QSR NUD*IST : Qualitative Solutions & Research Pty. Ltd. –

Non-numerical Unstructured Data Indexing Searching and Theorizing.

Intended Audience: This software is intended for researchers performing qualitative data analysis.

  1. Range of Functions:

You can begin your project with data, with a preconceived coding structure or with a coding structure from a previous project. Each project has two parts: the documents system, or group of documents that you import, and the index system, which consists of the coding that you create and the searches you perform.

QSR NUD*IST provides the following functions:

You can import text documents such as interview transcripts as ASCII text. Imported documents must be formatted with hard carriage returns for text units and an asterisk at the beginning of each text section. It is up to the researcher to determine these units. For example, a text unit can be a line or a paragraph. Each new speaker can be a section unit. For photos or data that cannot be imported as ASCII text, you can log them as "external" documents and code and write memos as with imported documents. You can annotate your data by inserting tagged annotation lines and write separate memos about documents.

Coding concepts or categories are referred to as "nodes." NUD*IST will keep a simple list of "Free Nodes" if a relationship between your coding concepts has not yet been defined. If your coding concepts are in a hierarchical or parent/child relationship, you can create "Index Nodes" to show this relationship. The coding (or node) database allow you to view all or a specified group of text coded a certain way.

You can also search all documents, memos and document annotations for simple text strings or for a pattern of text. (See Appendix A for text search pattern options.)

The index search capacity correlates with the "Theorizing" element of NUD*IST. With 17 provided search operators (see Appendix B), the software manual claims that the user can "build and test theories." These operators serve as queries to investigate relationships or cross-tabs between coded concepts or items. As with any software program, the power of this feature will be only as good as the coding and thinking done by the researcher.

  1. How could this software be useful for ERGies?
  2. Overall, this software would be quite useful for ERGies as an electronic database for qualitative data and coding. It allows the researcher to easily group or sort all data coded a certain way and/or the results of searches. The report feature allows you to print out or save this sorted data for cutting and pasting. The coding process is flexible and can be revised as often as necessary.

    If structured interviews are formatted with questions as sub-headers, so the question and its answer is a section NUD* IST has a special command (Qnodes) that will automatically code each question and its response. Similarly, interview data in tables separated by tabs with document names as the first column can be automatically coded as they are imported.

    Before investing time and money in this software, however, the updated version of NUD*IST, called NVIVO, should be investigated. NUD*IST projects can be imported into NVIVO. NVIVO has an upgraded rich text editor so you're not limited to ASCII text and does not require you to format your documents with sections or text units. With NVIVO, memos are full status documents that can be coded and linked to other documents and hyperlinks can be made between documents, photographs, videos and spreadsheets. Apparently, the hypertext/linking feature of NVIVO also allows for a documenting of relationships between coding concepts that is not necessarily hierarchical.

  3. How does the software design affect or restrict the research process?
  4. NUD*IST design has both procedural and conceptual limitations for the research process, but neither of these restrictions is overbearing. Procedurally, data must be formatted as ASCII text documents (.txt) with hard returns as "text units" and asterisks (*) as sub-headers. This formatting can be time consuming and requires the researcher to make decisions about text units and section units before the data is coded when it is difficult to make an informed decision. Additionally, as with any coding software, using NUD*IST can bind you to having all of your data stored in this format. NUD*IST projects can be imported into NVIVO, but there is still always the risk that the QSR company will fold and ten years from now, you may be stuck with data locked into an archaic format for which no software help is available.

    Conceptually, NUD*IST coding structure has only two options: unstructured ("free nodes") or in a tree like hierarchy ("index tree"). This software cannot accommodate conceptual maps that are not hierarchical (i.e. ones that look more like a "spaghetti diagrams" with lots of arrows connecting different aspects.) Apparently NVIVO can accommodate these relationships, but I do not have exposure to this updated software. Additionally, in terms of index searching, NUD*IST comes with 17 built in operations. (See Appendix B) If you are interested in searching for a relationship not represented by these, you must write your own code.

  5. Overall Evaluation:
  6. Design and Execution: The database is complex and it takes a while to learn all of the features, but once you are familiar with the software, text importing, coding and searching is fairly straightforward. As mentioned previously, the hierarchical structure and ASCII text requirement can be limiting. Overall I give QSR NUD*IST a GOOD rating.

    Functionality: This software is very useful for database, coding and text searching: I would rate it VERY GOOD for functionality in these areas. However, more experience is necessary to evaluate the index searching feature.

  7. Other Notes


APPENDIX A

Pattern Search Operators

The following diagram and table shows the default characters used to denote special pattern matching requirements in the Pattern Search operation.

Special characters which appear as shown below using the Documents, Search Text, Special Characters menu option can be customized to suit the search requirements.

Character Name Meaning

^ Left Anchor When used at the start of a pattern, it forces the pattern to match at the beginning of a text unit. When used elsewhere in a pattern it is an ordinary character.

$ Right Anchor When used at the end of a pattern, it forces the pattern to match at the end of a text unit. When used elsewhere in a pattern it is an ordinary character.

[] Alternation start and end Encloses alternatives in a pattern. An alternation pattern matches an expression if one of the alternatives in it matches the expression.

| String alternation divider Separates alternative strings in a pattern. The match is always case sensitive, even if the case sensitive option is not selected. The match is also exact; that is, you cannot use patterns as alternatives, only strings. This is called a string alternation. If there are no dividers inside the brackets, then each character inside the brackets is treated as a one-character alternative. This is called character alternation. The alternative that can match nearest the start of a text unit is the one that will succeed. Also, at any given point in the matching process, the alternatives will be tried in order from the left to right.

- Character range alternation When used as the second character in a character range alternation, it makes the first and third characters specify a range of characters for alternation, in the order specified for characters in the computer’s character set. Note digits occur in the order 0-9, letters in the order A-Z and a-z. When used elsewhere, it acts as an ordinary character.When used as the second character inside an alternative in a string alternation, it forces that alternative to be a character range alternation not a string alternative.

~ Negated character alternation When used as the first character in a character alternation, the match is with any character not in the remaining characters in the character alternation. When used elsewhere, it acts as an ordinary character. When used as the first character inside an alternative in a string alternation, it forces that alternative to be a negated character alternation not a string alternative. Negation can be used with character range alternation. In this case the match is with any character not in the range. This is called negated character range alternation.

. Spanning Wild-card Matches any one character.

# Zero or one repetition wild-card Allows the preceding character or grouping to match or be ignored (match 0 or 1 times).

* Zero or more repetition wild-card. Allows the preceding character or grouping to match any number of times in a row, or not to match at all (match 0 or more times).

+ One or more repetition wild-card Allows the preceding character or grouping to match any number of times in a row, (match 1 or more times).

() Grouping start and end Any pattern or string inside parentheses is a grouping for repetition wild-card purposes; that is, if a repetition wild-card occurs immediately after a grouping, the wild-card applies to the whole grouping. Note alternation, in square brackets [], counts as a grouping in the same way.

\ Escape If the escape character is placed in front of a special character in a pattern, then the special character is treated literally; that is as an ordinary character. If it is placed anywhere else it is ignored.

 


APPENDIX B

Index Searching Operators

To give you an idea of just how "powerful" these theory building and testing tools are, below is a list of the search operators available in QSR NUD*IST. This is copied directly out of the help files so it also gives you an idea of how intelligible (or unintelligible) the help files are.

1. At Least

The At Least Index Search operation requires you to nominate 2 or more nodes and also a minimum number of nodes. These can be any nodes in the Index System - Tree, Free, Document Annotation etc.

The effect of the At Least operation is to operate a section at a time, or a document at a time (depending upon what you nominate) and then retrieve all the sections/documents which have some coding from at least the number of the nominated nodes. Note that the coding can be anywhere within the same section or document and does not have to have any other overlap etc.

This is useful to detect clustering of coding.

For example, the At Least 2 of nodes with coding relevant to Theatre, Opera, Ballet, Concert would find all sections referring to more than one performing art.

2. Collect

The Collect Index Search operation requires you to nominate 1 node. This requires a hierarchical structure and would normally be a node in the Index Tree.

Merges (unions) all the index references at the node and at all nodes in the sub-tree below it

3. Excluding Docs

The Excluding Docs Index Search operation requires you to nominate 2 nodes. These can be any nodes in the Index System - Tree, Free, Document Annotation etc.

Index references for the first node, but excluding documents coded by the second node.

4. Followed By

The Followed By Index Search operation requires you to nominate 2 nodes. These can be any nodes in the Index System - Tree, Free, Document Annotation etc.

You also need to nominate the amount of "nearness" to be checked, i.e. a certain number of text units, in the same section, or in the same document.

Finds all pairs of passages of text where the a range of text units coded at the first node is "near" to range of text units coded at the second node - according to the selected nearness criteria -and also entirely before it.. Note that this finds each passage in entirety, not just individual text units which are near to each other.

The information returned will depend upon whether you chose to save the first node, the second node or both.

Followed by is similar to Near and allows you to detect, and retrieve passages where a selected 2 topics are discussed in proximity and in a nominated order.

5. If Inside

The If Inside Index Search operation requires you to nominate 2 nodes. These can be any nodes in the Index System - Tree, Free, Document Annotation etc.

The effect of the If Inside is to retrieve any passages coded by the first Node nominated providing that the second Node codes at least all of that passage (and optionally text before and or after)..

This differs from Intersect in that If Inside considers the passage as a whole, not individual text units.

For example, If Inside angry, question2 would find all passages relating to anger providing it was in the context of responding to question 2.

6. If Outside

The If Outside Index Search operation requires you to nominate 2 or more nodes. These can be any nodes in the Index System - Tree, Free, Document Annotation etc.

The effect of the If Outside is to retrieve any passages coded by the first Node nominated providing that there exists an entire passage coded by the second Node inside the first.

This is the reverse of the If Inside search.

7. Including Docs

The Including Docs Index Search operation requires you to nominate 2 nodes. These can be any nodes in the Index System - Tree, Free, Document Annotation etc.

Index references for the first node, but restricted to documents coded by the second node.

8. Inherit

The Inherit Index Search operation requires you to nominate 1 node. This requires a hierarchical structure and would normally be a node in the Index Tree.

Merges (unions) all of the index references in the ‘ancestor’ nodes of the given node — the ones above it in the path from it up to the root of the index system.

9. Intersect

The Intersect Index Search operation requires you to nominate 2 or more nodes. These can be any nodes in the Index System - Tree, Free, Document Annotation etc.

The effect of the Intersect operation is to find the passages of text which are coded by every one of the nominated nodes.

For example, the Intersect of nodes with coding relevant to Men, Over 40, Smoker would find just those passages relating to men over the age of 40 who smoke.

10. Just One

The Just One Index Search operation requires you to nominate 2 or more nodes. These can be any nodes in the Index System - Tree, Free, Document Annotation etc.

The effect of the Just One operation is to find all text units coded by just one of the nominated nodes, i.e. which are not also coded by one or more of the other nodes.

For example, the Just One of nodes with coding relevant to Smoker, Drinker would find all passages relating to tee-total smokers and also non-smoking drinkers

11. Less

The Less Index Search operation requires you to nominate 2 or more nodes. These can be any nodes in the Index System - Tree, Free, Document Annotation etc.

The effect of the Less operation is to find all text units coded by the first mentioned node, but which are not also coded by one or more of the other nodes.

For example, the Less of nodes with coding relevant to Smoker, Drinker (in that order) would find all passages relating to tee-total Smokers

12. Matrix

The Matrix Index Search operation requires you to nominate 2 nodes. These require hierarchical structure and would normally be in 2 distinct sub-trees in the Index Tree

Take all the children of the first-named node, and all the children of the second-named node, and apply the named operator (such as Intersect) in pairs to one of the first group with one of the second group. The result is a Matrix node containing the coding from each such pairing operation in a rectangular matrix.

13. Near

The Near Index Search operation requires you to nominate 2 nodes. These can be any nodes in the Index System - Tree, Free, Document Annotation etc.

You also need to nominate the amount of "nearness" to be checked, i.e. a certain number of text units, in the same section, or in the same document.

Finds all pairs of passages of text where the a range of text units coded at the first node is "near" to range of text units coded at the second node - according to the selected nearness criteria. Note that this finds each passage in entirety, not just individual text units which are near to each other.

The information returned will depend upon whether you chose to save the first node, the second node or both.

This is useful to detect, and retrieve passages where a selected 2 topics are discussed in proximity.

14. Not In

The Not In Index Search operation requires you to nominate just one node. This can be any node in the Index System - Tree, Free, Document Annotation etc.

The effect of the Not In operation is to find all text units which are not coded by the nominated node. Since in general, nodes will only code a small part of a document, then Not In may retrieve a very large number of text units.

Not In is often used in conjunction with other index search operations (especially in command files). For example, the Not In of the node containing coding interviews of a Female could be used to find all passages of text by Males.

15. Overlap

The Overlap Index Search operation requires you to nominate 2 or more nodes. These can be any nodes in the Index System - Tree, Free, Document Annotation etc.

The effect of the Overlap operation is to find all text units coded by any of a given set of two or more nodes (like Union), except that is a find of a text unit range contains index references from only one of the set of nodes, it is rejected.

Overlap means strictly that two text ranges must have at least one text unit in common.

For example, the Overlap of nodes with coding relevant to Smoking, Drinking would find all passages where smoking and drinking have been discussed together (in the same text unit).

16. Union

The Union Index Search operation requires you to nominate two or more nodes. These can be any nodes in the Index System - Tree, Free, Document Annotation etc.

The effect of the Union operation is to find the passages of text which are coded by any one or more of the nominated nodes.

For example, the Union of nodes with coding relevant to Masters, Doctorate, Bachelor would find all passages relating to people with a university degree.

17. Vector

The Vector Index Search operation requires you to nominate 2 nodes. The second requires a hierarchical structure and would normally be in the Index Tree whilst the first could be any nodes in the Index System - Tree, Free, Document Annotation etc.

Just like Matrix, except the operator is applied in pairs to the first node address itself (not its children) and the children of the second node address. This gives a one-row multi-column matrix.

For all 17 above:

The results are saved as a new node in the Index Search area and also are placed for any immediate use on the Node Clipboard.

Please refer to the Restriction topic for more information on how Restriction can be applied to this operation.