Producing Data II

1
Producing Data II

• Experiments

2
Explanatory and Response Variables

• A response variable measures what happens to the
  individuals in the study.
• An explanatory variable explains or influences
  changes in a response variable.
• In an experiment, we are interested in studying
  the response of one variable to deliberately
  imposed changes in the other (explanatory)
  variables.

3
Experiments Vocabulary

• Subjects
• Individuals studied in an experiment
• Factors
• The explanatory variables in an experiment
• Treatment
• Any specific experimental condition applied to or
  imposed upon the subjects If there are several
  factors, a treatment is a combination of specific
  values of each factor

4
Case Study
Effects ofTV Advertising
Rethans, A. J., Swasy, J. L., and Marks, L. J.
Effects of television commercial repetition,
receiver knowledge, and commercial length a test
of the two-factor model, Journal of Marketing
Research, Vol. 23 (1986), pp. 50-61.
Objective To determine the effects of repeated
exposure to an advertising message (may depend on
length and how often repeated)
5
Case Study

• Subjects a certain number of undergraduate
  students
• All subjects viewed a 40-minute television
  program that included ads for a digital camera

6
Case Study

• Some subjects saw a 30-second commercial Others
  saw a 90-second version
• Same commercial was shown either 1, 3, or 5 times
  during the program
• There were two factors length of the commercial
  (2 values), and number of repetitions (3 values)

7
Case Study

• The 6 combinations of one value of each factor
  form six treatments

8
Case Study

• After viewing, all subjects answered questions
  about recall of the ad, their attitude toward
  the camera, and most importantly, their intention
  to purchase it the response variable.

9
Comparative Experiment DesignSubjects ?
Treatments ? Response

• Experiments should compare treatments rather than
  attempt to assess the effect of a single
  treatment in isolation so as to find out which or
  intended treatment is better.
• Problems when assessing a single treatment with
  no comparison
• Conditions better or worse than typical
• Lack of realism
• Subjects not representative of population
• Placebo effect (power of suggestion)

10
An Industrial Experiement

• The yield of a product produced by a chemical
  reaction depends on the temperature and the
  stirring rate in the vessel in which the reaction
  takes place. An engineer investigates the effects
  of combinations of two temperatures (50C and 60C)
  and three stirring rates (60, 90, 120 rpm) on the
  yield. She will process two batches of the
  product at each combination of temperature and
  stirring rate.
• Subjects, response variable? How many factors and
  treatments? How many subjects?

11
Comparative Experiment Design

• How to design an experiment to study the impact
  of nicotine patch (by a drug company) on the
  cessation of smoking.
• Find a number of smokers, divide them into two
  similar groups. Individuals in one group are
  instructed to use nicotine patch Individuals in
  the other group use a dummy patch (placebo)
• The latter group is called control group.
• Response variable?

12
Control Group

• Without control group in the nicotine patch
  experiment, how can you decide the effectiveness
  of the nicotine patch? It couldve been that the
  participants knowledge that hes been treated by
  a doctor improves his spirit and determination to
  stop smoking (placebo effect or power of
  suggestion)
• Control group creates a comparison group and
  enables us to control the effects of lurking
  variables on the outcome. Control group should
  consist of similar individuals as the other group
  (achieved by randomization).
• Example time spent on study and students final
  grades (hypothetical experiment) smart group and
  not-so-smart group.

13
Randomized Comparative Experiments

• In the nicotine patch experiment what happens if
  we assign old smokers to try out nicotine patch
  and young smokers to take dummy patch?
• To avoid bias, when we assign individuals
  (subjects) for treatment, we want to make sure
  that the assignment of individuals to treatments
  is random.
• Each treatment should be applied to similar
  groups or individuals (removes lurking vbls).
• Assignment of treatments should not depend on any
  characteristic of the subjects or on the judgment
  of the experimenter.

14
Drinking Tea Regularly Slows the Growth of
Cataracts

• How to design an experiment to test this claim?
• Test on rats. Preparation inject 18 rats with a
  substance that causes cataracts. Then feed these
  rats tea extracts.
• One group of 6 fed black tea Second group of 6
  fed green tea Third group placebo.
• Response variable growth of cataracts over next
  two months.
• Outline of the design

Treatment 1 Black tea extract
Group 1 6 rats
Compare Responses the growth of cataracts
Random Assignment by SRS
Treatment 2 green tea extract
Group 2 6 rats
Treatment 3 control
Group 3 6 rats
15
Completely Randomized Design

• In a completely randomized design, all the
  subjects are allocated at random among all of the
  treatments.
• Can compare any number of treatments (from any
  number of factors).
• Effects of TV advertising page 200 (60students)

Treatment 1 30 seconds,once
Group 1 10 s
Compare effects of TV advertising
Group 2 10 s
Random Assignment
Treatment 2 30 seconds, 3 times
Group 3 10 s
Treatment 3 30 seconds, 6 times
Group 4 10 s
Treatment 4 90 seconds, once
16

• Control use a control group if no natural
  comparison treatment is available.
• Randomize remove assignment bias.
• Use enough subjects large sample size lows the
  possibility of extreme situations, which are not
  representative

17

• If the response indicates a result that doesnt
  happen by accident, its statistically
  significant.
• Strong association doesnt imply causation In a
  well-designed experiment, a statistically
  significant association does imply causation.

18
Double-Blinded Experiments

• If an experiment is conducted in such a way that
  neither the subjects nor the investigators
  working with them know which treatment each
  subject is receiving, then the experiment is
  double-blinded
• To control response bias (from respondent or
  experimenter)

19
Double-Blinded Experiment

• Example An experiment testing the effectiveness
  of a pill on ulcer with a control group should be
  double-blinded
• In an experiment testing whether listening to
  Mozart or relaxation tape with a control group
  improves performance, individuals get all three
  treatments in random order the subjects know
  what they listen to but the investigators dont
  know in which order single-blinded

20
Matched Pairs Design

• Compares two treatments
• Technique
• Choose pairs of subjects that are as closely
  matched as possible
• Randomly assign one treatment to one subject and
  the second treatment to the other subject
• Sometimes a pair could be a single subject
  receiving both treatments
• Randomize the order of the treatments for each
  subject
• Example Coke or Pepsi? Page 210. Randomized
  design give subjects randomly Coke or Pepsi
  Matched Pairs Design give each subject both Coke
  and Pepsi without markings.

21
Block Design

• A block is a group of individuals that are known
  before the experiment to be similar in some way
  that is expected to affect the response to the
  treatments.
• In a block design, the random assignment of
  individuals to treatments is carried out
  separately within each block.
• A single subject could serve as a block if the
  subject receives each of the treatments (in
  random order)
• Matched pairs designs are block designs

22
Pairing or Blocking
Men, Women, and Advertising, Page 211

• Compare effectiveness of three television
  advertisements for the same product, knowing that
  men and women respond differently to advertising.
• Three treatments ads (need three groups)
• Two blocks men and women
• Randomizing ignores the difference between men
  and women toward advertising

23
Pairing or Blocking
Men, Women, and Advertising
24
Pairing or Blocking

• Pairing or blocking
• To reduce the effect of variation among the
  subjects
• Different from a completely randomized design,
  where all subjects are allocated at random among
  all treatments