Bio-
researchers should take the logic of their endeavours graphy of F. A. Hayek. University of Chicago Press,
seriously and perhaps the above assertion should not London.
simply be dismissed out of hand. Coffey A. & Atkinson P. (1996) Making Sense of Qualitative
Data: Complementary Research Strategies. Sage, London.
This paper discusses a relatively underexplored
Cormack D. (ed.) (2000) The Research Process in Nursing,
topic, abductive inference in qualitative research, and 4th edn. Blackwell Science, Oxford.
it suggests that the acknowledged fallibility of this Davies M.B. (2007) Doing A Successful Research Project:
form of reasoning, a fallibility illustrated by the gen- Using Qualitative or Quantitative Methods. Palgrave Mac-
erality problem, must, because it effects belief forma- millan, Basingstoke.
Downward P. & Mearman A. (2007) Retroduction as mixed-
tion, belief justification and hence research use
methods triangulation in economic research: reorienting
value, be recognized. This is not an attack upon quali- economics into social science. Cambridge Journal of Eco-
tative work (kindred problems infest quantitative nomics, 31(1), 7799.
studies). Rather, since more conceptual work needs to Evers C.W. & Wu E.H. (2007) On generalising from single
be done here, the paper simply aims to provoke and case studies. In: Philosophy, Methodology and Educa-
stimulate discussion. tional Research (eds D. Bridges & R. Smith), pp. 199213.
Blackwell Publishing, Oxford.
Gordon W., Morton T. & Brooks G. (2005) Launching the
Acknowledgements Tidal Model: evaluating the evidence. Journal of Psychi-
atric and Mental Health Nursing, 12(6), 703712.
While the views expressed herein are not necessarily Gray J. (2007) Enlightenments Wake: Politics and Culture at
theirs I would like to thank the following people for the Close of the Modern Age. Routledge Classics, Oxon.
Hacking I. (2007) On not being a pragmatist: eight reasons
kindly giving their time, advice, and encouragement
and a cause. In: New Pragmatists (ed. C. Misak), pp. 3249.
during the writing of this paper: Professor Margaret Oxford University Press, Clarendon Press, Oxford.
Miers, Abby Sabey, Dr Theresa Mitchell, Paul C. Snel- Haig B.D. (2005) An abductive theory of scientific method.
ling and, very importantly, the unidentified reviewers. Psychological Methods, 10(4), 371388.

2012 Blackwell Publishing Ltd

Nursing Philosophy (2012), 13, pp. 244256
256 Martin Lipscomb

Haig B.D. (2008) An abductive perspective on theory con- Peirce C.S. (1998a) Pragmatism as the logic of abduction
struction. Journal of Theory Construction and Testing, (Lecture VII Harvard Series). In: The Essential Peirce:
12(1), 710. Selected Philosophical Writings Volume 2 (18931913)
Hayek F.A. (2006) The Constitution of Liberty. Introduction: (ed. N. Houser), pp. 226241. Indiana University Press,
I.M. Stelzer. Routledge Classics, Oxon. Bloomington and Indianapolis, IN.
Hempel C.H. (1945) Studies in the logic of confirmation (I). Peirce C.S. (1998b) Sundry logical conceptions. In: The
Mind: A Quarterly Review of Psychology and Philosophy, Essential Peirce: Selected Philosophical Writings Volume
LIV(213), 126. 2 (18931913) (ed. N. Houser), pp. 267288. Indiana Uni-
Hempel C.H. (1966) Philosophy of Natural Science. versity Press, Bloomington and Indianapolis, IN.
Prentice-Hall, Englewood Cliffs, NJ. Peirce C.S. (1998c) The nature of meaning (Lecture VI
Howson C. (2000) Humes Problem: Induction and the Jus- Harvard Series). In: The Essential Peirce: Selected Philo-
tification of Belief. Oxford University Press, Oxford. sophical Writings Volume 2 (18931913) (ed. N. Houser),
Josephson J. (1997) What is abduction? Abductive reasoning pp. 208225. Indiana University Press, Bloomington and
information page abductive inference in reasoning and Indianapolis, IN.
perception. Available at: http://www.cse.ohio-state.edu/ Polit D.F. & Beck C.T. (2006) Essentials of Nursing Research:
lair/Projects/Abduction/abduction.html [accessed 27 Methods, Appraisal, and Utilization, 6th edn. Lippincott
January 2011]. Williams and Wilkins, London.
Koons R. (2011) Defeasible reasoning. The Stanford Ency- Rholm M. (2010a) Abductive reasoning and the formation
clopedia of Philosophy, Winter 2009 edn. (ed. E.N. Zalta). of scientific knowledge within nursing research. Nursing
Available at: http://plato.stanford.edu/archives/win2009/ Philosophy, 11, 260270.
entries/reasoning-defeasible/ [accessed 27 January 2011]. Rholm M. (2010b) Theory development and the logic of
Lawson A.E. (2002) What does Galileos discovery of Jupi- discovery. International Journal for Human Caring, 14(3),
ters moons tell us about the process of scientific discov- 3642.
ery? Science and Education, 11(1), 124. Richardson R. & Kramer E.H. (2006) Abduction as the type
Lawson A.E. (2010) Basic inferences of scientific reasoning, of inference that characterizes the development of a
argumentation and discovery. Science Education, 94(2), grounded theory. Qualitative Research, 6(4), 497513.
336364. Sther B. (1998) Retroduction: an alternative research strat-
Lipscomb M. (2012) Questioning the use value of qualita- egy? Business Strategy and the Environment, 7(4), 245
tive research findings. Nursing Philosohy, 13, 112125. 249.
Nursing and Midwifery Council (NMC) (2008) The Code: Stanford K. (2011) Underdetermination of scientific theory.
Standards of Conduct, Performance and Ethics for Nurses The Stanford Encyclopedia of Philosophy, Winter 2009
and Midwives. NMC, London. edn. (ed. N. Zalta). Available at: http://plato.stanford.edu/
Parahoo K. (2006) Nursing Research: Principles, Process and archives/win2009/entries/scientific-underdetermination/
Issues, 2nd edn. Palgrave Macmillan, Basingstoke. [accessed 19 January 2011].
Peirce C.S. (1992) Deduction, induction and hypothesis. In: Thagard P. (1993) Computational Philosophy of Science.
The Essential Peirce: Selected Philosophical Writings MIT Press, Bradford, Cambridge, MA.
Volume 1 (18671893) (eds N. Houser & C. Kloesel), pp. Wilson M. (2005) Institutionalism, critical realism, and the
186199. Indiana University Press, Bloomington and critique of mainstream economics. Journal of Institutional
Indianapolis, IN. Economics, 1(2), 217231.

2012 Blackwell Publishing Ltd

Nursing Philosophy (2012), 13, pp. 244256
Copyright of Nursing Philosophy is the property of Wiley-Blackwell and its content may not be copied or
emailed to multiple sites or posted to a listserv without the copyright holder's express written permission.
However, users may print, download, or email articles for individual use.