An Introduction to Marine Recreational Information Program Data

NOAA Fisheries’ Marine Recreational Information Program works with state and regional partners to produce the recreational fisheries statistics that help scientists and managers assess and maintain sustainable U.S. fish stocks. In accordance with NOAA’s Information Quality Guidelines, program staff work to provide those who are affected by science and management decisions with the information needed to understand the data that informed them. This includes providing open access to:

• The data used;
• The analytical methods applied; and
• The assumptions and statistical procedures employed.

Guidance for Data Users

While this webpage provides an overview of our data products, production schedule, review process, and use considerations, fisheries analysts and stock assessors are encouraged to download the MRIP Data User Handbook for more detailed information about downloading, exporting, querying, and performing custom analyses of our recreational fishing data. Data users are also encouraged to watch recordings of the 2021-2022 MRIP Data User Seminar Series.

Data Products

NOAA Fisheries’ Marine Recreational Information Program provides open access to its recreational fishing microdata and its catch and effort estimates.

• Microdata are the recreational fishing information gathered through our recreational fishing surveys.
• Estimates are the calculated statistical values produced from these microdata.
• Data is a term that encompasses both microdata and estimates.

There are several ways to access the Marine Recreational Information Program’s data products. You can:

• Visit our Recreational Fishing Data Downloads page to access our public-use datasets and the template programs we have developed to support custom domain analyses.
• Use the MRIP Query Tool to filter recreational fisheries statistics by time series, geographic area, species, mode, and other characteristics.
• Submit custom data requests.

A complete list of updates to MRIP data, estimates, and queries can be found on our Recreational Fishing Estimate Updates page. To be notified of these updates, subscribe to our email service.

In accordance with NOAA Fisheries’ Data and Information Management Policy Directive (PDF, 5 pages), metadata that describes the “what, where, when, how, and who” of the Marine Recreational Information Program’s data holdings is available on InPort: a centralized repository of NOAA Fisheries’ data documentation.

Data Availability

Data Collection Schedule

Different recreational fishing surveys are administered in different regions at different times. In the Northeast, for example, the agency’s Access Point Angler Intercept Survey (APAIS) is not administered during known periods of low fishing activity. The table below indicates which of the Marine Recreational Information Program’s surveys are conducted during which two-month sampling periods, or waves.

Note: The APAIS and For-Hire Survey (FHS) do not sample headboats south of Virginia, as these vessels are covered by the Southeast Fisheries Science Center’s Southeast Region Headboat Survey. The APAIS and FHS are not administered in Hawaii, but Hawaii does implement an angler intercept survey alongside the FES. Data collection efforts are suspended in Puerto Rico as the territory rebuilds following Hurricane Maria (September 2017).

Estimation and Publication Schedule

Preliminary microdata and estimates from our general surveys, which include the APAIS, FES, and FHS, are published approximately 45 days after the end of each two-month sampling period (April 15, June 15, August 15, October 15, December 15, and February 15). Final microdata and estimates from these surveys are published on or around April 15 of the following year.

Preliminary microdata and estimates from the LPS are published approximately two months after the end of each one-month sampling period (i.e., June estimates are published in September, July estimates are published in October, etc.). Final microdata and estimates from the LPS are published in April of the following year.

Note: Once the agency’s Recreational Fishing Survey and Data Standards are fully implemented, we will no longer publish estimates that are specific to each sampling period. Instead, we will publish cumulative estimates every two months, beginning with the first survey administration of the survey year. Data users who choose to create estimates not available through our Query Tool may do so using the microdata and custom domain analysis programs available on the Recreational Fishing Data Downloads page and the guidance provided in the MRIP Data User Handbook. However, custom, user-produced estimates are the responsibility of the user and should not be considered official MRIP products.

To be notified of updates to MRIP data, estimates, and queries, subscribe to our email service.

Learn more in Section 2.2 of the MRIP Data User Handbook

Data Review

To minimize the potential for error in our estimates, we follow a number of best practices for quality assurance and quality control.

To support quality assurance, which works to prevent invalid data from entering our system, we:

• Ensure field interviewers are trained.
• Ensure data entry is intuitive.
• Conduct validation interviews.
• Build checks into the estimation process to help identify invalid or mismatched data.

To support quality control, which works to detect and correct errors that make it into our data, we:

• Manually review our estimates for potential errors at each step of analysis. The manual review process often begins with a high-level look at catch and effort estimates. This can be an effective way to spot systematic changes that may indicate an error in the estimation run, such as mistakes in data entry or errors in the estimation program itself. Our internal estimate review team includes statisticians with the Office of Science and Technology, as well as Regional Office and Science Center staff. Regional partners are able to flag unusually high or low estimates based on their familiarity with local fisheries and fish stocks.
• Use statistically backed approaches to systematically identify outlier estimates. When an outlier is identified, statisticians must consider several factors to help them investigate the reason for a sudden increase or decrease and determine whether corrections must be made. For example: Is the estimate based on a small sample size, or influenced by an unusual datapoint? Did a weather event or change in fishing regulations cause a spike or sharp decline in related fishing activity? Are there other spikes in the time series?

Depending on the results of this review, statisticians can generally take one of three actions:

• Correct an identified error in the underlying data or estimation program. For example, if we find that a catch record was attributed to the wrong species, or attributed to one angler when it was reported by a group, a correction would be made.
• Adjust the sample weight to minimize the impacts of an influential, non-representative data point. For example, if we verify the accuracy of an unusual catch record with a state sampler, but determine that it is uncharacteristic of the catch that is generally reported for that species, its sample weight may be adjusted, or “trimmed,” to reduce the impact of this non-representative catch. It is important to note that this process does not only adjust one data point: it also redistributes the “excess” sample weight to other data points within the estimation domain. This action must therefore be applied carefully to ensure an adjustment to one sample weight does not create additional outliers elsewhere.
• Take no action, if investigations have not identified errors and there are no indications the data are not representative. In these circumstances, data users may wish to reduce an outlier’s influence by exploring alternative estimation methods.

Note: The actions that Office of Science and Technology data scientists and statisticians can take to adjust estimates are limited. Simply deleting data that seem unusual or revising results to appear more “realistic” would compromise the scientific integrity of our recreational fisheries statistics. However, data users familiar with local fisheries and fish stocks may employ additional techniques to ensure estimates meet their specific needs, including model-based approaches to “smooth” trends in the time series.

Data Use Considerations

Data users should be aware of the limitations of our data.

Revisions

Preliminary estimates may be revised before they are published as final, and final estimates may be revised if errors are found. The direction and magnitude of such revisions cannot be predicted. When substantial revisions are made, subscribers to our email service are notified and notes are posted to the MRIP Query Tool and Recreational Fishing Estimate Updates webpage.

Precision

Percent standard error, or PSE, is a measure of precision that is published alongside all of our point estimates. Estimates should be viewed with increasing caution as PSEs increase beyond 30%. Once our Recreational Fishing Survey and Data Standards are fully implemented, the MRIP Query Tool will not include an estimate when its PSE exceeds 50%. (A PSE of 50% or above is considered highly imprecise.)

Sample Size

Small sample sizes may result in imprecise estimates. Catch estimates for rare-event species, for example, are often less precise than catch estimates for commonly caught species. But when we group year, state, wave, or mode estimates together, sample sizes increase and precision improves. For this reason, our estimates are best viewed in aggregate: annually and at the state or regional level.

Time Series

Generally speaking, estimates from Maine through Mississippi may be compared across extended periods of time, because calibration methodologies have been applied to account for changes in survey design and sample coverage over the years. However, we advise caution in using the National Summary Query to make such long-term comparisons for estimates in Louisiana, California, Oregon, Washington, and Hawaii.

Weight Estimates

In some cases, landed fish may not be represented in weight data. This can occur when no fish were observed, or when observed fish were too large for a weight measurement to take place. Furthermore, weight estimates published in the MRIP Query Tool may differ from weight estimates published by the Southeast Fisheries Science Center, which follows a different weight estimation procedure for South Atlantic and Gulf of Mexico managed species. More information about how weight estimates are produced can be found in the Weight Data entry of the Recreational Fishing Data Glossary.

Learn more in Section 3 of the MRIP Data User Handbook

More Information

• Recreational Fishing Data Downloads
• Recreational Fishing Data and Statistics Queries
• MRIP Data User Handbook