GRIDmeter
  • GRIDmeter™
  • Methods
    • GRIDmeter
    • FLEXmeter
  • Code
  • Blog

Testing Comparison Group Size and Stability

9/15/2020

0 Comments

 
Picture

high-level takeaways

Picture
Comparison groups can create uncertainty for multiple reasons:
  1. They are not exactly representative of the treated group(s)
  2. There are random events that lead to long term differences in outcomes (multiple equilibria)
Comparison Group implementation will need to vary depending on if savings are annual or hourly marginal and whether comparison group is assembled prior to or following program enrollment

Portfolio and comparison group sizing matter. There is a very predictable uncertainty reduction as group sizes increase.

what we now know

Picture
Program tracking with annual savings calculations and portfolios of 3,000+ heterogeneous buildings. Random sample of 3,000+ non-participant buildings from eligible population

Program tracking with annual savings calculation with portfolios of 1,000+ relatively homogeneous buildings
.  Random sample of 3,000+ non-participants from targeted population

Program tracking with marginal hourly savings calculations with smaller numbers of homogeneous buildings
. Stratified sample of 3,000+ non participants from targeted population

 Savings reconciliation after enrollment has closed and full treatment population is known, especially needed where program enrollments are substantially different than comparison pool. Stratified sample that maximizes sample size relative to error between treated and non-treated groups

0 Comments



Leave a Reply.

    Archives

    December 2020
    October 2020
    September 2020
    August 2020

    Categories

    All

    RSS Feed


Contact the Project Team
©2020 Recurve Analytics Inc. "Recurve" is a trademark of Recurve Analytics Inc. All rights reserved.
​Privacy Policy. Terms of Service.
Picture
  • GRIDmeter™
  • Methods
    • GRIDmeter
    • FLEXmeter
  • Code
  • Blog