Capture-Recapture Flashcards
When is capture-recapture most useful?
when it is not practical to count all individuals; hard to reach populations
What is the capture-recapture method?
1) A sample of 𝑛1individuals is obtained from the population.
2) Each of these individuals is marked.
3) The individuals are returned to the population.
4) Later, a second sample is taken from the population (i.e. 𝑛2 individuals), and we see which of these are marked from the first sample.
What are the capture-recapture assumptions?
1) The population is closed, i.e. constant
2) Cases sampled on both occasions can be identified and matched.
3) Each case has an equal chance of being included in each sample
4) Independence between the capture and recapture processes, i.e. cases captured as part of sample 1 should have the same probability to show up in sample 2 as those not captured in sample 1.
5) most studies will violate assumptions to some degree, but its effect can be quantified.
How is the contingency table laid out for capture recapture?
> yes yes in top left
a, b, c, x row wise
How do you calculate x in capture recapture?
x = bc / a
How do you calculate N in capture recapture (Peterson estimator)?
n1n2 / a
> where n1 is total in first sample
n2 is total in second sample
overestimated total number of cases when sample sizes are small
How do you calculate N in capture recapture (Chapman estimator)?
𝑵=(𝒏𝟏+𝟏)(𝒏𝟐+𝟏)/(𝒂+𝟏) −𝟏
> N should be rounded to nearest whole number
What is sensitivity?
the conditional probability of the case-finding method actually finding an individual, given that the individual is present in the sample (true positive)
How do you calculate sensitivity in capture-recapture?
Sensitivity of method 1 = 𝒏𝟏/𝑵.
Sensitivity of method 2 = 𝒏𝟐/𝑵.
> sensitivities between 0.7 and 1 are reasonable
How do you calculate confidence interval for capture sensitivities?
𝑝̂±1.96√((𝑝̂ ( 1−𝑝̂ ))/𝑁)
> where 𝑝̂ is the estimated sensitivity and N is the estimated total population size
What happens if population is not closed in capture recapture?
a)probability of recapture is reduced, N will be overestimated, sensitivity underestimated
What happens if cases cannot be reliably identified and matched in capture recapture?
1) true matches may be missed = overestimation of N, underestimation of sensitivity
2) false matches may be created = underestimation of N, overestimation of sensitivity.
What happens if all individuals do not have equal probabilities of capture?
N is underestimated, sensitivity overestimated
how can assumption of independence between capture probabilities be achieved?
a) the use of different site locations in the same area where study is taking place
b) use of separate teams of individuals to obtain samples
