FR2060_201003_ombsuppstmt2

Supporting Statement 2

Attachment 2

FR 2060

Overview of the Sample Selection Process

Sampling Procedures

For the consumer component of the survey, the population of interest consists of all households located in the geographic area that the Reserve Bank defines as the market. The unit of observation is the household. The respondent will be the financial head of the household or a financially knowledgeable spouse.

For the business component of the survey, the population of interest consists of all business enterprises that had revenues of less than $1,000,000 in the past year and that had an establishment in the designated geographic market area. However, there may be rare occasions where information from business enterprises with more than $1,000,000 in annual revenue would be considered necessary. The unit of observation is the business enterprise. The respondent will be an individual who is responsible for the lending relationships and financial decision‑making in the firm. That individual may be the owner, chief executive officer, president, treasurer, or comptroller.

A random sample will be drawn from listings in the local telephone directory using a random number table to select a page and then to select an entry on the page. The residential listings will be used for the consumer survey. The business white pages will be used for the business survey. In the event that business white pages do not exist, the yellow pages will be used.

The number of listings selected should be large enough to yield fifty completed household interviews and twenty-five completed business interviews. The proportion of blank listings (for example, listings with the number not working or that do not belong to a small business or household) and the response rate must be estimated to determine the number of listings to be selected.

A small number of screening questions will be required to determine whether a listing is an element of the population. If the survey yields substantially fewer than the desired number of completed interviews, perhaps because of an underestimate of the number of blank listings, then the first sample may be supplemented by a second random drawing. New listings, however, will not be allowed to substitute for refusals. Not at home will be treated as a refusal but efforts should be made to recontact.

Sampling Errors

The following tables present sampling errors, for the target sample sizes, across various hypothetical population sizes and across selected percentages of household or business respondents that do something asked in the survey

Table 1

Sampling Errors for Survey of Consumers

(Sample Size of 50 Households)

Sampling error (plus or minus, in percentage points) for a population of:

Percentage of households that do something asked in the survey	2,000 households	4,000 households	10,000 households	20,000 households
90	9	9	9	9
70	13	13	13	13
50	14	14	14	14
30	13	13	13	13
10	9	9	9	9

Table 2

Sampling Errors for Survey of Business

(Sample Size of 25 Businesses)

Sampling error (plus or minus, in percentage points)

for a population of:

Percentage of businesses that do something asked in the survey	120 businesses	240 businesses	600 businesses	1,200 businesses
90	11	12	12	12
70	17	18	18	19
50	18	19	20	20
30	17	18	18	19
10	11	12	12	12