B.2.b.iii Calculate sample sizes. The assumptions above and results of the test interviews were inserted into the equations above, which were then solved to arrive at a proposed sample size. It was determined that a sample size of 250 for both participants and nonparticipants will yield the desired results. Stratification Variables Stratification increases the precision of estimates compared with a simple random sample of a target population. In stratified samples, the target population is divided into non-overlapping groups, known as strata, from which separate samples are drawn. The goal of stratified sampling is to choose sample sizes within each stratum in a manner designed to obtain maximum precision in the overall estimate for the population. Each survey group included in Part B contains stratification. Specifically, For the HVAC Contractor Surveys, the first stratification variable is climate conditions (as defined by a North/South distinction). Within the North and South strata, there is a further stratification or grouping by timeframe for one of three training periods. Exhibit B-2-2 below depicts these strata. Exhibit B-2-2 Distribution of Participating Contractors by Period and Region Climate Region South North Grand Total Column Percent For the Commercial Office Equipment Managers Surveys the first stratification variable is For the ENERGY STAR Household Survey, the stratification variable is ENERGY STAR message saturation (defined as low, medium, or high). Using accepted public relations resources and methods, the program tracks, by the Nielson DMAs, the number of consumer impressions from PSA and media efforts. DMAs were developed by Nielson for planning and analyzing the results of publicity and advertising campaigns. Nielson estimates the number of households in each DMA using U.S. Census statistics. Low message saturation: Areas that received only the national-level ENERGY STAR promotions from EPA/DOE. B.2.b.iv Medium message saturation: Areas in which national-level efforts were supplemented by additional EPA/DOE target market outreach (PSAs and media outreach) that achieved at least 500 GRPS (gross rating points based on counts of media "hits"). High message saturation: Areas in which utilities or other third party organizations based a publicity and/or rebate program on the ENERGY STAR label. This third party publicity had to include at least 2 of the following: bill inserts; paid ads; retailer promotion/program; or rebates. Sampling Method The sampling method used for each survey group is listed in Exhibit B-2-1 above. These methods are described in more detail below. ENERGY STAR Household Survey The ENERGY STAR Household Survey will use a staged probability sample with stratification for varying levels of publicity efforts, which are enumerated above in Section B.2.iii. Specifically, this survey sample will include the following steps. The sample will be selected from the 57 largest (of 213) DMAs. It will then be sorted into three categories representing low, medium, and high exposure to the ENERGY STAR message (See Section B.2.iii above). Over 70 percent of all U.S. households are contained in these 57 DMAs. Four DMAs will be selected from each of the three publicity categories using random methods for a total of twelve (four times three) strata. This is a standard procedure in staged probability sampling. Approximately, 7,500 individuals will be selected in areas corresponding to ZIP codes for the chosen DMAS. The 7,500 individuals will be selected randomly from a large pool of individuals contained in residential drivers' license data bases, such that all 12 strata will be equally represented (625 individuals in each stratum). It is anticipated that 1,000 of the 7,500 individuals will respond to the survey. HVAC Contractor Surveys The HVAC Contractor Surveys use a staged probability sample with stratification for climate conditions (North/South distinction) and one of three training periods. Segmenting the contractor samples by climate zone was necessary to capture the range of primary business interests (air conditioning versus heating) among the participating contractors. Based on discussions with researchers at Lawrence Berkeley National Laboratory, it was decided to limit the number of climate zones used in the analysis to two. * Mapping of SIC groups to these categories is based upon the system used American Refrigeration Institute (ARI) cooling zones were used as the method for characterizing regions. The continental United States was divided into Northern (cooler) and Southern (warmer) regions. The Northern Region contains ARI cooling zones 1-4. The Southern Region contains ARI cooling zones 5-12. The border between the two regions is roughly coincident with a 5,000 cooling degree hour isobar at 78 degrees F. That isobar, or cooling line, runs east from the Virginia/North Carolina border, swinging north to the lowa/Missouri border, then south to the Utah/Colorado border, then north again to San Francisco. The isobar for 5,000 heating degree hours with a 65 degree reference point closely parallels that cooling line. The implementation of the dealer training program can be divided into three distinct periods for purposes of analysis. During the first period-November 1996 through April 1998—most of the tools provided to contractors for calculating equipment operating costs and the financial consequences of various financing approaches were based on paper charts and calculation routines. Also, during this period, one approved ENERGY STAR financing plan was available. During the second period-May 1998 through October 1998-financial computation and presentation tools were computerized and substantially upgraded. However, no qualified financing programs were available during that period. During the most recent period-November 1998 to March 1999—the financial and presentation tools have continued to be refined. Also, there are currently a number of financing options available to customers. Some are officially designated as Energy Star programs. Others are vendor financing programs under which ENERGY STAR equipment can be financed. In constructing the sample, provisions were made for adequate representation of both climate regions and all three program periods. Specifically, quotas of 80 dealers for Periods 1 and 2, and 90 for Period 3 (for a total of 250) were set. This will provide sufficient representation of the periods for purposes of comparison while providing a good representation of the entire population of participating contractors. Sample results will be weighted to reflect the relative weights of the groups of participant contractors defined by each program period and climate region. The next step in the sample development process was to choose compact geographic areas from which to draw the participant contractor sample First, the list of participating contractors was grouped into three-digit ZIP Code areas. Using a ZIP Code map and the tabulation of contractors, concentrations of participants which might have been large enough to affect the local market were identified. As a rule of thumb, 3-digit ZIP code areas with at least 10 participants, or groups of contiguous ZIP Code areas with 25 or more participants were identified. Specific contractors will then be randomly selected from the zip code list that will be grouped according to training period. Non-participant contractors will be drawn randomly from telephone and Dun & Bradstreet listings for the three-digit ZIP codes encompassed by the areas selected for the participant sample. Commercial Office Equipment Managers Surveys In order to meet market segmentation objectives for these samples, "sample design segments" were defined according to computer density and a broader classification of establishment type. These segments were developed using the data on the number of computers, SIC code, and size (number of employees). The final sample design segments are listed in Exhibit B-2-3. Specifically, the design segments listed in Exhibit B-2-3 were developed by characterizing establishments by their two digit SIC classification coded within the broader commercial or industrial categories. The broader commercial or industrial classification was based upon the system used by the Commercial Building Energy Consumption Survey (CBECS). The SIC codes related to agriculture and government were excluded. Public schools and other establishments maintaining local government services were included in the sample. Estimating an establishment's computer density index, or the number of computers per employee, was completed by first assigning a "cell category" to every commercial and industrial establishment, according to its two-digit SIC code and employment size category combination. The index was determined by: Estimating the number of computers for each cell based on summed averages from the detailed (3 or 4 digit SIC coded) MarketPlace site computer data; and then, After examining the range of the computer density index values, high, medium, and low ranges were constructed as follows: high density (density >1.0); medium density (0.5 < density < 1.0), and low density (density < 0.5). Once the sample design segments listed in Exhibit B-2-3 were established, the data were further sub-segmented by establishment size. For this analysis, only establishments with 25 or more computers, according to the MarketPlace characterization, were included. Fewer than seven percent of establishments with fewer than 100 employees had 25 or more computers. Establishments in these employment categories were therefore dropped. Exhibit B-2-4 shows the results of this analysis. |