Sometimes, researchers divide the items on a questionnaire in half to make
two versions, a technique known as split-half reliability. Researchers use the
Spearman-Brown formula to compare the two halves. Split-half testing is used
to establish internal consistency. Item to total calculations are also used to test
for internal consistency. Each item on the instrument is compared to the total
score obtained. Strong items have high correlations with the total score. Items
with low correlations to the total score are examined, and decisions are made
to change or eliminate them. When the level of measurement is dichotomous,
researchers determine internal consistency using the Kuder-Richardson coef-
ficient (K-20). This correlation compares all items at the same time. Cronbach’s
alpha is the most common method used by nursing researchers to assess internal
consistency. This method can be used when data are interval level or higher.
Like the Kuder-Richardson, all items are compared simultaneously to obtain a
single correlation (Trochim, 2000). Computer software is required to perform
these sophisticated tests.As nurses appraise the methods section of any research article, it is important
to consider the issues of validity and reliability (Roberts & Priest, 2006) before
making practice decisions. The report should contain information about the
validity and reliability of each instrument used. References for the original
development of the instruments should be provided. Occasionally researchers
include instruments in articles that allow readers to observe how the questions
align with the concepts that were being measured.Appraising Data Collection in
Quantitative Studies
When reading the methods section of any research article, nurses should
determine that each instrument is described and the reliability and validity
are reported. The level of measurement should be noted for each variable
measured. Nurses should also appraise whether instruments represent
the concepts and variables being operationalized. Interventions should be
detailed so that nurses know what was done in the research. If a pilot was
done prior to the study, its instruments, interventions, and findings should
be presented. If subjects are grouped, the author should describe how group-
ing was accomplished.Many quantitative studies fall short of significant findings if there are holes
in the methods section. If the instruments do not adequately measure the
concepts or variables, or if they lack validity and reliability, the study may be
flawed (Ioannidis, 2005). Box 10-1 lists questions that nurses should consider
when appraising the methods section.KEY TERMS
interrater reliability:
A test for instrument
reliability when
two observers
measure the same
event. Scores are
correlated, and
strong positive
correlations indicate
good reliability
split-half reliability:
A test for instru-
ment reliability in
which the items
are divided to form
two instruments.
Both instruments
are given and the
halves are com-
pared using the
Spearman-Brown
formula
item to total
correlation: A test
for instrument
reliability in which
each item is
correlated to the
total score; reliable
items have strong
correlations with
the total score
Kuder-Richardson
coefficient: A test
for instrument
reliability for use
with dichotomous
items; all items are
simultaneously
compared using a
computer
Cronbach’s alpha:
A test for instrument
reliability used with
interval or ratio
items; all items are
simultaneously
compared using a
computer270 CHAPTER 10 Collecting Evidence