Estimation Methods Overview
To estimate total recreational catch, we expand catch rate—or the number of fish caught per fishing trip—by effort—or the total number of fishing trips taken.
At its core, estimating total recreational catch involves expanding catch rate, or the estimated number of fish caught per angler trip, by effort, or the estimated number of fishing trips taken. Weighting each catch record and fishing effort questionnaire allows the data we collect to represent themselves and the broader population of anglers we weren't able to sample. In this way, we’re able to draw conclusions about the full recreational fishing community without having to collect information from each member of that community.
To follow real data points through this estimation process, watch From an Interview to an Estimate: Understanding How NOAA Fisheries Estimates Recreational Catch. Download this presentation (PDF, 38 pages) or find more detailed information about these estimation methods in our Survey Design and Statistical Methods Manual.
Catch Rate
Sampling Angler Trips
Catch rate, or the estimated number of fish caught per angler trip, is determined using data collected through the Access Point Angler Intercept Survey (APAIS). During each APAIS interview, state samplers record:
- The location, date, and interview time.
- The mode of the angler's trip (e.g., shore, private or rental boat, or for-hire vessel).
- The general area where the angler fished (e.g., inland, state territorial sea, or federal Exclusive Economic Zone).
- The species, number, and disposition of the angler's catch (e.g., observed harvest, reported harvest, or released alive).
Before an angler’s catch data can inform our estimates, these data must be weighted. Statistical weighting is a standard practice used by government agencies and private institutions around the world. Weighting ensures our sampled units are able to represent themselves and the broader population of anglers we weren’t able to sample. In other words, weighting allows an individual catch record—for example, two summer flounder caught in September by a New Jersey angler fishing from a private boat in federal waters─to represent the catch of other anglers taking similar trips—fishing during the same two-month sampling wave, in the same state, mode, and fishing area. In this way, we’re able to draw conclusions about the full recreational fishing community without having to collect information from each member of that community.
Weighting Catch Rate Data
There are three steps, or stages, to producing the sample weight that will be applied to a recreational catch record.
Stage one weights are based on the probability that a particular site assignment will be selected for sampling. Site assignments are unique combinations of fishing sites, dates, and time intervals during which field interviewers conduct the APAIS. Some fishing site-day-time combinations are more likely to be selected for sampling than others, based in large part on the amount of fishing expected to occur during that assignment.
Stage two weights are based on the duration of the site assignment. Site assignments consist of one six-hour window, which field interviewers spend at either one high-activity site or two low-activity sites. If a field interviewer is on-site for the full length of their six-hour assignment, the stage two sample weight is one. If a field interviewer isn’t on-site for the full length of their assignment—for example, because extreme weather forced them to leave—we “upweight” their assignment based on the time they did spend sampling. This ensures our samples are representative of every hour within a given day.
Stage three weights are based on the proportion of interviews a field interviewer is able to complete. If an interviewer is able to collect information from every angler they observed during their assignment, the stage three sample weight is one. If a field interviewer isn’t able to intercept every angler trip—for example, because there are too many anglers at a site—we “upweight” their assignment based on the number of trips they did intercept. This ensures our samples are representative of all of the anglers that fished from a site during an assignment.
The final sample weight is equal to the stage one weight multiplied by the stage two weight multiplied by the stage three weight. It can be understood as the number of additional angler trips that will be represented by the trip that was sampled. This weight is multiplied by the catch data collected during that same site assignment to produce a weighted catch record for each interview.
Estimating Catch Rate
Once the weighting process has been completed for every angler interview, we place each weighted data point into domains, or groups that share the same characteristics. These domains are defined by:
- Year
- Two-month sampling wave (e.g., January-February, March-April, May-June, etc.)
- Geographic sub-region
- State
- Fishing mode (e.g., shore, private or rental boat, or for-hire vessel)
- Area fished (e.g., inland, state territorial sea, or federal Exclusive Economic Zone)
- Species
- Catch type (e.g., observed harvest, reported harvest, or released alive)
We estimate catch rate within these domains as a weighted average, beginning at the most granular level. First, we determine the catch represented by our sample by adding all the weighted data points within a domain together. Then, we determine the number of trips represented by our sample by adding each data point’s sample weight together. Finally, we determine catch rate, or the number of fish caught per angler trip, by dividing the catch by the trips represented by our sample.
Fishing Effort
Sampling Households
Shore and private boat fishing effort, or the estimated number of fishing trips taken from these two fishing modes, is determined using data collected through the Fishing Effort Survey (FES). Each FES questionnaire asks recipients to record:
- The number of days they went saltwater fishing from shore in their state of residence during a specified time period.
- The number of days they went saltwater fishing from a private or rental boat that returned to shore in their state of residence during a specified time period.
Before an angler’s effort data can inform our estimates, these data must be weighted. Statistical weighting is what allows an individual fishing effort report—for example, two shore trips and four private boat trips taken in October by a New Jersey household—to represent the fishing activity of other households in the same state.
Note: The following process applies only to estimating shore and private boat fishing effort. For-hire fishing effort is determined using data collected through the For-Hire Survey (FHS) and regional for-hire trip reporting programs. More information about for-hire effort estimation can be found in our Recreational Fishing Survey Design and Statistical Methods Manual.
Weighting Fishing Effort Data
There are three steps to producing the sample weight that will be applied to a particular FES questionnaire.
Base weights are based on the probability that a particular household will receive our questionnaire. Some households are more likely to be selected for sampling than others, based in large part on known increases in fishing activity among households that are closer to the coast and home to someone who possesses a recreational fishing license.
Sampling households based on their anticipated levels of fishing activity helps ensure our sample is efficient: Sampling licensed households more often makes it more likely that the households we survey will report some number of fishing trips. Sampling unlicensed households ensures we’re capturing fishing activity that may be occurring by people who aren’t required to or otherwise don’t have a fishing license.
Next, we adjust our base weights to account for known differences in response rates among households that are closer to the coast and home to someone who possesses a recreational fishing license. This ensures our sample is representative of all of the households that received our survey, rather than only those households that responded to it.
Last, we adjust our weights one more time to account for potential differences between our sample frame—which is limited to households serviced by the U.S. Postal Service—and an independent demographic data set reported by the U.S. Census Bureau’s American Community Survey. This common statistical technique ensures our sample is representative of all of the households in surveyed states, rather than only those households on our sample frame.
The final sample weight can be understood as the number of additional angler trips that will be represented by the trip that was sampled. This weight is multiplied by the effort data collected through a particular questionnaire to determine weighted effort.
Estimating Fishing Effort
Once the weighting process has been completed for every FES questionnaire, we place each household’s weighted data into domains, or groups that share the same characteristics. These domains are defined by:
- Year
- Two-month sampling wave
- Geographic sub-region
- State
- Fishing mode
We estimate fishing effort within these domains, beginning at the most granular level. First, we determine the total effort represented by our sample by adding the weighted household data within a domain together. Using data collected through the APAIS, we adjust this estimate in order to:
- Proportion trips by “area fished,” because our catch rate data is proportioned as such.
- Account for out-of-state fishing trips, which aren’t captured through the FES.
Once these adjustments are made, we have our estimate of shore and private boat fishing effort, or the number of fishing trips taken from these two modes.
Total Recreational Catch
We multiply weighted catch rate by weighted effort from corresponding domains to determine total recreational catch. We produce catch and effort estimates by calendar year, several fishing-year options, cumulative estimates, geographic sub-region, state, fishing mode, area fished, species, and catch type. Because the precision of our estimates generally improves as sample sizes increase, our estimates are best viewed at higher levels of aggregation.
Our staff practice quality assurance and control measures before our estimates are published. We check for errors in data entry and investigate any unusual changes in catch and effort trends. We also involve NOAA Fisheries’ regional offices and science centers, who have local on-the-ground knowledge, in the review of preliminary estimates before they are published. Preliminary wave estimates may be revised before they are published as final. All wave-level estimates are considered preliminary until made final in April of the following year. Learn more about our estimate publication schedule and estimate review process.
These catch estimates are critically important for science and management needs: Stock assessment scientists use recreational catch estimates to inform their understanding of stock size and sustainable harvest levels. Fisheries managers use this information to set regulations that help ensure access to fishing and promote the long-term health of our fish populations. Taking a few minutes to participate in a recreational fishing survey is one of the most important things anglers can do to support fisheries science, management, and the sustainability of a great American pastime.