In order to be included in our final algorithm, a given variable (also referred to as a Use Detail) must meet the following conditions:
- Must be consistent with the ENERGY STAR approach. The normalization in the ENERGY STAR score is intended to account for business activity or building use. To accomplish this goal equitably, we adjust for characteristics that explain how a building operates (e.g., the hours of operation or the number of workers) but not why it operates a certain way (e.g., what type of lighting or HVAC system it has). You can learn more about this approach at: www.energystar.gov/ENERGYSTARScore.
- For example, if percent LED lighting were included in our regression equation then our resulting predictions would adjust for expected lower energy use in those buildings, making it harder for a building with more LEDs to earn the ENERGY STAR. Conversely, this approach would mean that buildings with 0% LEDs would only have to use less energy than others with 0% LEDs. So, a building with higher energy could end up with certification if it were the “best of the buildings without LEDs.” Being the best among a group with a less efficient technology is not the same as being the best relative to the national population.
- Must be statistically significant with respect to energy. In some cases, there are variables that are philosophically consistent with the ENERGY STAR approach, but they are not found to have a statistically significant effect on energy consumption. If there is no statistically significant relationship, EPA will not perform an adjustment.
- Must have adequate data for analysis in a national survey. There are also variables for which EPA did not have adequate data to make a determination. In this case, the national survey may not have data on certain aspects of operation. In the absence of adequate data, EPA cannot adjust for these characteristics. These characteristics could be re-visited during a future ENERGY STAR score revision, as more data becomes available.