Northern Prairie Wildlife Research Center

What is statistics? One good definition (Savage 1977) is that statistics consists of trying to understand data and to obtain more understandable data. Research biologists obviously use statistics during data analysis, but actually they use statistics, whether they realize it or not, during several other stages of the research process. The key to a good data analysis is having "good" data to analyze. This is crucial; good data can always be reanalyzed, but even the best analysis of poor data will usually be unsatisfactory. A requisite for obtaining good data is a clearly defined research objective. By thinking statistically, biologists are able to formulate a broadly stated research problem in terms of explicit, addressable questions. This in turn leads to considering the population under study, identifying appropriate sampling or experimental units, defining relevant variables, and determining how those variables will be measured. Such critical thinking is needed to design an effective study and to determine sample size requirements. Thought should also be given to how the data will be analyzed, what complications might arise during the data collection stage, and what can be done to handle them.

An appreciation of statistical issues is valuable when reviewing the literature on the topic one intends to address in a new study. Which of the extant studies are solid and provide a useful basis for further work? Which are based on weak data, inappropriate analyses, or unsubstantiated interpretations, and therefore should be given little credence? These considerations are equally important to a wildlife manager who is reviewing research findings and judging their relevance to particular management situations.

Statistical thinking is also important during the data collection phase of a research or management study. A statistically minded biologist will be better able to detect unanticipated problems with the study design than will a biologist who is statistically unaware. This can be as simple as recognizing an important but unaccounted-for source of variation and deciding to record an additional variable that can be used as a covariate during data analysis. Or, the biologist may realize early on that 2 or more experimental units are not responding independently of one another and take necessary steps either to replace one of the units or at least to account for the lack of independence during data analysis. An appreciation of statistics will help the investigator recognize when sample sizes are insufficient to achieve objectives. In the most extreme situation, this can lead to the early termination of a study that is destined to fail and a savings of resources that would otherwise have been wasted.

Knowledge of statistics is obviously helpful during the data analysis and
manuscript preparation stages. Knowing which analyses to perform, and why,
along with understanding and assessing the assumptions underlying those analyses
and being able to interpret the results are of critical importance. In addition
to accurately describing the statistical methods in the manuscript, biologists
need to know such things as if a *P*-value is appropriate or not and
when to report a standard deviation as opposed to a standard error. Managers
will call upon their statistical understanding to help them interpret results
of research studies or management evaluations and to determine how much confidence
they should place in those results.

The final juncture at which statistical knowledge can be useful is when considering comments from reviewers, whether they be referees of manuscripts or teams reviewing management strategies. Some comments by reviewers, including those that pertain to statistical analyses, are of enormous value; others may have little or no utility. No one knows the data better than the person who collected, analyzed, and interpreted them. Being able to judge the appropriateness of reviewer comments and recommendations is essential for choosing an appropriate analysis and developing a high-quality manuscript or resource management plan.

Previous Section -- Introduction

Return to Contents

Next Section -- Recommended quantitative education for wildlife students