Sign in or register
for additional privileges

Using NVivo: An Unofficial and Unauthorized Primer

Shalin Hai-Jew, Author

You appear to be using an older verion of Internet Explorer. For the best experience please upgrade your IE version or switch to a another web browser.

Creating Relationships between Project Items (Nodes, Sources, and Other Entities)

A core aspect of qualitative, mixed methods, and multi-methodology research involves the exploration of associations (and more precisely) relationships and relational dynamics.  NVivo enables the formulating of such relationships in a multi-step process:

  1. defining relationship direction (association, causal in either direction, reciprocated, or other); 
  2. defining relationship types, and 
  3. applying these relationship types to the nodes and other entities in the project. 

The association tie is just a line without arrowheads to link nodes.  A causal tie uses an arrow at one end or the other.  A reciprocated association has arrowheads on both ends of the link connecting the nodes. 

Relationship types show up in the word(s) applied to the edge or link. 

The nodes represent egos (individuals) or entities (groups). 

Or, a researcher may simply create relationships (which are node types), without classifying the relationships as a certain type.  Creating relationships as types is helpful when that relationship dynamic applies to multiple contexts.  A singular sort of relationship does not have to belong to any time and may be indicated simply as a relationship. 

Postulating Relationships

In various types of qualitative and mixed methods research, researchers hypothesize or postulate various types of “relationships” between phenomena. These may be mere associations, such as co-occurrences of certain phenomena with each other.  There may be correlations of various strengths; these may be positive or negative correlations. There may be time relationships, such as precursor events and post-event occurrences; there may be synchronous events. There may be proxemic or distant relationships based on interactions and lengths of interactions—in social network analyses. There may be causal relationships.  Etc. 

For researchers, it may help to begin postulating some of these relationships early on the in coding in order to have these articulated, so they may be elaborated on, tested, and assessed. In the coding phase, just positing that there may be a relationship does not suggest any commitment to the idea (absent evidence or facts).  It is only something to be explored.

How Relationships may be used in NVivo

Assigning relationships to nodes not only helps articulate the potential association(s).  Within NVivo, such links may be used to create visual relational models (such as node-link diagrams). Within the Models area, researchers may also manually draw such models with entities and relationships / interrelationships.  

Relationship Types as a Classification in NVivo

"Relationships Types" are a kind of classification in NVivo.  They are defined by name, description, direction (associative, one way, or symmetrical), and the entities at the two ends of the connection.  Relationships are generally defined as connections between one entity to another (in a dyadic sense).  The dyadic nature of the link may be more limiting except for the fact that a node may be connected with another using a number of different defined relationships. This layering of relationships may be fairly complex if there are multiple interrelationships.  One limit to the uses of the layering of relationships may come from the fact that visual clarity starts to degrade with the larger numbers of defined relationships.  The entities may be nodes or concepts; they may be sources.  Relationship types and the applied relationships are generally defined by the researcher.  

Identifying (Representing) Relationships between Nodes

In the current version of NVivo, to create a postulated or known relationship between two defined entities, one can start at the top menu with the Create tab.  Then select the Relationship button.  Then define the needed elements in the window.

The other visuals here are from a prior version of NVivo.  (These will be updated to the present version as time allows.)

Researchers may set up Relationship Types (categories of relationships) in NVivo.  
  1. Click "Classifications" in the lower left corner.  
  2. Click the "Relationship Types" at the upper left in the Navigation View.  
  3. In the ribbon at the far upper right, click "Relationship Type".  
  4. A popup window opens with a new name for the relationship and the Direction of the relationship (associative, one way or symmetrical). [An association is indicated with a line with no line ends.  A one-way relationship is an arrow pointing from one node to another.  A symmetrical relationship is indicated by a line with arrows on both ends.  An associative graph is considered an "undirected" graph.  A graph with links that have arrows on one or both ends is a "directed" graph.  Graphs may contain nodes that are all of a type, for a unimodal graph.  They may have nodes that represent two different categories of entities as two-mode graphs.  There are even more complex types of graphs that are multimodal.] 
  5. Click "OK" once that relationship is defined. 
  6. Now, that relationship may be applied to relationships between nodes.  

Defining (Code) Nodal Relationships

They may also Create -> Relationship -> Named Relationship Type -> and select the nodes to place in relationships. This is where the researcher may name the node-based variables in the dyadic (two-node) relationship.  

Researchers may structure relationships between nodes by subsuming some nodes under others (treating some nodes as "children" to "parents"). They may place nodes in certain folder structures. 

Merging Relationship Types

1. To re-code several relationship types into a combined or merged one, go to the Classifications button at the bottom left of the Navigation View.  Then, click "Relationship Types" above (within the "Classifications View").  

2. In the List View in the main window, click on the particular relationship that you want merged with another.  
3. In the ribbon in the Home tab, click "Cut" in the Clipboard group.  

4. Then in the List View, identify the relationship you want to merge the cut relationship into, and right click that.  In the dropdown, click the “Merge Into Selected Relationship Type”.  The cut relationship will be integrated with the new relationship, and all prior coded relationships under the “cut relationship” label will be updated to the merged one.  

5. If you changed your mind about the cut relationship, place your cursor into the List View and right-click and copy.  The original cut relationship will no longer be grayed out or merge-able with another relationship unless you reinstate the above process.  (The CTRL + Z also works to undo a sequence of prior actions.)   

Defining a Singular Relationship

Finally, there is the option of defining a basic relationship.  This may be done from any setting in the Navigation View. 

Go to the Create tab in the NVivo ribbon.  Click on Relationship in the Nodes group. 

The New Relationship window opens.  Here, the researcher defines the From and To elements. 

When clicking on the Select button, the researcher will be presented with a Select Project Item window. This enables him or her to select from any sub-folder level object such as a node or a source or an entity.  This process is followed for both the "From" and the "To" in the relationship. 

The user then selects a relationship type.  Or he or she may choose just a direction of relationship.  A color may be used to indicate the relationship.  (Colors may be used to indicate a certain "type" of relationship if the researcher wants.) 

Click "OK."  The relationship is connected.

To see the new relationship, in the Navigation View, go to Nodes, and go to the Navigation folder.  The linked relationships are indicated there. 

Why do relationships (between variables) matter?  Relationships may be an indicator of an association, a causation (causal relationship), or a mechanism in a model.  Relationships may indicate dynamics.  They may indicate levers that may be changed to affect outcomes.  They may enable the definition of systems (represented in models). 

In quantitative research, associations are based on statistically significant associations.  There are computational ways to identify such potential associations in qualitative research, too.  The significance measures relate more to interrater coding agreement and some numerical tendencies (but not p-values). 
Join this page's discussion (1 comment)

Discussion of "Creating Relationships between Project Items (Nodes, Sources, and Other Entities)"

Very helpful

Thanks so much for this page - it's the clearest description I've found and I've come back to it countless times during my coding.

Posted on 20 February 2018, 11:04 pm by John Bankier  |  Permalink

Add your voice to this discussion.

Checking your signed in status ...

Related:  Ingesting "Internal" Source ContentsFuture Look: Data Repositories for NVivo-based Data Sets?Setting up a Qualitative or Mixed Methods Research Project...ScreenshotNCapture and the Web, YouTube, and Social MediaComplementary ToolsData Query: Coding Comparison (Advanced) and Cohen's Kappa CoefficientStarting a New NVivo ProjectCiting Published and Unpublished Sources in NVivoIntroData Query: Compound Query (Advanced)Research Journaling in NVivoThe NVivo User InterfaceManual Coding in NVivoA Research Workflow with NVivo Integrations"Using NVivo" CoverA Simplified Timeline of Qualitative and Mixed Methods Research (as a semi-recursive process)Using Demographics to Further Explore Interview, Survey, or Focus Group DataSome Types of Data Visualizations in NVivoTeam or Group Coding in NVivoData Query: Matrix Coding QueryWhat is NVivo?Ingesting "External" Source Contents (Think "Proxy")Data Query: Group Query (Advanced)To Go with NVivo or Not? (and Partial Checklist for Stopping Usage of NVivo)Creating Codebooks in NVivo (through Reports...through Share)Conducting Data Queries in NVivo (Part 1 of 2)qualitative researchDrawing Relational ModelsData Query: Word Frequency Count QueryData Query: Coding Query"Autocoding" through Machine Learning from Existing (Human) Coding PatternsDisambiguating Data VisualizationsData Query: Text Search QueryConducting Data Queries... (Part 2 of 2)Analyzing Social Media Data in NVivo"Autocoding" through Styled or Sequentially-Structured Textual Datamixed methods researchDownloading, Installing, and Registering the SoftwareCopy Project Window in NVivo 10 (Windows)