Understanding issues and limitations of Auto-Coding in Nvivo

By Divya Dhuria & Priya Chetty on June 16, 2017

As pointed out in the previous article, the process has its own set of issues and limitations, which might create certain obstacles in the nodes created and therefore needs to be taken care of. This article, therefore, highlights such issues and limitations along with their anti-dote to generate desired results.

Issues associated with Auto-coding

It may happen that auto-coding creates two nodes for a similar question. For example in the case research, Auto-coding in Nvivo has created two nodes for a question, “what do u think about student’s performance?” segregating the responses of participants in two separate nodes 5 and 3, respectively as shown in the below figure. If not observed before proceeding with data representation and analysis steps, this will lead to generating and interpreting limited information.

Two Nodes for a Single Question
Two Nodes for a Single Question

In such cases, merge two similar nodes into one single Node.

  1. Select the concerned Node
  2. click on Home (Ribbon)
  3. Click on Cut as shown in the below figure
Home icons for Merging Nodes in Nvivo
Home icons for Merging Nodes in Nvivo
    1. Select the other concerned Node
    2. Click on Home (Ribbon)
    3. Click on Merge as shown in the below figure
Home icons for Merging Nodes in Nvivo
Home icons for Merging Nodes in Nvivo

Limitations of Auto-coding

Both Auto-coding and Manual-coding make the list of nodes and code the content from transcripts. However, there are differences in the nodes made out of these two procedures. Auto-coding procedure in comparison to manual coding has several limitations in line, the only exception being its time-saving feature.

From the procedure of formatting of Transcripts in auto coding, it is yet to be seen how Auto-coding does not code all the responses in their respective Nodes.

For example, the below figure represents the Nodes from Auto-coding. We can see that each of the Nodes represent a question in the transcripts, and for that reason we have 15 different Nodes. We can also see that each of the Nodes has eight sources and reference, which means; these Nodes contain eight different responses, which corresponds to the number of respondents in our study-8. However, while looking on the Sources and References against each Node, we can see that Node on question number 10 has one less source or reference. This means, Auto-coding procedures has skipped one response for question 10.

List of Nodes for each question through Auto-Coding
List of Nodes for each question through Auto Coding

Comparing the auto-Coding output with manual coding, we can see that in the case of manual coding, no response has been left to be coded in their respective nodes. The below figure shows the list of all the Nodes in Manual-Coding and also the number of sources and references which are eight in all nodes. That means the Auto-Coding procedure does not necessarily Code all the responses.

List of Nodes for each Question through Manual Coding
List of Nodes for each Question through Manual Coding

In the case of qualitative analysis with bigger sample size, the chance of auto coding procedures leading to insufficient responses in Nodes is even higher. Therefore, Manual-Coding is a more accurate procedure for a small number of participants. However, in the case where the number of respondents is very high, Auto-coding is more preferable for saving time.

Following generating nodes for appropriate interpretation of responses based on themes and attributes, it is important to classify the nodes and generate memos as part of the data processing approach. This will help to record ideas inferences and insights for the understanding of material used in the research.

Discuss