7.3 Quasi-Experimental Research

Dave Lewis

21 Sept 202019:26

EducationalLearning

32 Likes 10 Comments

TLDRThis script delves into quasi-experimental designs, contrasting them with true experiments due to their lack of manipulation and control. It clarifies the concept of quasi-independent variables, often misunderstood, and outlines various quasi-experimental designs like non-equivalent groups and pre-test post-test. The script also addresses the limitations and validity threats inherent in these designs, such as assignment bias and maturation, and highlights the benefits of combination designs in mitigating these issues. Examples illustrate the application of these designs in real-world research, emphasizing the importance of accurate over merely positive results.

Takeaways

😀 'Quasi' means resembling, but it often refers to something that isn't truly what it appears to be.
🤔 Quasi-experiments resemble true experiments but lack key elements like manipulation and control.
👥 Quasi-independent variables, such as gender, are used to differentiate groups without being manipulated.
📊 Non-equivalent groups design compares different groups without random assignment, introducing assignment bias.
⏳ Pre-test post-test designs measure changes over time within the same group, but have time-related threats to validity.
🕰️ Interrupted time-series design takes multiple measurements before and after an event or treatment, minimizing threats to validity.
💊 Combination designs merge non-equivalent groups and pre-test post-test designs to reduce confounds.
🎶 Contextual factors, such as music, can influence memory recall, as shown in a combined design study.
👥 Pre-test post-test control group designs compare treated and untreated groups to eliminate time-related confounds.
🧠 Replication is crucial in scientific studies to ensure the reliability of findings.

Q & A

What does the term 'quasi' mean?
-'Quasi' literally means 'resembling'. It is often used to describe something that appears to be a certain way at first look but doesn't actually meet the criteria upon closer inspection.
What is a quasi-experiment?
-A quasi-experiment is a study that resembles a true experiment but does not fully meet the criteria of manipulation and control.
What are quasi-independent variables?
-Quasi-independent variables are non-manipulated variables used to differentiate groups or scores being compared in a quasi-experiment. An example would be gender in gender studies.
Why can 'quasi' be problematic in research?
-'Quasi' can be problematic because it confuses people by making a study appear experimental when it is not. This can lead to misunderstandings about the nature and validity of the research findings.
What is a nonequivalent groups design?
-A nonequivalent groups design is a type of quasi-experimental design where different groups of participants are formed without random assignment. This is essentially a between-subjects nonexperimental design.
What is a pre-test post-test design?
-A pre-test post-test design is a quasi-experimental design where the dependent variable is measured before and after a treatment is implemented. It is also known as a non-experimental within-subjects design.
What are some threats to validity in pre-test post-test designs?
-Threats to validity in pre-test post-test designs include history, maturation, testing effects, instrumentation changes, and regression to the mean.
What is an interrupted time-series design?
-An interrupted time-series design is a quasi-experimental design where multiple measurements are taken at intervals before and after a treatment or event. This helps to balance out irregularities and provides more accurate measurements.
What is a non-equivalent control group design?
-A non-equivalent control group design is a quasi-experimental design where additional control groups are added to help eliminate extraneous variables and improve the study's internal validity.
What is a combination design in quasi-experimental research?
-A combination design, also known as a mixed design or non-experimental between and within-groups design, combines elements of non-equivalent groups and pre-test post-test designs. It includes both between-subjects and within-subjects factors to eliminate order effect confounds and reduce individual differences.

Outlines

00:00

🔬 Understanding Quasi-Experiments and Their Limitations

This paragraph explains the concept of 'quasi' in research, particularly in the context of quasi-experiments. It clarifies that quasi-experiments resemble true experiments but lack the critical elements of manipulation and control. The paragraph delves into the nature of quasi-independent variables, using gender as a prime example, and discusses the common oversight of researchers in not specifying the non-manipulated status of these variables. It also touches on different types of quasi-experimental designs, such as nonequivalent groups design and pre-test post-test design, and their inherent inability to establish cause-and-effect relationships due to the lack of manipulation.

05:06

📊 Challenges and Solutions in Non-Equivalent Group Studies

The second paragraph discusses the inherent challenges in non-equivalent group studies, such as assignment bias and its impact on a study's internal validity. It suggests the use of control groups in a non-equivalent control group design to mitigate some of these issues. The paragraph also explores the pre-test post-test design, highlighting its vulnerability to various threats to validity, including history, instrumentation, testing effects, maturation, and regression to the mean. An example of a political advertisement's effectiveness evaluation is provided to illustrate the potential for external influences on results.

10:08

📚 The Utility of Interrupted Time-Series Design in Quasi-Experiments

This paragraph introduces the interrupted time-series design as a variation of the pre-test post-test approach, offering a more nuanced view by taking multiple measurements over time. It explains how this design can help minimize threats to internal validity by averaging out anomalies that may occur between measurements. The paragraph also contrasts this with the limitations of a simple pre-test post-test design, which may omit critical information about trends or natural progressions that could account for observed changes. An example of data analysis is provided to demonstrate how a treatment's effect might be misattributed without a comprehensive view of the data.

15:13

🧬 Combination Designs: Enhancing Quasi-Experimental Research

The final paragraph discusses combination designs, which integrate elements of both non-equivalent groups and pre-test post-test designs to enhance research methodology. It outlines the structure of such designs, involving at least one between-subjects factor and one within-subjects factor, and how they can reduce confounds related to order effects and individual differences. The paragraph provides an example of a study on memory retrieval in different emotional contexts, illustrating how combination designs can yield insightful results. It concludes by mentioning the pre-test post-test control group design as another effective approach to control for time-related confounds.

Mindmap

Keywords

💡Quasi-experiment

A quasi-experiment is a research design that resembles a true experiment but lacks the manipulation and control over variables that a true experiment would have. It is central to the video's theme as it sets the stage for discussing different types of non-manipulative research designs. For example, the script mentions 'quasi-experiment' as something that looks like an experiment but isn't, due to the lack of control over the independent variables.

💡Quasi-independent variables

Quasi-independent variables are non-manipulated variables used to differentiate groups or scores in a study. They are key to understanding the limitations of quasi-experimental designs, as they cannot be controlled by the researcher. The script uses 'gender' as a classic example of a quasi-independent variable, noting that studies comparing genders are quasi-experiments, not true experiments.

💡Manipulation and control

Manipulation and control refer to the ability of a researcher to influence and regulate the variables in an experiment. The video emphasizes that the absence of these elements is what distinguishes quasi-experiments from true experiments. The script explains that quasi-experiments lack these criteria, which is a fundamental aspect of their design.

💡Internal validity

Internal validity is the degree to which a study's design and execution allow for accurate cause-and-effect conclusions. The video discusses how quasi-experiments can suffer from issues like assignment bias, which can compromise their internal validity. The script points out that without manipulation, there's a risk of confounds that can affect the study's validity.

💡Non-equivalent groups design

Non-equivalent groups design is a type of quasi-experimental design where different groups of participants are compared without the researcher's control over group assignment. The video uses this concept to illustrate one way quasi-experiments can be structured. The script provides the example of a study on aggressive behavior in the workplace, comparing males and females.

💡Pre-test post-test design

Pre-test post-test design is a within-subjects quasi-experimental approach where the same group is measured before and after a treatment or event. The video explains how this design can help understand changes over time. The script describes how this design can be affected by various threats to validity, such as history and maturation.

💡Combination design

A combination design is a quasi-experimental design that includes elements of both non-equivalent groups and pre-test post-test designs. The video highlights its advantages in reducing certain confounds and the need for fewer participants. The script provides an example of a study on memory retrieval, comparing the effects of happy and sad music on recall.

💡Assignment bias

Assignment bias refers to systematic differences between groups that can affect the outcome of a study due to non-random assignment. The video discusses how this bias can be inherent in quasi-experiments. The script mentions that without random assignment, there can be fundamental differences between groups that introduce confounds.

💡Replication

Replication in research refers to repeating a study to confirm or refute the original findings. The video stresses the importance of replication for ensuring the accuracy of results. The script mentions that even if a study appears to show an effect, it should be replicated to be certain of its validity.

💡Interrupted time-series design

Interrupted time-series design is a quasi-experimental approach that involves multiple measurements taken over time, with a treatment or event applied in between. The video explains how this design can help minimize threats to internal validity. The script illustrates this with an example of how researchers might measure the effects of a treatment by taking multiple measurements before and after its application.

💡Regression to the mean

Regression to the mean is a statistical phenomenon where extreme scores tend to move closer to the average upon retesting. The video mentions this as a potential threat to the validity of pre-test post-test designs. The script explains that repeated testing can lead to results that appear to show improvement or decline simply due to statistical regression, not actual change.

Highlights

The term 'quasi' means 'resembling' but not fulfilling the criteria of manipulation and control in experiments.

Quasi-experiments have quasi-independent variables that differentiate people into groups without manipulation.

Gender is a classic example of a quasi-independent variable used in non-manipulated studies.

Confusion often arises from the misuse of 'quasi' in research, with researchers not specifying the nature of their variables.

Quasi-independent variables are not abbreviated differently from independent variables, leading to potential misunderstandings.

Quasi-experimental designs include non-equivalent groups and pre-test post-test designs, differing from true experiments in the manipulation of variables.

Non-equivalent groups design compares non-manipulated groups, such as males versus females, without random assignment.

Aggressive behavior in the workplace was found to differ between males and females in a naturalistic observational study.

Quasi-experiments cannot determine causation due to the lack of manipulation, only differences.

Assignment bias is an inherent problem in non-equivalent group designs due to the lack of random assignment.

Control groups can be introduced in non-equivalent control group designs to minimize the effect of extraneous variables.

Pre-test post-test designs measure changes over time within the same group of participants, subject to various validity threats.

Validity threats in pre-test post-test designs include history, instrumentation, testing effects, maturation, and regression to the mean.

Interrupted time-series design involves multiple measurements over time to minimize threats to internal validity.

Single case designs apply time-series methodology to individual subjects, such as tracking a patient's progress.

Combination designs integrate elements of non-equivalent groups and pre-test post-test to reduce confounds and increase efficiency.

Memories are more easily retrieved in the same context, as demonstrated in a combination design study involving music and mood.

Pre-test post-test control group design compares treated and untreated groups over time to control for time-related confounds.

Transcripts

Browse More Related Video

QUANTITATIVE Research Design: Everything You Need To Know (With Examples)

Study Designs (Cross-sectional, Case-control, Cohort) | Statistics Tutorial | MarinStatsLectures

Research Design: Choosing a Type of Research Design | Scribbr 🎓

Types of Experimental Research Designs - CRD RBD RCBD LSD FD - Research Methods - Practical Research

2^k Factorial Designs Experiment - ANOVA Model

Factorial Designs: Main Effects & Interactions