What types of coding can you do is a dataset?
There are four types of codable datasets used for different purposes within DiscoverText:
- a) mutually exclusive codes
- b) non-mutually exclusive codes
- c) user defined codes
- d) hierarchical codes
The Default Coding Settings
By default, DiscoverText is set up for mutually exclusive codes. The coder(s) must pick the code that is the best fit for the item. The default "standard coding" type means all coders assigned to that dataset will code the same items. This is the basis of our inter-rater reliability measurement and adjudication process. Use the default settings to test new code schemes and train coders.
Non-Mutually Exclusive Codes
Check the box next to "Allow coders to select more than one code" to create a code set where coders can select more than one code per item. Using the "Standard" coding type with non-mutually exclusive codes you can still measure inter-rater reliability and adjudicate differences between coders.
By enabling the user-defined options when creating a dataset, coders can manually create coding choices as "candidates" for inclusion in the final project.
Coding schemes that emerge from the data inductively can be discovered by creating candidate codes. It is a way to signal preliminary thinking. There is no candidate function in the software. It is more a state of mind related to where you are in the project.
Hierarchical Code Sets
When you create a dataset, you can create a hierarchical code set, which is a multi-keystroke coding option. This is a useful way to code a complex dataset.
For example, in a code set about American political issues, the top level code indicates the geographical region, such as the Pacific Northwest. The next level indicates the type of issue, such as education.
For further information, please see: Coding 101