Introduction
QUARC



Software for the analysis of qualitative data - what is that?

CAQDAS

    The acronym CAQDAS stands for Computer-assistend qualitative data analysis software. Unlike statistical analysis packages, CAQDAS does not analyze data. The software is neither more nor less than a tool that supports the process of qualitative data analysis.


What is CAQDAS good for?

    The use of software frees the user from those tasks that a machine can do much more effectively, like searching for words or strings in a text or retrieving coded text passages.

    It becomes easier to analyze data systematically. Also, new ways of analyzing qualitative data become feasible. Even large volumes of data can be structured very quickly and clearly presented .


How does it work?

The basic method can be described as Code-and-Retrieve. Text passages are marked and coded with one or more codewords. These coded text passages can then be retrieved, reviewed, put together in different ways, re-coded if deemed necessary, etc. Coded segments can be retrieved by either using single code words in a search or by combining multiple code words using logical functions. The search process is the stage where questions are asked. Answers will likely be found during the entire time one is immersed in the data, and the analysis progresses step by step. Unfortunately, this process is not as straight forward as it might sound. There will be coding, retrieving, thinking about what one has retrieved, more coding, more retrieving, more thinking, some writing in between, etc. – and without the computer, we had to do it all by using our hands and heads......  

Some additional elements of this process are (or might be):

  • Writing memos and comments  
  • Searching for words, a string of text or text patterns.   
  • Manipulating variables that allow you to combine qualitative and quantitative material, e.g. some programs allow you to use variables as selective criteria in a search  
  • The graphical representation of categories and theoretical concepts in the form of networks or hierarchies.    
    •


    What are the advantages of computer supported qualitative data analysis?

    • Speed  
    • Greater flexibility  
    • More complex analysis processes become feasible  
    • The software offers new possibilities for analyzing qualitative data.
    • With the support of software, data can be analyzed more systematically and according to specific criteria.  
    • Computers make it easy to handle large amounts of data (even though this can also be a disadvantage).  
    • The use of computers increases the status and believability of a qualitative research project (a sad fact – but this is often still the case).    


    What are the disadvantages of using this software?

    • There is the danger of loosing touch with the data
    • You can perform a quick but dirty analysis
    • Not everything can be put into a machine-readable form (e.g. the marking of text passage with colored pens and according to personal preferences).  
    • The positive effect of gaining more manageability can become lost due to high degrees of complexity.  
    • The purpose of using the software can be misunderstood and lead to dull and meaningless analysis results.


    What are the major steps in computer supported qualitative data analysis?

    • Reading the data  
    • Coding and de-contextualizing  
      • a. on the textual level
      • b. on the conceptional level  
    • Re-contextualizing ( = the process of retrieving coded text segments, of exploring these text segments, and of further questioning the retrieved results)  
    • Writing memos and comments    
    • Creating links between various elements of the analysis (e.g. between code words, between text passages, between memos/comments and code words, between documents and memos, between memos and memos, etc.) either within the package itself or outside of it.     
    • The visualization of results            


    What type of software is available for the analysis of qualitative data?

    • "Normal" word processing software  

      • Word processing software allows goal directed but simple searches for words and strings within documents. It is possible, but only in very limited ways, to apply keywords or codewords to certain parts of text. Text passages that are identified as belonging to a certain category can then be saved to different data files and it is possible to sort data that way. Thus, qualitative data analysis is possible with the support of word processing packages, however not much is gained over doing it by hand.  

    • Text Retrievers and Text Managers (e.g. Sonar Professional, FolioVIEWS)  

      • These programs are specifically designed to find words and strings of text in your data. They can do that much faster than word processors since their internal structure is different. In addition, they offer a number of complex search functions and operators. For example, it is possible to mark and sort data as well as to attach memos and comments. Some of these programs support detailed content analysis, offer keyword-in-context display, word lists and counting functions. What they do not support is the coding of user defined text passages.    

    • Rule-Based-Systems (e.g. HyperReserach)  

      • Code-and-Retrieve functions are at the heart of rule-based-systems with a focus on deductive and case-oriented analysis. The analysis process is based on the formulation of rules. These rules are subjected to tests and are then either verified or falsified by the data.    

    • Logic-Based-Systems (e.g. AQUAD)  

      • Similar to rule based systems, logic based systems are most often used for deductive analysis, but they work a bit differently. Logic programming and the computer language Prolog are the basis ofsuch programs. Code-and-Retrieve functions are supported including a number of additional search operations; for example, the possibility to search for positive and negative cases. The results generally are presented in the form of numbers.        

    • Classic Code-and-Retrieve Programs (e.g. The Ethnograph, winMAX)  

      • Nowadays, these programs are no longer simple code-and-retrieve programs. For example, they also offer word and string searches, the capability to write memos and to visually represent the coding scheme in the form of a sideways hierarchical tree structure, the use of variables, the export and import of tables, etc.    


      COMMENT:
      The following programs are often classified as "theory builders". Since this classification can easily be misunderstood, different categories describing these programs are used:  

    • The Index-Based-Approach (Nud*ist)  

      • One feature that is very particular to Nud*ist is the display of the coding scheme or index system in the form of a hierarchical tree structure. The Nud*ist tree is upside-down, with its root at the top. It is possible to freely manipulate the tree, thus the tree nodes and the segments coded at those nodes can be moved. Search results can be added under a new node and become part of the tree. The capability to search text and to write memos and annotations are also integral parts of the program.    

    • Conceptual Network-builders  

      - Based on Code-and-Retrieve (Atlas.ti):

      In addition to facilitating the code and retrieve process, Atlas.ti is capable of exploring data within semantic networks. Other advanced features are also available. Unique to Atlas.ti are the various ways of visualizing and linking the various objects and elements of a project.  

      - Specialized Graphical Display (e.g. Inspiration, Decision Explorer):

      These programs do not offer a code-and-retrieve function. They specialize in model building and diagramming. Inspiration and Decision Explore can be used in conjunction with Nud*ist as a way of combining a code-and-retrieve approach with the graphical representation and manipulation of the results.    


    What are the important functions and features one should look for?

    • Data Entry

      • Is a specific format required or can I import data directly from a word processing package. Is it possible to edit the data once they are imported?    

    • Coding

      • What is the smallest chunk of text that can be coded? Is interactive coding possible? Can I use the mouse for coding, and how are the already coded text segments and the code words displayed? Can I use free codes (i.e. codes that are not connected to any text passages), and is automatic coding possible?    

    • Supported Analytic Approach and Method

      • Is the program variable or case oriented, does it lend itself to an inductive or a deductive type of analysis, is it based on a more hermeneutic, grounded theory, content analytic, etc. approach?    

    • Memos, Comments and Annotations

      • For what kind of objects can they be written and where? How do I know that I have attached or inserted a memo, a comment or an annotation to an object or into a document, is it made visible? Can I link memos, comments, and annotations to other objects?    

    • Search Functions

      • What can be searched and how are the search results displayed? Can I search for words, strings of text or text patterns? When searching for coded text segments, what kinds of operators are available? Can I choose code words from code word lists? Are there options to allow searches only in particular documents or according to specified criteria?    

    • Output and Display Options

      • Can the output be sent to the screen, printer and disk? Can I select only parts of a result for reviewing, editingor saving? Is there an easy way to access the context of search results? Is it easy to identify where the retrieved text passage is coming from, how specific is the source tag?    

    • Links

      • What kind of objects can be linked?    

    • Conceptional Theorybuilding

      • Is it possible to do this within the program? How are links represented - in form of hierarchies, network views, or matrixes?

    • Does the program offer any form of graphical representation?
    • Log Function
    • Can the various steps of an analysis be traced?    
    • Is Team Work supported?  
       
    • Can the software run in a Network environment?    
    • User-friendliness, general support, e-mail discussion groups?  
       

    A few more questions that are important to answer before buying a particular program:  
     

    What kind of computer user I am?

    Do I need the program only for the current project?


    If so, this means that it is possible to choose exactly that program which fits the specific requirements of the current project. If one is buying for the long term, more general considerations might be more important.    

    What is the nature of my data?  

    • What kind of data do I have, just transcribed interviews, or also videos or graphic material or some audio material that is not transcribed?
    • Is the material structured or open ended?
    • How is the study designed, is it a single or multi-case design?
    • Is the amount of data set fixed from the beginning or is there a need to add more data during the analysis process?  

     
    What kind of analysis do I want to conduct?  

    • Explorative or verifying, inductive or deductive?
    • An iterative or a one pass analysis?
    • Is the coding scheme set from the beginning or will it be developed throughout the analysis process?
    • How fine-grained should the analysis be?
    • How important is the context?
    • Should the output include both quantitative and qualitative elements?

    How important is it to stay close to the data?

    How much money can I spend, and what are the technical requirements?

    What kind of features should a good program offer?

    • Flexible methods of data management  
    • Simple and interactive coding  
    • A large variety of simple search options including text searches and searches for coded segments  
    • A complete display of hits and detailed source tags  
    • A variable context option and free manipulation of the search results, e.g. the capability to mark parts of the search result, and to edit, copy or save it.  
    • Capability to write memos and comments and the possibility to link these to other objects of the project.  
    • Log functions  
    • Display of conceptional schemas like networks  

    (following Hubermann und Miles (1995), see Literature)

    QUARC - Qualitative Research & Consulting, Dr. Susanne Friese, Dr. Susanne Friese, Fallingbosteler Strasse 1, D-30900 Bissendorf, Germany
    Phone: 49-(0)30 520 041 372 Fax: 49-(0)30 520 041 373 E-mail: info@quarc.de