Share
Explore BrainMass

Assessing Research Designs

Topic 1: Quasi Experimental Designs
A quasi-experimental design is similar to an experimental design as they both include an experimental and control group. The difference is that a quasi-experimental design lacks the element of randomization (Goodwin & Goodwin, 2013). The most common is the non-equivalent groups design, which means the participants cannot be randomly assigned to experimental, and control groups.

One disadvantage of not being able to use random sampling is there is potential for creating non-equivalent groups. A second potential hazard of non-equivalent groups is that it could cause a threat to internal validity. If the experiment does not begin with equivalent groups than the researcher cannot be certain that the independent variable was responsible for the change in the experimental group randomization (Goodwin & Goodwin, 2013). It's a possibility that confounding variables could have contributed to the change. Quasi-experimental designs are valuable because they are often used to test the effectiveness of a treatment. This allows for researchers to make inferences about the cause and effect relationship of particular treatments. Another advantage is that often quasi-experimental designs prove to have strong external validity because the results can be generalized to current populations and settings randomization (Goodwin & Goodwin, 2013).

An example of a quasi-experimental design would be to examine the effects of smoking on respiratory functioning. The two groups, which are not randomly assigned, are individuals who smokes 1 pack a day and individuals who smoke 2 packs a day. These two groups could also be compared to a control group, which would be individuals who do not smoke at all.

Goodwin, C. J.& Goodwin, K.A. (2013) Research in psychology: Methods and design (7th ed.). John Wiley & Sons.

Comment/Discuss:

Topic 2: Small N Designs
Carolyn:
An example of an applied behavioural analysis design can easily be found in the teaching of students with ADHD, as ABA is a highly supported teaching tool. One study had students without ADHD help their fellow students with ADHD to stay on task and increase their productivity. Whenever students with ADHD were on task and concentrating, the confederate students would provide positive reinforcement. And when students with ADHD were off-task or not concentrating, the confederates would avoid eye contact and verbal exchanges.

One example of a case study is that of Phineas Gage. Perhaps his is one of the most well-known case studies in psychology. An iron rod completely penetrated Gage's brain leaving him with a catastrophic brain injury and yet he survived. His personality was greatly affected, however, and his subsequent behaviour became the subject of case study.

The strengths of small N studies include the following: they successfully reflect the behaviour of individuals, provide detailed analysis that is not found in other designs, can provide "prototypical descriptions" (Goowin, 2010, p.443), generate new research ideas through induction and they can help falsify weak theories.

The limitations of small N studies include the following: they are difficult to control extraneous variables, subject to experimenter bias, results may not generalize causing difficulty with external validity, and they rely on memory (ies) which is easily affected over time.

Reasons to opt for small N designs instead of statistical methods: participants may be difficult to locate, hard to obtain, too expensive or require long training periods.

Goodwin, C.J. (2010). Research in Psychology: Methods and Design (6th ed.). Danvers, Massachusetts, USA. John Wiley & Sons.

Comment/Discuss:

Topic 3: Observational Method
Shannon:
Observational methods are divided into naturalistic observation—studying the behaviour of people or animals in their natural environments with no researcher manipulation or intervention—and participant observation—the researcher will join in or interact with the group (Goodwin & Goodwin, 2013). Behaviour observed in natural conditions without any manipulations is deemed more credible than that seen in labs; however, this methodology is descriptive in its outcomes and cause and effect cannot be determined. It is limited by the restrictions in controlling for extraneous variables, and the researchers must be restrained and thoughtful in their conclusions. Observer bias can also affect the results by: 1) the researcher having preconceived ideas about what will be observed; 2) collecting massive amounts of data that will then need to be reduced to something more practicable with which to work (Goodwin & Goodwin, 2013). Participant reactivity can also be problematic. It occurs when the participants know they are being observed which (knowingly or unknowingly for the participant) influences their behaviour (it is also referred to as the Hawthorne effect) (Goodwin & Goodwin, 2013).

If we were to study sharing behaviour in children, we may choose a naturalistic observational method and observe through a two-way mirror to minimize and participant reactivity (Goodwin & Goodwin, 2013). We would need to operational definitions to reduce biases of the observers and use time sampling or event sampling to collect our data strategically thereby reducing the potential for mass amounts of data being collected (Goodwin & Goodwin, 2013). We might want to become a participant observer if our research involves learning more about a particular group, which might mean interactions with the group members potentially influencing the perceptions of the participant observer (Goodwin & Goodwin, 2013).

References

Goodwin, C. J. & Goodwin, K. A. (2013). Research in psychology: Methods and design (7th ed.). Hoboken, NJ: John Wiley & Sons.

Comment/Discuss:

Topic 4: Survey research
In probability sampling, each individual in a distinguishable group or population has a likelihood of being chosen for the sample, which is a section of that larger group. Researchers hope to be able to make inferences with regard to larger populations based on the information obtained within the sample (Goodwin & Goodwin, 2013). Ideally the sample should be indicative of the larger population you are hoping to gain insight into. Gaining knowledge of the characteristics of the individuals in the sample is important to then match them to those within the larger population. A sample that is chosen directly rather than self-selected (dependent on who chooses to participate) tends to be less prone to bias (Goodwin & Goodwin, 2013).

A simple random sample is when each individual within the population has an identical possibility of being chosen for the sample. This type of sample can be effective and is sometimes chosen for ethical reasons (Goodwin & Goodwin, 2013). An example of this would be drawing straws to decide who would be chosen. Some of the potential difficulties with this type of sample are that the sample may not reflect specific aspects that are to be explored. It is also not practical with very large populations (Goodwin & Goodwin, 2013). A stratified sample accounts for different subsections of the population that are important to the sampling. There can be different layers or strata represented within the randomly chosen sample. The researcher must decide which strata must be accounted for within the sample (Goodwin & Goodwin, 2013). A cluster sampling randomly chooses a group or cluster of individuals with shared attributes. This is used in large population samplings when simple or stratified samplings are not appropriate (Goodwin & Goodwin, 2013).

References

Goodwin, J. G., & Goodwin, K. A. (2013). Research in Psychology: Methods and design (7th ed.). John Wiley & Sons, Inc.

Comment/Discuss:

Solution Preview

Topic 1: Quasi Experimental Designs
A quasi-experimental design is similar to an experimental design as they both include an experimental and control group. The difference is that a quasi-experimental design lacks the element of randomization (Goodwin & Goodwin, 2013). The most common is the non-equivalent groups' design, which means the participants cannot be randomly assigned to experimental, and control groups.

One disadvantage of not being able to use random sampling is there is potential for creating non-equivalent groups. A second potential hazard of non-equivalent groups is that it could cause a threat to internal validity. If the experiment does not begin with equivalent groups than the researcher cannot be certain that the independent variable was responsible for the change in the experimental group randomization (Goodwin & Goodwin, 2013). It's a possibility that confounding variables could have contributed to the change. Quasi-experimental designs are valuable because they are often used to test the effectiveness of a treatment. This allows for researchers to make inferences about the cause and effect relationship of particular treatments. Another advantage is that often quasi-experimental designs prove to have strong external validity because the results can be generalized to current populations and settings randomization (Goodwin & Goodwin, 2013).

An example of a quasi-experimental design would be to examine the effects of smoking on respiratory functioning. The two groups, which are not randomly assigned, are individuals who smokes 1 pack a day and individuals who smoke 2 packs a day. These two groups could also be compared to a control group, which would be individuals who do not smoke at all.

Goodwin, C. J. & Goodwin, K.A. (2013) Research in psychology: Methods and design (7th Ed.). John Wiley & Sons.

Comment/Discuss:

Quasi-experimental research designs are also often referred to as non-randomized and also involve pre and post study designs. Such studies are often employed in educational research, as it is often impossible and sometimes unethical to randomly assign participants (students) in / to such settings.

The principal investigator or researcher tries to control for differences between the non-randomly assigned groups using common methods such as matching, and statistical control. So for example, say a researcher was interested in comparing the effect of an enhanced visual compartment to curricula could use a pre and posttest and compare effects before the enhanced segment and after to compare the effect if any.

Quasi designs seem to be much better than just pre-experimental studies however in that they utilize the actual comparison of groups. Yet I would say they tend to fall short in randomization. Also, with quasi-experimental designs even though the grades can be compared to decipher if there is a difference between the two groups Pre and post) before and after the study, we really cannot state for sure the difference is related to the experiment/experience itself: As such it could be another influence /other confounding variables.

References

Goodwin, J. G., & Goodwin, K. A. (2013). Research in Psychology: Methods and design (7th ed.). John Wiley & Sons, Inc.

Psychology Online ...

Solution Summary

The expert assesses the research designs for quasi-experimental designs.

$2.19