(A) Experiment Vs Observational Studies/Sample Survey: ESGC 6112 - Experimental Design - Lecture 1

ESGC 6112 – Experimental Design – Lecture 1
(A) Experiment vs Observational Studies/Sample Survey
Observational studies/sample survey:

(1) data collected under conditions which already exist.
(2) e.g. Max temp, total rainfall, annual income of certain
households, number of telephone calls in a day
(3) Finite population
Experiment:
(1) to see how does the response/output change when we
deliberately changes the conditions.
(2) E.g. In a chemical process change the pressure/temperature, in
an agricultural experiment change the dose of fertilizer; in an
office change the air-cond temperature and note the
productivity; in a pharma experiment, change the level of a
carcinogenic drug and count the number of tumors.
(3) Infinite population of response.
(B) What is Experimentation
 Experimentation is part of everyday life. Eg.: Will leaving 30

minutes earlier than usual in the morning make it easier to find a
legal parking space at work? How about 20 minutes earlier? Or
only 10 minutes earlier? – We’re frequently interested to learn if
and how some measure of performance is influenced by our
manipulation of the factors that might affect that measure.
 An Experiment is an inquiry in which an investigator chooses the

levels (values) of one or more input, or independent, variables, and
observes the values of the output, or dependent, variable(s).
 The purpose of experimental activity is to lead to an understanding

of the relationship between input and output variables, often to
further optimize the underlying process. That is to (i) improve
process yield; (ii) reduce variability; (iii) reduce development time,
(iv) reduce overall cost, etc
 An experimental design is the aggregation of independent
variables, the set of levels of each independent variable, and the
combination of these levels that are chosen for experimental
purposes.
 The core of an experimental design is to answer the 3-part question:
- Which factors should we study?

- How should the levels of these factors vary?
- In what way should these levels be combined?
(C) Guidelines for Designing Experiments
1. Recognition of and statement of the problem

- Develop objectives of experiment & overall objective
- List of specific problems/questions to be addressed & relate
directly to overall objectives
2. Selection of the response variables/dependent variables/output

variables
- ensure that the right performance measure is chosen – the
quantity that is really the important one; eg. In an experiment to
help develop a better cover for a specialty catalog – choice of
dependent variable is the sales volume that results rather than
the brightness of the cover.
- translate output variables to measurable quantities; eg:
subjective – how do you measure taste or appearance; units of
measurement
3. Choices of factors (input or independent variables), level and range

- cause-and-effect diagrams to identify factors affecting the
dependent variable;
- factors such as people factors, equipment factors, environmental
factors, method factors, and materials factors etc…
- minimize the number of factors under study – a higher number
of factors under study is associated with an increased size of an
experiment, this in turn increases the cost and time of the
experiment.
- If the response of the dependent variable to the factor is linear, 2
levels of the factor will suffice, but if the response is non-linear,
one needs more than two – if the factor is measured on a
numerical scale, 3 levels will suffice.
- Identify potential synergy (interaction) between different
factors:
An interaction, or synergy, is the combined effect of factors
considered separately. – the total is > the sum of the parts.
4. Choices of experimental design

- Sample size (no. of replicates)
- Suitable run order
- factorial designs – the experiment comprises varying
combinations of levels of factors, are in general vastly superior
to one-at-a-time designs, in which the level of each factor is
varied, but only one factor at a time.
- Determine how much “fractioning” is appropriate – When there
are large number of combinations of levels of factors, inevitably
only a fraction of them are actually run. Determining which
fraction is appropriate, and the best subset of combinations of
levels of factors to make up this fraction, is a large part of the
skill in designing experiments.
- Consideration of blocking, which is controlling factors that may
not be of primary interest but that if uncontrolled will add to the
variability in the data and perhaps obscure the true effects of the
factors of real interest.
5. Performing the experiment

- Monitor the process carefully
- Pilot/trial run
- The order of running the combinations of levels of factors
should be random.
6. Statistical analysis of the data

- Able to tell whether an observed difference is indicating a real
difference – go through hypothesis testing.
- ANOVA – analysis of variance – whether the level of a factor
(or interaction of factors) influences the value of the output
variable.
7. Conclusion and recommendation

- Verify the conclusions. If we are attempting to determine
which factors affect the dependent variable, and to what degree,
our analysis could include a determination as to which
combination of levels of factors provides the optimal values of
the dependent variable.
- Evaluates whether the results make sense – economic
evaluation (or cost/benefit analysis) of the results.
(D) Experiment-Design Applications in Management
1. Corporate Environmental Behavior
2. Supermarket Decision Variables
3. Financial Services Menu
4. The Qualities of a Superior Motel
5. Time and East of Seatbelt Use: A Public Sector Example
6. Emergency Assistance Service for Travelers

(A) Experiment Vs Observational Studies/Sample Survey: ESGC 6112 - Experimental Design - Lecture 1

Uploaded by

Copyright:

Available Formats

(A) Experiment Vs Observational Studies/Sample Survey: ESGC 6112 - Experimental Design - Lecture 1

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

(A) Experiment Vs Observational Studies/Sample Survey: ESGC 6112 - Experimental Design - Lecture 1

Uploaded by

Copyright:

Available Formats

ESGC 6112 – Experimental Design – Lecture 1

(A) Experiment vs Observational Studies/Sample Survey

Observational studies/sample survey:

(B) What is Experimentation

 Experimentation is part of everyday life. Eg.: Will leaving 30

 An Experiment is an inquiry in which an investigator chooses the

 The purpose of experimental activity is to lead to an understanding

 The core of an experimental design is to answer the 3-part question:

- Which factors should we study?

(C) Guidelines for Designing Experiments

1. Recognition of and statement of the problem

2. Selection of the response variables/dependent variables/output

3. Choices of factors (input or independent variables), level and range

4. Choices of experimental design

5. Performing the experiment

6. Statistical analysis of the data

7. Conclusion and recommendation

(D) Experiment-Design Applications in Management

1. Corporate Environmental Behavior

2. Supermarket Decision Variables

3. Financial Services Menu

4. The Qualities of a Superior Motel

5. Time and East of Seatbelt Use: A Public Sector Example

6. Emergency Assistance Service for Travelers

You might also like