Detection and Correction of Pseudoreplication in Recreational Fishing Surveys

Pseudoreplication is an error of statistical analysis and interpretation. Hurlbert (1984) defined pseudoreplication as “the use of inferential statistics to test for treatment effects with data from experiments where either treatments are not replicated (though samples may be) or replicates are not statistically independent”. Four types of pseudoreplication were identified: simple, temporal, sacrificial and implicit. These definitions are useful for addressing issues in manipulative experiments but are inadequate for identifying similar statistical problems in the field of probability-based sample surveys. In order to identify pseudoreplication when it occurs in probability-based sample surveys (e.g. recreational fishing surveys) I propose the following definition. Pseudoreplication is the failure to use the structure of a probability-based survey design in a statistical analysis or interpretation of the survey information. These statistical and interpretation errors occur in three main ways by:

(1) Inappropriate pooling of Primary Sampling Units (PSU’s);

(2) Inappropriate disaggregation of PSU’s; and

(3) A lack of statistical independence among PSU’s.

I will provide examples of these different types of pseudoreplication with respect to probability-based recreational fishing surveys and discuss the consequences for the estimation of fishing effort and catch. Some suggestions will be provided on potential ways to correct for pseudoreplication in recreational fishing surveys.