Response from Tania Lang to Comments on 2004 Salary Survey
15 Jan 2005 - 4:05am
See comments prefixed with <Tania> in response to your questions.
Myhill, Carl S (GE Energy) wrote:
> ...Whilst there are some interesting data in the report, I am pretty appalled
> with the report as it has been presented, particularly the fact that so much
> of the data are not represented. > > Here are a couple of points which sprang out: > > - the word 'average' is used throughout the report. Which average is being > used? Mean? Median? Mode? Presumably the Mean is being reported but this is
> not clear because often with salary data the Median is used, particularly by
> governments. e
<Tania> As per the UPA salary report that you referred to, we used the
term "Average" rather than "mean", "median" or "mode". I believe the
word "average" used throughout the report is equivalent to "mean".
Average is the formula that I used in Excel (i.e. AVERAGE(E5:E7))
> > - there is very little attempt to provide any 'descriptive statistics' > saying anything meaningful about the spread of the data. Not even basic > things like the highest and lowest values. What is the standard deviation? > What are the quartile ranges? Perhaps a picture of the range of values > reported? The UPA did a pretty good job of this in their 2000 report >http://www.upassoc.org/upa_publications/upa_voice/survey/2000_survey.html > (I seem to recall the 2003 version being very good as well, but can no > longer find it online). <Tania> Carl, Pabini and I are not statisticians and hence detailed
statistical analysis was not performed. Basic statistics are provided
and results where anomalies have been adjusted are noted in the comments
(for example, areas of small sample sizes). As highlighted in the
Introduction, we request that readers "...use the information in this
report only as a general guide. In many instances, sample sizes were
small-therefore, the data may not be generally representative." We also
mention in the methodology section, that the data was cleansed and any
obviously incorrect values that would skew the results were excluded
e.g. data from students or casual workers with extremely low salaries.
> Salary surveys are a very useful tool for us as individual professionals > looking at our own situations, and also when we're recruiting. I am > extremely disappointed with this effort, which gives me very little > information I can actually use. <Tania> Carl, thank you for your comments. I am sorry you feel unable to
use this data. Perhaps if you'd like to fund another effort we could
accommodate you. The Salary report was done on a voluntary basis with
limited resources. As independent consultants we had neither the time or
resources available to conduct the detailed statistical analysis you
seem to require. Certainly given time, resources and funding we could
have produced something to satisfy your specific business needs. We hope
the report has been of some value to the broader audience.
<Tania> As an independent consultant I have had an interesting year,
worked full time, organized several local HCI events, renovated a house
and had a baby in addition to spending several weeks working with Pabini
on gathering data and producing this report. We've been happy to do this
even though we reap little reward apart from increased traffic to our
sites. We hope the data presented has been of some use to the usability
<Tania> Since you have vast resources of the GE corporation at your
disposal you may not appreciate that independent consultants don't have
the resources you possess to achieve and report on the level of
statistical analysis of the salary survey data you require.
<Tania> I would be absolutely delighted if you would like to volunteer
your personal time and/or funding (presumably from your HR department
budget) to conduct a 2005 salary survey. Maybe then you could produce
what you would consider a less "appalling" report that would be of more
value to the usability community.
> PS Perhaps the authors could publish the raw data so we can slice it up > ourselves <Tania> Carl, to get the raw data into a "presentable" format for
processing that would satisfy the statisticians amongst the community
would involve fusing data spread across worksheets and files.
Unfortunately, I do not have the time or resources to do this. However,
I am happy (within reason) to answer any specific questions or provide
specific raw data on an individual basis.