5. Interpret
5.2 Statistical assessment
Overview Introduction Statistical Assessment Mechanics of Interpretation Finding Meaning in the Data Results and Next Steps
Typically, the inference design will specify the calculation of population means, sample variances, and distributional metrics (e.g., the cumulative distribution function, run-timing curve, or order statistics such as the 25th and 75th percentiles). In intensive fisheries (those with high harvest rates and active management), managers will often need timely, high-resolution escapement measurements in order to make corrections to the fishery's intensity so as to achieve the management objectives. However, these timely estimates will not be the final official statistics that will become a part of the history of the fish stock. In such cases, the basic measurements might be hourly escapement counts. For in-season management, hourly counts (measurements) might be summed produce daily or weekly escapement counts or perhaps the cumulative escapement up to a date (metrics). Then at the end of the season these daily counts can be further summarized and also tracked through time. For a single year, the collection of summary statistics (means, totals, confidence intervals, and so on) and other associated numerical products are what we call the indicators. To continue with the example of a system with daily escapement counts, the final statistical summary can be used to judge whether the season will be within the desired escapement goal range. The consistent feature of this assessment is that it involves calculation without judgment or conclusion.
By the time an analyst sits down to think about what the study really means, the study has already been designed, the data have already been collected and summarized, and most importantly, the question that the study was launched to answer has already been asked. At this stage the analyst needs to take what information is actually available and use that information to make an inference. But first, it might be helpful to think again a little bit about the potential strength of inference that can be drawn from the data that was just collected -- although this strength of inference should have been first considered in the planning stages of the study. Actually, the analyst may have to go back to this strength-of-inference topic several times through the course of the study. Once the data have been summarized and once the analyst has a clear idea of how strong an inference is reasonable for the particular study design, next comes the hard part -- interpretation.
