Supporting_Statement_Part B

Supporting Statement for Continuing Collection RE: Visitor Survey for Cape
Hatteras National Seashore
OMB Control Number 1024-0258
B. Collections of Information Employing Statistical Methods
The agency should be prepared to justify its decision not to use statistical methods in any case where such methods
might reduce burden or improve accuracy of results.

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
government units, households, or persons) in the universe covered by the collection and in the corresponding
sample are to be provided in tabular form for the universe as a whole and for each of the strata in the proposed
sample. Indicate expected response rates for the collection as a whole. If the collection had been conducted
previously, include the actual response rate achieved during the last collection.

The potential respondent universe for the visitor interviews includes all visitors 18 years of
age or older using the ocean-side beaches at Cape Hatteras National Seashore (CAHA) from
August 2009 through November 2009 (under the emergency clearance) and between
December 2009 and July 2010 (under a regular PRA clearance). This includes visitors who
use the ocean-side beaches between 6:00 a.m. and 6:00 p.m. Visitor groups will be
intercepted based on a sampling plan (described in section B.2) stratified by season, location
in the park, day of the week, and time of the day.
Table B1 presents the approximate visitation based on National Park Service (NPS) public
use statistics for November 2007 to October 2008. It’s assumed that the visitation patterns
observed between November 2007 and October 2008 are representative of the visiting
behavior expected for the study period. We expect that the large majority of visitors will be
on the beach at some point between 6:00 a.m. and 8:00 p.m. For the visitor survey, we expect
a response rate of 60%, based on a 74% response rate to a 2002 visitor survey at CAHA that
involved a shorter on-site contact, followed by a mailback survey.
Table B1. Potential Universe for the Visitor Survey.
Sampling months

ALL

Estimated Size of Respondent
Universe

Sample Size

2,112,126

2,000

Expected
Response
Rate
60%

Estimated
Responses
1,200

2. Describe the procedures for the collection of information including:
*
*
*
*
*

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

Visitors will be sampled on the ocean-side beaches. The sampling plan will be stratified by
season, day of week, time of day, and type and location of beach. Scheduled sampling days
will be based on data supplied by nine housing rental companies on Hatteras Island for
occupancy rates by week for 2007 and counts of visitors on beach sections collected as part
1

of an on-going effort by RTI International to document baseline vehicle use and the number
of visitors on different sections of the beach. The occupancy rates and counting data will be
used to estimate the proportion of visitors (including pedestrians) on the beach each week of
the year, and selected weeks will be sampled proportionately based on the occupancy data.
Weekends will be sampled at twice the rate of weekdays because information from park staff
and from counts of vehicles using the beach indicate that there are about twice as many
visitors on the weekend as on weekdays. Sampling days will be divided into three four-hour
time periods: from 6 a.m. to 10 a.m., from 10 a.m. to 2 p.m., and from 2 p.m. to 6 p.m. Two
four-hour periods will be randomly selected for each day on which the survey is conducted.
The beaches will be divided into sections varying in length from 1/16th mile to ½ mile (most
of the sections will be 1/8th mile), based on the expected number of groups on different
sections of the beach. GPS coordinates will be available for each segment. The beach
segments will be clustered according to location. One cluster of beach segments will be
selected for each four-hour interview time period to allow for easy driving or walking
between segments. Interviews will be conducted at three beach segments during each fourhour time period. The start time for surveying on each beach segment will be randomized
across the four hours.
The visitor survey will be conducted by a team of two interviewers using handheld devices to
administer in-person computer-assisted questionnaires. The process is as follows:
• Use GPS receivers to identify the beginning of the section of beach.
• Interviewer 1: Start by walking along the dune line and counting the number of
groups of people on the beach and in the water until the end of the segment is
reached.
• Based on a pilot test in August 2009, we calculated the number of surveys per hour
the interviewers are likely to be able to complete. After counting the groups, the
number of groups is entered into a formula in the handheld computer survey device.
• If the number of groups is less than or equal to the number of interviews that are
expected to be completed every hour, then every group in the segment will be
interviewed.
i. Interviewer 2: start interviewing at the beginning of the segment
ii. Interviewer 1: after counting, start interviewing from the end of the segment
• If there are more groups than can be interviewed in the allotted time, then the formula
will tell the interviewers to interview every Nth group and randomly select a group
between 1 and N to start with. The interviewers will walk in a serpentine pattern,
starting at the dune and moving toward the water interviewing every Nth group.
When the interviewers reach the water, they will return to the starting point and
repeat the selection process until the time limit is reached.
• To interview:
i. Approach a group;
ii. Ask who is in the “group,” including members who might be in the water;
iii. Ask to interview the member of the group 18 years of age or older who has the
next birthday;
iv. Conduct the interview

2

•

For safety reasons, do not interview individuals in moving vehicles, but do interview
groups sitting near parked vehicles

I
Estimates will be produced using a design-based approach that incorporates the sampling
weights to estimate the population parameters. If non-response is in an issue, then appropriate
methods, such as weighting-class adjustments, will be used to reduce the potential bias induced
by the non-response.
Approximately 25% of the 1/8-mile beach segments will be surveyed (about 60 miles of beach
and 30 days of interviewing in four segments each day). We will be sampling the segments with
replacement.
The design effect was calculated assuming that six beach segments will be selected at random for
each sampling day using a stratified sampling design. It also was assumed that one adult member
of every group of beach users in the segment at the time of the survey will be selected at random.
This results in a “census of groups of beach users” in the selected beach segment.
The proposed 30 days of interviewing will yield 180 (30 x 6) beach segments (which result in
clusters of beach users). The 180 beach segments will be allocated proportionately to the number
of segments in the two strata (Stratum 1 = Hatteras and Bodie with 49 beach segments, and
Stratum 2 = Ocracoke with 18 beach segments). After combining all this information, the
proposed stratified random sample of 180 beach segments and census of groups within selected
beach segments results in a design effect of 1.01
Note that a sample of size n drawn using simple random sampling from a population will usually
be more efficient (i.e., have smaller sampling errors) than a sample of the same size drawn by
means of a sample of pre-existing clusters in the population. Stratification tends to reduce the
sampling errors of the estimator; therefore, it tends to increase the efficiency of the sampling
design.
The total number of samples is constrained not only by the available budget, but by consideration
of the precision that can be expected. Approximate precision estimates may be obtained,
assuming the survey design is simple random sampling. Under this assumption, the confidence
interval half-width (precision) can be estimated using the following equation:
Half − width = Z 1−α * estimator variance

where Z1−α is related to the level of confidence required for the estimate (Cochran, 1987). Given
that our proposed survey design considers stratification to account for the different types of
beaches, the actual precision estimates are expected to be better (smaller confidence intervals)
than those obtained with the equation above.
The visitor survey has two objectives: 1) to estimate the percent of the population that will
change visit behavior under a proposed alternative, and 2) to estimate the mean number of days
in a trip under a proposed alternative. (Trip length has relevant economic consequences.)

3

For the first objective, if the true proportion of the population estimated to change its
behavior under a proposed alternative is 40%, then to produce estimates within 10% of this
true value (95% confidence level, 80% power), we need about 377 participants (Table B2).
Table B2. Sample Sizes Required to Estimate True Proportions at Specific Power Levels and
Confidence Intervals.
Precision (Half-width)
True Proportion 0.05 0.06 0.07 0.08 0.09 0.1 0.11 0.12 0.13 0.14
0.10
566
393 289 221 175 142 117
99
84
73
0.15
801
556 409 313 248 201 166 139 119 103
0.20
1005
698 513 393 311 252 208 175 149 129
0.25
1178
818 601 460 364 295 244 205 175 151
0.30
1319
916 673 516 407 330 273 229 196 169
0.35
1429
993 729 559 441 358 296 249 212 183
0.40
1507 1047 769 589 466 377 312 262 223 193
0.45
1555 1080 793 608 480 389 322 270 230 199
0.50
1570 1091 801 614 485 393 325 273 233 201
From Cochran W.G. 1987. Sampling Techniques. New York. John Wiley & Sons

0.15
63
89
112
131
147
159
168
173
175

0.16
56
79
99
115
129
140
148
152
154

0.17
49
70
87
102
115
124
131
135
136

0.18
44
62
78
91
102
111
117
120
122

0.19
40
56
70
82
92
99
105
108
109

For the second objective—to estimate the mean number of days in a trip under a proposed
alternative—we can use an estimate based on the coefficient of variation (CV), which is the
standard deviation divided by the mean, i.e., the standard deviation as a percent of the mean.
The smaller the standard deviation as a percent of the mean, the lower the CV. If we expect
high variability in responses (i.e., the range in number of days varies substantially in the
population), then we should be considering population CVs in the range of 75% to 125%.
This results in samples that vary in size from 9 to 1539, depending on the desired precision
(Table B3).
At CAHA, the final sample size should be the larger of the two sample sizes that satisfies
both the first and second objectives of the visitor survey. This means the sample size for the
visitor survey should be closer to 1,539 than to 377.
Table B3. Sample Sizes Required to Estimate a Parameter at Desired Precision Levels and
Coefficients of Variation.

Population
CV
25%
50%
75%
100%
125%
150%

5%
98
387
867
1539
2403
3460

Confidence level =

Percent Relative Error of the Estimated Mean
10%
15%
25%
50%
26
13
6
3
98
45
18
6
219
98
37
11
387
173
64
18
603
269
98
26
867
387
141
37

75%
3
4
6
9
13
18

0.95

4

0.2
36
51
63
74
83
90
95
98
99

3. Describe methods to maximize response rates and to deal with issues of non-response. The accuracy and
reliability of information collected must be shown to be adequate for intended uses. For collections based on
sampling, a special justification must be provided for any collection that will not yield "reliable" data that can
be generalized to the universe studied.

Data from the survey will be used to inform the economic analysis of the ORV management
alternatives. The analysis will focus on the economic impacts of the alternatives relative to
the baseline conditions under the two no-action alternatives (years 2007 and 2008). In order
to analyze the new alternatives, we must select parameters to quantify changes in factors that
affect visitors (such as ORV restrictions). Because of the uncertainty inherent in forecasting
future activity, the economic analysis will estimate scenarios covering a range of possible
impacts.
The data from the visitor survey is one source of information that will be used to establish the
range of possible values for different parameters. Other sources include publicly available
secondary data on visitation and economic conditions. The visitor survey and other data
sources will be combined to create reasonable ranges for the parameters in the economic
model. We plan to use several sources of information, because no single source provides all
the data we need, and each source has positive and negative features. To reduce the concern
over the possibility that visitors will answer the survey strategically to influence the outcome
of the economic analysis, the questions in the survey and comparisons with other data
sources will be used to identify strategic responses.
To maximize response rates to the visitor survey, the instrument was based on input from a
number of parties with survey research experience and members of the public and park staff
familiar with CAHA visitors. This input also was used to make sure the questions could be
understood, the minimum numbers of questions were asked, and the questions collected the
data needed for the analysis. The interviewers will be carefully trained in how to sample and
make contact with visitors and how to convert initial refusals into completed interviews.
Non-response rates will be monitored throughout the year of data collection to identify and
correct problems.
We will take several steps to identify non-response bias and correct for bias in our analysis.
The interviewer will attempt to ask refusers their Zip code or country of residence and also
ask (or observe) if the visitor has ever driven a vehicle on the beaches at CAHA. In addition,
any easily observable traits of the visitors (such as group size and presence of children), as
well as time of day, day of week, and location of interview will be recorded on a log for each
contact for a later comparison of respondents with non-respondents.. This information, along
with a code for the interviewer, will be analyzed using a probit or logit model to identify
systematic patterns in non-response. In any regression analysis, a two-step selection model
(such as a Heckman selection model or propensity model) will be used to help correct
regression results for selection bias.
4. Describe any tests of procedures or methods to be undertaken. Testing is encouraged as an effective means of
refining collections of information to minimize burden and improve utility. Tests must be approved if they call
for answers to identical questions from 10 or more respondents. A proposed test or set of tests may be submitted
for approval separately or in combination with the main collection of information.

5

The survey instrument was developed with input from contractor staff, including economists,
survey research experts, and statisticians. National Park Service staff, including staff from
the Social Science Division, and members of the Regulatory Negotiating committee for
CAHA, also were consulted about the survey and survey administration. A scoping trip to
CAHA took place in October 2008 to identify beaches and beach uses. This information was
used to assess the feasibility of different sampling methods. A pre-test took place in
December and January involving 9 visitors to evaluate and improve the procedures and the
survey instrument. Finally, a pilot test was undertaken in August 2009 under the emergency
clearance to test the final procedures and instrument.
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.

RTI International, a contractor to NPS, designed the survey and will collect and analyze the
data. The project manager is Dr. Carol Mansfield (919-541-8053). Other RTI staff
participating in the survey design and data collection are Dr. Brenda Muñoz, Mr. Brian
Evans, and Dr. Christine Poulos.

6