Exploring the Logic of Experimental Design 

For many years, experiments have been a basis research in pursuit and new knowledge. For any experiment, there are three characteristics namely: independent variables, controls, and dependent variables. Use of controls and independent variables makes all the difference between experiments and other research strategies.

Chapter exercise (p.244)

Question 2

Just like applied research basic research is executed to expand knowledge a swell as understanding of developing and or assessing theory. Its significance centers on data for comprehension’s sake (Jackson, 2012)

Question 4

The statement is poorly punctuated and grammatically incorrect.

Question 6

1. a) Exam performance
2. b) Subjects
3. c) Traditional manner—alone using notes they took during class lectures and

Interactive groups with notes from class lectures

1. d) Yes independent variable is manipulated because it is dependent on extrinsic factors

 

Question 8  

1. Age
2. Time
3. Age and reaction time
4. Yes, independent variable is manipulated because it is dependent on extrinsic factors such as time.

Chapter Exercises (8)

 

2)

 

1. a) This is a one-tailed test
   b) H0: No difference between new and other toothpastes in mean number of cavities in the general population: μ = 1.73
   Ha: New toothpaste reduces cavities compared to other brands μ < 1.73
   c) d) e) f)
   We cannot use the z-test (or the t-test) because we need the standard deviation (besides the mean) of the new brand toothpaste.
   sample size=60
   sample mean = 1.5
   sample standard deviation (or population standard deviation) — missing
   hypothesisized mean = 1.73

 

 

4)

Type II error:

Critical t: ±2.0687 at DF= 23
Critical t: ±2.1604 at DF= 13

 

Purpose of an Experimental Design

In analyzing experimental design, it is necessary to take the perspective of a statistician. Statisticians conduct an experiment with emphasis on experimental conditions, changes and any practical importance of the changes.  In order to meet the above requirements, there is need for experimental design. This is the basis of modern science, which relies heavily on well-executed experiments to establish causal relationships. Kirk defines experimental design as an assignment for assigning experimental units to treatment levels and the statistical analysis associated with the plan. Kirk continues to give the various elements that comprise an experimental design (Brown, 2010). These are:

1. Statistical hypothesis,
2. Determination of dependent variable, independent variable and nuisance variable,
3. Identification of randomization procedure,
4. Identification of experimental units and sample population, and
5. Statistical analysis

Advantages and Disadvantages of an Experimental Design in an Educational Study

When considering the benefits and drawbacks of an experimental design in an education study set up, it is necessary to consider the different types of designs. Before embarking on an experiment, researchers need to choose an experimental design type. There are two categories of experimental design namely dependent design and independent design (Kovera, et al. 2010). The advantages and disadvantages of each of these experimental design types will advise on the choice the researcher makes. Before highlighting these factors, here are some considerations to make when choosing an experimental design.

1. Ethical issues are a major consideration especially on subject testing and independent variable manipulation.
2. For experiments involving human beings, there is the inherent problem of fatigue effects and practice. This leads to increased boredom and improved results in consequent studies respectively.
3. Doubling of data is possible in cases where there are insufficient subjects and they have to be used several times.
4. When dealing with confounding variables, a paired design makes the perfect choice. Confounding variables represent anything other than the independent variable that might affect experimental results.

 

Internal Validity Claims or External Validity Claims?

In experimental design, there are two common terms namely internal validity and external validity. According to Yu and Ohlund, internal validity determines whether the condition or treatment of an experiment has any effect on the outcome, and looks for evidence the claim. On the other hand, external validity looks at the ecological validity of the condition or treatment.  From the above descriptions, it appears that both external and internal validity claims are done in different settings. To analyze the two options, this paper sough the insight of Yu and Ohlund in their piece Threats to Validity of Research. According to this article, the question of internal validity and external validity is material to experimental and “real world” settings respectively.

Interestingly, research relies heavily on both these aspects of experimental design. This takes the discussion to the two types of studies; effectiveness and efficacy studies. According to Yu and Ohlund, efficacy studies are the types conducted in the laboratory while effectiveness studies are done in the outside environment (Yu & Ohlund, 2012). Outside environment in this case refers to natural setting where the external validity of an experiment comes under scrutiny.                                                                                                                  To get a clear determination of which is suitable between internal validity claims and external validity claims, this paper brings into perspective the nature of experiments in the medical world. During clinical trials, patients are carefully selected. The argument is that if done haphazardly, the research results of the process may be jeopardized by the uncontrolled variables. The situation in the real world is significantly different. Patients have all sorts of diseases in varying degrees. The meaning of all this is that a drug could be so good in an internal validity experiment yet fail in the external environment. Therefore, it is important for researchers must find a healthy balance between internal validity claims and external validity claims if the results of a study are to be effective. This healthy balance is a situation whereby emphasis is on real-life situation aspect, which is represented in external validity claims (Salkind,  2010).

Control and Control Group in an Experiment

Regardless of how technical an experiment might be, the experimental design bears several features that include;

1. Proper controls to be compared with the experimental group
2. Personal opinion and bias are eliminated
3. Adopts the concept of replication whereby the experiment is repeated severally to ensure reliable statistical analysis.                                                                                   When conducting an experiment, controls are necessary for other factors other than the ones that the experiment is interested in. This is the essence of an experiment control group.  Normally, an experiment is divided into two groups namely the experimental group and the control group, and a change is only introduced in the experimental group. In so doing, the control group becomes the baseline to measure the changes introduced in the experimental group (Trochim & Donnelly, 2008).

Confounding Variables (Confounds) In an Experiment

Apart from the dependent and independent variables, there is the inherent likelihood of other factors affecting the results of the experiment. Such factors are referred to as confounds, confounding variables or random variables. For example, if the aim of an experiment is to measure which of two methods is more effective in teaching a student to read. In this case, the independent variable is the teaching method, while the dependent variable is reading ability.

In this example, a number of confounds could affect the experiment. Firstly, having different instructions to the student could alter the results of the experiment. This confound is overcome by ensuring all students get standardized instructions. Secondly, giving different materials could introduce the effects of a confounding variable. To avoid this, the experimenter may standardize the instructions given to the students taking part in the experiment. Then there are the variations between the participants, for instance in terms of age. This issue can be solved by using students from a specific age group. Alternatively, the age structure of the students can be selected in such a way that it is similar for groups in the experiment.

Balancing Between Causation and Correlation

Concepts of causation and correlation are usually confused especially when causal and correlation relationships are involved. However, there is a difference between the two research concepts .In other words, there exists a causal relationship if one thing occurs because of a second one.  Cause is an important aspect in research because it sometimes leads to logical conclusion. If an event causes another to occur, it suffices it to say that such event must first occur.

The definition of correlation as defined in the Merriam-Webster online dictionary is: “it is a relation existing between phenomena or things or between mathematical or statistical variables which tend to vary, be associated or occur together in a way not expected on the basis of chance alone. Therefore, in correlation, two or more things relate in such a way that there is a mathematical description of their variables.

From the two definitions, causation and correlation are fundamentally different concepts. However, there is the likelihood of a correlation implying causation. For example, consider the link between cancer and smoking. Years of research have continually pointed to the correlation between smoking and incidences of cancer. This correlation is so strong that most researchers see this as a causal relationship.

Comparison of Between-Participants, Within-Participants, and Matched-Participants Designs

Here, the paper addresses the question: does smiling cause mood to rise. In within-participants design, participants are subjected to a disgusting video to spoil their mood. After 5 minutes, they are instructed to smile and their mood level checked. The same participants are subjected to annoying music for 5 minutes, and their mood is measured. The main advantage of this design is that the error variance is reduced. In addition, there is the power in number of subjects. However, the design has a weakness of high chances of “carryover effects.” For example, participants in this experiment may be emotionally depleted by the second experiment. This experiment may also be impeded by lack of enough time (Pole, et al. 2010).                                                                                                                                      When it comes to between-participants experiment, the exposure to a disgusting video and out-of-sync music is done to two groups of participants. One group listens to the music while one watches the video. Their mood changes are then measured to ascertain the effect of the dependent variable, which in this case is the mood. This design has a major drawback in the lack of natural anchor. In other words, the amount of information the researcher can get regarding the independent variable and its effects is limited. However, a researcher who is running on a low budget would find this design very suitable because it requires limited materials and time (Kovera, et al. 2010).

 

In the matched-participants design, matched set of the students are distributed randomly. In this example, the students could be matched for race and age. This design is advantageous because the experimenter has better control of unwanted effects on the music and video, which are the independent variables. This design may only fail to add value to the experiment if the matched characteristics bear no effect on the variables.

Reference

Brown, J. Scott. (2010). “Variable.” Encyclopedia of Research Design. SAGE Publications.

Jackson Sherri (2012). Research methods and statistics: a critical thinking approach,          Belmont, CA:Wadsworth Cengage learning.

Kovera, Margaret Bull. (2010).“Confounding.” Encyclopedia of Research Design.. SAGE            Publications.

Pole, Jason D., and Susan J. Bondy.(2010). “Control Variables.” Encyclopedia of Research          Design. 2010. SAGE Publications.

Salkind, Neil J. (2010).“Dependent Variable.” Encyclopedia of Research Design. SAGE               Publications.

Trochim W. M. K., & Donnelly, J. P. (2008): Chapter 7 & Pages 186-191

 

Yu, C., & Ohlund, B. (2012). Threats to validity of research design. Retrieved from http://www.creative-wisdom.com/teaching/WBI/threat.shtml