Essential statistical knowledge for nursing research

Statistical knowledge is essential for nursing research as it allows researchers to analyze and interpret data in a meaningful way. One of the fundamental concepts in statistical analysis is descriptive statistics, which provides a summary of the main characteristics of a dataset. Descriptive statistics include measures such as mean, median, mode, and standard deviation, which can be used to describe the central tendency and variability of a dataset. Understanding these measures is important for summarizing the data in a clear and concise manner.

---

Excerpts from:

Berman, N. & Gullíon, C. (2007) Working with a Statistician. In Walter T. Ambrosius (Ed.) Topics in biostatistics  pp. 489-503. Humana Press Inc.

“In some projects, you may not need to work with a statistician. For instance, if you have done a simple study, with 1 design factor and 1 outcome measure, and your hypothesis and data clearly match the assumptions and purpose of a statistical test that you know how to do and to interpret, you may be able to complete and publish your research without the aid of a statistician.

Most studies and data are not so simple. Methodological considerations in choice of design, measure, and analytic methods can rapidly exceed the scope of a fundamental text such as this. This does not mean that this text is not useful—a foundation in statistical thinking and methods is invaluable for working with a statistician and understanding the rationale for methodological choices. Such a foundation also provides a common language, which can greatly facilitate communication.

A variety of situations may lead you to consult a statistician. If your data do not exactly match the assumptions outlined for a statistical test, you may need to identify an alternative analytic approach. A statistician will know about a great many more methods than can be covered by a book such as this, such as more complex or specialized statistical procedures. Alternatively, the best strategy might be to transform your data—but how do you select the optimal transformation? What are the considerations in choosing to transform versus using an alternative statistical test? A statistician can help you answer these questions.

In addition to dealing with these issues in planning an analysis, a statistician can contribute to other aspects of a research project. A statistician can help you relate your research goals to the methods in this text and to understand what results can be expected. She may also be able to help you learn to use and understand a statistical software package, if you would like to do your own analysis. If you work with a statistician from the beginning of a project, as suggested below, she can help with the design of your study and offer ways to facilitate the implementation, including data collection, and final analysis.

The best time to seek advice from a statistician is early in the planning of a project that will involve data collection and analysis. Statisticians are trained and experienced in developing study designs and can help develop a plan that will answer the research question while also making optimal use of your resources. A statistician can help translate your study goals into testable hypotheses and can define analytic methods that will successfully test them. Using power analysis, (they) can determine the number of subjects or analytic units needed to obtain a definitive answer to your question. (They) can also advise you on various methods for randomization and can create an appropriate randomization schedule for your study.

If you do not involve a statistician in study design and implementation, you may design a futile study, that is, one in which the data cannot answer your research question, even in the hands of a skilled statistician. For instance, you might plan to enroll too few subjects to be able to answer your research question definitively. A competent power analysis will tell you if the sample size is sufficient and may lead you to rethink the whole study.

It is important to realize that input from a statistician can add value at every stage of a study, particularly if the optimal methods require more expertise than you have to implement and interpret properly. The objective of study design is to use no more time and resources than necessary to obtain a valid, useful answer to your hypothesis question.

A statistician who is a collaborator typically is a full partner in your research. (They) will become familiar with many aspects of your field, may read extensively in the literature of your field, will expect to work with you on a study from beginning to end, and will maintain an interest in your research even when (they are) not responsible for a specific task. If you sought NIH funding for your project, this person would be listed in the Key Personnel as a co-investigator.

Alternatively, a statistician may limit (their) involvement to serving as a consultant on your study. A consultant typically is someone who is more distant from the project, who is responsible for a limited scope of work. (They) will answer questions when asked but is not involved in the study when not performing clearly specified tasks. This does not mean that the consultant does not care about the success and integrity of a study, but that the time and level of detail to which (they) attend is limited.“