Thanks very much to the following (in alphabetical order) who replied to
my query (at the end of this mail):
Brian G Miller
The responses (in 'random' order) were:
From the information you have provided I can deduce:
1. Probability of being involved on the pilot and use the
technology for their normal work =0.4*0.27=0.108
2. Probability of being involved on the pilot and not use the
technology for their normal work = 0.4*0.73=0.292
3. Probability of not being involved on the pilot and use the
technology for their normal work=0.6*0.27=0.162
4. Probability of not being involved on the pilot and not use the
technology for their normal work =0.6*0.73=0.438
So if the question "Are people who have been involved in pilots more
likely (than those who haven't been involved in pilots) to use the
technology in their daily work?" you need to compare options 1 and 3
above. That is 1. 10.8% with 3. 16.2%. So given the sample size
(randomly selected from the population) is 176, we can carry out a one
sided proportion test to determine if it is the case that in the
population people who were involved in the pilot were more likely to use
the technology at work. To do so I have attached an excel spreadsheet
with tests the two proportions under the 5% level of significance. From
the spreadsheet you can conclude that there is no evidence to suggest
that those who have been involved with a pilot are more likely to use
the technology at work than those who haven’t.
You have 2 classification factors, pilot/no pilot and your binary
outcome. You have a 2x2 table, and can test for association by a
chi-square test. You can/should also quantify the association, e.g. as
an odds ratio.
All this is calculated by simple arithmetic, and is easy to do in Excel.
I gather that you have a group of people who have NOT been involved
in pilots, and a group who have used pilots.
For each person, you can tell if they are using the technology in
their present work, or not. No partials, here - you can tell yes, or
no. Before you review the list.
Looking at each group, you can determine a percentage of the group
that uses technology in present work.
At this point, it looks to me like a comparison of two proportions.
A t test would do that job.
Suppose that you measure the _degree_ of using the new technology in
normal work. On say scale of say 0 to 10. Now you would have a
reasonable measure, and it (probably) would be close enough to use a
regular t test.
You need to cross-tabulate the two binary variables (i.e. involvement in
pilot and using technology as part of normal work) and conduct a two-way
chi-square test of assocation on the resulting frequencies. I believe
CHITEST command will do this.
You need to organise your data so that it is in the
form of a 2 x 2 contingency table; something like
Yes a b
No c d
where a, b, c and d sum to 176. The test you need is
then one for association in a contingency table. I
have no idea whether this can be done in excel, but there
are lots of online programs which will do it. See
and go to Contingency tables.
You want a contingency table I believe:
(think it is called a "Pivot table" in Excel)
with columns "pilot", "normal" and rows "yes", "no" (would use
technology) and then in the 4 cells the count of people
1) who did the pilot and would use the tech
2) who did the pilot and wouldn't
3) who use it normally and would
4) who use it normally and wouldn't
then look at the Pearson statistic for the table as to whether there is
any relationship between pilot/normal and yes/no
[My original mail]
Ken Masters wrote:
> Hi All
> Please can I be advised on the right tests to run (in Excel, if
> I have a small sample (176) that has been involved in using a particular
> 40% have been involved in pilots of that technology
> 27% use that technology as part of their normal work
> I want to determine if there is any relationship between being involved
> in a pilot, and using the technology as part of their normal work,
> because I have a fair number of people who have been involved in pilots
> who do not use the technology as part of their work, and a fair number
> who use the technology as part of their normal work, but have never been
> involved in any pilots. The type of question I’m looking to have
> answers for is "Are people who have been involved in pilots more likely
> (than those who haven't been involved in pilots) to use the technology
> in their daily work?" If these tests can be run in Excel, that would
> really be useful.
> Any assistance will be much appreciated.
> Ken Masters