I am doing a study in which in a first phase I want to teach participants features of novel social groups. Specifically, participants will learn about a group A and a group B, each of which has 14 features. So group A for example, may be healthy, from a Midwestern town, from a close family, etc, while group B will have 14 unique features.
So, participants will randomly be shown each of the 24 features independently (some assigned to A some assigned to B). Their task will be to categorize each feature as A or B. At first they will just be guessing but overtime they will learn them. So I will use a scaled item with the responses as A or B for reach feature. Then they will get feedback of correct or incorrect (using the skip to function). However, the issues is I want to teach to 100% correctness for all of the 28 items consecutively. In other words, they can not move on to the next phase unless they get all of the feature correctly categorized one after the other.

The only way I can think of doing this is having a ton of blocks of the 28 features (i.e., more trials than it would take any participant to learn the features). There would have to be some sort of fixed random order though so that I could use the skip-to-function to indicate they had gotten them all correct (I would prefer this just to be random however which is part of the issue). So for example if they get the feature categorized correctly it tells them so and then goes to the next feature in that block. However, if they get it incorrect, then it would go to the next block and start all over again. Does this make sense? If they get all of them correct then it would skip to phase 2.

Is there a better way to do this?
As always thank you!
Amanda