An algorithm to correct the survivorship bias in capture-recapture models: Problems and implications

Date: Wednesday 31 January 2024, 2.00PM
Location: Online
Local Group Meeting

Share this event

The RSS Manchester local group will host an online seminar with Dr Blanca Sarzo (University of Valencia) who will speak about correcting survivorship bias in capture-recapture models.

“Survivorship bias” arises when conclusions are drawn conditional on only the surviving individuals, whilst failing to correct for those individuals who have not survived. This issue has been well studied in many fields such as economy, construction, forestry, health, etc. (Czeisler et al., 2021; Cooke et al., 2003) but less well explored within the context of capture-mark-recapture studies (CMR). The survivorship bias can be immediately observed in individual heterogeneity models that are commonly fitted to capture-recapture data. This bias is manifested within these studies in that weaker individuals are more likely to die at younger ages compared to stronger individuals who may survive longer within the study period. In other words, the weaker individuals are less likely to be observed within the study period, compared to the stronger individuals, potentially leading to biased estimates, dependent on the sampling strategy.

To illustrate this, we present a simulation study where an algorithm to correct the survivorship bias is implemented for two different individual heterogeneity CMR models: (i) Mah model (individual heterogeneity and age effects on survival probability), and (ii) Math (individual heterogeneity, age and time effects on survival probability), considering that individuals may have different ages when they are first marked. We highlight the necessity of careful interpretation of the model parameters as well as the problem of survivorship bias when individual heterogeneity is included within this context.

Cooke, B., W. Miller, and J. Roland. (2003). Survivorship bias in tree-ring reconstructions of forest tent caterpillar outbreaks using trembling aspen. TreeRing Research 59, 29–36.
Czeisler, M. E., J. F. Wiley, C. A. Czeisler, S. M. Rajaratnam, and M. E. Howard. (2021). Uncovering survivorship bias in longitudinal mental health surveys during the covid-19 pandemic. Epidemiology and Psychiatric Sciences 30.

Dr Blanca Sarzo has a background in Ecology and a MSc and PhD in Statistics from the University of Valencia (department of Statistics and OR). In 2021 she was granted with a postdoctoral Fellowship from the University of Valencia which was performed at University of Edinburgh with Professor Ruth King. At present she is working as postdoctoral fellow in the Foundation for the Promotion of Health and Biomedical Research at Valencia Region. 

Contact Antonia Marsden