Chapter 8

Question 1

What is the essence of experimental design?

Answer 1

exposing participants to 2 different conditions that are identical except for the independent variable of interest, which you’ve manipulated.

Question 2

What must we do with other variables in experimental research? How?

Answer 2

we must keep them constant through various means including direct experimental control and random assignment

Question 3

What can a researcher conclude if the scores for the dependent variable are different between the groups (the levels of the indpendent variable)?

Answer 3

The independent variable caused the changes in the dependent variable.

Question 4

What is the example of controlling for confounds in the crowding and cognitive performance study?

Answer 4

When we are interested in crowding, group size becomes a potential third variable. The researchers want to study crowding not the number of people in the room per se: So the number of people in the room is not a variable of interest. However, in the design, as crowding increases so does the number of people in the room. the independent variable (crowding) and the variable of no interest (number of people) become completely intertwined and their influences cannot be separated. The number of people could explain why the people in the crowding condition did worse on the cognitive test. Maybe they were more distracted by the presence of other people than those in the uncrowded condition (who only have one other person in the room). A solution to this might be to test all participants in groups of ten, and vary crowding by the seating arrangement (spread out or tightly packed). Now you have controlled the number of people in the 2 conditions.

Question 5

What is it called when the results of an experiment can be confidently attributed to the independent variable.

Answer 5

high internal validity

Question 6

What is the simplest possible experimental design?

Answer 6

what thathas only 2 variables

the independent variable (experiment vs control condition)
And the dependent variable.

Question 7

What are the 2 broad classes of an experiment?What are these 2 broad classes also known as?

Answer 7

the between subjects design (also known as the independent group design)
The within-subjects design (also known as the repeated measures design

Question 8

What is the between subjects design?

Answer 8

An experimental design in which different participants are assigned to each level or condition of the independent variable. Also called an independent groups design.

Question 9

What is the within subjects design? how are the experimental and controls distributed?

Answer 9

An experiment in which the same participants experience all levels of the independent variable (i.e., all conditions). Also called a repeated measures design.

The experimental and controls are equally distributed within all participants

Question 10

What is the frist step in designing a between subjects experiment? What must the procedures used do?

Answer 10

to assign participants to the levels of the independent variable.

procedures used must create equivalent groups and must eliminate any potential selection differences

Question 10

What are the three steps of the between subjects design?

Answer 10

(1) obtaining two approximately equivalent groups of participants

(2) introducing different levels of he independent variable to participants

(3) measuring the dependent variable

Question 11

What are selection differences?

Answer 11

Differences in the type of participants who make up each group in a between-subjects experimental design.

Question 12

When doing an experimental design, should the people selected to be in the experimental condition differ in any systematic way from from those selected for the control condition?

Answer 12

NO.

Question 13

Give an example of what owuld result in a selection difference. What would this mean for the study?

Answer 13

If most high-income participants are assigned to the experimental condition, and most low income participants are assigned to the control condition.

This would mean that income is confounded with the independent variable.

Question 14

In a between-subjects experiment, how do researchers ensure that the participants in each condition are approximately equivalent?

Answer 14

using either random assignment, or a matched pairs design.

Question 15

How do you do a matched pairs design?

Answer 15

groups are made equivalent by first selecting pairs of participants who score the same (are matched) on some variable of interest, and then using random assignment to determine which person in which pair will experience which condition.

Question 16

For within-subjects experiments, can we do a matched pair design?

Answer 16

the groups of participants for each level of the independent variable are already equivelent by virtue of the design.

Question 17

Can researchers use 2 different amounts of the independent variable but still have the independent variable be present in both conditions? Examples?

Answer 17

Yes. This is sometimes done.

Examples would be more reward in one group and less reward in the other, or more treatment one group compared to the other.

Question 18

Why must we use the same measurement procedure for the dependent variable for both conditions in an experiment?

Answer 18

so that they can be compared.

Question 18

What is usually used to assess the difference between groups in an experiment? What must be done for us to draw conclusions from the statistical analysis?

Answer 18

a statistical significance test would typically be used

the experiment must be well designed, confounding variables must be eliminated

Question 19

What does random assignment ensure? What does this increase the likelihood of in a large sample of participants?

Answer 19

that the groups will be approximately equivalent in terms of a whole host of participation characteristics, such as income, intelligence, age, or political attitudes

This increases the likelihood that nuanced variables (variables not of interest) related to participant characteristics will be approximately equally distributed across conditions.

Question 20

How many participants do you need in each group for random assignment to establish group equivelence and to detect effects?

Answer 20

This is a complicated question. It is generally safest to collect as many people as you possible can, to maximize the effects of random assignment, to detect smaller effects, and to generalize to the population from which you’ve drawn your sample. As a starting point, consider 50 people per condition for a simple 2 condition between subjects experiment as a bare minimum. You should aim for much larger samples ideally

Question 21

When might a researcher decide to add a pretest measure? What does a pretest do? What do the researchers do with the pretest scores?

Answer 21

When the researcher needs to be extra cautious that random assignment to condition created groups that were equivalent on some particular variable (ex: extraversion).

A pretest measures the variable researchers are concerned about before any experimental manipulation.

The pretest scores for the 2 groups and then compared to ensure that the two groups were approximately equivalent on the critical variable, before the manipulation was introduced.

Question 22

Why would a researcher use a pretest to measure levels of the key dependent variable?

Answer 22

To ensure that participants in bot hconditions had approximately equal levels of the construct presumed to be effected by the independent variable and to more accurately measure the change in the construct.

Question 23

What is a pre-test post -test design? Why is it used?

Answer 23

An experimental design in which the dependent variable is measured both before (pretest) and after (posttest) manipulation of the independent variable.

Makes researchers absolutely sure that the 2 groups were equivelent at the beginning of the experiment for a crucial variable.

Question 24

What is a posttest -only design? When would we typically do this?

Answer 24

A true experimental design in which the dependent variable (posttest) is measured only once, after manipulation of the independent variable.

in between subjects designs

Question 25

is it pretest usually necessary?

Answer 25

No not usually if the participants have been randomly assigned to 2 groups.

Question 26

What are three main reseaons why a researcher may add a pre-test?

Answer 26

(1) to counter problems associated with a small sample size
(2) to select appropriate participants
(3)When participants might drop out of a study

Question 27

What happens to the effectivness of random assignment as sample sizes get smaller?

Answer 27

while random assignment is likely to produce equivelent groups, as sample size decreases, it becomes less likely that the groups will be approximately equal.

Question 28

Why might a researcher want to select participants to include in an experiment through pre-test? What would happen once the target population were identified? What problem might this selection method introduce?

Answer 28

possibly to find the lowest or highest scorers on a measure of smoking, math, anxiety, or prejudice for exmaple.

Once the target populations are identified they would be randomly assigned to the experimental or control group

This may introduce the problem of regression towards the mean which is another threat to the internal validity of an experiment.

Question 29

Why might pre-tests be useful when there is a good posibility that participants will drop out of an experiment?

Answer 29

to ensure that those who are left do not differ across conditions (producing a selection bias)

Question 30

What is the name of the problem that arises from participants chosing to leave a study?

Answer 30

selective attrition

Question 31

What is selective attrition? What is it a threat to?

Answer 31

selective attrition
The loss of participants by way of individuals choosing to drop out of an experiment. Selective attrition is a threat to internal validity when participant dropout results in a difference between conditions on some participant characteristic, causing a confound.

Question 32

What studies are most likely to be affected by participant drop out?

Answer 32

longitudinal studies.

Question 33

What are possible reasons why peopel dropping out would likely not result in differences between conditions? What are some reasons this might result in a difference?

Answer 33

When they drop out for reasons outside of the experimental manipulation such as llness,

if people are dropping out for some reason related to the manipulation, this selective attrition may cause a difference between groups (a selection bias)

Question 34

What is an example of when participants who are leaving for reasons related to the manipulation lead to a selection bias? How might this mess with internal validity?

Answer 34

Imagine a study of a treatment to reduce smoking. One possibility is that the heaviest smokers, those most addicted, in the experimental condition (treatment intervention) might be more likely to leave than those in the control condition (no treatment). Therefore, when the post test is given, only the light smokers would remain in the experimental condition so a comparison between groups would show less smoking in the experimental group even if the program had no effect. Selective attrition then becomes an alternative explanation for the results

Question 35

What do pretest scores allow researchers to determine if there are high selective attrition rates?

Answer 35

allow them to determine whether selective attrition is a plausible alternative explanation for differences between groups, by examining the scores on the crucial variable for those who left and those who remain/

Question 36

What is a disadvantage of a pretest? What might this create?

Answer 36

a pretest can sensitize participants to what you are studying, enabling them to figure out the hypothesis. (it may create a demand characteristic, another threat to internal validity)

They may then react differently to the manipulation than they would have without pretest

Question 37

What becomes difficult when a pretest influences the way participants react to a manipulation?

Answer 37

it is very difficult to generalize these people’s results to people who have not recieved a pretest. AKA the independent variable might not have an effect on the real world where pretests are rarely given.

Question 38

What is a way that researchers can try to avoid or measure the potentially detrimental influence of a pretest on the results of a study?

Answer 38

(1) if awareness of the pretest is a problem, the pretest can be disguised using deception. One way to do this is the embed the pretest in a set of irrelevant measures, so that it is not obvious that the researcher is interested in a particular topic

(2) another way is to examine the influence of a pretest directly, by include the presence or absence of a pretest s another condition, embedding the levels of the independent variable within those 2 groups.

Question 39

Why would someone use a matched pairs design?

Answer 39

to make groups approximately equivalent for between subjects experiments

Question 40

What is a matched pairs design? What is it sometimes also known as? What variables might be involved?

Answer 40

In a between-subjects experiment, a method of assigning participants to conditions in which pairs of participants are first matched on some characteristic (a participant characteristic) and then each member of the pair is randomly assigned to a condition.

Also known as a yoked pairs design

The matching variable can either be the dependent variable itself or another variable that is strongly related to it.

Question 41

For context for the yoked pairs design, what is a yoke?

Answer 41

a piece of wood that connects two animals (such as 2 oxen) working in a pair

Question 42

When a researcher achieves equivelancy of their groups, what is reduced?

Answer 42

the chance of selection bias

Question 43

When is matching most likely to be used in an experiment?

Answer 43

When it is not possible to collect a large sample , making simple random assignment less likely to be successful in creating equivalent groups

Question 44

What are some of the negatives to matching procedures? When do researchers often feel that the matching pairs design is worth while.

Answer 44

They can be expensive and time consuming, depending on the number of characteristics used for matching.

Sometimes these efforts are worthwhile only only when you believe that the matching variables are strongly related to the dependent variables.

Question 45

Do all participants experience all conditions in a within subjects design? What does this ensure?

Answer 45

yes.

this ensures that the groups for each variable are absolutely identical, removing all possibility of a selection bias or differences in participant characteristics between groups.

Question 46

In a within subjects design how are people measured? Can pre-tests also be used in a within subjects design?

Answer 46

participants are measured on the dependent variable after being in each condition of the experiment

Yes.

Question 47

What are 3 advantages of the within subjects design? Give an example.

Answer 47

A major advantage is that fewer participants are needed because everyone participates in all conditions. (Imagine you need 50 participants per conditions. For a between subjects design you would need 100 participants but for a within subjects design you only need 50)

• When participants are scarce, or when it is costly to run each participant, within subjects design can maximize the amount of data collected. (ex: research relying on specialized machinery resulting in costs of hundreds of dollars per participant (fmri), or unique populations that take years f searchign just to find a few people who meet the criteria like people with rare conditions)
• They are exptremely sensitive, able to detect small differences between conditions. When the exact same people are in both conditions less of the variance in the data is attributed to erorr.

Question 48

What is a situation where an experiments would have something known as an error in the scores?

Answer 48

if you are testing the impact that certain reading techniques on memory there will be a lot of variability on the scores of each group because everyone starts with different memory capabilites (some good some bad). This is because people differ. There are individual differences in memory so there is a range of scores in both conditions. Because we are not interested in these individual differences for the purposes of this particular study, this variability can be characterized as a part of the error in the scores (note that it is not wrong or bad just not of interest). We cannot explain this variability and it effects our statistical analyses. Just like when we consider reliability of measures, we want as little error as possible.

Question 49

Why are within subjects designs good for studies that may be susceptible to error in the scores?

Answer 49

because there are no individual differences between the two conditions. In the case of the memory study, you are seeing the affect on the individual which leads to more accurate statistics. This is because in a within subjects design we have measured these systematic individual differences and separated them from random error and from the effect of the independent variable.

In the within subjects design in this case we would be measuring the difference between an individuals scores. not every difference score will be the same so there will still be some unexplained random error but we are better able to detect the effect of an independent variable on the dependent variable if one exists.

Question 50

What is the major challenge with the within subjects design?

Answer 50

the different conditions must be presented in a particular order. Order can become a confound if left unadressed.

Question 51

What is an order effect?

Answer 51

In a within-subjects design, the effect that the order of conditions has on the dependent variable.

Question 52

What are three types of order effects in a within subjects design? What do they pose a threat to?

Answer 52

• a practice effect
• a fatigue effect
• a contrast effect

they pose a threat to the internal validity of the study

Question 53

What is a practice effect?

Answer 53

When participants perform better over the course of a study simply because they are more experienced with the tasks; particularly problematic in within-subjects designs.

Question 54

What is a fatigue effect?

Answer 54

When participants perform worse over the course of a study simply because of diminished effort or fewer resources due to the passage of time; particularly problematic in within-subjects designs.

Question 55

What is a contrast effect? What could be an example of when this would happen?

Answer 55

In a within-subjects design, occurs when participants’ responses in a later condition are affected by a particular experience they had in an earlier condition.

An experiment where the independent varaible is crime and the dependent variable is punishment. After reading about murder in the first condition, reading about theft in the second condition might make this crime of theft seem much milder to participants than it would have if it were presented the first or on its own. People might punish the theif much less harshly when it follows the description of murder as a result.

Question 56

What are the 2 ways to deal with the effects in within subjects designs?

Answer 56

(1) using counterbalancing techniques
(2) ensuring that the time between conditions is long enough to minimize the influence of the first condition on the second.

These strategies can also be combined

Question 57

What is counterbalancing?

Answer 57

A method of controlling for order effects in a within-subjects design, by either including all possible orders of presentation for conditions (complete counterbalancing)or randomly determining the order for each participant.

Question 58

What does counterbalancing do? Does it remove the potential influence of order effects? What does it allow us to determine about order effects?

Answer 58

any effect of order is equally distributed between the two conditions.

No but it ensures the influences are spread out equally among groups.

This way any difference in conditions cannot be due to an effect of order. (order influences both conditions equally)

It also allows us to determien whether any order effects are happening

Question 59

What is the equation for the complete counterbalancing potential orders with 3 variables?

Answer 59

3 (3! = 3 X 2 X 1 = 6)

Question 60

What is the equation for the complete counterbalancing potential orders with 4 variables?

Answer 60

24 (4! = 4 X 3 X 2 X 1 = 24)

Question 61

How many possible orders are there in an experiment with 10 conditions with complete counterbalancing?

Answer 61

3,628, 800

Question 62

What is a technique that would allow people to control for most order effects without having to run ll possible orders (within subjects study)? What technique is this known as more generally?

Answer 62

to employ the latin square design

Partial counterbalancing

Question 63

What is the latin square design? How does it work?

Answer 63

A technique to control for order effects without having all possible orders.

it uses a limited set of all possible orders carefully instructed to ensure that

(1) each condition appears first, second, third, and so on

(2) each condition occurs directly before and directly after each condition exactly once.

Question 64

STUDY THE LATIN SQUARE: ORDINAL POSITIONS OF WITHIN SUBJECTS PAGE 160.

Question 65

What is the point of time intervals between the presentation of different conditions? Can it be over a period of days?

Answer 65

deals with order effects.

Yes.

Question 66

What problem could extended time intervals create?

Answer 66

participants will have to commit to the experiment for a longer period fo time which can make it more difficult to recruit volunteers, and selective attrition may become a problem

Question 67

What are the 2 major advantages of within subjects designs?

Answer 67

(1) fewer participants are required to complete the experiment

(2) removing error variance associated with participant characteristics means that any differences between conditions will be easier to detect.

Question 68

What is a specific situation in which between subjects designs might be preferable?

Answer 68

when experimental procedures are expected to produce relatively permanent changes, such as surgical procedures like the removal of brain tissues or some psychotherapies.

Question 69

What is one way to consider whether a between subjects or a within subjects design is preferabel?

Answer 69

Think about how we would encounter variables in every day situations.

we sometimes encounter variables in a between subjects fashion (ex: how the characteristics of a defendent will affect jury trials, the jurers focus on a single defendent in the trial.

Other things we encounter in a within subjects fashion (employers looking over job applications)