Plant-eders,
Thanks for all of the thoughtful comments on using Excel or other
data analysis programs in classrooms. Responses were split about
equally between Excel users (some liked it more than others) and
users of other programs. Many of the former recommended tutorial
sessions in or outside of class time (a very good idea that we will
do). Thanks Eric Ribbens for alerting us to the known problems of
using Excel for statistics. One of my colleagues here suggested that
we use Excel and SPSS, the former for data manipulation and graphing,
and the latter for its statistical functions. Many of our students
learn SPSS in a Stats class so it would make a lot of sense to also
use in in our Bio courses.
Other programs that were mentioned include: Kaleidagraph, SigmaPlot,
SigmaStat, Cricket Graph, DeltaGraph, Minitab, and Prism
(http://www.graphpad.com). GraphPad (maker of Prism) has several
free online statistical calculators at their URL above.
One of our (Plant-ed) own, John Hewitson, has been involved in
writing a program called the Warwick Spreadsheet System - a suite of
macros running under Excel (version 4 or later, PC or Mac). It looks
interesting but I have not tried it out yet. John Hewitson said,
>We have tried to make complex tasks simpler, add useful facilities
>which Excel does not normally have and use Excel for modelling and
>simulations. It even includes Biologists' favourite statistical
>tests made simple. The "normal" facilities needed by Biologists are
>presented in simple menus with the full power of Excel available as
>"others" so as not to confuse novice users, yet not limit the power
>of Excel. Graphing data is especially easy. We have tested the
>system with young children but used it extensively amongst
>undergraduates and secondary school pupils. We reckon that our
>tutorial should enable anyone to become a reasonably competent user
>in an hour or so.
>>See it at: http://members.aol.com/aberdareco>>The extensive printed material (Users Guide, Worksheets for Biology,
>Chemistry, Physics and Astronomy) is all available to read over the web at
>>http://website.lineone.net/~aberdareco>>These sites include reviews from journals, demonstrations, details
>of the facilities and how to obtain a demonstration CD (which
>unlocks with a passcode number to give the full version) from
>aberdareco at aol.com
Below is a list of the longer individual comments sent only to me.
As mentioned by Bill Williams, an amazingly similar discussion on
Excel took place simultaneously on the CUR list. Some of their
comments are at the bottom.
Thanks again!
Jon
==========
Actually I find Excel is steep for the instructor... you can create a
spreadsheet where students just drop data in two columns and the
spreadsheet does a t-test on the fly without the students writing any
formulas or running any macros. You save the spreadsheet to the drive
and students open it and use it without having to hard-wire it.
I have a non-linear regression template that uses the Excel Solver
for the reiterative part of that analysis. Again the students have a
somewhat longer instruction set for using it, but it isn't too
onerous. It took me forever to figure out how to do the spreadsheet,
but now that it's done, the student use is pretty simple.
My philosophy on stats is that students need to know how to use
stats, which tests to select, and how to interpret the
results...rather than learning how to hard-wire tests or to do them
by hand. I think there is value in hard-wiring and hand
number-crunching just as there is value in constructing a car from
the bolts on up, yet we don't have to build a car to drive one. So I
teach students how to drive rather than to build...just one point of
view for undergraduate instruction.
==========
We use Excel for most of the bio courses. Previous versions of Excel
were difficult to produce good science graphs, but it is much better
now. Also, nearly all of the students here seem to have MS Office
already on their computers, so they have it accessible after lab to
complete the data analysis. You are right that the learning curve
can be steep, but if they start with simple things (calculating mean,
std. dev, graphing 1 variable) then moving into complex formulas is
more intuitive. Some of the more statistically based programs make
doing simple calculations difficult and that is where learning Excel
is important. It can do much of the needed stats, or can with
plug-ins.
==========
We are taking the opposite approach. Our students will learn to use
Excel! In some cases they come to the university already up to speed
or have contacts in dorms that help them. We also teach Excel
through an online tutorial I wrote, and by gradual introduction of
graphing and data analysis components.
==========
I just bought Kaleidagraph because that steep learning curve with
Excel was driving ME crazy. I can't yet say one way or the other if
it's appropriate for the use you intend, but KaleidaGraph, SigmaPlot
and DeltaGraph have free downloads from their websites, which I found
useful. Test driving, so to speak.
==========
We've been using Cricket Graph (at least for the spreadsheet and
graphing part), and it will do some simple statistics. I think part
of the problem is that software companies think that they have to
introduce ever-more massive programs, when something basic is fine.
The companies (and sometimes faculty) forget we need roughly the same
thing each year - if we want students to make a scatter plot and then
find a regression line, we don't need anything fancier. I don't want
to sound like a Luddite, and I know companies think there is more
profit in a massive program that is upgraded every year. So college
software and hardware budgets get blown out of the water. Isn't there
a niche marketer somewhere who could make a program to meet the needs
you expressed without requiring a G4 with 128 Megs of memory? Yes,
it's great that the students no longer have to figure out the
regression equation on a calculator. This is more than you asked for!
==========
We use Jandel products....SigmaStat and SigmaPlot. They are
spreadsheet programs, and there is a bit of a learning curve to
surmount. Advantage is that you get publication quality stats and
graphs.
==========
The simplest by far and most useful at undergrad is InStat, Instant
statistics. It is incredibly simple and highly intuitive. Goes up to
one way anova and T tests and has excellent interpretation. It is not
pretty but it works. Also does Mann Whitney etc post tests. It does
open testing of the variance difference and makes simple statements
such as: if variances are significantly different one should not
accept means as different.
The best simple graphics program I find is Cricket graph otherwise
known as Graph 3. Again very simple and intuitive but produces v
nice graphs analyses regressions and slopes, but does no stats. Not
sure if either of these are still supported I've had mine some years
now.
==========
There are some nice tutorials on Excel that are available on-line.
To see a number of them, go to scrtec.org, choose Trackstar, and
search for "Excel". (The Trackstar site contains mini-websites
created by teachers for their students.)
Just today I had my students walk through a PowerPoint tutorial on
the TrackStar site. Most of the students had the hang of it within
15 or 20 minutes. I think PowerPoint is actually harder than Excel,
too! The students had a real sense of accomplishment afterwards, as
well.
If you don't have enough time in your class, or access to a computer
lab, you could assign the tutorial as homework. Just a thought.
Good luck!
==========
Here is what we do in our undergrad (Jr College) majors' Botany
class: We tell students at the beginning of the semester to have a
scientific calculator handy because they will need it later
on in the semester. (Most will have this for their chem and math
classes anyway).
When we start experiments and lab reports, we divide the class
into teams of 4 or 5 people. We advise them to bring their calculator
manual to class. They do basic mean and standard deviation on their
results in class with help from us or from their teammates. The
most complicated thing we do is 2-variate regression. Most figure out
how to do this on their calculators.
We tell them to graph their regression experiment by hand, NOT on
the computer. We suggest all frequency polygons be done by hand. We
allow them to use their (or the campus') computers only for bar
graphs.
After many years of assisting in Botany, and being a reader of the
resulting lab reports, I can tell you that computer graphs mostly
suck big time. The students do not know how to adjust the axes, or
label them properly. They do not understand what a graph is until
they can do one by hand (IMHO).
I remember spending hours trying to figure out linear regression in
Excel, which I can accomplish in seconds on my old HP calcullator.
I realize that students are supposed to learn how to use these
computers and their software, but By Neddy Jingo sometimes the old
ways are the best and fastest!!!
==========From the CUR list==========
Excel's "inaccuracies" include such things as accurately computing
SDs to "only" seven decimal places against a benchmarking standard,
F-ratios in oneway ANOVAs to "only" eight decimal places, and so on.
There are other problems besides these, but they are unlikely to make
much difference in our ordinary uses of Excel (unless you OFTEN run
10th-degree polynomial regression analyses).
==========
For a program whose authors are *serious* about statistics, see Igor,
<http://www.wavemetrics.com/>, a graphing and data-analysis package
originally written for the Mac but ported to the PC a few years ago.
I use it as the primary graphics and data-analysis tool for my
research, but I'm not sure I would recommend it for widespread
undergraduate use. The learning curve is steep, but I understand some
schools have success with it.
==========
Jukka Lehtonen, in Finland, has a nice web page explaining how to use
Excel to plot standard curves. The examples are very clearly written
at an intro level, without assuming prior knowledge of Excel. Perhaps
this could be a starting point, with some discussion on how to
analyze the organic data. The URL is
http://www.btk.utu.fi/molmol/student/excel_intro.html
==========
I teach the use of Excel as a laboratory component of my Instrumental
Methods class. Instead of giving students a handout, or demonstrating
to them how to use the program (something they will quickly forget in
any event), I simply give them a typical assignment, which may even
include analyzing their own laboratory data, and have them go at it
while I act as a consultant. I give them a few suggestions to get
started, and then the students generate their own questions, which I
answer. When the students do it for themselves, they learn what is
necessary very quickly. A three-hour session is usually more than
sufficient to get students up to speed with the basics. Fancier
options are taught on a need-to-know basis.
==========
I was informed of 3 really good websites with instructions and
tutorial on using excel.
http://s9000.furman.edu/mellonj/excel.htm#contenthttp://chem.mathsci.usna.edu/plebechem/excel_tutor/homepage.htmlhttp://www.btk.utu.fi/molmol/student/excel_intro.html
As well as a number of great handouts as well as ideas for using excel.
Thank you so much.
==========
I have been using excel or lotus 123 with classes for about 10 years.
We start our first term general chemistry classes on the spread
sheets and demand, graphs, calculations and functions. We have a
handout, but I will assure you that a one-on-one at a computer for
about 10 minutes after the student has struggled works the best. If
you can take a word attachment to an email email me directly. Put a
subject on the message.
==========
--------------------------------------------------------
Jonathan D. Monroe Associate Professor
Department of Biology, MSC 7801 office: 540-568-6649
James Madison University fax: 540-568-3333
Harrisonburg, VA 22807 email: monroejd at jmu.eduhttp://csm.jmu.edu/biology/monroejd/jmonroe.html
--------------------------------------------------------
---