Exam 2 Quantitative Designs Flashcards
What is quantitative design?
All dealing with numbers. basic or fancy. all look at numbers somehow
What are the major categories of quantitative design when looking at factors such as control, manipulation, bias?
Experimental, quasi-experimental, nonexperimental
What are the major categories of quantitative designs when looking at time as the factor?
Cross-sectional, longitudinal, retrospective
When can we use a quant. design?
- To describe phenomena
- To explain relationships and differences
- To predict relationships
- To determine causality
Experimental designs
Highest level of evidence for individual studies
- make comparisons of before and after
- make comparisons between groups
Experimental designs
MUST Demonstrate
Randomization
Control
Manipulation of variables
Experimental designs
Notations used
R(randomization), O (observation), X(treatment or intervention)
Randomized Control Trial (RCT)
A clinically focused study of any experimental design
Randomized Control Trial:
- Involve a large number of subjects
- Strict inclusion guidelines
- Random assignment to treatment and control groups
- comparable on key characteristics
- Treatment (I.V.) given to intervention group
- D.V. measured for both groups
Types of experimental designs
Basic
2 groups pretest/post-test
2 groups pretest/post-test
- Tx group : R O1 X O2
-Ctl group : R O1 O2
Can examine between and within subjects (groups)
Types of experimental designs
Basic
2 group post-test only
- Tx group : R X O1
- Ctl group : R O2
Types of experimental designs
Basic
Soloman 4 group
1st Tx and Ctl group: R O1 X O2
2nd Tx and Ctl group: R X O2
Types of experimental designs
Basic
Multiple group
Multiple Tx groups and 1 Ctl group
Types of experimental designs
Factorial
Usually two or more treatments to individuals in the treatment group
- Usually between subjects/groups design
- Examines main and interaction effects
Types of experimental designs
Crossover designs
Subjects serve as on control
I.Vs are manipulated
I.Vs are randomized in terms of order administered
O1 O2 R X1 O3 O4 (-X) O5 O6 X2 O7 O8
(Example of longitudinal design)
Quasi-experimental designs
- Can manipulate the I.V.
- May lack randomization
- May lack a control group
- Less able to make claim of probable cause and effect
Quasi-experimental designs:
Nonequivalent control group pretest-posttest
- COMPARISON group is not selected by random means
- Individuals are self selected into the groups (comparison rather than control group)
- Causes a problem with analysis because of the selection bias.
- Tx group : O1 X O2
- Ctl group: O1 O2
Quasi-experimental designs:
Time series
- Can manipulate the I.V.
- No randomization
- No control group
- Examine one group over time
O1 O2 O3 X O4 O5 O6
Also, Longitudinal design
Quasi-experimental designs:
Preexperimental designs
Can manipulate I.V. but no control and no randomization
Quasi-experimental designs:
Preexperimental designs
One group pretest-postest
O1 X O2
Quasi-experimental designs:
Preexperimental designs
Nonequivalent groups posttest only
- Tx group : X O2
- Ctl group: O2
Non-experimental designs
- Tend to focus on personal beliefs, experiences, etc
- Cannot manipulate the I.V., cannot have a control group, cannot randomize
Non-experimental designs are used to:
- Describe phenomena
- Explain relationships/differences among variables
- Predict relationships/differences among variables
Non-experimental Descriptive Designs
- To provide a picture of situations as the naturally happen.
- To develop theory, identify problems with current practice, justify current practice, make judgments, or determine what other in similar situations are doing.
- There is no I.V. or D.V., just research variables
Non-experimental Correlational Designs
- May describe, predict, or detect relationships.
- Examine relationships among variables (how do the variables covary? covariance is NOT causation.
Non-experimental Correlational Designs
Statistical correlation shows:
- Strength of relationship
- Direction of relationship
- Significance of relationship
Non-experimental Correlational Designs
Descriptive Correlational
Examines the degree of relationships between variables as described (correlation coefficients)
-Test nondirectional hypotheses
Non-experimental Correlational Designs
Predictive correlational
- Examines the degree of relationships between variables as described
- Tests a directional hypothesis
- Used to determine the amount of variance (influence) predictor variables have on an outcome variable (multiple regression)
Non-experimental Correlational Designs
Model testing correlational designs
Causal modeling and path analysis
- Many variables involved as predictors
- Examines how much impact each on the outcome variable or on other predictor variables
Ethics
Appropriate informed consent
Design must maximize benefit and minimize risk (unforeseen events may require the study to be halted)
Manipulation of the I.V. may not be ethical
Self selection into groups may be most ethical approach
Control groups must receive standard care
Those who drop out of study (treatment group) must still receive standard care
What is the difference between an experimental study and a quasiexperimental study?
Randomization
What is the difference between an experimental study and a non-experimental study?
Experimental study has control group, non-experiemental does not.
Research variables instead of true DV/IV
Don’t manipulate the variables in non-experimental, no randomization in non-experimental
What do the following symbols indicate?
R
X
O
R-Randomization
X-Treatment
O-Observation
Tx group: R O1 X O2
Ctl Group: R O1 O2
Experimental pretest/postest
1st tx and ctl group: R O1 X O2
2nd tx and ctl group: R X O2
Soloman four group
Tx group: O1 X O2
ctl group: O1 O2
Quasi-experimental
Nonequivalent control pretest/posttest
Why are non-experimental studies useful?
Shows relationships or differences between groups
What is the difference between a non-experimental descriptive and a non-experimental correlational design
Descriptive…means, modes, frequencies, etc
Correlational…looks at correlations…pearson r
