¶ … statistics have on shaping healthcare policy and guiding evidence-based practice, it is critical that researchers understand how to present the results of their studies. It is also critical that healthcare workers develop strong skills in statistical literacy, so that the results of studies are not misconstrued. Not all research results are generalizable to a population outside of the sample. Even the most carefully constructed research designs need to be critically analyzed. Similarly, care must be taken when communicating statistical results to a general audience. The American Statistical Association (1999) outlines eight main areas of ethical concern. Those areas of concern include the following:

• Professionalism

• Responsibilities to employers or funders

• Responsibilities in testimony or publications

• Responsibilities to research subjects

• Responsibilities to research team colleagues

• Responsibilities to other statisticians

• Responsibilities regarding allegations of misconduct

• Responsibilities of employers or clients to the integrity of research

In the healthcare setting, each of these ethical duties is relevant, but it is the latter that may be most relevant to daily work for practitioners who read, gather, disseminate, discuss, and interpret research findings and often implement those findings into evidence-based practice. It is therefore critical that statisticians be aware of the impact their work has on public health.

Statisticians are supportive of creating the "evidence-based society" and evidence-based organizational culture in healthcare (p. 5). However, statisticians are also in the unique position of having to offer warnings to healthcare practitioners, administrators, and pharmacists eager to deliver new products and services to the patient population. Statisticians deal with uncertainties and probabilities, whereas non-statisticians, even within the medical science arena, seek clear-cut, black-and-white answers.

When it comes to actual research methods and design, the role of statisticians is more immediately apparent. For example, statisticians have an ethical responsibility to honestly and objectively interpret raw data, regardless of the substantive content of a research hypothesis. A statistician with access to participant personal information has an ethical responsibility to preserve and safeguard privacy and confidentiality. The issues all researchers face when conducting experiments including informed consent, remain salient. Ascribing to statistical ethics...

Statisticians need to disclose the precise methods of data analysis used, including the software they use, and what data that needed to be excluded from the study and why. Research design should reflect the best interests of providing accurate results, not necessarily according to the desires and preferences of funders. Specific types of research designs and methodologies present unique ethical considerations. For example, cluster randomized trials have become increasingly common in healthcare and particularly in public health research. Cluster randomized trials help determine clustered effects of an intervention such as a vaccine.
The ethic of responsibility to research participants, as outlined by the ASA (1999) includes issues related to informed consent. Informed consent is different in cluster randomized trials versus other research designs. There are two basic types of cluster randomized trials: cluster-cluster, where the entire cluster is being measured, and individual cluster trials, where the individual participants are being measured. The type of design will vary depending on the research question and hypotheses. Informed consent protocol may differ depending on which of these cluster randomized trial methods is used. Of course, this simplistic classification ignores the plethora of other types of cluster randomized trials but general rules of thumb for informed consent create common sense guidelines for statisticians.

Regardless of type of cluster randomized trial, the primary areas of ethical concern include units of randomization, units of intervention, and units of observation ("Medical Ethics and Statistics," n.d.). Units of randomization imply different types or levels of consent. Individual consent alone may not be enough. Consent might need to be gleaned from community organizations and other group stakeholders like families or professional organizations. Care must be taken when presuming a leader of any organization speaks on behalf of the entire group or community, as this is not always the case. However,…