Page 325 - Six Sigma Demystified

Part 3 S i x S i g m a To o l S 305

After discussion of the options, each member should choose his or her top
N options, where N is the number of cards handed out per member. Write one
option on each card, not repeating any option; then write the rank order on
each of the N cards. It’s best to weight using the “larger is better” criterion. For
example, if there are six cards per member, then a member’s top vote should
be rated 6 and the least favorite selection (of the six) should be rated 1.
Based on all the rankings from all the participants, you calculate how often an
option was selected and the ranked weight for each option. For example, six options
for software implementation were evaluated by a team. Since the number of
options is fewer than 20, each team member was provided four index cards. Each
member selected his or her preferred four options and then ranked those options,
with a rank of 1 being the least desirable option (of the four). See Table T.13.

TAbLe T.13 Nominal group Technique
Software Times Total
Options Bob Jim Tim Jill Selected Score
Internal 2 1 4 3 7
Existing 1 4 2 4 4 11
Vendor A 1 1 1
Vendor B 4 3 3 2 4 12
Vendor C 3 2 3 3 8
Vendor D 1 1 1

Interpretation

The result is that we can concentrate our efforts on the items selected most
often or with the highest ranked weight.
In the preceding example, based on both the number of times an option was
selected and the total score, the “Existing” software and that from “Vendor B”
are the two favored options. Note that when the two evaluation methods (total
score and number of times selected) do not agree, or when the scores are close,
it is usually preferred to evaluate more than one leading option.

Nonparametric Test on equality of Means

Nonparametric tests may be used in place of traditional hypothesis tests on the
equality of two means when distributional assumptions cannot be met. A non-