17 October, 1997
17 October 97
I spoke to the weather people today. They told me that it may be a week
or more before they have the forecast back online. Today was a fair
weather day. It was cold and overcast most of the time but it cleared
substantially in the evening.
I was able to do another protein assay on the starfish tube feet. These
data were better, but the yield is not what I'd like it to be. It's very
possible that I will have to modify the protocol substantially and treat
the tube feet with some chemicals which will free up more protein when I
homogenize the tissue. It will be valuable to repeat today's experiment
with the same procedure (at least one more time) to make sure the numbers
are reproducible.
When samples are taken, and measurements made, the mean (average) is a
valuable measure of central tendency. But how can one know the degree to
which the mean is a true representation of the population being sampled?
One way is to calculate a statistic called the coefficient of variation.
This number is expressed as % and represents the standard deviation divided
by the mean. The standard deviation is the average of the deviation of the
measurements from the mean. In our laboratory, data must have a
coefficient of variation (CV) of about 7% or less. Data with CVs higher
than 7% aren't considered reliable.
In today's experiment about half of my samples had a CV of less than 7%.
This tells me that still I have some work to do in perfecting the
technique to the point that I can have confidence in my data. I don't
think I'll make modifications in the general procedure until I have CVs of
less than 7%.
Things to ponder:
Scientists always perform statistical tests to help them interpret their
data. Suppose you were doing a science fair project in which you wanted to
learn about the effects of a certain drug on plant growth. You design your
project so that you have two groups of the same age and type of plant.
Each group has the same number of plants (30 each) grown and cared for
under exactly the same conditions. The only difference would be that the
experimental group receives the drug while the control group does not.
When you conclude the experiment you measure the height of each plant.
Next, you calculate the mean for each group. Suppose it turns out that
experimental group has a mean of 22.5 cm while the control group has a mean
of 20.3 cm.
1. Based on the information given, could you conclude that the experimental
group had a higher mean due to the effects of the drug?
2. The answer to #1 is NO! Can you think of some reasons why this is so?
In order to know if the effect was due to the drug a type of statistical
test is needed. This hopefully illustrates why scientists must carefully
analyze their data using statistical analysis.
Contact the TEA in the field at
.
If you cannot connect through your browser, copy the
TEA's email address in the "To:" line of
your favorite email package.
