SUMMARY OF THE RESEARCH

PROBLEM STATEMENT:
Do energy conservation programs reduce the use of natural gas for heating?

BACKGROUND:
To estimate the energy savings that could be attributed to two programs designed to improve home energy efficiency

HYPOTHESIS:
Program participants will use consume less natural gas

MEASUREMENT OF VARIABLES
Dependent variable: Amount of natural gas used
Independent variable(s): Program participation
Control variable(s): age of house; floor area; number of stories; number of household members; income; price of natural gas; gas is used for heating water; winter severity

RESEARCH DESIGN:
This study uses pre- and post-program measures of energy use for five groups; this is a quasi-experimental design since there is no random selection of participants nor random assignment to levels of treatment or treatment groups.
 

Group	Time 1	Program	Time 2	Time 3
MECS Audit	O1	X	O2	O3
MECS non-audit	O1		O2	O3
PUCIP Audit & Loan	O1	X	O2	O3
PUCIP Audit	O1	X	O2	O3
PUCIP non-audit	O1		O2	O3

SAMPLING:
245 MECS-audit and 107 MECS non-audit; 68 PUCIP audit & loan; 56 PUCIP audit; and 105 PUCIP non-audit. Audit and audit & loan participants were volunteers; non-audit participants were recruited from the same areas (method not described).

INSTRUMENTATION:
Natural gas records; energy audit reports; mailed survey of household demographics; telephone survey of conservation actions and energy-related attitudes.

DATA COLLECTION/ETHICS:
Data on natural gas usage collected for 4 years; mail and telephone surveys administered once during the 4-year project.

DATA ANALYSIS:
Statistical models were developed to control for differences among households on control variables; regression was used to estimate effects of program participation

CONCLUSIONS:
All groups showed some energy savings but program participants saved more than non-participants. Audit participants saved 2-3% while audit & loan participants saved 14% over non-participants

CRITIQUE
Possible Threats to Internal Validity

History:
controlled by comparison groups

Maturation:
controlled by controlling for age of home; but no control for home remodeling during the study

Testing:
collection of data could have influenced comparison groups

Instrumentation:
gas company record-keeping did not change over the project; mail and telephone surveys are assumed to be reliable and valid;

Regression Artifact:
Audited households consumed more gas than non-audited households in the pre-program period, so they could be expected to save more than non-audited households over the course of the study

Selection bias:
Participants were self-selected volunteers, who could be expected to have different levels of motivation than non-participants

Experimental Mortality:
Only households with complete data (4 years of gas records, completed mail and telephone survey) were included; unknown what percent of the total households this represents

Design contamination:
Non-participants could have taken energy conservation measures on their own or learned about them from participating neighbors

Possible Threats to External Validity
unique program features:
May only be applicable to customers of these two companies in Minnesota, and the two communities were quite different.

experimental arrangements:
Only volunteers participated
other threats: