Deck 13: Xperimentation

Question 1

To be useful to management, a causal relationship must be persistent enough to make management action worthwhile.

Answer 1

True

Question 2

When there is a high degree of association between an action and an outcome, the order in which the action and the outcome occurred is irrelevant to the question of causality.

Answer 2

False

Question 3

Experiments are used to detect or to confirm causal relationships and to quantify them.

Answer 3

True

Question 4

Experiments are defined as studies in which implementation involves an intervention by the observer beyond that required for measurement.

Question 5

Descriptive studies can usually be used in place of experiments to infer and to confirm the existence of a causal relationship.

Question 6

A high degree of association between two variables, such as frequency of advertising exposure and respondent advertising recall, establishes a causal relationship between the two.

Question 7

The simple "try-it-out" design can be represented by the following notion: X O

Question 8

Laboratory experiments are conducted in a natural setting, where the respondents are usually not aware that an experiment is taking place.

Question 9

A researcher studying the effectiveness of advertising copy tested various groups of randomly selected respondents.After the first interviews, the researcher modified the questionnaire for use with the remaining respondents.The results of this study may be compromised by the maturation effect.

Question 10

In an advertising effectiveness experiment, a researcher is faced with the question of whether sales increased because of advertising or because of economic or competitive changes in the marketplace. This question involves the internal validity of the experiment.

Question 11

The history effect occurs when events external to an experiment affect the responses of the subjects.

Question 12

The history effect primarily threatens the internal validity of an experiment.

Question 13

The internal validity of an experiment depends on the extent to which competing explanations for the results are avoided.

Question 14

For a study to be called an "experiment," it is necessary that the researchers have complete control over all of the relevant variables.

Question 15

Laboratory experiments give the researcher greater control over the experiment but less internal validity than do field experiments.

Question 16

External and internal validity go hand in hand.An experiment lacking in one is probably lacking in the other.

Question 17

In marketing studies, the primary source of history effects is the behavior of competitors.

Question 18

Because the validity issue in any experiment is so important, the researcher is justified in going to any expense to guarantee validity.

Question 19

Field experiments tend to have much greater external validity than laboratory experiments.This means that to a greater extent their causal inferences can be generalized from the experimental environment to the actual marketing environment.

Question 20

A causal relationship between an action and an outcome has been confirmed whenever there is evidence of a strong association between the two and evidence that the action preceded the outcome.

Question 21

A researcher studying the effects of several control variables is faced with the problem of increasing sample size and increasing cost.The usual solution is the before-after "try-it-out" design.

Question 22

Use the following information for questions
Members of an anti-smoking group needed to test the effectiveness of
an education pamphlet which they intended to circulate among the
public.They first interviewed a group of respondents to find out
their attitudes towards the detrimental effects of smoking.Next,
the respondents were given the pamphlet to read.Finally, the
respondents were interviewed again, for the second time that day.

-This study is an example of the before-after "try-it-out" design.

Question 23

Maturation involves changes in respondents that are a consequence of time, such as aging, or getting hungry or tired.

Question 24

This figure demonstrates an interaction effect between the two control variables, advertising and price:

Question 25

A marketing research manager, who wants to reduce the cost of experimentation, decides to use the Latin square design.This decision is based on the assumption that any interaction between variables can be ignored.

Question 26

A matched control group may be represented as M X 0₁ -------- M O₂

Question 27

Use the following information for questions
Members of an anti-smoking group needed to test the effectiveness of
an education pamphlet which they intended to circulate among the
public.They first interviewed a group of respondents to find out
their attitudes towards the detrimental effects of smoking.Next,
the respondents were given the pamphlet to read.Finally, the
respondents were interviewed again, for the second time that day.

-The internal validity of the smoking study above is threatened by the mortality effect.

Question 28

A non-matched control group may be represented as X O₁.

Question 29

Randomization should always be used when the number of dimension on which groups are to be matched is small two or three..

Question 30

One approach to reducing selection bias is to match the control and the experimental groups.

Question 31

As the sample size decreases, the degree of matching achieved by randomization will increase.

Question 32

A researcher is using a full factorial design to test the effects of four variables, each of which has two levels.The least number of conditions to be tested is 2 x 2 x 2 x 2 or 16.

Question 33

In the following before-after randomized control group design, the output of interest would be the difference between 0₁ - 0₃ and O₂ - 0₄. R 0₁ X 0₂ --------------- R 0₃ 0₄

Question 34

Use the following information for questions
Members of an anti-smoking group needed to test the effectiveness of
an education pamphlet which they intended to circulate among the
public.They first interviewed a group of respondents to find out
their attitudes towards the detrimental effects of smoking.Next,
the respondents were given the pamphlet to read.Finally, the
respondents were interviewed again, for the second time that day.

-Because the respondents in the smoking study above had become familiar with the measuring instrument, the results of the study are threatened by the instrumentation effect.

Question 35

Use the following information for questions
Members of an anti-smoking group needed to test the effectiveness of
an education pamphlet which they intended to circulate among the
public.They first interviewed a group of respondents to find out
their attitudes towards the detrimental effects of smoking.Next,
the respondents were given the pamphlet to read.Finally, the
respondents were interviewed again, for the second time that day.

-The internal validity of the smoking study above is threatened by the before-measure effect.

Question 36

The following is a Solomon Four-Group design.Because groups III and IV do not have a before-measure, they cannot be regarded as experimental groups. R 0₁ X 0₂ --------------- R 0₃ 0₄ --------------- R X 0₅ --------------- R 0₆

Question 37

Direct mail tests of advertising appeals are an ideal setting for the randomized control group design.

Question 38

The factorial design allows a researcher to determine the effect of two variables in one experiment.It can determine main effects and interactive effects.

Question 39

The before-after randomized control group design eliminates the before-measure effect.

Question 40

In non-matched control group designs, one group is exposed to treatment while another is not.The response variable is measured in the two sample groups at the end of the experiment.This solves the problem of selection bias.

Question 41

Trend studies are most useful when a researcher is testing hypotheses concerning an individual's response to a control variable.

Question 42

Consider the following table from data on advertising recall and attitudes: $\chi$ ² .25 $\chi$ ² test Significant at .001 level $\phi$ .27 The logical implication of the table is: advertising -> attitudes.

Question 43

Causation is established when two variables vary concomitantly.

Question 44

Strong association is sufficient evidence for predicting the dependent variable.However, if management desires to influence the level of the dependent variable, then a better understanding of the nature of the relationship between all variables of concern is necessary.

Question 45

If it is impossible for a researcher to find an unaffected subgroup to use for a control, the use of a one-group before-after design might be unacceptable because of the reactive effect of the before observation.In such cases, a separate random sample is sometimes interviewed for the before measure.

Question 46

Even if a true experiment is conducted, there is no way to adequately control for an unidentified intervening variable.

Question 47

The value of the correlation coefficient, a statistical measurement of association, is sufficient evidence to establish or reject causality between two variables.

Question 48

One reason that an association between advertising and attitudes may be explained by an ownership variable is the difference between the composition of ownership groups with positive and negative attitudes.

Question 49

In the absence of a true experiment, there is no way to adequately control for an identified intervening variable.

Question 50

All the hypotheses in a descriptive study must be fully developed before the data analysis stage.

Question 51

Continuous panel studies are especially useful in gaining insights into choice behavior for product categories in which stable sales often obscure large but compensating gains and losses of individual buyers.

Question 52

With the Solomon Four Group design, it is possible to measure the interaction of the treatment and before-measure effects by looking at 0₂ - 0₄) - 0₅ - 0₆)).

Question 53

If the composition of two ownership groups with respect to positive and negative attitudes) is identical, an association between advertising and attitudes would not be affected by the introduction of the ownership variable.

Question 54

Progressive Marketers Incorporated found no third variable involved in the association between sales and advertising.It can, therefore, be concluded that the relationship is of the form: advertising $\rightarrow$ sales.

Question 55

Trend studies and continuous panel studies are two examples of time series designs.

Question 56

The value of the correlation coefficient, a statistical measurement of association, is necessary evidence to establish causality between two variables.

Question 57

Progressive Marketers Incorporated has found a strong relationship between the number of automobile accidents in the quarter just ended and its sales level for the upcoming quarter.Their marketing researcher has, however, warned that the relationship is spurious and can be explained by an intervening variable.Under the circumstances, automobile accident data are of no use to the management in planning their strategy for the upcoming quarter.

Question 58

Until a causal link has been established between two variables, a researcher who finds a high level of association between the variables will not be able to make predictions.

Question 59

The before-measure effect is a potential threat to the internal validity of continuous panel data.

Question 60

If variables X and Y are strongly associated, then a causal relationship between X and Y must exist.

Question 61

The advantage of randomization in experimental control is that it will control for both known and unknown variables.

Question 62

A test-retest or parallel forms approach will help determine

A)construct validity
B)concurrent validity
C)reliability
D)face validity
E)none of the above

Question 63

Which of the following designs corrects for all forms of biases in responses?

A)one shot framework
B)pretest-posttest descriptive framework
C)pretest-posttest causal framework
D)Solomon four group design
E)none of the above

Question 64

An advantage of ___________in experimentation is that no knowledge is necessary of extraneous variables that could possibly influence the results.a

A)randomizing
B)eliminating
C)equalizing
D)balancing

Question 65

The main advantage of experimental research over non experimental research is

A)it is easier to control the dependant variable
B)cause and effect relations can be tested
C)interaction between variables do not confound the study
D)one can collect more accurate information on the problem
E)none of the above

Question 66

An intervening variable causes spurious association.

Question 67

From a data analysis viewpoint, it is impossible to distinguish between an intervening variable and a spurious association.

Question 68

Variables, other than the manipulated variable, that affect the response of test units are called extraneous variables.

Question 69

Classical designs can be further categorized into three groups: Pre-experimental, Quasi-experimental, and true experimental groups.

Question 70

The group of subjects exposed to the experimental treatment is termed as control group.

Question 71

Which of the following is true about history effects?

A)They involve changes in respondents that are a consequence of time.
B)They result from events internal to the research design.
C)They are not diminished by using a nonmatched control group design.
D)The longer the time period involved in the experiment, the greater the likelihood that history effects will account for
Observed results.
E)None of the above

Question 72

One of the advantages of a nonmatched control group design is the control of

A)interaction and maturation effects.
B)maturation and instrumentation effects.
C)history and maturation effects.
D)instrumentation and mortality effects.
E)selection bias.

Question 73

Which of the following is more likely to provide evidence of a causal relationship between two variables?

A)an experiment
B)descriptive research
C)exploratory research
D)cross-sectional analysis
E)non experimental research

Question 74

The "try-it-out" design

A)is not considered to be an "experiment."
B)is not likely to be threatened by history effects.
C)applies the experimental treatment to a subject or group and measures the results.
D)can be represented by the notation O X.
E)none of the above.

Question 75

To infer a causal relationship, which of the following types of evidence does a researcher need?

A)Evidence of a strong association between an action and an outcome.
B)Evidence that the action preceded the outcome.
C)Evidence of a high level of internal validity.
D)Evidence that there are no strongly competing explanations for the relationship between an action and an outcome.
E)All of these

Question 76

An experimental design which helps identify all forms of the "pre-test" bias is

A)before-after randomized control group
B)traditional matched-group design
C)time series
D)solomon four-group
E)all of the above

Question 77

When many variables are involved and theory is undeveloped, the only way to identify interactions is to perform cross-tabulation on all conceivable combinations of variables.

Question 78

The question of whether the results of an experiment apply to the real world refers to

A)external validity.
B)internal validity.
C)history effects.
D)interaction effects.
E)none of the above.

Question 79

One way to address causality is to consider the time lag between cause and effect

Question 80

Randomization is the procedure in which the assignment of subjects and treatments to groups is based on some criteria.