Dear Steve,

In answer to your questions:

> (a) ability to output case by code
> matrix for analysis in spss (e.g., analyzing co-occurrence of codes),

As far as output to SPSS goes, NVIVO is far superior to Atlas.  In Atlas you
are just limited to exporting a table to SPSS where each case represents a
quotation (or passage).  There is a way in SPSS to aggregate the quotations
so that you can get each case to represent a document but it is a very
awkward interface.

NVIVO has a very slick interface with SPSS.  In addition, there are a whole
variety of matrices you can produce - not just case by code.  You can
produce matrices to look at patterns in coding between people who have
different demographic characteristics and more.  The interface works two
ways which is nice for studies which combine qualitative and quantitative
studies.  For example, you may have done a survey with the people you have
had in-depth interviews.  You can import from SPSS the answers to closed
ended questions in the survey.  If you had asked an attitude question scored
1 to 5, you could contrast in NVIVO those who answered 1 with those who
answered 5 with what they had to say about a particular theme in the
in-depth interview.  You cannot do this at all in Atlas as it is only
possible to export out of Atlas into SPSS and not the other way round.

> and (b) ability to code network data, as in looking through text and
> coding a social tie such as friendship between specific pairs of
> individuals. For example, a sentence like "John called his friend
> Michael" is coded for presence of John, presence of Michael, and
> friendship link between John and Michael.

With either Atlas or NVIVO you can code the way you describe above.  The
differences lie in the way you can retrieve that information in a meaningful
way.  With NVIVO you would use its Search Tool once its coded.  You would be
able to pull out friendship restricted to Michael and John.  You would get
the text that relates to that. The results are not just a report which you
can print out and read but are also given as a node (in NVIVO speak) which
means it can be part of your database.  As it becomes part of your database
you can then restrict your analysis to that text and ask further questions
about Michael and Johns' friendship.  This does not happen in Atlas -
results are just a report.  Also the Query Tool has some problems and it is
not as powerful as NVIVO's Search Tool.

However, in Atlas you have a Network Tool - so you could go through a
transcript and link passages together - e.g. this passage contradicts this
other passage but supports this other one.  You can customise the links so
you can label them whatever you want.  Then you can see these passages in a
network view with the links labelled - taking  the passages out of the
linearity of the text and display them as a network.  This is Atlas'
strength - this feature does not exist in NVIVO although it has a modeller
which works differently.

Below is a comparison of Atlas and NVIVO.

Comparison of Atlas and NVIVO

Atlas's strength is in its network tool. So I recommend it when the focus of
the analysis is on the different kinds of links there can be between
categories or when you want to be able to pull out that this bit of text
contradicts that bit of text but supports this other bit.  You can see that
all displayed in a network view which makes relations within the text very
transparent.  You can also pull out sections of the video and display how
different sections in the video are related to each other in the network
view.

NVIVO's strength is its Search Tool which allows you to cut your data in
different ways - pull out different combinations of themes to let you
reflect on what is going on.  It also supports rich text (which Atlas does
not at the moment) which means that you can format your transcripts the way
you want -
using different fonts and font sizes, bold, italics, underline, colour to
mark up your transcripts. You can also start your documents directly in
NVIVO - bypassing the wordprocessor completely.  This is very handy for
observations which you can type directly into NVIVO and later add to them.
It is also useful to keep your research journal in NVIVO.  You can then use
hyperlinks to the points in your data which support the analysis you are
developing. NVIVO also has a modeller where you can draw models.
You can not edit your transcripts or add to them once they are in ATLAS.

Both Atlas and NVIVO run only on pcs.  They both require a high
specification
machine.  I recommend a minimum of 128
MB RAM (they will run on less but will be very slow).  Also for both
packages,you should have a high resolution screen (at least 1024 by 768
pixels) and the larger the screen the better.

NVIVO is not recommended at the moment for projects with a lot of data e.g.
hundreds of documents.  It works best at the moment with projects of up to
300. This will also be a problem for Atlas when the new version comes out
because it will support rich text.  It is the rich text feature which takes
up a lot of computer resources.

NVIVO and Atlas handle videos and audios differently. In Atlas you can input
a video or audio into your project and code by tape count.  But you just get
the selected bit played back at you which may be difficult to manage -
depending on how you intend to handle this type of material.  NVIVO is
better if you are going to transcribe a video or audio as you can then link
to
extracts of the video from important parts of the transcript.  Again it
depends very much on what you want to do with your analysis.

Atlas and NVIVO handle photos differently. In Atlas you can
identify segments of the picture and either code it or write comments on it.
For NVIVO you would have to write a description of the photo which you
could code - then you could insert a hyperlink to be able to open up and
see the particular photo in question.

NVIVO has a more sophisticated way to handle cases and restrict your
analysis to just one case or across cases. It is possible to do this kind of
analysis in Atlas but it is less elegant.  It is also easier in NVIVO to
analyse individuals over time.

NVIVO has a very slick interface with packages which support spreadsheets
e.g. Excel or SPSS.  So it is very easy to import for example information
about your respondents e.g. age, educational level, type of learning
disability etc or to import the
answers to any attitude surveys etc that may have been used in the study.
You can also export any tables which NVIVO produces back into a statistical
package (if it makes sense to do so).  Atlas has an awkward interface with
SPSS and you can only export a limited amount of information. So I would
recommend NVIVO if you are doing a study which combines qualitative and
quantitative approaches.

Recommendation

At the end of the day, I would go for ATLAS if your focus is on the links
between your categories or links between different parts of the text. I
would go for NVIVO if you will use the Search Tool  a lot to pull out
different combinations of codes and restrict the analysis to different
groups of analysis. You would do this for thematic analysis . (ATLAS has an
equivalent tool called the Query Tool but it is not as powerful and does not
work as well as NVIVO's search tool.)  I would also go for NVIVO if you want
to be able to edit directly  the text and use rich text and hyperlinks
between documents and write up your journal, memo and observations directly
in the software.

I hope this helps.

Silvana

Dr. Silvana di Gregorio
SdG Associates
Research & Training Consultants
Tel/Fax: +44-(0)20-8806-1001
E-mail: sdg@sdgassociates.demon.co.uk
Web-site: www.sdgassociates.com