Table B-III

DOE-SPEC-1142-2001

Table B-III

Residuals for Day 5

Day5 controls

-0.1753

0.4976

0.1991

-0.4286

Day5 controls

0.0307

0.0757

-0.0306

-0.2510

Day5 controls

-0.4056

0.4434

0.1815

-0.6916

Day5 Be1

-0.0295

0.0321

-0.0726

0.0295

Day5 Be10

0.0420

0.8046

-0.7850

-0.0419

Day5 Be100

0.1963

0.2029

-0.1963

-0.6022

The median absolute value of the residuals is 0.1963, and

Sm = 1.48 * 0.1963 * √(24/20) = 0.3183,

B.5 Calculate the standard error of the Ln(SI):

StErr[Ln(SI)] = Sm * sqrt(pi/2) * constant.

For this assay design (12 control wells and 4 beryllium stimulate wells)

the constant is 0.577. If, for example, 8 control wells had been used,

the constant would be 0.612 (see EHP96 page 960 for details).

On Day 5 the standard error [Ln(SI)] = 0.3182531*1.253*0.577=0.230.

B.6 Divide the Ln(SI) by its standard error to obtain the standardized Ln(SI):

standardized Ln(SI) = Ln(SI) / standard error[Ln(SI)] .

For Day 5 Be100 the standardized Ln(SI) = 0.7197/0.230 = 3.13. Dividing each Ln(SI) by

its standard error results in a statistic that is in "standard measure," having mean 0 and

standard deviation 1 (see B.10.6, p. 631), i.e., a standardized Ln(SI).

B.7 The results of the calculations for each treatment on Day 5 and Day 7 are recorded.

B.8 Assumptions and Alternative Methods of Analysis for the Tritiated Thymidine BeLPT.

B.8.1 The LAV analysis is based on the assumptions that:

Ln of the well counts are normally distributed;

standard deviations of Ln counts are constant within harvest days;

multiple outliers may be present in the Ln well counts; and

if "responder cells" are present, an increase in cell proliferation relative to the

control wells will occur in cultures with beryllium.