02.03.2013 Views

Downloadable - About University

Downloadable - About University

Downloadable - About University

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.

312 Decisions involving groups of individuals<br />

model which was too complex to be useful. Two methods of combining<br />

individual estimates of unknown quantities are considered below.<br />

Taking a simple average of the individual judgments<br />

First, let us examine the situation where the individual group judgments<br />

can be regarded as being unbiased (i.e. there is no tendency to overor<br />

underestimate), with each person’s estimate being equal to the true<br />

value plus a random error which is independent of the errors of the<br />

other estimates. In these circumstances it can be shown that taking the<br />

simple average of the individual estimates is the best way of aggregating<br />

the judgments. The reliability of this group average will improve as<br />

the group size increases because the random error inherent in each<br />

judgment will be ‘averaged out’. However, each additional member of<br />

the group will bring progressively smaller improvements in reliability,<br />

so that a point will be reached where it will not be worth the effort or<br />

cost of extending the group because a sufficiently reliable estimate can<br />

be achieved with the existing membership.<br />

The situation described above is rarely encountered in practice. Generally,<br />

the members of the group will produce estimates which are<br />

positively correlated. For example, if one member has overestimated<br />

next year’s sales there will be a tendency for the other members to do<br />

likewise. This is likely to occur because group members often have similar<br />

areas of expertise or because they all work in the same environment<br />

where they are exposed to the same sources of information. If there is<br />

a high intercorrelation between the judgments of the group members,<br />

then little new information will be added by each additional member<br />

of the group and there may be little point in averaging the judgments<br />

of more than a small group of individuals. For example, Ashton and<br />

Ashton 2 conducted a study in which a group of 13 advertising personnel<br />

at Time magazine were asked to produce forecasts of the number of<br />

advertising pages that would be sold by the magazine annually. When<br />

the simple average of individuals’ forecasts was used, it was found that<br />

there was little to be gained in accuracy from averaging the forecasts of<br />

more than five individuals.<br />

Taking a weighted average of the individual judgments<br />

When some members of the group are considered to be better judges than<br />

others then it may be worth attaching a higher weight to their estimates<br />

and using a weighted average to represent the group judgment. For

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

Saved successfully!

Ooh no, something went wrong!