In general, we
do not recommend non-statisticians invest in statistics software.
We beleive researchers should concentrate their valuable time and
efforts on their own area of area of expertise and leave the
statistical analysis to statisticians. At MedStatStudio we can
aid researchers to tabulate their data in an appropriate format to
allow error free transfer to our statistics software model.
Usually, this entails either a web based interface directly to our
database, or a simple coma seperated spreadsheet of data.
general, we consider this question similar to asking which type
of Computer Aided Design Software is best - if you are a
professional you already know, and if you are an amateur no
software will make you into a professional.
What statistics software do you use?
At MedStatStudio, we
find that no statistical software package is adequate for all
phases of statistical analysis. In general, we use MySQL as a
database for the data collection and manipulation, often with a
PHP/HTML web interface. Data analysis is usually performed using
'R a language and environment for statistical computing'. Results
are often returned using the FastRWeb/Rserve packages for direct
viewing on the web interface.
Which textbook in statistics do you recommend?
believe it is usually not necessary nor efficient for researchers
to attempt to gain an in-depth knowledge of statistical theory.
Throughout the research process, consultants at MedStatStudio will
provide references for inclusion in your scientific manuscripts,
and for your reading if you enjoy getting a more in depth
knowledge of the methodology.
Can I have the raw data?
Of course, the raw data is
property of the researcher. At MedStatStudio we empasize
transparency in all phases of the research. Regardless of how your
data is processed - even when using our web interface - we are
always happy to return the raw data to you, and would be happy to
provide the data in a format convenient for you.
What are some hidden costs of using amateur
Time. Researchers time is expensive. Researchers who are
not educated in statistics are often very slow at performing data
analysis. This is particularily true among clinicians who also
engage in research. In contrast, professional statisticians can
often perform the same work in a fraction of the time.
Software. High quality statistics software is costly.
Often in both true dollar value of obtaining the software, and in
terms of time to learn its use. Use of non-statistical software,
such as spreadsheets, which are not suitable for statistical
analysis, both wastes time and leads to inferior results.
Professional statisticians have already invested the money and
time for the software platform, and this cost is not passed on to the customers.
Lost Opportunity. When research is difficult or expensive
to perform, many center choose avoiding research as a solution.
This can lead to many lost opportunities for quality improvement.
Furthermore, even when research is performed the methodology may
be poor: the study design may be incorrect, or the wrong
statistical test may be chosen. This often leads to difficulty in
seperating he noise from the signal and failure to
extract meaningful conclusions from the data.
Failure to Maximize Results. Poor methodology can lead to
studies which are far more difficult and costly to perform than is
necessary. For example, one-factor-at-a-time studies are far less
efficient than factorial studies. Modern statistical methodology,
such as well established quality control methods such as six-sigma
are usually superior to in-house designs. Local researchers often
waste hours and hours inventing a study design, only to find that
a professional statistician could have informed them quickly
Wasted Resources. Poor study design, especially data
dredging and failure to account for multiple hypothesis leads to
false conclusions. History is full of examples of findings
initially though to be important, but later in subsequent studies
found to be insignificant. Often, money and resources are wasted
on these developments before they are eventually abandoned when
they fail to meet expectations. Correct study design at the
outset can drastically lower the chance of making such false
Costly Revisions. Small errors in initial study design can
often lead to the need for costly revisions, or, at its worst,
repeating the entire research study. Even such trivial tasks as
tabulating the data, when done incorrectly, can lead to wasted
time on revisions.