Hybrid Coding: Deductive Themes to Inductive Child Codes
This methodology is a sophisticated way to manage large datasets by allowing the human researcher to control the major themes (deductive) while using the AI to quickly explore the fine-grained details (inductive).
Deductive Coding is "top-down" where researchers leverage their expertise, research questions, or existing theories, and create buckets (parent codes) before they even start analyzing the text.
Inductive Coding is "bottom-up" where researchers create new codes that emerge from the text itself.
Phase 1: Deductive Coding (The Human Analyst)
In the first phase, the researcher defines the high-level boundaries of the project
Phase 2: Inductive Coding (The AI Assistant)
Once the high-level codes are created, we can use the AI Assistant to perform the labor-intensive task of breaking down the text within those containers.
| Code | Transcript Text |
|---|---|
| Content Warning |
Finding the Ruby: The Trick (1) "Before we begin a warning, this podcast contains descriptions of sexual violence. Listen to the discretion is advised." Finding Ruby: The Trap (2) "This podcast contains descriptions of sexual violence, listener discretion is advised, and it's worth noting that in this episode we are journeying alongside Ruby into the online sex trafficking den where she was held against her will all those years ago. If you can though, stay with it, hope, and light, and a whole lot of joy is coming around the corner." Finding the Ruby: The Trespass (3) "Before we begin a warning, this podcast contains descriptions of sexual violence. Listen the discretion is advised." |
| Illusion of Opportunity |
Finding the Ruby: The Trick (1) "A message lights up the screen, it's like a beacon, hello, and then seconds later, would you like a job? The words are almost luminous, they seem to expand into that small space and fill it with new energy." "FBI Agent" Stole My $600K...Then Vanished "I've been a caregiver all my life in various forms and shapes. I was a nurse and a social worker. My sense of compassion and desire to be helping other people is what brings me satisfaction and actually I I felt an excitement towards knowing more and being involved actively.."
Phony Home Contractors Steal Over $450,000 "...it was two men who said they were doing construction on a home nearby and noticed something was wrong with this roof they could fix it quick and cheap they said the fix turned out to be anything but quick or cheap David found himself targeted by an international crime ring" |
| Code | Search Query | Rationale |
|---|---|---|
|
Emotional Impact
(run against all transcripts) |
fear OR safe OR scared OR safety | This query targets the psychological and emotional dimension of the victim's experience. The goal here is to move beyond the events of the scam and code for its impact. By searching for these specific emotional and security-oriented terms, we can gather all data related to the survivor's feelings of vulnerability, the threats (either real or perceived), and the lasting loss of security. This is critical for understanding the human cost of these crimes. |
|
Financials
(run only against YouTube video transcripts) |
money OR bank OR check OR account | This query is designed to capture the logistical and mechanical components of the scam. In qualitative analysis, it's not enough to know that a scam occurred; we need to understand the process. This query group allows us to isolate all text segments where the victim describes the transfer of assets, the methods the criminals demanded, and the financial institutions that were involved. It helps us build a clear, evidence-based map of how the fraud was physically executed. |
The AI Assistant is a feature built into NVivo that uses generative artificial intelligence. The AI Assistant can do two things: Summarize and Suggest Child Codes.
This specific feature reads all the text inside a single parent code that you select. It then looks for common sub-topics or themes within that text. It suggests these new, inductively-found topics as child codes for you to review, accept, or reject. This is Phase 2 of our hybrid method: using the AI to find the detailed themes inside our broad, human-defined one.
For this demonstration, we will only use the 'Financials' code we just created.
(Note: If we ran the AI on our 'Emotional Impact' code, the query could potentially include sensitive text from all transcripts, which the AI's content filters would block.)
To run the feature:
Right-click on the 'Financials' parent code.
Navigate to AI Assistant.
Select 'Suggest Child Codes'.
Review the inductive sub-themes the AI has discovered from the text.
Copyright © Baylor® University. All rights reserved.
Report It | Title IX | Mental Health Resources | Anonymous Reporting | Legal Disclosures