apologies for the late reply, I currently work on stone-age equipment.
Still, I feel misunderstood, so I want to clarify some things:
Quoting Lyn Richards <[log in to unmask]>:
> [Sob], no, it certainly didn't need software to create as "probably the
> most frequent misguidance of research ... the temptation to substitute
> analysis with coding"!!
I take your point here, thankfully I am too young to have seen loads of
coding done "by hand." Just, it seems too tempting to assume the analysis
has been done, when there is a aesthetically neat coding tree and colorful
ribbons in the margins rather than piles of paper and cards scattered on
So, even if CAQDAS have exposed the peril of coding substituting for
analysis (a deed I was shamefully unaware of), they certainly have not
solved the problem. And if its a good CAQDAS it will indeed never solve
this problem, because if they would, they would not effectively assist the
> Sorting stuff into heaps is always easier than tackling analysis. We all
> do it - this isn't a smug comment about novices! What I call "topic
> coding" hits all of us when data build up; for very good reasons,
> whatever our epistemologies and theoretical approaches, we do the more
> clerical job of managing what we do know whilst we try to find out what
> we don't. But software has also clearly distinguished different sorts of
> and purposes for coding - and I think that's hugely helpful. It
> separates out the storing of descriptive coding (attributes) and "topic
> coding" (this is about that) from more "analytical coding" (I'm
> gathering here material about a hunch, or some idea that I will revisit,
> rethink, recode, re-view). Every researcher, whatever their
> Epistemologies and Theoretical Approaches, is helped by being able to do
Two things: I am not so sure, if "analytical coding" can really be that
sharply distinguished from "descriptive coding". Anytime you deem something
relevant for coding, and be it "who's talking" or "which sex/gender," it
carries some implicit analytic idea.
Second, of course, every researcher is helped by tools, which allow him or
her to effectively perform certain tasks, which under certain circumstances
become relevant. Just like any researcher is helped by having SPSS at their
disposal. I still would not advise somebody doing Discourse Analysis to use
SPSS. And I still would say, that at times, Excel and/or
Cardfile beats any CAQDAS for sorting some types of data in certain ways.
> The goal of all qualitative methods was always to keep
> thinking at the more abstract level.
I am not so sure about this. There is loads of quantitative research that
necessarily remains on a very abstract level (how more abstract can you get
than pure rational choice), while there is a lot of self-
professed qualitative work that does not adhere to a very generic theory at
I would even hazard a guess that there is an elective affinity between
quantitative methods and abstractions, because in order to pile up numbers
you usually must come up with some very generic categories to lump together
all sorts of very different social phenomena.
> And the computer allows us to keep
> asking questions, to code on to new more abstract categories and think
> about those, so first coding can be much more a first step.
In fact, as Susanne already pointed out, it *is* more than a first step,
and that can be a virtue as well as a vice.
On the plus side, it forces us to order things in a certain way, which is a
prerequisite for any theory, but on the other hand, as I already mentioned,
in particlular "in vivo" coding easily leads to inductive theorizing.
> And I think we need clearer thinking about software tools and their
> relation to users. I am hugely saddened by anyone asserting that because
> qual software will, inter alia, code, and coding can be misused, a
> researcher is better off not using software.
Can you please point to the place, where I advise against using software?
This was certainly not my intention. All I am advising against is *starting
out with a software* and then choose the appropriate method.
Let's face it: NVivo does a pretty lousy job computing structural equation
models and SPSS is not very good at coding videos. These are extreme cases,
but there are borderline cases, too. These borderline cases are less
obvious and therefore more interesting. All I did was to hazard a typology,
which software is appropriate for which method.
With the exception of Conversation Analysis, I think my typology is still
standing, btw, I am yet to see a Comparative Research in the vein of Charles
Ragin and his associates, which would be better served with CAQDAS than with
fs/QCA and a spreadsheet and/or database program. I also still wait for
examples to use CAQDAS efficiently in *all* types of Discourse Analysis.
Don't get me wrong: I am not against using CAQDAS in principle for these
methodologies, I just don't see, how it can be done.
Even in Conversation Analysis, a mthodology I know next to nothing of, it
seems that coding videos was the biggest plus for CAQDAS. Again, I am
unsure, if Transana would not do an equally good or better job than most
CAQDAS, some of which do not even handle videos.
> Or worse, that they should
> not use what qual software does do because it doesn't do things it
> doesn't do!
If the things CAQDAS do are irrelevant for certain methodologies, or, if
they can be performed better by alternative software, why should you use
them for these methodologies?
If I say that SPSS is not suited for latent class analyis, do I advise
against its use in regression analysis? Of course not. So, when I say that
for many comparative analyes fs/QCA is a better choice, that also does not
mean that I say throw out CAQDAS for any analyses.
> MSWord does a heap of things that bug us all and it doesn't
> create books.
Fair enough, but you are probably aware that there is an article (in FQS
AFAIR), which asseses -- quite favorably -- MS Word not only for writing
articles, but as a tool for content analysis. Would you say that we should
therefore dispense of CAQDAS and instead write little Word macros for the
tasks we need to perform? I don't hope so. For most textual analysis, there
are much better tools than Word, but that does not make Word a bad program
(to be sure, there are other things that do make it a bad program).
> But would you tell your student to handwrite (and
> re-write) a thesis on paper because they might be distracted by the way
> Word will Show Changes if they turned on that option? Surely you'd show
> them turn it off, or how to use it sometimes, and benefit from it? And
> you'd not suggest that the word processor won't help them because it
> won't write the thesis?
The difference is, that *regardless of methodology*, you are supposed to
write papers, but the functions CAQDAS perform are not entirely methodology
neutral. Sure, it's a good thing, when students are familiar with CAQDAS,
just as it is a good thing, when they are familiar with statistical software
packages, or with Word. But that is besides the point I am trying to make
thomas koenig, ph.d.
department of social sciences, loughborough university, u.k.