FINAL - Supporting Statementt B_4-28-14.rtf

Final Supporing Staement B United State Mint Low Denomination Coin Survey

SUPPORTING STATEMENT B

UNITED STATES MINT

QUANTITATIVE PUBLIC OPINION RESEARCH - U. S. COINAGE PRACTICES – LOW DENOMINATION COINS

B. Collections of Information Employing Statistical Methods

1. Describe (including a numerical estimate) the potential reporting universe and any sampling or other respondent selection methods to be used.

The potential respondent universe for this survey is all U.S. adults age 18 and over. The sample frame will consist of all residential telephone numbers – both cell and landline – in the United States. Numbers will be generated by first selecting a block of telephone numbers – a set of 100 numbers with the same first eight digits – that has at least one listed residential number. Within a block, the specific number to be called will be generated by randomly selecting the last two digits of the number. Landline Blocks will be assigned to one of eight strata depending on their geographic location (Census Region) and whether they are located in a metropolitan statistical area (MSA) or not in an MSA. In the cellphone frame, blocks will be assigned to one of four strata, based on census region. The stratification is illustrated in Table 1.

Landline

Cell Phone

Northeast

Midwest

South

West

Northeast

Midwest

South

West

MSA

Non-MSA

MSA

Non-MSA

MSA

Non-MSA

MSA

Non-MSA

When calls are made to landline phones, the specific individual to be interviewed will be randomly selected from among the individuals who reside in the household. This will be done by asking to speak with the household member who most recently had a birthday. Because it is presumed that cell phones are used by only one individual, no such most recent- birthday adjustment is needed on calls made to cell phones. Individuals with both a landline and a cellular telephone will be interviewed in each frame, and the frames will be combined using design weights as described in response to question 2.

2. Describe the procedures for the collection of information including: (a) Statistical methodology for stratification and sample selection, (b) Estimation procedure, (c) Degree of accuracy needed for the purpose described in the justification, (e) Unusual problems requiring specialized sampling procedures, and (g) Any use of periodic (less frequent than annual)data collection cycles to reduce burden.

Data will be collected by telephone by the survey research firm, IPSOS using a random digit dialing (RDD) approach as descrived above. Ipsos will conduct the calling using the RDD lists procured from GENESYS. Individuals willing to participate will be asked a few screening questions to determine eligibility and the interview will proceed. (i.e. over age 18). 

Within each stratum, Ipsos will attempt to secure the proportionate number of completes. Given differential response rates, this might not be possible, in which case Ipsos will introduce design weights. The sample will be weighted to 16 cells, reflecting region and phone status.

Landline
Northeast		Midwest		South		West
Landline Only	Landline + Cell	Landline Only	Landline + Cell	Landline Only	Landline + Cell	Landline Only	Landline + Cell
3.06%	18.61%	3.25%	18.81%	5.08%	29.48%	3.54%	18.17%

Cell Phone
Northeast		Midwest			South		West
Cell Phone Only	Landline + Cell		Cell Phone Only	Landline + Cell	Cell Phone Only	Landline + Cell	Cell Phone Only	Landline + Cell
12.53%	5.2		12.67%	10.86%	19.85%	17.04%	12.24%	9.61%

IPSOS will collect basic demographic information from all respondents. This information will then be used to assess the sample for any biases and, if needed, develop post-stratification weights. Ipsos will review the sample for these variables and compare it against known targets drawn from the U.S. Census Current Population Survey (CPS) for such variables as gender, age, education level, household size, urbanicity, household income, and race/ethnicity. If it appears that there are differences, these targets (and census region) will be used to develop the sample weights using a rim weighting (raking) method, the basic idea of which is straightforward. The process begins with one of the target variables (say, gender) and adjusts the respondent-level weights to bring the dataset into alignment with known targets. In this case, the known targets will be drawn from the U.S. Census Current Population Survey (CPS). The algorithm then moves to a second variable (say, age) and adjusts weights to match the U.S. population proportions in the various age categories. The process proceeds iteratively until a stable set of weights is obtained that fits all the demographic variables to within some tolerable level of strain.

IPSOS will obtain 1000 completed interviews. This sample size will allow variances of +/- 10% in equally-sized regions to be evaluated.

3. Describe methods to maximize response rates and to deal with issues of non-response.

Several steps have been taken to maximize the survey response rate. First, in drafting the survey questionnaire, the research team has taken length into account, keeping interview length to 12 minutes or less. Hopefully, this will limit the number of respondents who drop out before completing the questionnaire.

Second, in conducting the survey, anyone who initially refuses to participate will be recontacted after approximately 3-5 days to see if their cooperation can be obtained by a second effort. These second calls will be made by experienced interviewers who have shown strong performance and low refusal rates.¹

Third, for each number that is called, up to seven attempts will be made to reach someone there. These call-back attempts will be staggered over different times of the day and over different days of the week, sometimes calling on a weekday evening, sometimes during the day on a weekday, and sometimes during a weekend. In addition to increasing the response rate, making multiple attempts to reach someone at each number should increase the representativeness of the resulting sample if members of certain demographic groups are more difficult to reach.

Additionally, for each telephone number where an interview is completed, data on how many calls were needed before the interview was completed will be collected. Using these data, it will be possible to examine whether those who are more difficult to reach have different characteristics – and different experiences – than those who were reached more easily.

Post-stratification weights will be developed to address differential unit response rates for different demographic subgroups (as described in the response to question 2). Ipsos will also review survey questions to determine if there is substantial item non-response (judged as greater than 20% on-response to a specific questionnaire item). In such cases, an imputation approach, such as “nearest neighbor hot deck” based upon key demographic or behavioral questions, will be considered.

4. Describe any tests of procedures or methods to be undertaken.

No formal testing is planned.  The questions have been reviewed by researchers at US Mint and IPSOS. IPSOS will administer the survey instrument to several employees who are not engaged in the project to test the flow of the survey, evaluate the clarity of the questions and to validate that the survey is within the desired length. Initial interviews will be monitored by Ipsos and US Mint staff.

5. Provide the name and telephone number of individuals consulted on statistical aspects of the design and the name of the agency unit, contractor(s), grantee(s) or other person(s) who will actually collect and/or analyze the information for the agency.

The statistical aspects of the design have been developed by IPSOS.  The key project staff with respect to statistical design is:

Zachary Lewis, MA (Social Sciences, the University of Chicago) Associate Vice President, 202-420-2027

Alan Roshwalb, PhD (Statistics and Business Administration, University of Michigan), Senior Vice President, Ipsos Public Affairs, 202-420-2029

1Refusal conversion recontacts will not be undertaken if initial refusals were judged “irate” or “firm” by the interviewer.

1