Making more complex models with NVivo

[austinb]

I am trying to do some data visualizations and am stumped on two points:
1) How can my model show how many references a node contains;
2) How can my model show nodes that contain references which have ALSO been coded to other nodes?

I have a parent node entitled “People interviewed” with about 30 child nodes that are the names of my interviewees. I have a second parent node entitled “Climate change impacts” which has about 40 child nodes representing specific impacts that were identified by my various respondents. I want a model that depicts the impacts noted by each person. So, (Person A) with a bunch of links to the nodes that represent the various impacts that Person A noted.

I realize this would be easier to do if I were trying to create the model using the sources for each interview (as opposed to the nodes), but many of my interviews were with multiple respondents, so I created a child node for each respondent (child node of the parent node “People interviewed”) and coded the sections of their interview accordingly.

EXAMPLE: We could model Thomas (as a NODE) against selected parent nodes AND their child nodes . This would show nodes that contain text which is also coded to the node “Thomas.” These nodes could be represented against a background of all nodes that are children of the selected parent nodes.

See attached for an example in which the YELLOW circles represent nodes that are ALSO coded to Thomas and the WHITE circles represent selected nodes that are NOT coded to Thomas. This provides some visual context for Thomas’ content.

If NVivo can’t do this, does anyone have suggestions on some software that I might be able to use that could?

Thanks!
Austin

[QDATRAINING Admin]

 

I’m really sorry for the delay in responding to your post as the whole team is in Ethiopia this week working with the Ethiopian Public Health Administration. We had no internet connection since Tuesday and have only arrived back in the capital Addis Ababa just now. 

You problem is easily solved! Create a matrix for Thomas. In the column put the child nodes for "Climate change impacts". In the row, put Thomas! After you run the matrix, transpose if you prefer that view. You will have a list of Thomas's codes AND the empty ones for Thomas. Next, close the matrix and copy it. Create a folder called say, "climate change impacts by case". and paste your matrix as nodes. Be sure to tick the option 'include empty matrix cells'; otherwise you will only get Thomas's coded content and you want both. 

Now you have the option to drop Thomas into the modeller or export him to Excel where you may have more graphical options.

If you find this difficult to execute, let us know and we will demonstrate for you on-line.

Kind regards,

 

 

[Anonymous]

Hi Ben,
Thanks so much for this explanation. I’ve followed your instructions and gotten almost to the place I need, but not quite. I’ve pasted my matrix as nodes (including empty matrix cells) and then dropped that into the modeller. But, the result I get is Thomas (in the center) connected to EVERY node from my “climate change impacts” parent. There is no differentiation between which nodes contain Thomas’ references and which don’t. Furthermore, the nodes all show as children of Thomas, as opposed to children of parents which are in turn children of Thomas. So, in short:

1) How can I make my model show which nodes contain references from Thomas and which don’t — and even better, how can I show HOW MANY references a given node contains?

2) How can I create a model for Thomas that is like the one in my attached example in which I have a number of “category”/parent nodes shooting off from Thomas and then the child nodes of those parents shooting off in turn.

Many thanks Ben!

Austin

[QDATRAINING Admin]

Hi Austin,

First of all, thanks for using the forum rather than e-mail because it saves us having to constantly repeat answers and it shares your expieriences with other researchers who may learn from it. 

I would address part 1 outside of the modeller and part 2 inside. For part one, I would export Thomas as a list to Excel. This will give you a table with Thomas’s codes or citations and the weighting or frequencies of those citations (references). From there I would visually represent that using the tools in Excel. Even a simple pie chart should give you what you need but there are other graphical options in there too and this would be a quick and easy solution to showing Thomas’s concerns, lack of concerns and their weightings.

For part 2, I would use relationship nodes. We have extracted Thomas from the thematic codes he shares with others. You are quite right when you say that they now show up in the modeller as Thomas’s codes because they are Thomas’s codes. There is currently no relationship between them and the parents of the thematic codes that he shares with others. I would create relationship nodes between these and define the relationship as say, ‘sub-category of’, I could then drop them into the modeller and use the ‘include relationships’ option to display the relationship you need to show. I would then colour code the populated and non-populated codes. I would then simply manipulate the model and it would look the same as the example you sent (except with real data behind it).

All that said, I fully accept that there are real limitations to the modeller. It has not developed very much since version 7 (for example we had to go to Excel to address the first part of your enquiry). Some people create basic models in NVivo and then export to products with better graphical abilities such as Smart Draw, Visio or even PowerPoint.

Hope this helps!

[Author]

Posts