I have a very tightly controlled paradigm and comparing/computing d' from individual trials in the presence/absence of a different prime proceeding each trial. Accuracy is typically low on this discrimination task, but we find reliable statistical differences among conditions. However, when converting accuarcies to d's, they are all less than 1 (range: .3-.9). I still find the predicted statistical differences between conditions, but are the results meaningless since d's less than 1 are not above threshold? Thanks in advance.
“meaningless”? That depends on context and the scientific question. They’re above zero so there is discrimination. There is no magic threshold at one.
Thanks, Michael. I tend to agree, but I’ve had reviewers question values less than 1 (or even higher in some cases).
Odd. The only times I’d question low d’s is if they are low relative to other pubs or if their CIs are too broad. Good luck.
- Badges
- Science topic
- Similar topics
- Philosophy
- Esthetics
- Interpretation