The Paperwork Reduction Act and Customer Surveys
As stated
elsewhere in the NOAA PRA guidance, the normal clearance process for
information collections takes from 2-5 months, depending upon whether a
proposed rule is involved. In an effort to reduce this delay for certain
types of customer surveys, NOAA has obtained a "generic clearance" for
them. The clearance makes it possible to get approval, within a few weeks, for
eligible surveys. You cannot use any of the "cleared questions"
linked below without going through this fast-track clearance process.
NOAA's OMB Desk Officer has set some new parameters for
use of the generic clearance process. The questions in the generic
clearance fall into two categories, quantitative and qualitative. The
quantitative questions ask the respondents to rank services or products
on some number scale. A number of NOAA offices have requested approval
for including quantitative-type questions on their surveys without
planning to conduct a survey that follows a statistical sampling plan.
The results of the quantitative questions could not therefore be
extroplated (in a valid way) to the overall customer population. Offices
got around this by saying in their justifications that they did not
intend to extrapolate that data, but planned to just use it to indicate
problem areas, places where there should be improvements, etc. OMB will
not longer accept this approach, since despite these statements the data
can easily be misused to represent a program's/office's
customers.
Future requests seeking to use quantitative questions will
need some description of a statistical sampling plan in order to get
approval.
The
Surveys:
- There
are two sets of cleared
questions, one with quantitative (ranking) questions about
satisfaction levels, and one with questions about uses of products,
suggested formats, and other issues.
- You do
not have to use either questionnaire in its entirety - you can select
only those questions that fit your needs.
- You can
adapt the questions to fit your particular program needs as long as
the basic objectives of the surveying effort are consistent with the
approved questionnaires. You can consult with the NOAA Clearance
Officer about how much you can vary from the examples.
- OMB does
not regard Web-based surveys as an acceptable method of sampling a
broad population (because it is only sampling people with Web access).
Web-based surveys must be limited to services provided by Web.
How You
Use the Generic Clearance:
- Select
and adapt the survey questions to fit your needs.
- Provide
a copy of the draft survey to the NOAA Clearance Officer for review.
The survey must provide certain PRA information to respondents. See
the last page of the cleared
questions for an example. On Web surveys, only the OMB number and
expiration date have to appear on the survey screen; the rest of the
information can be provided via a link.
- If told
by the NOAA Clearance Officer that the survey does properly fit under
the generic clearance, you need to prepare a clearance
request.
- The
first page of the request is a form. You need to fill out the
following fields: NOAA subagency (e.g. your Line Office), the title of
your survey, an estimated annual number of respondents, the estimated
hours or minutes per response, a contact point for questions about the
survey, and the signature of a program official. Leave the other
fields blank.
- The
second page of the request is a narrative answering the four questions
provided in the request guidance.
- As of
July 2005, with the approval of the renewal request for 0648-0342, Part B (Statistical analysis) of the supporting
statement for regular PRA submissions now also needs to be
completed even if you are not going to perform statistical
analysis on your data. Part B questions overlap with the
questions in the clearance request – covering more than statistical
analysis per se - so some copying and pasting can be done, but
you need to read each question carefully to be sure you are answering
all parts of it. See Supporting Statement
Instructions for information on responding to the Part B
questions. As noted in those instructions, ALL survey requests now
need to include answers to Part B, regardless of actual use of
statistical analysis.
- You then
send your request to the NOAA Clearance Officer, who sends it to DOC
to forward to OMB.
- OMB has
two weeks (but may take longer) to approve or reject the
request.
Please note - you cannot just use the questions from the
generic surveys without obtaining OMB approval through the 2-week review
process.
Problems with Customer
Surveys:
When people
learn of the need to get OMB approval for customer surveys, a common
reaction is that the requirement is an unnecessary piece of bureaucracy
when an office just wants to get feedback from its customers.
Experience, however, has shown that some plans for customer surveys are
among the worst in the agency. Part of the problem is that people doing
customer surveys often aren't familiar with surveying techniques and how
to analyze the resulting data. They often do not think out how the
information will be used, and if it will be valid for that use. The most
common reason for OMB rejection of plans is that the data produced will
have little or no value.
Let's look
at an example. An office wants to survey its customers for a product.
There are 100 customers and the office expects 15 of them to actually
respond to the survey. The questions in the survey are numerical
rankings of satisfaction with the product or related service. The
question here is what the resulting information will mean. Will the
office produce a report that says 60% of customers are happy with the
product? But if it was really only 9 people (60% of the 15 responding)
who were happy, how do you know that you can extrapolate the results to
all of the customers? Usually saying that 60% of respondents were happy
doesn't necessarily mean much either if the response rate is low. Even
questions that don't have numerical answers may produce little of value
for these reasons.
So how do
you avoid this problem? Generally you should look at one of three
approaches:
- Plan to get a response rate of 65-70% of your customers. The
data obtained can then usually be extrapolated to all of your
customers. The problem, of course, is that unless you have a close
relationship with your customers it can be hard to get this high a
response rate.
- Statistically sample your customers. This allows you to focus
on a smaller number of people and still obtain useful results. The
problem with this approach is that you need a firm basis in statistics
to use it, and you need to know a good bit about your customers in
advance to properly formulate a sampling plan.
- Don't
ask questions that need to be extrapolated to all customers. For
example, don't use numerical rating questions. You can ask customers
about things like what additional formats or features they might like
to see, or similar questions where a certain level of interest could
justify action even if you aren't sure what the overall customer
population wants.
The bottom
line is that you really need to think about these issues and consult
with the NOAA Clearance Officer to avoid clearance problems with a
customer survey.