Select Page

attached

Second, post at least one reply to a group member in which you: 

1.  Answer the following questions about your group member’s original study that they designed:

· What is the independent variable?
· How many levels are there for the independent variable? 
· What is the dependent variable?
· What is the confound?
· How could this confound be fixed? 

A new mosquito repellent had been announced to protect from insects bites. To test the effectiveness of this new product, the company sprayed a normal amount of repellent to a volunteer’s  arm while the other arm is not sprayed. Then the volunteer place both arms into a cage full of mosquitoes. After 30 minutes, the number of mosquito bites are counted.

Chapter 6

Conducting a Good Experiment I:
Variables and Control

Overview of this Chapter
In this chapter, we will look at variable construction, focusing our attention on …
Part One: The Nature of Variables
Part Two: Operationally Defining Variables
Part Three: Independent Variables
Part Four: Extraneous Variables (Confounds)
Part Five: Dependent Variables
Part Six: Nuisance Variables
Part Seven: Controlling Extraneous Variables
Part Eight: An Eye Toward The Future

Part One
The Nature of Variables

The Nature of Variables
Variables (which I know we have talked about a lot!), are events or behaviors that can assume two or more values. Variables focus on a wide-range of behaviors or events, such as …
temperature, humidity, chill factor, etc.
weight, age, height, etc.
attitudes, anxiety, confidence, etc.

These variables can be based on …
1). research tradition
2). theory

Variables Can Come From Research
1). Sometimes, variables are based on research tradition
A researcher may replicate (or replicate with extension) a prior research study using similar DV’s and IV’s
A researcher who wants to look at frustration might look at prior studies to see how other researchers measured this variable
They can ask questions (“How frustrated are you?”)
They can measure behaviors (Did the participant hit the wall with his fist?)
They can look at physiological reactions (Did the participant have increased blood pressure)

Variables Can Come From Theory
2). Variables may be based on a theory
Several years ago, psychologist Neil Vidmar proposed Generic Prejudice Theory, a theory based on the idea that people have “generic prejudices” about people who commit certain crimes
Vidmar argued that jurors are more likely to rely on the type of crime charged than the actual evidence presented in a legal trial. That is …
If defendants are charged with sexually abusing a child, Vidmar believed that the mere accusation of child abuse would weigh more heavily than the actual case facts

6

Example: Generic Prejudice Theory
Using Vidmar’s theory, Wiener and Winter created variables designed to see which was more powerful: “generic” versus “specific” (case-fact) prejudices.
We created different trial transcripts, with some based on sex abuse while others focused on robbery
Thus “Type of Trial” was an independent variable
We also looked at guilt as a dependent variable, as well as mock-juror’s feeling about the cases

7

Independent Variables (again)
Before moving on, I want to once again reiterate the difference between independent and dependent variables
The independent variable is the variable we control
Such variables are “independent” because they stand alone. They do not depend on other variables
In research, we compare levels of the independent variable
Sometimes we manipulate these variables (we give some people electrical shocks while we do not shock others) and sometimes we simply measure them (men versus women).
Either way, we have some control over the IVs

Dependent Variables (again)
The dependent variable is the variable we measure (response)
The DV changes as a result of the IV’s, thus DV’s depend on the IV
In research, we can have several DV’s
Frustration, ease of task, number correct, time to complete a task, etc.
These can be yes/no, scales, or fill-in-the-blank
But how do we create and define variables? Let’s find out …

Part Two

Operationally Defining Variables

Operational Definitions
Operational definitions “define” the independent and dependent variables in terms of the operations needed to produce them
Imagine we are studying learned helplessness, and we are interested in seeing whether giving people puzzles that are impossible to solve increases their levels of frustration
How do we define frustration here?
Here, we can make it simple. We can ask participants, “How frustrated are you with this task?”
But there are other ways to operationally define frustration, right? That is …

Operational Definition Example: Frustration
… we can define frustration by …
1). Asking different questions, like “How irritated are you?”
2). We can measure the participant’s blood pressure
3). Have others rate how frustrated the main participant was
4). Let participants retaliate against the experimenter who frustrated them (“Rate the performance of the researcher.”)
Let’s do a quick test on operational definitions …

A Task For You
Think about some of the following research studies
Try to figure out which variables need operational definitions

Research Idea # 1
A psychologist is interested in whether people are more likely to exhibit conformity when they are in situations that make them feel nervous and unsure of themselves.

What variables need to be operationally defined in this research idea, and how would you operationally define them?

Research Idea # 1: Definitions
Conformity feeling nervous and unsure all need operational definitions

How would you define these?
For conformity, maybe: “Perform the same behaviors as other people around them, like choosing the same entrée that all other diners chose despite having three different choices”
Nervous and unsure can be defined using rating scales: “How nervous were you? How unsure were you?” There are many other ways as well! Let’s try some more …

Research Idea # 2
A psychologist is interested in whether people are more or less creative in their work if they are pressured to be creative.
Creativity and pressure both need to be defined here!

Part Three
Independent Variables

Independent Variables
Time for some in-depth information on independent variables
In this section, we will discuss several types of independent variables (the variables you control / manipulate), including …
1). Physiological-based independent variables
2). Experience-based independent variables
3). Stimulus-based independent variables
4). Participant-based independent variables

Physiological IVs
1). Physiological-based independent variables manipulate the participants’ physiological state
For example, you might give some participants alcohol (and others a placebo that tastes like alcohol, but is not alcohol) to see how / whether alcohol impairs their ability to make accurate eyewitness identifications
Or you could give some participants a drug that alters their physiological arousal, and then see how they interpret their arousal when presented with bystanders who are acting like lunatics (Schracter & Singer actually did this in 1962!)

Experience-Based IVs
2). Experience-based independent variables involve manipulating the amount (or type) of training / learning a participant receives
For example, you might first give jurors a summary of a death penalty case and have them make legal decisions about the case. You can correct any of their misconceptions about the legal instructions, and then have them make decisions in a second trial.
When I did this in a study, I found that those who “practiced” were better able to understand jury instructions. They gave more accurate verdicts than those who did not practice

Stimulus-Based IVs
3). Stimulus-based independent variables involve manipulating some aspect of the environment
For example, remember the Bobo-doll study? Bandura showed kids a scene of an adult model either beating up a Bobo-doll (an inflatable doll that, when hit, returns to a standing position) or reading quietly next to the Bobo-doll
Those kids who saw an adult model aggressive behavior (hitting Bobo) mimicked this behavior more than kids who saw a neutral model

Participant-Based IVs
4). Finally, participant-based independent variables (age, gender, personality traits, etc.) focus on participant characteristics. Though not directly manipulated, they can still serve as IV’s
For example, a lot of my research involves sexual harassment.
Research in this areas shows that women tend to see more evidence of sexual harassment than men
This makes sense to you, right!
Every time I run a sexual harassment experiment, I measure participant gender, which I use as an IV in my later analyses

Pop Quiz 1 – Quiz Yourself
Natalie conducts an experiment in which she randomly assigns half the individuals to drink alcohol and half to drink tonic water that they think is alcohol. What kind of independent variable is being used?
A). Experience
B). Participant
C). Physiological
D). Stimulus

Answer 1: C
Natalie conducts an experiment in which she randomly assigns half the individuals to drink alcohol and half to drink tonic water that they think is alcohol. What kind of independent variable is being used?
A). Experience
B). Participant
C). Physiological (correct answer)
D). Stimulus

Part Four
Dependent Variables

Dependent Variables
Dependent variables are the variables we measure (responses)
In research, the experimenter must choose which DV’s to use, how to measure them, and decide how many to use
In this section, we are going to cover a few aspects of DV’s …
1). Selecting the DV
2). Recording and measuring the DV
3). Recording more than one DV
4). Characteristics of a good DV: Validity and Reliability
5). Range effects and the DV

Selecting the Dependent Variable
When selecting a DV, it helps to do a review of the literature to see what variables might be useful to measure
You can compare your study idea to prior studies to see how others measured something, and then use, adapt, or ignore those DVs in your own experiment
For example, doing a literature review can show you several ways to assess the effects of learned helplessness
We can look at frustration as a DV, though we could also look at depression scores, perseverance, apathy, etc. as possible DV’s

Recording and Measuring the DV
Many different ways!
A. Correctness: You might count the number of times that a correct or incorrect response occurs
B. Rate or frequency: You simply count the number of times a response occurs
C. Degree or amount: Find a single number that indicates a degree or amount (like a score on a Likert-type scale)
D. Latency or duration: Measure how quickly a participant responds, or how long the response lasts

More than One DV
Sometimes, you may want to record more than one DV
This will depend on whether additional DVs add to your understanding of a phenomenon
Prior research might help you determine how many DVs to measure.
In the Bobo-doll study, you can measure 1) if the kid hits the doll, 2) how many times the kid hits it, 3) how long the kid continues to hit the doll, 4) how aggressive the hitting is, 5) whether they hit the doll with more than their fists (like using other toys), etc. TONS of possibilities!

Validity
4). Characteristics of a good DVs – There are two important ones!
A. Validity: A variable is valid if it measures what it is supposed to measure
For example, if you measure people’s mathematical ability by giving them a history test, this test is not going to be a valid way of assessing their mathematical ability!
A valid DV must be directly related to the IV and measure the effects of the IV in accordance with the hypothesis.

Reliability
4). Characteristics of a good DVs – There are two important ones!
B. Reliability: A variable is reliable if it produces consistent measurements
For example, the SATs and GREs should tell us something about the skills of the test-taker.
If a test-taker retook the SAT again right after completing it a first time, their score the second time should be very similar to their score from the first time
That is, reliability = consistency
If scores differ dramatically, then the test is not reliable

4). Characteristics of a good DVs – There are two important ones!
We will discuss validity and reliability a lot more next semester in Research Methods and Design II, but keep these concepts in mind, as they will be important ideas when you write your first paper in a few weeks
Reliability and Validity

Pop Quiz 2 – Quiz Yourself
A researcher finds that she can consistently reproduce study results when she repeats the study design. She has found high:
A). reliability
B). validity
C). operationalization
D). theoretical importance

Answer 2: A
A researcher finds that she can consistently reproduce study results when she repeats the study design. She has found high:
A). Reliability (Correct answer)
B). validity
C). operationalization
D). theoretical importance

Range Effects
Note: You are not going to find the following information in your textbooks, but it is important, I want to briefly talk about another issue that may impact the variables in your study …
Range effects occur when variables have upper / lower limits
A. Floor effects refer to average scores being very low (e.g. on a scale of 1 to 9, most put 1 or 2 regardless of their condition)
B. Ceiling effects refers to average scores being very high

Why are Range Effects a Problem?

Range effects can affect data in two ways:
1. They can decrease the variability between two or more treatment conditions
2. They can decrease the variability within each condition
Let’s look at an example demonstration …

Range Effects: Demonstration
We are going to do a quick “demonstration” (though it will be more imaginary”) where I am going to present you with a food word. Your task is to figure out whether I am talking about a fruit or a vegetable
Before the demonstration starts, let’s talk about the DVs and the IVs for the design …

Range Effects Demonstration: The DVs
Fruit or vegetable? The DVs in our demonstration
Imagine you had to determine whether an item on a list of food was a fruit or a vegetable
Here, we can look at several different types of DVs
A. Correctness (the number of fruits you get correct and the number of vegetables you get correct)
B. Frequency (number of times you make a choice)
C. Latency or duration (how quick you are in picking out whether it is a fruit or vegetable)

Range Effects Demonstration: The IVs
5). Range effects occur when variables have upper / lower limits
Fruit or vegetable? The IV’s in our demonstration (2 X 3)
We are going to have two IV’s here
A. First, imagine you have either two seconds or four seconds to decide if it is a fruit or a vegetable!
Note: There are two levels to this first IV
Also, just imagine this IV (I will not time this)
B. Second, the food is presented as either the common name for the food, the botanical name, or as a picture
Note: There are three levels to this second IV

Ready?
The next 14 slides will contain the demonstration, Is it a Fruit or a Vegetable? Be sure to follow along?
Ready? Great!

Malus domestica

1. Is it a Fruit or Vegetable?

If you ONLY saw this botanical name (the botanical condition), would you think it was a fruit or a vegetable?!
Pretty hard, right? Few people would get it correct.

Now, imagine you were in the “common” name condition instead. Try to decide if the food on this next slide is a fruit or vegetable …
Remember to imagine you have 2 or 4 seconds to decide!
Malus domestica

2. Is it a Fruit or Vegetable?

Apple

3. Is it a Fruit or Vegetable?

Easier now, right!

But do you think you would be more likely to get this correct if given 4 seconds to decide or only 2 seconds?

Now, imagine you were in our picture condition …

Apple

4. Is it a Fruit or Vegetable?

Really, really easy now, right! It’s obviously a fruit. I think most participants would know this regardless of whether they had 4 seconds or 2 seconds to decide

Let’s try a few other examples …

5. Is it a Fruit or Vegetable?

Prunus persica

6. Is it a Fruit or Vegetable?

Peach

Prunus persica

7. Is it a Fruit or Vegetable?

Prunus persica
Peach

8. Is it a Fruit or Vegetable?

Prunus persica
Daucus corata sativus

Peach

9. Is it a Fruit or Vegetable?

Peach
Carrot

Prunus persica
Daucus corata sativus

10. Is it a Fruit or Vegetable?

Peach
Carrot

Prunus persica
Daucus corata sativus

11. Is it a Fruit or Vegetable?

Peach
Carrot

Prunus persica
Daucus corata sativus
Citrus limon

12. Is it a Fruit or Vegetable?

Peach
Carrot
Lemon

Prunus persica
Daucus corata sativus
Citrus limon

13. Is it a Fruit or Vegetable?

Peach
Carrot
Lemon

Prunus persica
Daucus corata sativus
Citrus limon

14. Is it a Fruit or Vegetable?

Peach
Carrot
Lemon
Cauliflower
Avocado
Peanut
Olive

Prunus persica
Daucus corata sativus
Citrus limon
Brassica oleracea botrytis
Persea americana
Arachis hypogaea
Olea europaea

15. Is it a Fruit or Vegetable?

Results

Conclusions
The red line on the graph shows a ceiling effect for the “picture” condition.
with all people knowing it is a fruit or vegetable regardless of whether they have 2 or 4 seconds …
The green line on the graph shows a floor effect for the botanical name condition
with almost no one knowing whether it is a fruit or vegetable regardless of whether they have 2 or 4 seconds to decide …
For the “common” name (blue line), people might respond differently on the DV (correctly, faster, more often) with the correct fruit or vegetable answer depending on whether they have 2 seconds or 4 seconds. Results might look like this …

More Conclusions
The point here is that if a study has ceiling or floor effects on the DV, you may not detect any differences in one of your IVs (time in seconds in this example). After all …
There was no variation for the picture condition (ceiling)
There was no variation for the botanical name (floor)
If you didn’t have the common name condition, you may never know there is a difference in 2 versus 4 seconds!
Thus researchers want to avoid making a task either too hard or too easy!

Part Four

Extraneous Variables (Confounds)

Section Overview: Extraneous Variables
In this section, we are going to talk about extraneous variables (which are also called confounds). This section looks at …
1). Defining extraneous variables
2). Examples of extraneous variables
3). Problems associated with extraneous variables
Type I Error
Type II Error

Extraneous Variables
Extraneous variables (EV’s) are variables (other than the IVs) that can influence the results of an experiment (the DVs)
If an EV is present, you have no way of knowing whether your IV is the main reason for your results or whether the EV (confound) is responsible for your results
Confounding refers to a situation in which the results of a study can be attributed to either the operation of an IV or the operation of an EV
EV’s vary between two groups. That is, you might have a control group and an experimental group, but there might be an unknown extraneous variable that impacts participant decisions more than your independent variable.

Extraneous Variables: Example
Imagine you decide to introduce a new hourly work schedule into a factory to see if the new schedule increases productivity
You find two factories (Factory A and Factory B) and have: Factory A workers work new 10 hour shifts, 4 days a week. Factory B workers work old 8 hour shifts, 5 days a week
You find that Factory A is more productive than Factory B!
Does this mean that the new 4 days a week schedule is better than the old 5 days a week schedule?
Think about this question for your next Pop Quiz …

Pop Quiz – Quiz Yourself
Which of the following is a reason for why Factory A might be better than B that is not related to the new hourly schedule?
A). Maybe Factory A was more productive than Factory B even before the new hourly schedule started
B). Maybe any change in schedule boosts productivity (that is, even six days at 6.66 hours per day would boost productivity)
C). Maybe Factory A has other incentives to work hard (like a productivity bonus) that is unavailable to Factory B workers
D). All of these are possible extraneous variables (confounds)

Answer: D
Which of the following is a reason for why Factory A might be better than B that is not related to the new hourly schedule?
A). Maybe Factory A was more productive than Factory B even before the new hourly schedule started
B). Maybe any change in schedule boosts productivity (that is, even six days at 6.66 hours per day would boost productivity)
C). Maybe Factory A has other incentives to work hard (like a productivity bonus) that is unavailable to Factory B workers
D). All of these are possible extraneous variables (confounds). That is, all of these factors might explain changes in the DV. The IV “New Hourly Schedule” may not have an effect at all!

Examples of Extraneous Variables
Let’s do another demonstration …
Instructions: Try to “spot the confound” in the following study designs. If the results actually supported the researchers’ hypothesis, what else might account for the results?
That is, what other potential explanations exist for the observed outcome?

Experiment #1
Perhaps as a result of studying for tests, you and your best friend develop a cold. You both call home, and while your mom suggests that you should eat some chicken soup, your friend’s mom suggests she eat tomato soup. You take your mom’s advice; your friend takes her mom’s advice. Let’s say you get better, but your friend does not. Can you conclude that chicken noodle soup is better for fighting colds than tomato soup?

It is possible, but the problem here is that you have two different types of soup and two different moms giving advice. Is it the type of soup or the specific mom who makes you feel better? Confound!

Experiment #2
A study of therapy effectiveness on substance abuse had the experimental group undergo therapy with a general psychodynamic approach administered by Dr. Elektra, while the other group participated in therapy with a more generalized cognitive approach administered by Dr. Ponder.
Let’s say the general psychodynamic approach is more effective. Is it fair to say this approach is better.
Not really. The doctor administering the therapy differs, so is the therapy approach better or is the specific therapist better?
(Either Elektra should do both, or Ponder should do both)

Experiment #3
A school psychologist hypothesizes that fear negatively influences test performance. To test this, one group of 100 students watches a video containing fear provoking images of spiders crawling on people. The other group of 100 students listens to soothing jazz music. Both groups take a test of their general knowledge.
Imagine they find differences between the groups. What (if anything) is a confound that might better explain the differences?

Experiment #4
The human resources (HR) department of a marketing research company wants to boost the morale of the accounting department. They have decided to try two different types of messages. For employees born January through June, Human Resources will place messages from their supervisor (e.g., “I know that we can count on you!”) on the employees’ paychecks. For employees born July-December, Human Resources will add generic accounting jokes on employees’ paychecks (e.g., “How many accountants does it take to change a light bulb? Depends, are we billing by the hour?”).
Confound One: Source of the message differs (Supervisor vs. HR)
Confound Two: The date of the employee’s birth may affect morale

Uncontrolled variability may make it difficult or even impossible to detect any real effects of the IV between the conditions
Think about the HR versus Supervisor “note” on paychecks. Suppose researchers think supportive acknowledgments are better than jokes, but results show NO differences! Why?
Seeing a supervisor note on their check may make some employees nervous, but HR support doesn’t bother them
Maybe those born July-Dec. have happy personalities to begin with – support simply brings others to their level

Problems Associated with Confounds

Not having a good grasp about the true effect of our IV’s might lead us to make errors in our conclusions. Remember these?
A. Type I Error
In this error, you conclude that your hypothesis was true when in fact it was false
B. Type II Error
With this error, you conclude that your hypothesis was false when in fact it was true
Let’s explore these errors a bit more

Type I and Type II Errors

Type I Error
A. You’ve probably heard the phrase “Type I Error” before
Think about the “Paycheck Note” study. Imagine that you find that employee morale does increase in the “supportive acknowledgement” condition, confirming your prediction
Unfortunately, you might conclude that support increases morale when in fact the real effect is employees knowing that their own supervisor “wrote the note”
Thus you conclude that support works when it really doesn’t work, and you thus have a Type I Error

B. I bet you’ve heard the phrase “Type II Error” before, too
Again, think back to our HR study
Maybe there really is a benefit to the “support” variable, but you may not see a difference between the groups because you did not adequately control the EVs
You thus fail to find significant differences when differences really do exist, giving us a Type II Error

Type II Error

Pop Quiz 3 – Quiz Yourself
A researcher develops a new study method to help students. He gets scores from students at Time 1, teaches them the new study method, and looks at their Time 2 scores. Since students did better at Time 2, he concludes that his new study method had a positive effect. What he doesn’t know is that the students stole the Time 2 answer key, which is why they improved! What error did he make in concluding that his new study method improved student understanding?
A). Type I Error
B). Type II Error
C). He made no error – the study method did help

A researcher develops a new study method to help students. He gets scores from students at Time 1, teaches them the new study method, and looks at their Time 2 scores. Since students did better at Time 2, he concludes that his new study method had a positive effect. What he doesn’t know is that the students stole the Time 2 answer key, which is why they improved! What error did he make in concluding that his new study method improved student understanding?
A). Type I Error (Correct Answer: You said there was an effect when there wasn’t)
B). Type II Error
C). He made no error – the study method did help

Answer 3: A

75

Part Six
Nuisance Variables

Nuisance Variables
Nuisance variables involve unwanted variables that can cause the variability of scores within groups to increase
These might sound a lot like extraneous variables, and this is partially true in the sense that EV’s make it more difficult to determine what is responsible for the observed results.
Yet nuisance variables and EV’s differ in one significant way
Nuisance variables work within conditions (while EV’s work between conditions)
Thus nuisance variables involve spreading scores out within the same group

Example of Nuisance Variables
Recall the Latane and Darley helping study (will someone help another participant going into a seizure?). Can you think of any other non-IV items that might be in this study?
First, consider an extraneous variable. You predict that hearing the seizure alone will spur helping while hearing the seizure in a group of four lessens the likelihood of helping.
If all “alone” participants are women and all “four person” participants are men, any differences that emerge may be based on gender, NOT on group size. An EV!

Example of Nuisance Variables (2)
Now, consider a nuisance variable. Same study, but now you have women and men in all conditions. Excellent! However, you don’t realize that you have several nurses in the “groups of four” condition. They respond really fast while non-nurses in that group respond slow. Average them, and the time to help is in the middle. Take the nurses out and that condition would be MUCH slower to help. It’s an NV!

Example of Nuisance Variables (3)
Other possible nuisance variables within the study:
1. Maybe one of the participants has a sibling who goes into seizures, and is knowledgeable about treating seizures
This spreads out “helping likelihood” within that condition (some help, some don’t)
2. Maybe a participant saw someone helping another person on their way to the lab, and this positive “modeling” behavior spurs that participant to also provide help
Once again, the helping response is spread out within the condition

Extraneous Vs Nuisance Variables
The difference between EVs and NVs is less important than their similarities – if either are present, it makes it difficult to determine exactly what effect your independent variable is having on your dependent variable
As experimenters, we need to control both
In the final section of this lecture, we will focus on several techniques that psychologists use to control the presence of extraneous and nuisance variables
Although these techniques can control both EV’s and NV’s, I am going to simply things by focusing on EV’s

Part Seven
Controlling Extraneous Variables (Confounds)

Controlling Extraneous Variables
In order for a study to have meaningful results, it is important to control extraneous variables (either get rid of them altogether or lessen their impact).
Our basic control techniques involve …
1). Randomization
2). Elimination
3). Constancy
4). Balancing
5). Repeated measures
6). Counterbalancing

Randomization
1). Randomization is a control technique that ensures that each participant has an equal chance of being assigned to any group in an experiment
Because all participants have an equal chance of being in any experimental group, their unique characteristics are equally distributed among all conditions (they cancel each other out!)
Yet even with random assignment, we do not know all of the variables that may influence the study, so it is possible some groups differ from others in meaningful ways.
Thus we can never be completely sure that randomization was effective (but we can be reasonably sure)

Assign participants to treatments randomly by “picking their name or number out of a hat” or by using a random numbers table
Such “control” helps establish cause-effect relationships
Again, this distributes (hopefully) odd characteristics among all of the conditions
You should get nurses in all of Darley and Latane’s study conditions, so extraneous variables are cancelled out – hopefully!
How do we use random assignment?

Does randomization guarantee that groups are equivalent at the start of a study?
A). Yes, because participants in all conditions share similar characteristics (women in both conditions, elderly in both, etc.)
B). Yes, because randomization ensures that all participants have been assigned to the condition best suited to their talent
C). No, it does not guarantee equivalency, though it does give us confidence that the two groups are mostly equivalent
D). No, randomization is not effective at making groups equivalent at all
Pop Quiz 4: Quiz Yourself!

Does randomization guarantee that groups are equivalent at the start of a study?
A). Yes, because participants in all conditions share similar characteristics (women in both conditions, elderly in both, etc.)
B). Yes, because randomization ensures that all participants have been assigned to the condition best suited to their talent
C). No, it does not guarantee equivalency, though it does give us confidence that the two groups are mostly equivalent (correct answer)
D). No, randomization is not effective at making groups equivalent at all
Answer 4: C

Elimination
2). Elimination is a control technique in which the researcher tries to remove all EV’s from an experiment
This may sound easy, but it is much more difficult to do when put into practice!
A. You can check your sample to see if they are doctors or nurses, and then tell them they cannot participate
Although this might get rid of some obvious “helpers”, where do you draw the line? People who have CPR training? People who are pre-med? People married to doctors who have marriage-based “doctor-empathy”?

Constancy
3). Constancy is a control technique in which EV’s are reduced to a single value that is experienced by all participants
In this technique, the experimenter tries to make the conditions identical for all participants (except for the IV) to cut down on potential differences in their experiences
You might use the same room at the same time of day, with the same temperature and the same humidity, and the same researcher to run all sessions
The more “constant” the procedure, the more control the research has over extraneous variables

Constancy: Controlling Characteristics
For participant characteristics, you may try to select a sample of people of similar age, gender, athletic ability, personality, etc.
You could even use twins to try to cut down on differences between participants!

Keep in mind, though, that controlling your population like this means that your results may only apply to those who resemble your study population!

Balancing
4). Balancing is a control procedure that achieves group equality by distributing extraneous variables equally to all groups
If you know what an extraneous variable is in your study, you can make sure that all conditions (the experimental and control conditions) have similar amounts of that extraneous variable
For my sexual harassment studies, I can make sure there are an equal number of men in the control and experimental group; I can also make sure that high and low hostile sexist men are in each group (that is, people who have hostile attitudes about women in the workplace vs. those without such views)

Repeated Measures
5). Repeated measures control techniques expose participants to all levels of the IV. That is, you have participants participate in two or more phases of an experiment.
In these repeated measures studies (also called within-subject designs), the participant acts as their own control group
You can look at their change in score at time one vs. time two. Since you use the same participant twice, participant characteristics are identical (since it is the same participant!)
But sometimes you might encounter order effects in these designs

Counterbalancing
6). Counterbalancing is a procedure for controlling order effects by presenting different treatment sequences, and it goes well with the repeated measures design
This may use within-subject counterbalancing, or presenting different treatment sequences to the same participant
Here, the same participant experiences all sequences (A and then B, followed by B and then A)
A researcher might be interested in looking at memory, so he gives subjects list A to read and then list B. Then, repeat! (they once again getting lists A and B, but this time in the order B – A). That is: A—B—B—A

Within-Subject Counterblanacing
This may use within-subject counterbalancing, or presenting different treatment sequences to the same participant
Here, the same participant experiences all sequences (A and then B, followed by B and then A)
A—B—B—A
Have participants solve the first list (A) followed by the second list (B) and then second (B) and first (A) again

Within-Subject Counterbalancing Issues
But a problem here involves exposing participants to the material on multiple occasions
Once they see a list at time 1, repeating it gets repetitive
They learn something in the first session, which carries over to subsequent sessions.
So here’s a solution – within group counterbalancing …

Within-Group Counterbalancing
An alternative is within-group counterbalancing: presenting different treatment sequences to different participants
Here, half of the sample experience one sequence (some participants read case A and then case B) while others experience the opposite sequence (read B and then A)
Some participants solve hard anagrams followed by easy while others solve easy anagrams followed by hard

96

Two Types of Counterbalancing
6). Counterbalancing
There are two different types of counterbalancing that I want to discuss now, and both have advantages and disadvantages
A. Complete counterbalancing
B. Incomplete (partial) counterbalancing

Complete Counterbalancing
A. For complete counterbalancing, all of the possible treatment sequences are presented
Complete counterbalancing is very easy for a two condition study, with ½ of the participants receiving treatment 1 first and ½ receiving treatment 2 first
Recall my homicide study, with participants reading two cases (one involving second-degree murder, the other involving manslaughter). I counterbalanced them …
Some read the manslaughter case first while others read the second-degree murder case first

Complete Counterbalancing Issues
Yet complete counterbalancing gets a bit more complicated as the # of levels each participant must complete increases!
Example: Homicide study
Now consider four levels
4 X 3 X 2 X 1 = 24 subjects just to get through one single fully counterbalanced order

Partial Counterbalancing
B. For incomplete (partial) counterbalancing, only a portion of all possible sequences are presented. This might be best for complex research designs
Partial counterbalancing includes only some conditions
Might be based on picking conditions at random OR
Might be based on the most important conditions …
Maybe you need 1st / 2nd degree murder to come in slots one OR two only, and voluntary manslaughter / involuntary manslaughter in slots three OR four only

Jeremy would like to do a study in which each participant is exposed to three different treatment conditions (Treatments A, B, and C). In order to do complete counterbalancing, Jeremy will need exactly ________ participants to get through one fully counterbalanced order.
A). 2
B). 4
C). 6
D). 12

Pop Quiz 5: Quiz Yourself!

Jeremy would like to do a study in which each participant is exposed to three different treatment conditions (Treatments A, B, and C). In order to do complete counterbalancing, Jeremy will need exactly ________ participants to get through one fully counterbalanced order.
A). 2
B). 4
C). 6
D). 12

1). ABC 2). ACB 3). BCA

4). BAC 5). CAB 6). CBA

Ok, here are all POSSIBLE combinations!
Answer 5: C

Repeated Measure Design Issues
Although counterbalancing is a useful control technique, it also introduces some issues unique to repeated measure designs
A. Sequence or order effects can influence results
B. Carryover effects may also influence results (possibly becoming confounds or EVs themselves)

Order Effects
A. Sequence or order effects can influence results
The position of a treatment in a series may determine, in part, participants’ responses on later items
Think about these questions…
1. How happy are you with your life in general?
2. How happy are you with your dating / marriage?
Would the order of these questions impact you?
Research shows those who get #1 first see life as less happy than those who get #2 first. Why?
Dating can be a happy thing in general – it “primes” people to see their life in general as happy as well! Order matters!

Order Effect Example
Imagine a study in which participants view a soccer player making successful or unsuccessful passes over 8 trials
½ see 4 successful passes then 4 unsuccessful passes
½ see 4 unsuccessful passes then 4 successful passes
Greenlees, Dicks, Holder, and Thelwell (2007) found that those in the former group rated the soccer player more positively than those in the latter group
The authors thought a primacy effect was involved

Carryover Effects Example
Think about a study involving an eyewitness asked to pick someone out of a lineup twice. Before the second time, they are told a co-witness agreed / disagreed with them
Our independent variable here is the co-witness (did he agree or disagree?)
Our dependent variable is the change in eyewitness confidence
BUT: If we see eyewitness confidence change from Time 1 to Time 2, is it due to the co-witness feedback, or is it due to a carryover effect from Time 1?
There are six factors to consider here …

6 Sources of Carryover Effects
1. Learning
2. Fatigue
3. Habituation
4. Sensitization
5. Contrast
6. Adaptation

Learning
1. Learning involves one treatment “carrying over” to impact performance on a second treatment
Maybe an eyewitness who picks someone out of a lineup is unfamiliar with lineup procedures the first time, but they improve their performance on a second lineup task
Perhaps they got better on the second task because they learned what to expect from doing Time 1, not because they rely on the co-witness feedback!

2. Fatigue from earlier treatments may similarly “carryover” and impact performance in later treatments. In our lineup case, fatigue may make lineup decisions worse at Time 2
Maybe the witness got tired of waiting before making a second lineup decision, and they weren’t as “fresh” when making the second identification as they were during the first identification
It’s an alternative explanation, right!

Fatigue

Habituation
3. In habituation, repeated exposure to stimulus may lead to unresponsiveness towards later stimuli
Our lineup witness may have a lot of fear the first time they ID the culprit in the lineup, but this fear may fade after seeing him repeatedly
They simply get used to that fear, but soon that fear goes away
Is performance affected by a co-witness or by habituation?

4. In sensitization, exposure to one stimulus may make the participant respond more strongly to another encounter with the same stimulus, because they are now sensitized to it
In contrast to habituation, the eyewitnesses’ fear may be high after seeing a lineup the first time. But upon seeing a second lineup, their fear is amped up even more than it was the first time around (they are too sensitive now)

Sensitization

5. Subjects may compare and contrast treatments
For example, our eyewitness may compare lineup 1 to lineup 2, which may affect their behavior
They may not feel as anxious the second time compared to the first, which may increase their confidence on the second task (but not necessarily their accuracy).

Contrast

6. If a subject undergoes adaptation (e.g., dark adaptation), earlier results may differ from later ones
This concept is a little tougher in a lineup study, but you may develop a tolerance to some situations
Think about a drug study where the participant has a high physiological response to the first administration of the drug, but then needs higher doses to reach that same level in future trials

Adaptation

Carryover Effects Can Happen Together
Something to keep in mind is that six these elements are not “mutually exclusive”. That is, several may operate at the same time (e.g. habituation and contrast may occur simultaneously)

A researcher wants to see if air traffic controllers pick up on potential plane collisions better when they have coffee. He gives them coffee halfway into their shift, and measures how quickly they note potential collisions. He finds they are better at the beginning of their shift, and concludes that coffee hurts their decisions. What carryover problem is involved here?
A). Habituation
B). Sensitization
C). Contrast
D). Fatigue
E). Learning

Pop Quiz 6 – Quiz Yourself

A researcher wants to see if air traffic controllers pick up on potential plane collisions better when they have coffee. He gives them coffee halfway into their shift, and measures how quickly they note potential collisions. He finds they are better at the beginning of their shift, and concludes that coffee hurts their decisions. What carryover problem is involved here?
A). Habituation
B). Sensitization
C). Contrast
D). Fatigue (correct answer)
E). Learning

Answer 6: D

116