Flygare2022 - Empirically defining treatment response and remission in obsessive-compulsive disorder using the Obsessive-Compulsive Inventory-Revised
- Year read:#read2022
- Subject: _thesis defense MOC OCI-R Psychometrics
- Bibtex: @flygare2022a
- Bibliography: Flygare, O., Wallert, J., Chen, L.-L., Cruz, L. F. de la, Lundström, L., Mataix-Cols, D., Ruck, C., & Andersson, E. (2022). Empirically defining treatment response and remission in obsessive-compulsive disorder using the Obsessive-Compulsive Inventory-Revised. In PsyArxiv Preprints. https://doi.org/10.31234/osf.io/vw9xe
- Hoarding subscale, OCI-12
- Yea would have been prefereable, but we will most likely not have access to item-level or subscale-level data from all trials. If anything this should mean that we underestimate the accuracy of the full OCI-R and that similar evaluations using the OCI-12 will have higher accuracy.
- This is with the “true” values as vantage point. How many true cases will we identify (sensitivity) and how many true non-cases will we exclude (specificity)?
- This is with our prediction as vantage point. When we say that there is a case, how many times is that prediction correct (PPV), when we say that there is no-case how often is that prediction correct (NPV)?
- Accuracy/Cohen’s kappa/AUC
- Accuracy is the proportion of correct classifications
- Cohen’s kappa ranges 0-1, where 0 is not better than what would have occured by chance, and 1 is perfect agreement. Kappa is higher if there are multiple available categories, so our binary classifications likely does not overstate the predictive power at least.
- 0–0.20 as slight, 0.21–0.40 as fair, 0.41–0.60 as moderate, 0.61–0.80 as substantial, and 0.81–1 as almost perfect agreement (Landis & Koch cut-offs)