01.06.2013 Views

Statistical Methods in Medical Research 4ed

Statistical Methods in Medical Research 4ed

Statistical Methods in Medical Research 4ed

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

there is none, or because a consideration of order<strong>in</strong>g is deferred until a later stage<br />

of the analysis.<br />

Suppose there are k groups of observations on a variable y, and that the ith<br />

group conta<strong>in</strong>s ni observations. The number<strong>in</strong>g of the groups from 1 to k will be<br />

quite arbitrary, although if there is a simple order<strong>in</strong>g of the groups it will be<br />

natural to use this <strong>in</strong> the number<strong>in</strong>g. Further notation is as follows:<br />

Group<br />

Number of observations<br />

1<br />

n1<br />

2 ... i<br />

n2 ... ni<br />

... k<br />

... nk<br />

All groups comb<strong>in</strong>ed<br />

N ˆ Pk iˆ1 ni<br />

Mean of y<br />

Sum of y<br />

y1<br />

T1<br />

y2 ... yi ... yk<br />

T2 ... Ti ... Tk<br />

y ˆ T=N<br />

T ˆ Pk iˆ1 Ti<br />

Sum of y2 S1 S2 ... Si ... Sk S ˆ Pk iˆ1 Si<br />

Note that the entries N, T and S <strong>in</strong> the f<strong>in</strong>al column are the sums along the<br />

correspond<strong>in</strong>g rows, but y is not the sum of the yi ( y will be the mean of the yi if<br />

all the ni are equal; otherwise y is the weighted mean of the yi, P niyi= P ni). Let<br />

the observations with<strong>in</strong> each group be numbered <strong>in</strong> some arbitrary way, and<br />

denote the jth observation <strong>in</strong> the ith group by yij.<br />

When a summation is taken over all the N observations, each contribut<strong>in</strong>g<br />

once to the summation, we shall use the summation sign<br />

P<br />

i, j<br />

or, <strong>in</strong> the text, P<br />

i, j . When the summation is taken over the k groups, each group<br />

contribut<strong>in</strong>g once, we shall use the sign<br />

P<br />

i<br />

or P<br />

i , whilst summation over the members of a particular group will be denoted<br />

by<br />

P<br />

j<br />

8.1 One-way analysis of variance 209<br />

or P<br />

j .<br />

The deviation of any observation from the grand mean, y, may be split <strong>in</strong>to<br />

two parts, as follows:<br />

yij y ˆ…yij yi†‡…yi y†: …8:1†<br />

The first term on the right of (8.1) is the deviation of yij from its group mean, yi,<br />

and the second term is the deviation of the group mean from the grand mean. If<br />

both sides of this equation are squared and summed over all N observations, a<br />

similar result holds:

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!