0052 ss rev 041709 Part B

SUPPORTING STATEMENT
MARINE RECREATIONAL FISHERY STATISTICS SURVEY
OMB CONTROL NO. 0648-0052

B.

COLLECTIONS OF INFORMATION EMPLOYING STATISTICAL METHODS

1. Describe (including a numerical estimate) the potential respondent universe and any
sampling or other respondent selection method to be used. Data on the number of entities
(e.g. establishments, State and local governmental units, households, or persons) in the
universe and the corresponding sample are to be provided in tabular form. The tabulation
must also include expected response rates for the collection as a whole. If the collection has
been conducted before, provide the actual response rate achieved.
The survey, which will be conducted in North Carolina, will utilize a dual-frame approach to
sample recreational, saltwater anglers. The sample universe is all saltwater anglers in NC. The
sample will be selected from state databases of licensed, saltwater anglers and address frames
derived from the USPS DSF. The license frames include all anglers who have a saltwater fishing
license for NC, and the DSF frames will include all households within the study areas.

Frame Size
Sample Size
Complete Screeners
Eligible (Active)
Anglers
Complete Diaries

NC License
Frame
540,0001
15,200
6,080 (40%)
2,432

NC USPS
DSF
3,496,6002
35,000
10,500 (30%)
1,050

Totals

50,200
16,580
3,482

1,4593

6303

2,089

1. Approximate number of licensed saltwater anglers as of 12/31/2008.
2. Estimated number of occupied housing units (Demographics USA, 2008).
3. Assumes that 75% of eligible respondents will agree to participate in the follow-up diary survey, and a subsequent response rate of 80%. of
the75%, or an overall response rate of 60%.

A primary goal of this study is to determine the response rates that can be achieved in a
recreational fishing mail survey. Based upon the survey literature, we expect response rates to
the screener questionnaire for the license frame and address frame samples to be approximately
40% and 30%, respectively. The higher response rate assumption for the license frame samples
is premised on licensed anglers being more willing to participate in the survey than the general
population.

1

2. Describe the procedures for the collection, including: the statistical methodology for
stratification and sample selection; the estimation procedure; the degree of accuracy
needed for the purpose described in the justification; any unusual problems requiring
specialized sampling procedures; and any use of periodic (less frequent than annual) data
collection cycles to reduce burden.
The survey includes a screener questionnaire to generally characterize respondents’ fishing
activity and collect fishing effort data, and a follow-up diary survey to collect information about
the types and numbers of fish caught. The screener questionnaire will be used to estimate fishing
effort, as well as identify a sample of anglers from which to collect more detailed catch
information in the diary survey. The diary survey will be used to compare catch characteristics
and catch rates between angler trips that would be accessible to field samplers (public-access
trips) and angler trips that would not be accessible to field samplers (private-access trips).
This study will use a mail survey as an alternative to the random-digit-dialing (RDD) telephone
survey for the household component of the dual-frame. The survey will be conducted in North
Carolina to evaluate the procedure and processes.
A major reason for examining the replacement of the RDD survey, the CHTS, with a mail survey
is the incompleteness of the RDD frame. Exclusions from the RDD frame are not completely
remedied by the existing dual frame approach since the license frame is also incomplete due to
licensing exemptions. The main sources of under-coverage in CHTS are households that do not
reside in coastal counties, households without landline telephone service (Blumberg and Luke
(2008) estimate this at 18% of households at the end of 2007), and households that have landline
numbers that are excluded in standard RDD list-assisted samples (Fahimi, Kulp, and Brick
(2008) estimate about 20% of all landline telephone households are not in the standard RDD
frame). Additional benefits of mail surveys are described in more detail below.
The purpose of this project is to test mail surveys as alternatives to telephone surveys in the dualframe approach. Generally, address frames are more complete and likely provide more accurate
contact information than telephone survey frames. The pilot test is intended to examine various
aspects of a dual-frame approach using a mail survey with samples selected from a general
household address frame and from an angler license frame. Specifically, the goals of the pilot
study are to, 1) assess the coverage of the independent sample frames (license and address
frames), 2) assess the response rates that can be achieved by using a mail survey from both the
general population and the license frame, 3) assess the timeliness of mail surveys for estimating
recreational fishing effort, and 4) determine if overlapping sample frame units (units on both
sample frames) can be identified by matching addresses, rather than relying on survey responses,
as is currently done in the dual-frame telephone surveys. The ability to match sample frames by
address is of particular consequence, as it will provide data for both respondents and nonrespondents, reduce the potential for measurement error, and reduce the length of the
questionnaire by eliminating the need for questions designed to identify overlapping frame units.
Dual-Frame Approach
The target population for the survey is all households that have at least one eligible angler. The
CHTS attempts to survey this population by means of an RDD sample of households that live in
2

counties along the coast. The dual-frame approach samples households that are defined as being
accessible by the union of the address frame and the license registration frame.
The address and license frames are overlapping. In fact, the license frame is essentially a subset
of the address frame (households in the license frame should also be on the address frame). More
generally, the union of the frames consists of three domains: households in address frame but not
in the license frame (S1), households in the license frame but not the address frame (S2), and
households in both frames (S12). If the address frame were complete, then S2 would be empty.
The completeness of the address frame (and the extent that S2 is null) will be evaluated.
Independent samples will be selected from the two frames, and estimates of total numbers of
participants and fishing effort will be made from each of the three domains. From the address
frame, estimates are made for S1 and S12; from the license frame estimates are made of S2 and S12.
Since both frames estimate the characteristics for the overlap domain (S12), these two will be
averaged to produce a more precise estimate for S12. The three estimates are then summed to
produce an estimate of the total population. More details are given on the frames and the
methods of sampling from the frames below.
The Address Frame
The address frame is derived from the USPS Delivery Sequence Files (DSF). Several
commercial organizations have obtained and enhanced the DSF, and these are available for
sampling purposes. The sampling for a mail survey includes all households, not just coastal
county households, thus eliminating one of the sources of undercoverage in the CHTS.
A stratified sample will be selected from the address frame, with different sampling rates in the
strata. Coastal counties will be included in one stratum, and the remaining counties will be in the
second stratum. Since a much higher rate of eligibility is expected in the coastal counties,
optimal allocation rules would result in higher sampling rates being applied in the coastal county
stratum than in the remainder stratum. For this study, the optimal allocation is not as relevant
since the goals of the survey are to better understand the response mechanisms in this new
approach rather than estimating specific population parameters.
The License Frame
The license frame is the list of anglers who are licensed to participate in saltwater fishing in NC.
The state maintains a list of licensed anglers as part of its regular administrative systems. Angler
lists will be pre-processed to identify anglers with duplicate listings and households with
multiple anglers (the survey will be at the household level and not specific to an angler living at
the household).
The license frame will be stratified into the same two geographic strata used in the address
frame, i.e., coastal counties and non-coastal counties. In addition, a third stratum that includes
non-resident anglers will be sampled. After counts of the number of licensed anglers in each
stratum have been determined, the sample will be allocated so that a large enough sample is
selected from each stratum to allow response rates to be estimated reasonably in each stratum.

3

One of the advantages of a mail survey over a telephone survey in a dual frame approach is that
determining the overlap (the households that could be selected from both frames) may be simpler
than in a telephone survey. The sample from the DSF will be matched to the license frame by
address (notice this is matching the DSF sample to the entire license frame not the license
sample). The goal is to determine if the overlap can be quantified without asking survey
respondents questions about their possession of a fishing license (this is likely a source of
measurement error in the current telephone survey approach). If successful, then there are
important consequences since these data are essential for computing the probability of selection
of the households. First, the data are available for both respondents and nonrespondents. Second,
the measurement error in determining the overlap domain can be greatly reduced. Third, this
procedure can be used to improve the allocation of the sample in future administrations (e.g.,
households on the license frame could be assigned a more appropriate sample size). Fourth,
questions designed to determine overlap among the frames would not be needed.
Estimation
As mentioned above, independent samples will be selected from the two frames to make direct
estimates of totals of the numbers of participants and fishing effort from each of the three
domains. Estimates of totals from the address frame will be produced for S1 and S12; while from
the license frame estimates will be produced for S2 and S12. The two totals for S12 will be
averaged to give a more precise overall estimate for S12. The three estimates will be summed to
estimate the total population.
The first step in estimation is to develop base weights that are the inverses of the probabilities of
selection for the units by frame. These standard weights will then be adjusted by the inverse of
the response rates within stratum, separately by frame, to account for nonresponse from within
each frame. The estimates produced from these weights will overestimate totals because the units
in the overlap are over-represented because they could be sampled from both frames. The final
step is to adjust the weights of the units in the overlap. A simple averaging of the two overall
domain estimates can be accomplished by dividing the weight of any unit in the overlap by two.
More sophisticated weighting could be carried out but this is not the main point of the survey and
the average has some benefits, notably simplicity of operation and explanation. More complex
schemes such as post-stratification of the weights will also be investigated.
Information collected through the angler diary survey will be further partitioned into domains
defined by the characteristics of the fishing trips. Two primary domains will be defined; 1) trips
that would be accessible to field samplers (fishing trips occurring at or returning to publiclyaccessible sites), and 2) trips that would not be accessible to field samplers (fishing trips
occurring at or returning to private-access sites such as private residences, community marinas,
private yacht clubs, etc..). Catch rate by fishing mode (shore, private boat) and species will be
independently estimated for each domain.

4

3. Describe the methods used to maximize response rates and to deal with nonresponse.
The accuracy and reliability of the information collected must be shown to be adequate for
the intended uses. For collections based on sampling, a special justification must be
provided if they will not yield "reliable" data that can be generalized to the universe
studied.
Standard mail survey protocol will be implemented (Dillman et al, 2008). Sampled households
will be mailed an advance letter describing the survey and requesting that the household
participate when the questionnaire is sent. Screener questionnaires, along with a $1 cash
incentive, will be mailed three days later. Households will be asked to complete the instrument
and mail it back in the material provided.
Approximately 2 weeks after the first mailing, a thank you postcard will be sent to all addresses.
Households that have participated will be thanked for their prompt response. Households that
have not yet responded will be reminded to participate. Three weeks after the initial contact, a
second questionnaire will be mailed to all households that have not responded. Additional
measures to increase response rates may include a final, specialized mailing by FedEx and/or a
follow-up telephone contact.
In addition to increasing response rates, the sequential reminders will allow us to compare
response variables among respondents who participate with varying levels or prompting (e.g.
early respondents vs. late respondents). This will help us identify and measure non-response
bias.
4. Describe any tests of procedures or methods to be undertaken. Tests are encouraged as
effective means to refine collections, but if ten or more test respondents are involved OMB
must give prior approval.
We plan to conduct focus groups and/or cognitive testing with up to 50 individuals to ensure that
instructions and survey instruments are clear. Mail survey questionnaires will be developed by
Dr. Nancy A. Mathiowetz, University of Wisconsin-Milwaukee.
5. Provide the name and telephone number of individuals consulted on the 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.
Statistical support was provided by the following:
Dr. J. Michael Brick, Westat, 301-294-2004
Dr. Nancy A. Mathiowetz, University of Wisconsin-Milwaukee, 414-229-2216
Dr. Dave Van Voorhees (301-713-2328) is Chief of the Fisheries Statistics Division, which
administers the MRFS Program.
The present contractor for the telephone survey is Macro International, Inc., of Burlington,
Vermont.

5