20150818AVU to Staff 78.doc

AVISTA CORPORATION
RESPONSE TO REQUEST FOR INFORMATION

JURISDICTION: IDAHO DATE PREPARED: 08/12/2015
CASE NO.: AVU-E-15-05/AVU-G-15-01 WITNESS: Tara Knox
REQUESTER: IPUC RESPONDER: Tara Knox / Curt Puckett and Dr. Roger L. Wright (DNV GL)
TYPE: Production Request DEPARTMENT: State & Federal Regulation
REQUEST NO.: Staff - 078 TELEPHONE: [phone]

REQUEST:

Please provide a source for Equation 3 on page 21 (error ratio) of the Electric Load Research Study presented in Ms. Knox' workpapers.
a. Why was this equation used, instead of the following?

b. What is the justification for a summation on standard deviation, rather than variance?

RESPONSE:

The best reference may be the California Evaluation Framework, especially Chapter 13. This chapter is self-contained and can be read independently of the other chapters. The error ratio is defined on page 332 as
Equation 1
.
Question: Why do we define the error ratio this way rather than as
Equation 2

And, what is the justification for a summation on standard deviation, rather than variance?
Response: If the planned approach were to use ratio estimation with a simple random sample design or a stratified sample design with proportional allocation, then the second definition (Equation 2) would be suitable. The first definition (Equation 1) is appropriate when ratio estimation is used with an optimally stratified sample design, as discussed in the preceding reference on pages 337-338.

To see why, refer to the equation given on page 338
.
Since we are interested in a large population, set c = 0, giving
.
In the case of simple random sampling or stratified sampling with proportional allocation, so that
.
In this equation, the key measure of variability is as in the second equation above.
In the case of an optimal sample design, so that
.
Here the key measure of variability is as in the first equation above.
In practice, we use a stratified sample design with near-optimal allocation of the sample to each stratum, so we use the first definition of the error ratio to guide our choice of sample size. The summation of the standard deviation throughout the population provides the basis for establishing the optimal stratification cut points allowing for an equal allocation of the overall sample size to each stratum.

Page 1 of 2

Page 1 of 2