Introduction to the Design and Analysis of Experiments
o Another statistics class....about what?
  - pick a # from 0-9
o How/what does DOE do for us as statisticians and people?
  - Seaworld Funcard
o Let’s talk about our won inquiry: bead jewelry
o the vocabulary
  - Factor treatment
  - Level
  - fixed vs. random
  - unit
  - response, measure, observation
  - factorial experiment
  - observational vs. experimental
  - replication
  - repeated measure
  - blocking
  - randomization
  - the model
  - assumptions
  - covariate
o “No method of analysis can save a badly designed experiment” Dean and Voss 
  on p. 13
  - ear temp 
o Designs
  - completely randomized factorial design
  - randomized comple block
  - randomized incomplete block
  - more than 1 block; nested
  - split plot



Bead Jewelry Making

Problem:  A manufacturer of bead rings can make any of 
3 sizes of rings.  Each contains 10 beads and materials costs of $1.  Each can be
sold for $5 and are equally demanded by the consumer public.  Which product
should she manufacture?

What principle question needs to be answered to address this problem?
(a)That is, what is our objective?

What causes us not to hit our target or ideal related to our objective each and
every time?
(b) Identify sources of variation.

When conducting an experiment who gets to do what?
(c) Choose a rule of assignment.

(d) What gets measured, how, will it be difficult; what can go wrong?

(e) Try the process out on a small number of persons -- pilot experiment.

Hypothesize a relationship between the experimental outcomes.
(f) That is, specify a model.

Once data is available how will it be analyzed?
(g) That is, outline the analysis

How many items/persons must be viewed?
(h) Determine the number of observations

Go over your plan again.
(i) Review and revise.

Other Considerations after settling on a Design

Is there a perfect bead size?  how would you find it? What practical
considerations should you consider?
How might the DOE change if we had a different cost for the beads?
There were other factors contributing to differences in times measured, should
we be interested in those in our jewelry production process?
If I had conducted this as an observational study by letting people pick out
their bead size and then done testing what unintentional effects might have
occurred?  Hint: people might pick beads which best fit their fingers.
Explain what a repeated measure would be in this experiment
What might happen if we did repeated measures?
Would it be good or bad to use a person as her own control, i.e.. block on
person?
Discuss measurement error 
What might we block by?
Did we randomize?
DATA RECORDING SHEET

Size: ________	Finger Length: _____________
                           First finger on left hand.  Measure from crease at palm, up the
center to tip 			with the ruler below. Report in centimeters (bottom
scale). 

Color: _______	Time: ____________________
		After stringing your 10 beads, look up at me marking time on the
board and record 			the number of seconds which elapsed while
you strung your beads.
                           


DATA MATRIX