Estimation Methods Overview

The Marine Recreational Information Program (MRIP) strives to be transparent about its work to collect recreational fishing data. While supplemental and specialized surveys can be used to collect data for select fisheries or during select fishing seasons, we administer three general surveys to collect the data that are used to produce our general estimates.

• The Access Point Angler Intercept Survey (APAIS) is an in-person intercept survey conducted at marinas, boat ramps, beaches, fishing piers, and other publicly accessible fishing sites in Hawaii and on the Atlantic and Gulf coasts. It is used to obtain information about the number and species of fish that are caught, kept, and released on saltwater fishing trips taken from the shore or a private boat.
• The Fishing Effort Survey (FES) is a mail survey that collects information from households in Hawaii and on the Atlantic and Gulf coasts. It is used to estimate fishing effort, or the number of saltwater fishing trips taken by anglers from the shore or a private boat.
• The For-Hire Survey (FHS) is a telephone survey that collects effort information from for-hire vessel representatives and estimates fishing effort from the for-hire sector. It asks for-hire operators to report vessel-fishing activity during a one-week reference period and to recount details from each trip (e.g., methods used, area fished, and species targeted). It is conducted on the Atlantic and Gulf coasts from Maine through Mississippi.

From these surveys, we can estimate:

• The species caught.
• The size, weight, and number of fish caught and kept.
• The number of fish discarded.
• Where people are fishing.
• How many people are fishing at a particular location.

We also strive to be transparent about our work to produce recreational catch statistics. Below, we’ve outlined the fundamental methods of estimating total recreational catch.

Basic Estimation

We produce estimates of total recreational catch by multiplying catch rate—or the number of fish caught per fishing trip—by effort—or the total number of fishing trips taken. A fishing trip can last any amount of time, from a few minutes to an entire day.

Catch rate can be expanded by effort to generate an estimate of total recreational catch. If, for example, our effort survey indicates there were 100 fishing trips in a particular region during a particular period of time and our catch survey indicates anglers in that region during that time caught an average of two cod per trip, our estimate of total recreational catch could be determined through the following calculation: 2 cod per trip x 100 fishing trips = 200 cod caught. It is important to note that this is a simplified example; our actual data collection and estimation processes are more complex.

While this is the most fundamental method of calculating total recreational catch, it is important to note that we do adjust our effort estimates to account for individuals who could not be surveyed. For example, because the FES only samples households in coastal states, estimates derived from this survey must be adjusted to account for trips taken by inland residents.

We produce estimates for all modes of fishing (including shore, private boat, and charter boat), all species caught, and three types of catch: observed fish that are caught, brought back to the dock, and identified by samplers (known as Type A catch estimates); reported fish that are caught, released dead, used for bait, or filleted (known as Type B1 catch estimates); and reported fish that are released alive (known as Type B2 catch estimates).

Basic Estimation Example

The following table uses three steps to demonstrate how total recreational catch might be estimated for anglers fishing from private boats in Massachusetts between May and October of 2011:

1. Begin with Type A, Type B1, and Type B2 catch-per-trip estimates; effort estimates; and effort adjustment factors.
2. Multiply each effort estimate by its associated adjustment factor to determine adjusted effort estimates.
3. Multiply each adjusted effort estimate by its associated catch-per-trip estimates to determine total recreational catch.

In this example, our estimates indicate that anglers fishing from private boats in Massachusetts between May and October of 2011 took 1,237,360 fishing trips and caught 413,520 fish: 3,516 observed fish were caught, brought back to the dock, and identified by samplers; 130,941 reported fish were caught, released dead, used for bait, or filleted; and 279,063 reported fish were released alive.

It is important to note these numbers do not include a percent standard error, which would represent our confidence in our point estimates. This measure of precision is similar to the margin of error used in polling, and is an important factor to consider when estimates are used in science or management.

Weighted Estimation

Weighting is a statistical method that ensures each sampled unit is properly represented in a final estimate. Weighting the data that contribute to our catch estimates allows us to account for the fact that some fishing sites are more likely to be selected as a sample location and some anglers are more likely to participate in a fishing survey.

In basic weighting, the assigned weight of a sampled unit is equal to the inverse of the probability that unit will be selected in a sample. If, for example, a sampled unit had a one in 10 chance of being selected in a sample, its assigned weight would be 10 out of one, or 10.

The probability that a particular fishing site will be sampled during a particular time of day can increase or decrease depending on how active that site is expected to be during that sampling assignment. In fact, survey sites, dates, and times are selected in proportion to the amount of fishing activity expected to occur there so we can ensure more active sites are sampled more often. This increases sampling efficiency—allowing our samplers to spend more time interviewing anglers at high-activity sites and less time idle at low-activity sites—and ensures our overall sample is representative of actual fishing activity.

Weighting Catch Estimates

Site Weight

Let’s imagine fishing sites in our sampling region can experience three different levels of fishing activity:

• Low activity, which sees about 10 fishing trips during a sampler’s assignment.
• Medium activity, which sees about 40 fishing trips during a sampler’s assignment.
• High activity, which sees about 100 fishing trips during a sampler’s assignment.

Let’s also imagine our sampling region is home to 40 low-activity sites, 20 medium-activity sites, and eight high-activity sites. The probability that each of these sites will be selected in a sample can be determined through two steps.

First, determine the expected level of fishing activity across your sampling region by multiplying the number of fishing trips expected to occur at each type of site by the number of each type of site in your region, then adding these totals together. For example:

• (10 fishing trips * 40 low-activity sites) +
• (40 fishing trips * 20 medium-activity sites) +
• (100 fishing trips * 8 high-activity sites) =
• 2,000 fishing trips.

Second, divide the expected level of fishing activity at each type of site by the expected level of fishing activity across your sampling region. For example:

• If our low-activity sites are expected to see 10 fishing trips each, the probability that one of these sites will be selected in a sample is 10 out of 2,000, or one in 200.
• If our medium-activity sites are expected to see 40 fishing trips each, the probability that one of these sites will be selected in a sample is 40 out of 2,000, or one in 50.
• If our high-activity sites are expected to see 100 fishing trips each, the probability that one of these sites will be selected in a sample is 100 out of 2,000, or one in 20.

Because the weight of our sites equals the inverse of the probability that site will be selected in a sample, our sites carry the following weights:

• Low-activity sites carry a weight of 200/1, or 200.
• Medium-activity sites carry a weight of 50/1, or 50.
• High-activity sites carry a weight of 20/1, or 20.

We call these primary stage weights.

Interview Weight

Samplers strive to interview all anglers returning from a fishing trip. However, high-activity fishing sites can be too busy for samplers to monitor the number of trips taking place and interview every angler returning from a trip. To account for this, we assign a weight for each interview based on the probability an individual angler will participate in a fishing survey.

Let’s imagine samplers were able to collect information for all 10 fishing trips at our low-activity sites. The probability that an individual angler at a low-activity site will participate in a fishing survey is 10 out of 10, or 1. In this case, the interview weight—which is equal to the inverse of this probability—is 10 divided by 10, or 1.

Now, let’s imagine samplers were able to collect information for 32 of the 40 fishing trips at our medium-activity sites. The probability that an individual angler at a medium-activity site will participate in a fishing survey can is 32 out of 40, or four out of five. In this case, the interview weight is five divided by four, or 1.25.

Finally, let’s imagine samplers were able to collect information for 40 out of the 100 fishing trips at our high-activity sites. The probability that an individual angler at a high-activity site will participate in a fishing survey is 40 out of 100, or two out of five. In this case, the interview weight is five divided by two, or 2.5.

We call these secondary stage weights.

Overall Trip Weight

The overall weight that should be assigned to each fishing trip at each type of site can be determined by multiplying the associated site weight and the associated interview weight. In our example:

• Low-activity sites carry a site weight of 200 and an interview weight of 1. Therefore, fishing trips at low-activity sites carry an overall weight of 200.
• Medium-activity sites carry a site weight of 50 and an interview weight of 1.25. Therefore, fishing trips at medium-activity sites carry an overall weight of 62.5.
• High-activity sites carry a site weight of 20 and an interview weight of 2.5. Therefore, high-activity sites carry an overall weight of 50.

Weighted Catch Estimates

Once you have determined the overall weight that should be assigned to each fishing trip at each type of site, you can calculate a weighted catch estimate for a particular species of fish at a particular type of sites and across your sampling region.

To do so, multiply the number of fish that were caught at a particular type of site by the overall weight of that site. In our example:

• If six black sea bass were caught at our low-activity sites, which carry an overall trip weight of 200, then our weighted catch estimate for black sea bass at low-activity sites is 1,200.
• If 30 black sea bass were caught at our medium-activity sites, which carry an overall trip weight of 62.5, then our weighted catch estimate for black sea bass at medium-activity sites is 1,875.
• If 34 black sea bass were caught at our high-activity sites, which carry an overall trip weight of 50, then our weighted catch estimate for black sea bass at high-activity sites is 1,700.

If we add these values together, we can determine the weighted catch estimate for black sea bass across our entire sampling region.

• 1,200 + 1,875 + 1,700 = 4,775

Weighted Catch per Fishing Trip

Once you have determined a weighted catch estimate for a particular species of fish across your sampling region, you can determine the weighted catch per fishing trip.

First, determine the sum of your weights by multiplying the number of interviews conducted at a particular type of site by the overall trip weight of that particular site. In our example:

• (10 * 200) + (32 * 62.5) + (40 * 50) = 24,000

Next, divide the weighted catch estimate for a particular species of fish across your sampling region by the sum of your weights. In our example:

• 4,775 / 24,000 = 0.199

In other words: on average, 0.199 sea bass were caught on each fishing trip in your sampling region.

More Information

• Recreational Fishing Survey Design and Statistical Methods Manual