Predicted Reliability is a statistically-derived formula that uses Power Circle Ratings (PCRs) from the J.D. Power Initial Quality Study (IQS) and the J.D. Power Vehicle Dependability Study (VDS) in order to reasonably predict a vehicle's reliability over time. IQS measures initial vehicle quality during the first 90 days of ownership and VDS measures long-term vehicle quality after three years of ownership.
View the 2016 Predicted Reliability Ratings
Historical Power Circle Ratings data is used to derive Predicted Reliability.
Initial Quality Study (IQS) and Vehicle Dependability Study (VDS) PCR data are first aligned by vehicle model years. These data are then split into 4 consecutive year subsets (e.g. 2014 to 2015, 2016 to 2017, etc.).
In developing the calculations, we take the following factors into account:
- In calculating Predicted Reliability, we use the most recent data from the previous three model years of study survey data.
- The recency of the IQS and VDS data, emphasizing newer data over older data.
- Correlation analysis indicates that although there is a high correlation between IQS data and VDS data. VDS data is a better predictor of long-term vehicle reliability than IQS data. As such, for the same model year, VDS data is used when available instead of IQS data..
In addition, the status of a vehicle is also considered. Is it a carryover model from the previous year, a redesign of an existing nameplate, or a new nameplate.
Among the three status types, predictions associated with a carryover model are slightly more accurate than predictions associated with a redesigned model. For new models, historical data at the model level are not available, so historical data at the make level is used to provide an indicator of how reliable a new nameplate is likely to be.
A series of regression models are developed for the predictions for different model years and types of vehicles.