Page 14 - The Empowered Nurse - Spring

empower_spring2020.qxp_Layout 1 6/23/20 11:52 AM Page 14

12
THE EMPOWERED NURSE

5

THE MIND

This column focuses on creating a better understanding of
interpreting statistical results and applications.

Beginning with the End in Mind tool been used in the population we want to study? Have the au-
thors identified particular benefits and/or limitations of its use?
When planning a quality improvement or evidence-based • What type of data will result from using this instrument?
practice project, or a research study, we consider a “problem” or Categorical: variables that typically don’t generate a
a clinical question that we specifically want to study. Once we quantity, for example, yes and no
have a good handle on what question or problem has piqued our Continuous: variables that may be represented across a
curiosity or raised our concern, we must first search existing lit- continuum
erature about the topic. What has been studied already? How
has the problem been solved elsewhere? What pitfalls and limita- The preference is always for instruments that will generate
tions have arisen from that work? Once these questions are con- a continuous score. We can generate categories from continuous
sidered, we determine where our “problem” or “question” fits data that increases our analysis options. A very simple example
into that prior work. This background work helps us to formu- of this idea is recording the number of clinic visits in the past
late a clear project or research question and can help us decide five years rather than simply recording whether or not (yes or
whether we can use existing evidence to solve the problem or no) a clinic visit occurred in the past five years. We can answer
whether we must generate new evidence. Becoming familiar the question whether or not there was a clinic visit by collecting
with the existing literature will also help us to choose an appro- the number of visits and we can glean so much more from the
priate theory or conceptual framework for our study or project continuous data. This idea works in the same way for psychoso-
and what the best design may be. For example, if very little cial measures (e.g., burnout, satisfaction, knowledge).
knowledge has been accumulated in the area of interest, we • Is there evidence for the validity of the instrument for the
probably do not want to immediately jump to a longitudinal in- purpose we will be using it for? Have there been a number of
tervention type of study. We may need to glean more qualitative prior reports of the reliability of the scores generated from the
or quantitative data about our phenomenon to better guide such instrument? If we are planning a longitudinal study, is there evi-
an intervention. In that case, we may decide to conduct a cross- dence that the scores resulting from the instrument are sensitive
sectional focus group and small quantitative study first. In the to change? These questions can be answered through systematic
case of quality improvement and evidence-based practice proj- literature review or by contacting the tool’s author.
ects, the best approach may be a small test of change using Plan- • Is the instrument copyrighted? Do we need to contact the
Do-Study-Act (PDSA) methods, or a pilot EBP project on one author for permission to use it? Do we need to pay for the use of
unit or in one shift. This article will focus on quantitative ap- the instrument?
proaches for research projects. If there is no existing instrument that meets our needs, then
we may need to begin the process of developing one. It is help-
Selecting a Study Instrument ful to consult with a specialist in measurement to guide us in this
The specific study or research question, our literature re- process. We need to ensure that we have created an item pool
view, and the conceptual model or framework that we select that will cover the conceptual area we are interested in. We also
guide us to the variables that must be measured. Optimally, the want to structure our questions appropriately and select a re-
literature will lead us to an existing instrument or survey that fits sponse format (categorical — e.g., 0=‘no’, 1=‘yes’; Likert-type
our needs. Questions we consider when selecting an instrument — e.g., 1=‘Not at all’ to 5=‘All the time’, etc.) that will give us
for data collection include: the data we need to answer your research question. To read
• Does it contain the variable(s) we want to investigate more about level of measurement, please see the article, “Why
(e.g., negative mood, patient satisfaction, burnout)? do we use statistics?” in issue 1 of the Empowered Nurse.
• How much has it been used in previous studies? Has the