del Assumptions
o what are they?
      ____
1. E(Yit)   =  m + t(i)
2. outliers?
3. independence?
4. constant variance?
5. normality?
o why important?
o what if violated?
o offer mainly visual/grahic Residual Analysis techniques
  and some tests
o obtain residuals
                         _____
  - e(it) = Y(it) - Y (i.)
  - standardized are 
   Z(it) = e(it) / sqrt (MSE)
   ala Dean  Z(it) = e(it)/sqrt (SSE/(n-1))
  - for us:                                                       Dean 
         Y(it)                                z(it)	L(it) Z(it)
	100			-.48	64	-.55
	93			-.78	76	-.90
	140			1.26	72	1.45
	105			1.16	79	1.34
	45			-1.45	71	-1.67
	85			.29	70	.34
	35			-.10	77	-.12
	40			.12	70	.13
	37			-.01	75	-.02
o proper model?
  - plot  z(it) vs. L(it) and check for pattern
  - plot z(it) vs. Y(it) and check for pattern
o outlier
  - 95% within 2 st.dev
  - recording errors?
o independence
  - if data collected sequentially or in
  proximity to each other the plot z(it) in order
  - check for pattern
o equal variance
  - graphical: plot z(it) in each treatment and
  view ranges
  - ad hoc test is  max(sample variance)
                         -----------------------------
                           min (sample variance)
  is less than 3