This glossary collects the core terms that appear throughout tidycreel’s functions, vignettes, and printed outputs. It is meant to be a quick reference: plain-language definitions first, then pointers to where each concept appears in the package workflow.
A creel survey is a survey of anglers, angling effort, and fish catch. In practice this means some combination of calendars, count observations, interviews, and species-level catch or length data collected over a defined season.
A design is the survey structure that tells
tidycreel what was sampled, when it was sampled, and how observations
should be grouped for estimation. In this package, the design starts
with creel_design() and is then enriched with additional
data layers such as counts, interviews, catch, or lengths.
The calendar is the table of dates and strata used to define the survey frame. It answers the question: what days were in scope for the survey?
A stratum is a grouping variable used to divide the
survey into more homogeneous sampling units — for example
weekday versus weekend.
Strata help estimation and planning by allowing
different parts of the survey to have different effort levels,
variances, or sample sizes.
A PSU is the primary unit that is sampled for
design-based inference. In tidycreel count designs the PSU is the
day: expanded effort counts the days in each stratum on
the calendar, so a PSU must identify a day. Sites or effort types
counted within a day are units inside the PSU (unit_cols in
add_counts()), summed into the day’s total. Bus-route
designs carry their own site frame with site inclusion probabilities.
Variance estimators depend on having enough PSUs per stratum.
The survey type identifies the data-collection
design. tidycreel currently supports types such as
instantaneous, bus_route, ice,
camera, and aerial. The survey type determines
which estimation path is used under the hood.
Count data are effort-related observations attached
with add_counts(). For instantaneous surveys each row is
the number of anglers seen at a randomly chosen moment. A count becomes
a day’s effort only when it is multiplied by the length of the period
the moment was drawn from, so give that period with
period_length_col; the effort is then in angler-hours.
Without it the estimate is a sum of counts, its unit is NA,
and a warning says so. Effort already totalled per day (like the bundled
example_counts) goes through
prep_counts_daily_effort() instead
(vignette("unit-rules")).
Interview data are party-level observations attached
with add_interviews(). These records typically include
catch, effort, trip status, and optional metadata such as angler type,
method, or species sought.
Catch data are species-level rows attached with
add_catch(). They expand interview totals into a long
format so species-specific rates and totals can be estimated.
Length data are fish-size observations attached with
add_lengths(). These may be stored as individual fish
lengths or as pre-binned release lengths, depending on the field
workflow.
A section is a spatial subdivision of the fishery,
registered with add_sections(). Section-aware estimators
let you estimate effort or catch for each part of a lake or river
separately.
Effort is the amount of fishing activity. In tidycreel this is often measured in angler-hours. Effort can be observed directly in count data, reported in interviews, or estimated over an entire season.
Angler-hours are hours fished multiplied by the number of anglers. This is a standard effort unit for creel work because it combines trip duration and party size into one comparable measure.
Party-hours are hours fished by a party, whatever
its size. An interview records a party, so without
n_anglers its effort is in party-hours, and a rate from it
is per party-hour. A party-hour rate times angler-hour effort is not a
catch; tidycreel warns when the two meet.
vignette("unit-rules") covers how each estimate’s
unit is derived.
A catch rate is catch per unit of effort. In
tidycreel this surface is exposed through
estimate_catch_rate().
A harvest rate is harvested fish per unit of effort.
In tidycreel this is estimated with
estimate_harvest_rate().
A release rate is released fish per unit of effort.
In tidycreel this is estimated with
estimate_release_rate().
The two catch-rate estimators. The ratio of means
divides total catch by total effort across interviews, and is used for
access (complete-trip) interviews. The mean of ratios
averages each interview’s own rate, and is used for roving
(incomplete-trip) interviews, where a clerk meets anglers in proportion
to how long they fish; each party is weighted by its size when
n_anglers is given.
vignette("estimator-choice") says which runs when.
CPUE means catch per unit effort.
HPUE means harvest per unit effort.
These are common fisheries abbreviations, but tidycreel’s exported
function names use the more explicit estimate_catch_rate()
and estimate_harvest_rate().
These are totals that combine estimated effort with estimated rates,
stratum by stratum. tidycreel provides
estimate_total_catch(),
estimate_total_harvest(), and
estimate_total_release() for these products. What they
total over is set by target (next entry).
target says what an effort or total covers. The default,
"sampled_days", is the total over the days actually
sampled. "period_total" expands each stratum to all its
calendar days, which is the season total in most reports.
"stratum_total" runs the same expansion on standard count
designs; the returned effort_target records which was asked
for. Expanded targets are not yet available for bus-route or ice
designs.
A complete trip is an interview where the angler has finished fishing. These records provide a full accounting of effort and catch for the trip.
An incomplete trip is an interview conducted before the angler has finished fishing. These interviews can still be useful, but they require more care when estimating rates because the trip outcome is only partially observed.
A refusal is a sampled angler or party who declines to be interviewed. Refusals matter because high refusal rates can bias summaries and estimates if participants differ systematically from non-participants.
Species sought is the primary species an angler reports targeting. This is used in summaries such as caught-while-sought and harvested-while-sought rates.
Design-based inference means uncertainty is computed
from the sampling design rather than from a fully specified population
model. tidycreel relies on the survey package for this
work.
A weighted estimate adjusts observed data according
to the survey design so that sampled observations represent the broader
fishery correctly. In this package, estimators such as
estimate_effort() or est_length_distribution()
use the internal survey design rather than simple raw tabulations.
An unextrapolated summary describes the sample as
observed, without survey-design weighting. Examples include
summarize_by_method() and
summarize_length_freq().
An extrapolated estimate projects from the sample to
the broader survey period or population using the survey design.
Examples include estimate_effort() and
estimate_total_catch() with
target = "period_total". A rate from
estimate_catch_rate() is design-based too, but it is a
rate, not a projection.
Variance is the sampling variability of an estimate. In practical terms it measures how much the estimate would vary across repeated samples under the same design.
The standard error is the square root of the variance. It is the most common uncertainty value shown next to an estimate.
Some estimates list the parts of their standard error in
se_components. A part that is absent was
not propagated: it does not apply, or its quantity was supplied without
a standard error and is treated as fixed.
NA means it applies and is unknown, which
makes se NA too. Zero means
it is known to be zero. vignette("refusals-and-na") shows
all three. One exception: on the bootstrap path of
estimate_effort_aerial_glmm() the visibility and
angler-ratio parts are inside se but not listed, because
the bootstrap replicates cannot separate them.
A confidence interval is a range of plausible values for the quantity being estimated, given the sample and the assumed estimation method.
Relative standard error is the standard error divided by the estimate, usually expressed as a proportion. It is a common precision target in survey planning.
Sample size planning means deciding how many days,
interviews, or other sampling units are needed before the survey begins.
tidycreel provides power_creel(),
creel_n_effort(), creel_n_cpue(), and
creel_power() for this work.
Power is the probability that a study will detect a meaningful change when that change is truly present. In tidycreel this is typically used for planning a future comparison in catch rate.
A design comparison is a side-by-side comparison of
estimates or precision from alternative survey designs or alternative
variance methods. compare_designs() provides this
surface.
Hybrid interviews are on-site interviews that mix
access-point and roving interviews within one creel survey — typically
access-point interviews for boat anglers and roving interviews for bank
anglers. Access and roving are properties of the
interview, not of the count: access interviews
intercept completed trips as anglers leave, roving interviews intercept
incomplete trips while anglers fish, and the two require different
catch-rate estimators. tidycreel records this on
add_interviews() via interview_type.
A hybrid design in the sense of
as_hybrid_svydesign() is a different thing: it combines two
or more count series covering disjoint parts of a
fishery, treating each as a stratum with its own within-day sampling
fraction, while all expand to the same population of days. It estimates
a period total rather than a sampled-day total. The totals may be added
only if the frames observe disjoint sets of angler trips. The frames are
named by a column you nominate with frame_col — an
angler-type column, typically — rather than by the interview vocabulary;
counts themselves are instantaneous, progressive, bus-route, camera or
aerial.
An instantaneous count survey samples counts at selected moments and uses those observations to estimate total effort over a season.
A progressive count survey moves through a route or
circuit over time rather than taking one instantaneous snapshot.
tidycreel handles this through add_counts() plus
progressive-count logic in the effort pipeline.
A bus-route survey samples access points or circuits with known inclusion probabilities and uses Horvitz-Thompson style expansion to estimate totals.
A camera survey uses automated image or timestamp data to index or estimate fishing effort. tidycreel supports both counter-style and ingress-egress camera workflows.
An aerial survey estimates effort from counts collected during overflights. The package supports both simple aerial estimation and a GLMM-based correction path for non-random flight timing.
Use this glossary as a map to the rest of the package:
vignette("tidycreel") for the core workflowvignette("interview-estimation") for interview-based
estimatorsvignette("unextrapolated-summaries") for raw interview
summariesvignette("survey-design-toolbox") for planning and
design comparison tools?creel_design for the main entry point into the
analysis pipeline