Q1. Example of interaction is drug dosage vs. task (simple or complex). In the case above, for the simple task, the higher the dosage, the shorter the time to accomplish the task, while for the complex task, the higher the dosage, the longer the time to accomplish the task. The statistical display of the effect of interaction between two or more variables is termed “factorial designs”.
Factorial design
Q2. Interaction occurs if the effect of one or more variables differs depending on the level of the other variable(s).
Q3. Examples of threat to internal validity
- Maturation effects: This occurs when changes in a score over time are due to naturally-occurring internal processes, for example, a math teacher of first-graders who wants to test the effectiveness of a new math curriculum. At the start of the year, the teacher issues a test, then exercises the curriculum and then later issues another test, usually at the end of the year. If the improvements are attributed to the new curriculum, then the validity of the results is threatened because the improvement might have occurred due to intellectual maturity.
- History effects: Involve an external event that occurs between two measurements. For example, a researcher, who is testing whether a particular chemical increases anxiety measures New York City residents’ anxiety on September 6, 2001, gives them drug, and measures them again 21 days later. The scores are likely to increase due to the terrorist attacks on September 11th. This is an external event.
- Testing effects: Occurs when the pretest influences the post-test. For example a situation when the post-test results are much higher than the pretest results, this is called the practice effect.
- Instrument decay: Occurs when the standards of measurement change over time. For example if a researcher grades speech performance before and after a training session, the results that are recorded after the session can be higher or lower than the results recorded before the session. This can be due to the change of the standards employed by the researcher rather than speech alteration.
- Statistical regression toward the mean (statistical regression): Occurs when the participants are chosen based on extremely high or low scores, like scoring very low or very high on an intelligence test. The interesting phenomenon the scores for the group will tend to be nearer to the mean if they are tested again. For example, consider a school admits students who score above 120 into a gifted class, which has a mean I.Q. score of 125. Regression toward the mean will predict that two months later, when the group takes the test again, their mean will be lower than 125. Likewise, if a group of students who score below 70 are placed into a special education class with a mean of 65 and are then tested two months later, their mean will be higher than 65. In all of the cases, there is a regression towards the population mean in the scores for the group.
Q4. Importance of Construct validity
Construct validity refers to the degree to which inferences can legitimately be made from the operationalization in your study to the theoretical constructs on which those operationalization were based. It is an assessment of how well you translated your ideas or theories into actual programs or measures. It is important because it can be used as the “truth in labeling” certain issues in a research.
Q6. Proactive approach involves trying to recall something you have memorized relatively recently and a separate. Therefore, the past memory muddles the one you wish to recollect. While retroactive approach involves trying to recall something you committed to memory in the past and some similar thing you have learned since that time. This impedes your ability to correctly recall what you want to mind.
Q7. Examples of alternative explanations in ex-post facto research
- Reverse causality: Rather than x causing y, y causes x.
- Common cause: Both the independent and dependent variables are the result of a third variable
- Other external independent variables: Other variables than those being studied may bring about the effect on y
Q8. Importance of using scatterplot/scatter grams
It is able to depict any relationships between variables that may be nonlinear. In addition, if the data is represented by a mixture model of simple relationships, such relationships can be visually evident as superimposed patterns.
Q9. Example of random selection and random assignment
Random selection: A researcher selects 100 names from a list of 40,000 Chinese men by placing all of their names in a hat. Each name is put on a separate piece of paper and names are drawn until 100 names have been picked. This can also be done by using special computer programs, especially when there are hundreds of participants to randomly choose.
Random assignment: In a study to help individuals quit drinking alcohol, researcher randomly assigned participants to one of two groups. In Group I, participants took a class to quit drinking alcohol. The classes took place each week for 8-weeks and included information about the merits of quitting drinking alcohol. In addition, participants in the class received strong social support from mentors or friends In the Group II; participants read a 4-page pamphlet created by the Center for Disease Control that explains the benefits of quitting drinking alcohol. The researcher randomly assigned participants to one of the groups. It was discovered that those who participated in the class and received backing from their friends were more likely to quit drinking alcohol compared to those in the other group that received and read from only the pamphlet.
Q10. Example of;
- Research problem: Usability
Poor usability leads to irritation and fatigue and it has an adverse impact on the usage experience. It can lead to loss of revenues in an on-line shopping web site. The poor usability of business applications has seen an increase in help desk costs. However, better usability makes our usage experience more fun and can increase productivity. A highly usable on-line shopping web site builds customer loyalty, tempts repeat visits, and increases its revenue-earning potential.
- Research hypothesis: The amount of sugary drinks ingested daily influences childhood obesity.
- Alternative hypothesis (H1): Childhood obesity is increased by the amount of sugary drinks ingested daily (tested hypothesis).
- Null hypothesis (Ho): There is no relationship between childhood obesity and the amount of sugary drinks ingested daily.
Q11. Randomized subjects control group posttest design vs. randomized subjects control group pretest-posttest design.
Randomized subjects control group pretest-posttest design is so called because it:
- Controls the timing of the independent variable and the group exposed to it.
- Controls the assignment of subjects to experimental and control groups through the use of table of random numbers.
- Controls all other conditions under which the experiment takes place.
This design involves the following steps:
- Randomly assign subjects to treatment or control groups;
- Administer the pre-test to all subjects in both groups;
- Ensure that both groups experience the same conditions except that in addition the experimental group experiences the treatment;
- Administer the post-test to all subjects in both groups;
- Assess the amount of change on the value of the dependent variable from the pre-test to the post-test for each group separately.
The above design is used for controlling for both threats to internal and external validity.
On the other hand, randomized subjects control group post-test design follows all the same steps as the randomized subjects control group pretest-posttest design except that it omits the pre-test. It can also be used for controlling for threats to internal and external validity. It eliminates the threat to internal validity of pre-testing by eliminating the pre-test. In addition, it creases the problem of experimental mortality by shortening the length of study.
Q12. How are Pearson’s, Rho, and Phi similar and different and related?
The similarity between Pearson r, Spearman’s Rho, and Phi is that all of them yield coefficients between -1 (a perfect negative correlation) and +1 (a perfect positive correlation); and their relationship is that they can all be computed using the same formula, that is, the formula for Pearson r. However, the difference between the three coefficients is that Pearson r is used with continuous variable measured on an interval or ratio scale, linear data and normally distributed; Spearman’s Rho is used with rank order data measured on an ordinal scale, can be used for non-linear; and Phi Coefficients is used when both variables are true dichotomous.
References
Altermatt, B. (2010, July 7). Threats to Internal Validity for Within-subjects Designs. Retrieved October 20, 2013, from vault.hanover.edu: http://vault.hanover.edu/~altermattw/methods/assets/Readings/Within_subjects.pdf
glimo.com. (2013). What is an interaction effect? Retrieved October 20, 2013, from glimo.vub.ac.be: http://glimo.vub.ac.be/downloads/interaction.htm
Mitchell, L. M., & Jolley, M. J. (2010). Research Design Explained, Seventh Edition. Wadsworth, USA: Cengage Learning Inc.
Owen, R. (2013). Retroactive Vs. Proactive Interference. Retrieved October 20, 2013, from ehow.com: http://www.ehow.com/facts_7269925_retroactive-vs_-proactive-interference.html
PPA. (2013). EXPERIMENTAL DESIGNS FOR RESEARCH. Retrieved October 20, 2013, from csulb.edu: http://www.csulb.edu/~msaintg/ppa696/696exper.htm
Trochim, M. K. (2008). Idea of Construct Validity. Retrieved October 2013, 2013, from socialresearchmethods.net: http://www.socialresearchmethods.net/kb/considea.php
