17 minute read
Teaching Activities
from Methods Toward a Science of Behavior and Experience 10th Edition William J. Ray TEST BANK & INSTRUCT
by ACADEMIAMILL
Ethics and randomization
Randomization is used in clinical trials and is necessary to appropriately evaluate a new treatment’s efficacy. Ask your students to consider this scenario: A close family member has been diagnosed with a serious, life-threatening disease. A new, experimental treatment is currently being tested and preliminary results show much promise. Your loved one is currently receiving the standard treatment that has questionable effectiveness so far. The question is: do you advise this person to forego the standard treatment and participate in a study in which random assignment is used? If they participate, they could receive a promising new treatment, but there is a chance that your family member would be assigned to a control condition and get a placebo.
Random selection and assignment
Demonstrate the effectiveness of randomization and the differences between random selection and random assignment by using your class as a population. Number the students in class from 1 to N. Use a random number table to randomly select a sample. Discuss how the results of your experiment can be generalized to the class as a whole. Randomly assign students from your sample to one or two experimental groups. Point out to students how various subject characteristics (e.g., gender, age, height, eye color, marital status) are spread across both groups and therefore are not confounding. In other words, the groups are approximately equal with respect to many variables. You can augment this activity by varying size of the sample randomly selected. Remind the students that small sample sizes pose problems.
Counterbalancing
Provide several experimental scenarios that did not use counterbalancing. What are the possible reasons for the obtained results? Point out how confounding variable could have affected the results possibly leading to a Type I or Type II error.
Wadsworth’s Research Methods Workshop http://www.wadsworth.com/psychology_d/templates/student_resources/workshops/resch _wrk.html
The Wadsworth’s Research Methods Workshop web site includes a workshop entitled Controls that contains information on how to eliminate factors that jeopardize the validity of research conclusions. The workshop begins byreviewing general strategies to achieve control, such as preparation of the research setting and random assignment of participants to experimental groups. Students are given practice in recognizing factors that introduce bias into a study. Counterbalancing techniques and use of replication as a control are also discussed.
Internet Resources
Research Randomizer (www.randomizer.org/index.htm)
This site provides a random number generator to use for random sampling or assigning subjects to groups. It also offers a tutorial to help students understand how randomness is used in selection and assignment.
Random Assignment to Study Social Programs (www.dss.gov.uk/asd/asd5/IH94.pdf)
Using random assignment to evaluate social programs has been criticized by some as being unfair and unethical since it necessitates assigning people to control groups. This document explores this issue.
Sampling in Survey Research (www.custominsight.com/articles/random-sampling.asp)
You will find information regarding the importance of sampling in survey research as well as a random sample generator.
Equating and Counterbalancing (http://psy1.clarion.edu/mm/teachRDE/C12/c12.ppt)
This PowerPoint presentation addresses matched designs and counterbalancing in the context of experimental design.
Suggested Readings
Campbell, D. T., & Stanley, J. C. (1963). Experimental and quasi-experimental designs for research. Chicago: Rand McNally.
An important reference on confounds and threats to internal validity.
Solomon, R. L. (1949). On extension of control group design. Psychological Bulletin, 46, 137-150.
More from the founder of the Solomon four-group design.
Research Activities for Students
The goals of the research activities are to: (1) relate Chapter 8 on an applied learning dimension, and (2) get you involved in research.
1. Design a Research Flowchart. Develop your research hypothesis from Chapter 8 into an experimental design. Your objective is to use a systematic, step-by-step process to test your hypothesis. Fill in the following information.
Step 1: What is the research hypothesis?
Step 2: What is the population of interest?
Step 3: How will you sample? Estimate the needed sample size.
Step 4: How will you assign participants to conditions?
Step 5: a. Manipulation of an independent variable? Experimental design Observation of differences between groups b. No manipulation of an independent variable? Correlational design Measurement of variables
1. How many independent variables?
2. How many levels of each independent variable?
Step 6: Operationally define your independent and dependent variables.
Step 7: Specify controls in your experimental design.
Step 8: What statistical analysis is planned?
Step 9: What level of alpha is selected?
Step 10: What are the expected findings?
2. Anorexic Patients. Random Assignment into Treatment Conditions
Using the Random Number Table. You are interested in investigating the effects of behavior modification therapy in the treatment of anorexia nervosa. The following participants, who have been coded to protect anonymity, are patients in the Eating Disorders Unit at the Community Hospital. Their weights and ages have been recorded. Your assignment is to randomly assign these participants to treatment conditions by using a random number table.
Objective: Randomly distribute subject characteristics between groups. Each participant should have an equal chance of being assigned to either the experimental or the control group.
Strategies Using the Random Number Table: There are many possible creative uses of the random number table. Here are two alternatives (Bordens & Abbott, 1991).
A. ODD-EVEN METHOD
1. Use the random number table in the appendix of your text.
2. Randomly start anywhere on the table.
3. Assign a random digit (using single digits from the random number table) to each participant.
4. a. Assign participants who were randomly assigned odd numbers to Treatment Group 1.
b. Assign participants who were randomly assigned even numbers to Treatment Group 2.
Note: If it is important to have the same number of participants in both groups, assign participants to Treatment Group 1 until its size equals one-half N. The remaining participants shall automatically be assigned to Treatment Group 2.
B. DIRECT METHOD
1. Use the random number table in the appendix of the text.
2. Randomly start anywhere on the table.
3. Designate numbers to represent treatment groups. For example let the number 1 represent the first treatment group, and let the number 2 represent the second treatment group.
4. Begin with the first participant and read the random numbers as single digits. As you come across either a "1" or a "2" assign that participants to the respective treatment group. Continue with the other participants.
Note: If it is important to have the same number of participants in both groups, assign participants to Treatment Group 1 until its size equals one-half N. The remaining participants shall be automatically assigned to Treatment Group 2.
Random Assignment
Group 1: Behavior Modification
Participant Age Weight
Group 2: Control Group
Participant Age Weight
Analyze the randomly assigned groups. Use a t-test to assess whether there are significant differences between these groups in terms of weight before the behavior modification treatment begins.
3. Identify Experimental Designs:
a. Investigators are interested in the affects of caffeine on Premenstrual Syndrome (PMS). Patients are randomly assigned to one of two conditions. Group 1 will be allowed to consume unlimited amounts of caffeine during a one-month period. Group 2 requires that patients will agree not to consume any caffeine products as specified in a written contract. At the end of the month-long study severity of PMS symptoms will be assessed.
Questions: b. Investigators are interested in the ability of computer technology to enhance visualization to promote T-cell production in children with cancer. Patients are randomly assigned to one of two conditions. Group 1 patients will receive a new computer visualization program that instructs the children to visualize the action of healing cells in their body which are aggressively fighting their cancer. Group 2 patients will not receive the computer program nor any visualization training. Before treatment begins, all patients' T-cell levels will be measured via blood tests. The experimental and control treatment conditions will be initiated, and the experiment is planned to proceed for three months. After three months all patients' T-cell levels will again be measured from blood samples.
1. What kind of experimental design is illustrated?
2. What is the independent variable?
3. What is the dependent variable?
4. Diagram this design.
5. What are the potential confounds?
Questions: c. Investigators are interested in the effect of positive self-statements on the selfesteem of women diagnosed with Battered Woman's Syndrome. All subjects will be randomly assigned to four groups. Group 1 will be given a self-esteem inventory prior to treatment. Subjects will then be instructed to use specific positive self-statements twice each day at determined times. The program will continue for four weeks. At the end of the four-week program each subject's selfesteem will be assessed, using the same self-esteem inventory previously administered. Group 2 will also be given a self-esteem inventory to determine a baseline measure. Subjects in this condition will not receive the positive selfstatement treatment nor any other treatment. After four weeks subjects will be reassessed on the original self-esteem inventory. Group 3 will receive positive self-statement training for four weeks and then will be administered the selfesteem inventory. Group 4 will be administered the self-esteem inventory at the end of the four experimental weeks. The self-esteem of all subjects will be compared to assess for treatment effects, pretest sensitization and history effects.
1. What kind of experimental design is illustrated?
2. What is the independent variable?
3. What is the dependent variable?
4. Diagram this design.
5. What are the potential confounds?
Questions:
1. What kind of experimental design is illustrated?
2. What is the independent variable?
3. What is the dependent variable?
4. Diagram this design.
5. What are the potential confounds?
4. Improving Pro-environmental Behaviors of Students. Develop an educational intervention to improve the pro-environmental behaviors of primary school students. Apply one of the experimental designs presented in Chapter 7. Summarize the investigation and address the following questions. Show all your work in your workbook.
Questions:
1. What kind of experimental design is illustrated?
2. What is the independent variable?
3. What is the dependent variable?
4. Diagram this design.
5. What are the potential confounds?
5. Application of Classic Experimental Studies. Use online references such as PsycINFO and PubMed to discover if any experimental studies have been used as evidence to change legislation or public policy.
Testbank
Multiple Choice
1. Experiments conducted in artificial environments may lack: a. internal validity. b. external validity. c. reliability. d. control.
ANS: B PTS: 1 REF: Introduction a. generalizability b. external validity c. relevance to the "real world" d. all of these
2. Control can help researchers make conclusions about the effects of an independent variable on a dependent variable. However, if control makes an experiment too unnatural, ____ may be compromised.
ANS: D PTS: 1 REF: Introduction MSC: WWW
3. In the example cited in the chapter regarding the study on estimation of time, the first question to be asked is: a. "Who will be our participants?" b. "Does the dependent variable have at least two levels?" c. "How do we assign participants to the groups?" d. "Is the difference between groups due to a confound?"
ANS: A PTS: 1
REF: Control Achieved Through Participant Assignment and Selection
4. The question of who our subjects will be in an experiment is most related to: a. where we are conducting our experiment. b. the likelihood of Type I errors. c. our experimental design. d. the topic being studied.
ANS: D PTS: 1
REF: Control Achieved Through Participant Assignment and Selection a. increases counterbalancing b. helps to control systematic confounds c. helps to make sure the groups are equated d. increases within-groups variance
5. What does random selection do?
ANS: B PTS: 1
REF: Control Achieved Through Participant Assignment and Selection a. random selection b. bimodal distribution c. random numbering d. analysis of variance
6. What technique allows for the control of systematic confounds?
ANS: A PTS: 1
REF: Control Achieved Through Participant Assignment and Selection
MSC: WWW
7. If we have conducted random selection accurately, we would be fairly confident that: a. there are equal numbers in subject groups. b. the independent variable has been operationally defined. c. the sample of subjects reflects the population that we are studying. d. confounding variables have increased in our experiment.
ANS: C PTS: 1
REF: Control Achieved Through Participant Assignment and Selection
8. If there is a confounding variable in an experiment, you cannot: a. know the cause of the outcome. b. obtain a significant result. c. select participants randomly. d. avoid making a Type II error.
ANS: A PTS: 1
REF: Control Achieved Through Participant Assignment and Selection
9. Participant selection procedures affect: a. internal validity. b. external validity. c. counterbalancing. d. experimental design.
ANS: B PTS: 1
REF: Control Achieved Through Participant Assignment and Selection
10. Participant assignment procedures affect: a. internal validity. b. external validity. c. counterbalancing. d. experimental design.
ANS: A PTS: 1
REF: Control Achieved Through Participant Assignment and Selection
11. If you wanted to randomly select fellow students in your class, you could: a. select those students on the left. b. pick those students sitting in front. c. use a random number table. d. select them as they come to class.
ANS: C PTS: 1 REF: Box 8.1 MSC: WWW
12. Assigning equal numbers of men and women to each experimental condition is an example of: a. random assignment. b. counterbalancing. c. proportional assignment. d. an equating procedure.
ANS: D PTS: 1
REF: Control Achieved Through Participant Assignment and Selection a. elimination procedure b. equating procedure c. counterbalancing d. all of these
13. Potential confounds can be eliminated from experiments by using ____ when assigning participants to conditions.
ANS: D PTS: 1
REF: Control Achieved Through Participant Assignment and Selection
14. If individuals are assigned to experimental conditions such that each condition appears equally often and each conditions precedes and follows other conditions equally often, then participant assignment has been: a. random. b. equated. c. counterbalanced. d. proportional.
ANS: C PTS: 1
REF: Control Achieved Through Participant Assignment and Selection
15. Using participants in different conditions that "match" on one or more characteristic is known as: a. randomizing. b. equating. c. counterbalancing. d. operationalizing.
ANS: B PTS: 1
REF: Control Achieved Through Participant Assignment and Selection
16. Potentially confounding order effects can be addressed with: a. randomization. b. equating. c. counterbalancing. d. none of these
ANS: C PTS: 1
REF: Control Achieved Through Participant Assignment and Selection
17. The counterbalancing procedure in assigning participants to either the experimental group (E) or control group (C) group would be best presented as: a. ECECECECECECEC b. EEEEEEECCCCCCC c. EECCEECCEECCEE d. CCCCCCEEEEEEEE
ANS: C PTS: 1
REF: Control Achieved Through Participant Assignment and Selection
18. If every person in the population has an equal chance of being selected for a study, then selection is: a. random. b. equated. c. counterbalanced. d. proportional.
ANS: A PTS: 1 REF: Randomization a. counterbalancing b. equating c. operationalizing d. randomization
19. A professor is advising you regarding the design of an experiment for a thesis. She asks you what you think is the best way to eliminate unintended participant assignment. What is your well-reasoned, educated response?
ANS: D PTS: 1 REF: Randomization
MSC: WWW
20. The best way to eliminated potential confounds that might result from participant assignment is: a. counterbalancing. b. randomization. c. elimination. d. equating.
ANS: B PTS: 1 REF: Randomization
21. If you are at the point where you are deciding who from the population shall be your participants, you should be using: a. factorial design. b. random assignment. c. Solomon's four-group design. d. random sampling.
ANS: D PTS: 1 REF: Randomization
22. Random assignment of participants to conditions has the advantage over other methods of assignment in that: a. it places equal numbers of participants in each group. b. it ensures that each condition precedes and follows every other condition equally often. c. it controls for known confounding variables. d. it controls for both known and unknown confounding variables.
ANS: D PTS: 1 REF: Randomization a. control, manipulation b. conditions, population c. subject pool, conditions d. order effects, equating
23. Random assignment is to ____ as random selection is to ____.
ANS: B PTS: 1 REF: Randomization
MSC: WWW a. counterbalancing b. equating c. cluster sampling d. randomization
24. As an experimenter you've identified some confounding variables that you want to control. But you also think there might be some unknown potentially confounding variables. What method of assignment would be most appropriate?
ANS: D PTS: 1 REF: Randomization
25. When assigning participants to groups, you should attempt to control for: a. as many individual differences as possible. b. at least two individual differences. c. just those individual differences that are potentially confounding based on commonsense and past research. d. sex differences.
ANS: C PTS: 1 REF: Randomization
26. Random assignment of individuals to groups: a. nullifies individual differences. b. reduces equating. c. allows the sample to be representative of the population. d. all of these
ANS: A PTS: 1 REF: Randomization
27. To assign participants to groups in a truly random fashion, you should: a. use a random number table. b. flip a coin. c. draw names out of a hat. d. any of these
ANS: D PTS: 1 REF: Randomization
28. If we wish the results of our experiment to generalize, we need to pay special attention to: a. confounding variables. b. participant assignment. c. participant selection. d. all of these
ANS: C PTS: 1 REF: Randomization
MSC: WWW
29. To select participants from a pool of potential participant, you should not: a. use a computerized random number generator. b. draw names from a hat. c. select names from a list without trying to pay attention to who the individuals are. d. use a random number table.
ANS: C PTS: 1 REF: Randomization
30. Internal validity in an experiment is affected by: a. participant assignment to groups. b. representatives of the sample. c. participant selection procedures. d. external validity.
ANS: A PTS: 1 REF: Randomization
31. External validity in an experiment is most related to: a. participant assignment to groups. b. representativeness of the sample. c. confounding variables. d. a pretest.
ANS: B PTS: 1 REF: Randomization
CONTROL: THE KEYSTONE OF THE EXPERIMENTAL METHOD 161
32. Control in an experiment can be achieved through: a. random assignment of participants to groups. b. experimental design. c. logic. d. all of these
ANS: D PTS: 1 REF: Control Achieved Through Experimental Design
33. The following experiment illustrates what design?
R Group A T M
R Group B T(zero level) M a. factorial design b. pretest-only control group design c. posttest-only control group design d. pretest-posttest control group design
ANS: C PTS: 1 REF: Control Achieved Through Experimental Design
34. A pretest-posttest control group design does not control for: a. effects of the pretest. b. mortality. c. history. d. maturation.
ANS: A PTS: 1 REF: Control Achieved Through Experimental Design
35. In a Solomon four-group design: a. participants who get a pretest also get a posttest. b. not all participants receive the pretest. c. not all the participants received the posttest. d. all participants receive a pretest but not all receive a posttest.
ANS: B PTS: 1 REF: Control Achieved Through Experimental Design
MSC: WWW
36. A classmate designs the following experiment. What advice would you give him to improve upon it?
Group A M T M
Group B M T(zero level) M a. to specify both groups as receiving zero level of T b. not to perform a pretest-posttest c. to randomly assign participants to groups d. to randomly select participants from the population
ANS: C PTS: 1 REF: Control Achieved Through Experimental Design
37. The following experiment illustrates what design?
R Group A M T M
R Group B M T(zero level) M a. factorial design b. pretest-only control group design c. posttest-only control group design d. pretest-posttest control group design
ANS: D PTS: 1 REF: Control Achieved Through Experimental Design a. a pretest b. random assignment of participants to groups c. control of threats to internal validity d. at least two levels of the independent variable
38. Which of the following is not a necessary characteristic of a true experiment?
ANS: A PTS: 1 REF: Control Achieved Through Experimental Design a. random b. 1 c. 2 d. 3
39. For a study to be considered a true experiment according to Campbell and Stanley, there must be at least ____ level(s) of the independent variable.
ANS: C PTS: 1 REF: Control Achieved Through Experimental Design
MSC: WWW a. Randomness, since being talkative can make this particular group more relaxed and better able to perform. b. Participant-maturation interaction, since being friendly to one group can introduce a confound. c. History, since this unintended interaction was not part of the independent variable, a confound was introduced. d. Testing, since the other group might feel cheated for being treated more formally by the researcher, and this introduces counterbalancing.
40. In an experiment, the researcher was unintentionally talkative to one of the participant groups. What best describes the threat to internal validity and why is it a threat?
ANS: C PTS: 1 REF: Control Achieved Through Experimental Design
41. You are analyzing a journal article that used the Solomon four-group design. You've identified the following groups:
R Group 1 Pretest Treatment Posttest
R Group 2 Pretest Posttest
R Group 3 Treatment Posttest
What would the fourth group look like?
I. R Group 4 Posttest
II. Group 4 Pretest Posttest
III. Group 4 Treatment
IV. R Group 4 Pretest Posttest a. I b. II c. III d. IV
ANS: A PTS: 1 REF: Control Achieved Through Experimental Design
42. If two groups are unequal at the start of an experiment with regard to the dependent variable, then the F ratio would be: a. 1 b. <1 c. >1 d. unknown
ANS: C PTS: 1 REF: Control as Related to the Logic of Experimentation
MSC: WWW
43. If two groups are equal with regard to the dependent variable at the start of an experiment before the independent variable is introduced, we can assume that the F ratio would be: a. 0 b. 1 c. >1 d. unknown
ANS: B PTS: 1 REF: Control as Related to the Logic of Experimentation
44. Before any treatment is introduced, the variation that exists between two groups that are equal is due to: a. chance. b. the F-ratio. c. the independent variable. d. the dependent variable.
ANS: A PTS: 1 REF: Control as Related to the Logic of Experimentation
45. If individuals in different groups are not treated identically except for the independent variable, then: a. within-groups variance is increased. b. the F ratio is decreased. c. a possible confound is introduced. d. between-groups variance is decreased.
ANS: C PTS: 1 REF: Control as Related to the Logic of Experimentation
46. If an extraneous variable affects all groups equally: a. within-groups variance is increased. b. it is more difficult to reject the null hypothesis. c. there is increased probability of a Type II error. d. all of these
ANS: D PTS: 1 REF: Control as Related to the Logic of Experimentation
MSC: WWW
SHORT ANSWER
1. Explain how random sampling impacts external validity.
ANS: Answer not provided.
PTS: 1 REF: Randomization
CONTROL: THE KEYSTONE OF THE EXPERIMENTAL METHOD 165
2. How do the elimination procedure and the equating procedure reduce potential confounds?
ANS: Answer not provided.
PTS: 1 REF: Control Achieved Through Participant Assignment and Selection
3. Describe the counterbalancing procedure as it relates to random assignment.
ANS: Answer not provided.
PTS: 1 REF: Control Achieved Through Participant Assignment and Selection
4. What is the difference between participant selection and participant assignment? How does each affect either internal or external validity in an experiment?
ANS: Answer not provided.
PTS: 1 REF: Randomization
5. What is the purpose of random sampling? How would you go about random sampling students at your college or university?
ANS: Answer not provided.
PTS: 1 REF: Randomization
6. Explain why randomization has an advantage over counterbalancing in assignment of subjects to groups.
ANS: Answer not provided.
PTS: 1 REF: Randomization
7. Can random selection or assignment lead to a situation that results in groups being very different even before an experiment begins? Explain you answer.
ANS: Answer not provided.
PTS: 1 REF: Randomization
8. Experimental design must allow for two functions. What are these functions and how can the design help the researcher achieve them?
ANS: Answer not provided.
PTS: 1 REF: Control Achieved Through Experimental Design
9. Outline the Solomon four-group design, indicating its strengths with respect to the control of confounds.
ANS: Answer not provided.
PTS: 1 REF: Control Achieved Through Experimental Design
10. What is a true experiment according to Campbell and Stanley?
ANS: Answer not provided.
PTS: 1 REF: Control Achieved Through Experimental Design
11. Discuss the logic of experimental design in terms of the F ratio and the null hypothesis. What is the effect on the F ratio of unequal groups at the start of an experiment?
ANS: Answer not provided.
PTS: 1 REF: Control as Related to the Logic of Experimentation
12. Why is the interpretation of experimental results "easy" if control is achieved by random selection, random assignment, and consistent treatment of subjects?
ANS: Answer not provided.
PTS: 1 REF: Control as Related to the Logic of Experimentation
CONTROL: THE KEYSTONE OF THE EXPERIMENTAL METHOD 167
