<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=utf-8">
</head>
<body text="#000000" bgcolor="#FFFFFF">
I always get a number of questions about Exp 2 and what to do for
the<br>
average MW if two of the trials are close and one isn't. Should you
include<br>
all three in the average in this case?<br>
<br>
Sometimes the answer is pretty obvious. If you have 150, 155, 220
hopefully<br>
you realize something may be wrong with the 220 since the other two
are so<br>
close. In this case the 220 would not generally wind up being
included in<br>
calculating the average MW. In the discussion section one would
list all three<br>
MWs and the average and explain the one MW was left out of the
calculation<br>
of the average. I suppose it could happen the 220 is the correct
one and somehow<br>
one got two wrong results out of three which were close but this
would be<br>
unusual (at least we hope). <br>
<br>
Sometimes what to do isn't quite so obvious. What if the results
were 150,<br>
160 and 190. Should the 190 be included. If you do the average
rounds to<br>
170 (2 s.f.). If you don't it rounds to 160 (2 s.f.). Which is
correct? I would<br>
probably go with the 160. However, to make a correct decision one
needs to<br>
do an error analysis and see if the 190 should be included in
getting the<br>
average.<br>
<br>
I have a link under the "Laboratory" link. This is a copy of
Appendix D<br>
in the current lab manual. Here's the direct link,<br>
<br>
<a
href="http://chemistry.osu.edu/%7Erzellmer/chem1220/lab/App_F_1220_lab_manual.pdf"><b>Treatment
of Numerical Data (Error Analysis, sig. fig., graphing)</b></a><br>
<br>
The discussion about how to determine if a data point can be ignored
in<br>
such cases is discussed in Section III, "Reporting Results" on pages
D-4<br>
through D-5.<br>
<br>
By the way, I can't tell you if leaving out one "bad" experimental
result will lead<br>
to a better or worse average. That depends on your results and how
careful<br>
you were being. It is quite possible for the two "good" results
(they are very<br>
close to each other) to actually be incorrect. Generally, if you
get two results<br>
which are very close to each other and one which isn't one would
expect the<br>
two which are close to be better. However, this isn't always the
case. It is<br>
quite possible you made a mistake and made that same mistake twice
so the<br>
two "good" ones aren't "good". Besides, I don't know what any of
the actual<br>
true results are for any of the experiments. Even if I did, I
couldn't tell you<br>
under penalty of death (well maybe not that severe a punishment but
close).<br>
<br>
For the post-lab data entry you have to enter your data as is (i.e.
all three<br>
MW values) and the average for all three (including the "bad" one).
Then<br>
in the data table in the report template, for the average list what
you get from<br>
just the two "good" MW values. Discuss this in the Discussion
section of the<br>
report. List all three MWs in the Discussion and the average based
on just the<br>
two good ones. Explain this was done and why. Include a calculation
in the<br>
sample calculations showing the "bad" value could be left out. You
can find<br>
how to do this at the link above (<br>
<br>
You might wonder if you can ask for a manual regrade of the results
grade.<br>
Any requests for regrading the results grade results in a 50%
reduction in<br>
the score, meaning the maximum score you could get would be 7 pts
rather<br>
than 14 pts. Also, your % error has to be pretty large before you
don't get<br>
any points. Finally, 14 pts equates to less than 0.3% of your total
course<br>
score and this is the only lab for which your results are graded.
I think<br>
about the only reason to bother considering a request for a manual
regrade<br>
is if you get zero points due to calculation mistakes you made in
the<br>
on-line data entry which results in getting zero points for
accuracy.<br>
<br>
By the way, if you do make calculation mistakes for the data entry
you<br>
need to correct them and report the corrected values in the report
or you<br>
could have additional points deducted. Your best defense against
making<br>
calculation errors for the data entry is do ALL the calculations and
double<br>
check them BEFORE doing the data entry. Then it's pretty much just
a<br>
matter of entering the numbers you have.<br>
<br>
Dr. Zellmer
</body>
</html>