Marczyk, DeMatteo, Festinger CH5

General Types of Research Designs and Approaches

Experimental Designs

Experimental Design:
(a) research design in which study participants are randomly assigned to experimental and control groups
(b) provides the highest degree of control over a research study,
(c) it allows the researcher to draw causal inferences with the highest degree of confidence (due to randomization).

Randomized Two-Group Design
- Experiments composed of two groups or two levels of an IV. Primary purpose of this design is to demonstrate causality (as opposed to
merely being correlated).

Type of Two Group Designs

Randomized Two-Group Posttest Only Design: Since participants have gone through randomization, there is theoretically no need for a
pretest. This design encompasses (1) random assignment (2) intervention and control groups (3) observations following the
treatment.

Randomized Two-Group Pretest-Posttest Design: This design allows the researcher to compare the groups on several measures following
randomization to determine whether groups are truly equivalent (through the use of baseline information). The disadvantage is
that the pretest might make the participants aware of the purpose of the study and influence their posttest results.

Solomon Four-Group Design: A combination of randomized posttest only and pretest-posttest two-group designs. Allows the researcher
to examine between-group differences at baseline without the results being influenced or confounded by the pretest
administration.
Factorial Design: Enables researchers to empirically examine the effects of more than one IV, both individually and in combination, on
the DV. Allows researchers to look at the interaction effect: result of 2 or more IVs combining to produce a result different from
those produced by either IV alone.

Threats to validity

1. If a study’s control group is inadvertently exposed to the intervention or when key aspects of the intervention also exist in the
control group.

2. Contrast effect: Occurs when one of the groups is perceived by participants as better or more desirable then the other. This may affect
the behavior of the participants in the less desirable group.

3. If there are substantial differences in the implementation of the experimental and control groups.

4. Participant dropout.

5.Randomization does not always work, especially if the sample size is small. Differences may not be distributed equally.

6.Logistical difficulty: Some experimental designs have unrealistic settings.

Quasi-Experimental Designs

1. Nonequivalent Comparison-Group Designs: When using this design, the experimenter attempts to select groups that are as similar as possible.
a. Nonequivalent Groups Posttest-Only (Two or More Groups): This design is composed of nonrandomized experimental and control groups. Disadvantages: low probability that any resulting between-group differences on the DV could be attributed to the intervention.
b. Nonequivalent Groups Pretest-Posttest (Two or More Groups): This design has 2 advantages over the previous one: temporal precedence of the IV to the DV can be established (allows for better interpretation), and the use of a pretest allows the researcher to measure between-group differences before exposure to the intervention (reduces threat of selection bias).
2. Interrupted Time-Series Designs:
a. Time-Series Design: Extension of a one-group pretest-posttest design by the use of numerous pretests and posttests. Periodic measurements are made on a group prior to the presentation (interruption) of the intervention to establish a stable baseline. Results in the establishment of normal fluctuation of the DV over time. This allows the researcher to more accurately interpret the impact of the IV. Four basic variations:
b. Simple Interrupted Time-Series Design: A within-subject design in which periodic measurements are made on a single group in an effort to establish a baseline. The IV can influence the series of observations after it has been introduced by either (1) changing in the level or (2) changing in the slope. Disadvantage: little control for alternative explanations for measured change.
c. Reversal Time-Series Design (ABA Design): Basic goal of this design is to establish causality by presenting and withdrawing an intervention, or IV, one to several times while concurrently measuring change in the DV. Begins with a series of pretest to observe normal fluctuations in baseline. “Reversal” refers to the idea that causality can be inferred if changes that occur following the presentation of an intervention diminish or “reverse” when the IV is withdrawn. Not applicable to all IVs and DVs.
d. Multiple Time-Series Design: The same as nonequivalent pretest-posttest design with the exception that the DV is measured at multiple time points both before and after presentation of the IV, or longitudinally. Allows researcher to make both within-group and between-group comparisons. Although nonrandomized, it helps rule out other explanations for the observed effect.
e. Longitudinal Design: Involves taking multiple measurements of each study participant over time. Purpose is to gather normative data on growth, to plot trends, or to observe the effects of special factors.
3. Single-Subject Experimental Designs: Seek to (1) establish that changes in the DV occur following introduction of the DV (temporal precedence) and (2) identify differences between study conditions. Rely on randomization to equally distribute extraneous variables. Eliminate between-subject variables by using only one participant. Control for relevant environmental factors by establishing a stable baseline of the DV.
a. Single Subject Reversal Design: Measures behavior during 3 phases: before intervention is introduced (A), after introducing the intervention (B), and again after withdrawing the intervention (B). Goals are to determine whether there is a change in the DV following introduction of IV and to determine whether the DV reverses or returns to baseline once IV is withdrawn. To rule out maturation or practice effects, it could be extended to ABAB design. Disadvantages: not all behaviors are reversible, and withdrawal of certain interventions may be unethical.
b. Single-Subject Multiple Baseline Design: Several behaviors of a single subject are monitored simultaneously. Once stable baselines are established for all behaviors, one of the behaviors is exposed to the intervention. Disadvantage: requires the use of relatively independent behaviors.

Nonexperimental or Qualitative Designs

1. Case Studies: In-depth examination of a single person or a few people. Goal is to provide an accurate and complete description of the case. Principal benefit is that they can expand our knowledge about the variations in human behavior. Must have 5 components: (1) research question, (2) propositions, (3) units of analysis, (4) determination of how the data are linked to the propositions, and (5) criteria to interpret the findings. Advantages: serve as a source for research ideas and hypotheses; helped develop therapeutic techniques; enabled scientists to study extremely rare and low-base-rate phenomena; and can describe and detail instances that contradict universally accepted beliefs and assumptions. Disadvantages: only descriptive; may involve a great deal of experimenter bias; and unlikely that the findings will generalize to other people with similar issues and problems.
2. Naturalistic Observation: Observe organisms in their natural settings. Four defining principals: (1) noninterference, (2) observation and detection of invariants (behavior patterns or other phenomena that exist in the real world), (3) useful for exploratory purposes, and (4) descriptive. Disadvantages: no control over settings, participants may not have opportunity to display behaviors phenomena the researcher is trying to observe, and the topics of study are limited to overt behaviors.
3. Survey Studies: Asking large numbers of people questions about their behaviors, attitudes, and opinions. Nine general steps to conduct a survey: (1) define a general objective, (2) develop a more specific objective, (3) determine the sample population (4) decide how the sample will be surveyed and develop questionnaire, (5) fieldwork: make decisions about the individuals who will administer the surveys, and about their qualifications, hiring, training, (6) content analysis: transform qualitative data into quantitative data, (7) analysis plan: conduct descriptive and correlational statistics on data (8) tabulation: data entry, and (9) analysis and reporting.
4. Focus Groups: Formally organized, structured groups of individuals (who usually share a particular characteristic, demographic, or interest that is relevant to the topic being studied) brought together to discuss a topic or series of topics during a specific period of time. Trained moderator is present to set ground rules, raise discussion questions, and maintain the focus of group discussions. Advantage: provides an open, fairly unrestrictive forum for individuals to discuss ideas and to clarify each others’ opinions; and it serves to crystallize the participants’ opinions. Disadvantages: focus group may not be representative of general population, participants’ opinions can be altered through group influence, and it’s difficult to quantify open-ended questions.

Test Yourself:
1. The most important element of a true experimental design is __ assignment?
2. If groups are perfectly matched on all known factors, the researcher can be certain that any group differences on outcomes are due to the independent variable. T or F?

(Answers: 1. random; 2. False, it is still possible that nay number of unknown variables may be responsible for the group differences)

Unless otherwise stated, the content of this page is licensed under Creative Commons Attribution-ShareAlike 3.0 License