MSN5300Week 5Faculty: Dr. K. Richards
Chapters 10, 11
Quantitative Research Designs
At the end of this class, students will be able to:
analyze relationship between research question and choice of research design;
distinguish between interventional and non-interventional research methods;
differentiate between descriptive and correlational research designs; and
describe 3 essential elements of experimental research.
Methodology
represents the major type of research used for a study
quantitative
qualitative
both
3
Quantitative Design
researcher’s choice of the best way to answer the research question
interventional versus noninterventional designs
Interventional design
– tests effect of an intentional action (intervention) on a measured result (outcome)
– experimental designs, quasi-experimental designs
Noninterventional design
– counts and measures characteristics about the phenomenon of interest and the study variables without intervention.
– descriptive designs, correlational designs
answer to research question comes from statistical analysis of numerical data
4
Quantitative Design (…con’t…)
5
Methods
details of the proposed research
include:
recruitment site(s) selected
subjects:
who
inclusion/exclusion criteria
how recruited and consented
data collection:
tools to be used
how data will be collected
how any interventions will be enacted
how data will be organized
data analysis:
how data will be analyzed
Copyright © 2017, Elsevier Inc. All rights reserved.
6
Methods
Random Selection Versus Random Assignment
random selection:
every element in the accessible population has an equal change of being selected
study results likely to be generalizable to the accessible population (good external validity)
random assignment:
subjects are assigned to one of the study groups, and every subject has the chance of being in a certain group
groups are very similar to one another (good internal validity)
7
Control Versus Comparison Groups
8
Experimental:
randomly assigned controls for the effects of extraneous variables
Quasi-experimental:
may or may not control
if they do control, groups are properly termed nonrandom control groups
if they do not control, they are comparison groups
Key Terms
causality
multiple
causality
probability
bias
measurement
9
manipulation
control
prospective versus
retrospective
validity
Causality
cause-and-effect relationship
interventional research tests causality
noninterventional research does not
correlation is not causation
10
Multiple Causality
2 or more variables combine to cause an effect
Probability
Probabilities:
likelihoods
expressed as percentages
Prediction
usually based on probabilities
offering an opinion or guess about an unknown or future event, amount, outcome, or result
11
Bias
to slant away from the true or expected, leaning to one side
potential sources:
operational definitions
measurement methods
researcher
study subjects
study sample
data
identify in advance to avoid tainting data
12
Measurement
value is assigned to variable
must apply to the concept studied
quality of raw data collected determined by:
choice of instrument
validity
reliability
13
Manipulation
move around or control
synonym for intervention
14
Answer: A
14
interventional research:
independent variable
also refers to control for the effects of potentially extraneous variables
noninterventional research:
enacted through:
determining exclusion criteria
operationalizing variables carefully
Copyright © 2017, Elsevier Inc. All rights reserved.
15
Control
Prospective Versus Retrospective
16
prospective = looking forward, “real time”
retrospective = looking backward
data collected and recorded in the past:
existent databases
previously collected research data
audit data
both retrospective and prospective data collection are appropriate for noninterventional research
External Validity
design-resultant decisions that affect the population to which research results can be generalized
due in large part to sampling strategy
random sampling enhances external validity
17
Statistical Conclusion Validity
18
correctness of the decisions that the researcher makes regarding statistical tests
selection of an adequately large sample so that true relationships among variables are revealed
use of the correct statistical tests, given the nature of the study variables
power analysis to avoid the threat of low statistical power in noninterventional research
Non-Interventional Research Designs
19
Descriptive Research
conducted in a natural setting
answers research question r/t incidence, prevalence, or frequency of occurrence of a phenomenon of interest
may be prospective or retrospective
20
Descriptive Design
describes phenomenon of interest and its component variables within one single subject group, sometimes called a cohort
21
Comparative Descriptive Design
two distinct groups are described and compared in terms of their respective variables
Longitudinal Designs
examine changes in the same subjects over time
may use correlational statistical methods, such as linear regression
if retrospective, data reflect measured increments of time in the past
22
Cross Sectional Designs
examine change over time different groups of subjects
purpose is to examine changes in a variable over time by comparing its value in several groups that are in different phases of a process
Interventional Research Designs
23
Treatment Fidelity
24
study intervention (treatment) must be chosen so that treatment fidelity can be maintained
treatment must be described fully so that research assistants or agency personnel know exactly how it is to be applied
treatment fidelity is easier when data collection occurs over a short time and number of data-collectors is few
Other Controls
Controlling measurement:
reliability and validity of all measurement tools provided in the research report
measures of dependent variables must be consistently implemented
25
Experimental Research
26
design:
pretest-posttest control group design
Pretest-Posttest Control Group Design
27
“true” experimental design
random assignment of consented subjects to either treatment or control group
measurement of dependent variables in all subjects
application of the intervention to treatment group only
post-intervention remeasurement of dependent variable again in all subjects
Quasi-Experimental Research
28
used when experimental design cannot or should not be used
lacking at least one attribute of experimental design:
researcher-controlled manipulation of the independent variable, or
traditional type of control group, or
random assignment of subjects to groups
image1.jpeg
image2.png
image3.svg
.MsftOfcThm_Accent1_Fill { fill:#4472C4; }.MsftOfcThm_Accent1_Stroke { stroke:#4472C4; }
