Академический Документы
Профессиональный Документы
Культура Документы
Many research questions and questions for evidence-based practice (EBP) are
about causes and effects
Although it might be easy to grasp what researchers mean when they talk about a
cause, what exactly is an effect? A good way to grasp the meaning of an effect is
by conceptualizing a counterfactual (Shadish et al., 2002)
A counterfactual is what would have happened to the same people exposed to a
causal factor if they simultaneously were not exposed to the causal factor
CRITERIA FOR ESTABLISHING CAUSE-AND-
EFFECT RELATIONSHIPS
The first criterion is temporal: a cause must precede an effect in time
Second, there must be an empirical relationship between the presumed cause
and the presumed effect
The third criterion for inferring a causal relationship is that the relationship
cannot be explained as being caused by a third variable
Coherence involves similar evidence from multiple sources, and the criterion of
consistency involves having similar levels of statistical relationship in several
studies
Biologic plausibility, that is evidence from basic physiologic studies that a causal
pathway is credible
EXPERIMENTAL, QUASI-EXPERIMENTAL,
AND NONEXPERIMENTAL DESIGNS
EXPERIMENTAL DESIGN
Experimental designs are considered the “gold standard” for intervention studies
because they yield the highest quality evidence regarding the effects of an
intervention
Experiments offer greater corroboration than any other research approach that, if
the independent variable is manipulated, then certain consequences in the
dependent variable may be expected to ensue
Through the controls imposed by manipulation, comparison, and—especially—
randomization, alternative explanations to a causal interpretation can often be
ruled out or discredited
DISADVANTAGES OF EXPERIMENTS
2. Time–Series Design
Involves collecting data over an extended time period, and introducing the
treatment during that period.
Although the time–series design does not eliminate all the problems of
interpreting changes in turnover rate, the extended time perspective
strengthens the ability to attribute change to the intervention.
The time–series design rules out the possibility that changes in resignations
represent a random fluctuation of turnover measured at only two points
QUASI-EXPERIMENTS
Advantages Disadvantages
There are various reasons for doing a non-experimental study, including situations
in which the independent variable inherently cannot be manipulated or in which it
would be unethical to manipulate the independent variable.
There are also research questions for which an experimental design is not
appropriate, such as studies whose purpose is description
CORRELATIONAL DESIGNS
Advantages Disadvantages
Involve the collection of data at one point in time (or multiple times in a short time
period, such as 2 hours and 4 hours postoperatively)
Crosssectional designs are especially appropriate for describing the status of
phenomena or relationships among phenomena at a fixed point
Retrospective studies are usually crosssectional: data on the independent and
dependent variables are collected concurrently but the independent variable
usually captures events or behaviors occurring in the past.
Sometimes used to study time-related phenomena, but the designs are less
persuasive than longitudinal ones.
The main advantage of cross-sectional designs is that they are economical and
easy to manage
LONGITUDINAL DESIGNS
Researchers who collect data at more than one point in time over an extended
period
Useful for studying changes over time and for ascertaining the temporal
sequencing of phenomena, which is an essential criterion for establishing
causality
Longitudinal studies involve collecting data from different people in a population
to examine trends over time
In a more typical longitudinal study, the same people provide data at two or more
points in time.
LONGITUDINAL DESIGNS
1. What is the strength of the evidence that a relationship exists between two
variables?
2. If a relationship exists, what is the strength of the evidence that the
independent variable of interest (e.g., an intervention), rather than other
factors, caused the outcome?
3. What is the strength of evidence that observed relationships are generalizable
across people, settings, and time?
4. What are the theoretical constructs underlying the related variables and are
those constructs adequately captured?
FOUR ASPECTS OF A STUDY’S VALIDITY
Statistical tests are used to support inferences about whether or not such a
relationship exists
Design decisions can influence whether statistical tests will actually detect true
relationships, and so researchers need to make decisions that protect against
reaching false statistical conclusions
INTERNAL VALIDITY
Internal validity refers to the extent to which it is possible to make an inference that the
independent variable is truly causing or influencing the dependent variable
Experiments tend to have a high degree of internal validity because randomization to
different groups enables researchers to rule out competing explanations
Threats to Internal Validity
- Temporal Ambiguity
- Selection
- History
- Maturation
- Mortality/Attrition
EXTERNAL VALIDITY
Construct validity involves inferences from the particulars of the study to the
higher-order constructs they are intended to represent
Construct validity is important because constructs are the means for linking the
operations used in a study to a relevant conceptualization and to mechanisms for
translating the resulting evidence into practice.
CRITIQUING QUANTITATIVE RESEARCH
DESIGNS
If the research purpose is descriptive or exploratory, an experimental design is not
appropriate.
If the researcher is searching to understand the full nature of a phenomenon
about which little is known, a highly structured design that allows little flexibility
might block insights
Research control as a mechanism for reducing bias, but in certain situations, too
much control can introduce bias