Linux is amusing, but this is why I teach with Excel

The other day, I wrote that I wanted to make things easier for my students by using the kinds of software that they were likely to have on their computers and the kinds that they are likely to see in the business and biotech world when they graduate from college.

More than one person told me that I should have my students install an entirely different operating system and download OpenOffice to do something that looks a whole lot harder in Open Office than it is in Microsoft Excel.

I guess they missed the part where I said that I wanted to make the students work a little easier.

Before I go farther, let me say I do understand. I have been a minority computer user for years. I spent ten years as one of the only two faculty members in my former 65 person department (Math and Science) who bucked the system by using a Mac, so I know what's it like to have people think that you're strange and unrealistic. I spent many years being warned not to expect any support or help from the computer crew and just as many years wondering what exactly the support crew did besides installing antivirus software and fixing our Exchange server.

And, gasp, I even have VMware on my Mac so that I can run a Linux virtual machine with my own instance of our software (which runs on Linux and other variations of UNIX).

But, I teach students who live and work in the rest of the world.

And, like it or not, the rest of the world is most often the world of Windows. I adapt to the extent that I need to. In the early 90's, I learned that it wasn't really fair to the students if I said, "sorry, I don't do Windows." If you're a conscientious instructor, that's not a realistic option. Your students live in a world where all the school computer labs are stocked with computers running Windows. The staff know how to deal with Windows (and not always that well, either). And, unless you're going to live in your parent's basement, or work as a programmer (somewhere other than Microsoft, of course), sometimes it's not so bad to at least learn what to do if you need to get a job at a gasp, "commercial company."

I am appalled by some of the attitudes that I've seen from other writers. "Universities are not supposed to prepare students for jobs." Clearly, these writers are not paying college tuition bills. Can they really believe that it's morally right to let students (or their parents) pay $160,000 for four years of college tuition and finish school without any marketable skills? (Okay, those are private school costs, but still!)

Let's take a little look into the world outside academia. Let's listen to the things that I hear from people in the biotech and software community.

A few years ago, I went on a tour of a lab at a community college in San Francisco where the students were learning how to graph data from agarose gels with Excel. A person from another company, standing next to me, said, "Wow, that seems pretty basic. Don't students do this in high school?"

Well, no. I've suggested that sort of thing only to be told that the high school teachers can't do this because some students don't have computers.

A few days ago, I described my problems with pivot tables to a woman at another software company. She said, and I paraphrase, "Wow, that seems pretty basic, don't they learn that in high school?"

Uh, no, they don't. And they don't learn this in college either, apparently.

From my perspective, the academic community treats ubiquitous programs like Excel like they treat typing. It's either "doesn't everyone know that?" or "oh, right, that's something students do in vocational training."

And the business and biotech communities seem to think that learning Excel is like learning spelling. Doesn't everyone learn how to use Excel? It's kind of a basic skill isn't it?

Some of my students are going to go work in the biotech industry. Some of them will have to work at companies where software systems have to be validated. And in those cases, the software that's been developed by professionals is more likely to get validated and approved for use in FDA regulated processes.

It's tough living in the real world and writing in the ideal one. And sometimes it's even tougher being a person who moves back and forth between.

More like this

Life has so many dimensions, why do we constraint our thinking to two way flat tables ?

This is not about windows/linux war, it's about using spreadsheets to teach. Why not exploring the several dimensions with matrix featured software ?

EJ

By Ernesto Jardim (not verified) on 11 Mar 2008 #permalink

Keith: Students do get a pretty good discount on software. It's probably cheaper for them to buy a copy of Office than it is to get a copy of RedHat Linux.

The real question is, does OpenOffice have the same graphing bug that we found with Excel on Windows?

EJ: matrix featured software? tell me more.

matrix featured software:
OpenSource: R, MuPAD, YACAS, Octave
Commercial: Mathematica, Matlab

I'm quite sure there are a lot more out there.

EJ

By Ernesto Jardim (not verified) on 11 Mar 2008 #permalink

The fundamental, unforgivable omission of the linear fit function in OpenOffice is why I still have to find a Windows computer for much of my daily work. We just have too many simple assays that require it.

I agree with EJ that there are decent packages specifically for graphing in linux (LabPlotis one that I have installed on my machine), but once again you run into the energy barrier of learning/teaching an entirely different user interface.

It's unfortunate, but I don't really see Excel being displaced from most labs (particularly those in industry) in the near future. If you want to teach your students using the tools they should expect to see, you'll have to sort out the compatibility issue. My only recommendation (and it's a suboptimal one) is to put the software information you'd like in the course syllabus and work from there.

Personally, I've always found that if someone understands the task they're doing and knows how to use computers generally then it takes some pretty abysmal software before they get stuck with anything but very complex tasks. They should be able to choose suitable software and get the job done in their own way, on their own.

Of course, that's an ideal you'd need a dedicated computing course to meet. If you're running a biology course then I presume you want them to pass and fail based on their skills in biology and not computing. But some people just don't 'get' computers, and teaching them MS Office by rote won't solve that -- it will just make them think they get computers.

You need to use some programs in college, but programs change. Whether or not they used Excel's advanced functions in college, a college graduate should be able to sit down with a book and learn these themselves. If they can't their college education has failed.

That said, there's nothing wrong with using Excel (except for its buggy statistics packages), and using a program that many of them might use afterwards is reasonable.

"Keith: Students do get a pretty good discount on software. It's probably cheaper for them to buy a copy of Office than it is to get a copy of RedHat Linux. "

Posted by: Sandra Porter

A comment - this can vary a *lot* between colleges. My niece in a community college would pay several times the rate that a student in Elite University next door would pay. (file under 'rich get richer, poor get poorer')

Barry: Most colleges will sell discounted software to students no matter where they attend school. The students just need some kind of student ID. If the software is cheaper at Elite University, your niece should do her shopping over there.

The same way you use excel but with a lot more flexibility and potential.

EJ

By Ernesto Jardim (not verified) on 11 Mar 2008 #permalink

Regarding the usage of matrix featured software. I didn't noticed there were so many posts in between.

EJ

By Ernesto Jardim (not verified) on 11 Mar 2008 #permalink

The proper question is whether to teach 'Excel' or an analytic technique.

A proper science education (IMHO) should not put the emphasis on training with a specific tool but rather on analysis technique and process. If that is done well, then the student should be able to use any suitable tool with ease.

There are very many contentious ideas here and the discussion could be long and difficult. Perhaps just one aspect to consider would be that tools change, especially software tools. To train people for current tools has its uses but education should be towards building new tools and towards being able to make effective use of tools not yet created.

Bryan: exactly! That's what we're doing. We have a long project analyzing the bacterial populations that were found on campus and we're using many different tools to ask questions about what we find.

I would prefer that the software not get in the way of the science, but since the students don't know how to use spreadsheet software and don't quite understand how to portray data correctly in a graphical form, I have to teach that part of it, too.

Life has so many dimensions, why do we constraint our thinking to two way flat tables ?

This is not about windows/linux war, it's about using spreadsheets to teach. Why not exploring the several dimensions with matrix featured software ?

Because starting simpler is better, and starting "more likely to be used beyond college" is better. Students struggle as it is to add error bars to simple bar charts in Excel; Mathematica is way too much for many to start with.

see, i'm still not at all clear about pivot tables. don't think i've ever actually used one yet.

in my job, the "don't they learn that in school" skill is database manipulation using SQL. folks who don't understand normalization, joins, and query efficiency give the rest of us no end of headaches. excel? good tool for the accountants, but i don't use it.

(i can also see the web-based application slowly rising on the far horizon. most of my DB manipulation gets done with a web-based admin tool these days, unless i write custom scripts for whatever task is at hand.)

By Nomen Nescio (not verified) on 11 Mar 2008 #permalink

Pivot tables are a very quick and easy way to count the number of times that things occur, at least that's how we use them.

If you want to see how this works for counting bacteria, here's an example in a video.

I think you're right about web apps. If Google Docs did pivot tables that's what I would use.

We're in the business of teaching students how to learn. Not locking them into a particular application. If you want to teach students how to learn, have them do a lab or tutorial as part of their coursework. To claim that Excel is somehow their saviour and that they cannot even use that is pretty nonsensical. All you are really doing is helping them learn how to rely on one particular software package.

A nice example of this is MatLab. Math students that learn how to use MatLab rarely know anything else exists; Mathematica, or even SciLab are completely lost on them. They rely on MatLab to such a degree that they have to buy the expensive software just to do their own personal research. That is precisely the sort of reason that particular software packages are generally avoided.

By Malificent (not verified) on 11 Mar 2008 #permalink

Malificent: You're not correct. We're already using 8 different programs in this course (iFinch, FinchTV, Excel, blastn, Entrez, JalView, Clustal, and Cn3D).

I hardly think that qualifies as locking students into one application.

Students do get a pretty good discount on software. It's probably cheaper for them to buy a copy of Office than it is to get a copy of RedHat Linux.

I'd be terribly surprised if the discount didn't apply to Red Hat too, but that's still a weird example to use - the vast majority of Linux distributions, including the popular Ubuntu, are absolutely free of charge to anyone and come packaged with all the other free software like OpenOffice.

Regardless, I think you're right for trying to teach them in Calc/Excel, even though they'd probably use something like Bioconductor if they did this for a living. Spreadsheet skills are fast to develop and portable, even if R gives better results and is easier once you learn it. You don't have to pivot the whole course around Excel though; you might offer a special (optional) spreadsheet-training session at the beginning, and throughout the rest of the course let them use whatever software they like as long as it gives the kind of output they need.

bsci wrote: "You need to use some programs in college, but programs change. Whether or not they used Excel's advanced functions in college, a college graduate should be able to sit down with a book and learn these themselves. If they can't their college education has failed."

That's only true if they have some idea of the range of advanced functions that exist. No one is going to sit down with a book to learn how to use pivot tables (or whatever) unless they can already see how that information will be useful to them, which means already knowing what pivot tables are FOR. And it never ceases to amaze me how few (very high-powered) people do.

I do agree it's important to teach along the lines of "We're doing this in Excel but any spreadsheet program should do something similar". Though frankly OpenOffice's awful clonky substitute for pivot tables is precisely why I pay money for Microsoft.

my $0.02, based on my recent undergrad experience...
When I got to biochem lab, I wanted to use open office, since that's what I was using for wordprocessing, ect. Microsoft Office was too expensive, even with student discounts (right now, the student version of office is $150 on Microsoft's website and $80 through my institution- with discount- for those that think $80 is nothing, particularly when you consider the cost of tuition ect., all I can say is I don't know how much money *you* had as an undergrad, but at no time was I in a position $80 wasn't a big deal).
Also, I *ran OpenOffice on my Windows machine just fine*.

Consider introducing GraphPad Prism if you have any students that are going to grad school. It's relatively user friendly/has good help files, and I wish I'd understood it better in my coursework.
Despite my preference for software doing what it was *designed* to do (and Excel was *not* intended for linear regression), I think Excel is a justifiable choice. A friend of mine (with far more CS experience than I'll probably ever have) managed to impress his boss to a ridiculous degree with a really very minor amount of Excel competency.

@ bsci- I think my college education failed ;-)
I managed to pick up Word almost entirely on my own (no book, limited use of help files, mostly just extended playing around with it)- Excel was utterly bizare when I first saw it. I'm still not sure I could learn it from a book. Mathematica, sure. But not Excel.

Pivot tables are nice, but there are other spreadsheet functions that work as well.

Here, I made a screencast showing that this isn't terribly difficult to do in Open Office:

http://www.youtube.com/watch?v=pqd7tC8fIBY

It's a little small - fullscreen might help:
http://www.youtube.com/watch/v/pqd7tC8fIBY

Also, FWIW, I disagree with the poster that said you should have them install Linux. I suggested Open Office precisely because it is cross platform and wouldn't require students to install a new OS. It's also 95% similar to Excel, so they would be using similar techniques (pivot tables are apparently one exception)

since the students don't know how to use spreadsheet software and don't quite understand how to portray data correctly in a graphical form, I have to teach that part of it, too.

It would be one thing if they were all excel gurus, but if you're starting from scratch anyway, then why teach them skills that they can't use at home? (without paying exorbitant prices for MS Office).

sometimes it's not so bad to at least learn what to do if you need to get a job at a gasp, "commercial company."

I'd argue that people with skills that aren't platform-dependent are vastly more hire-able. When you can walk in and say "I can comfortably use this software on any platform you throw at me", it's much more impressive. And again, Open office is 95% the same as excel, so they aren't going to face many problems working in an MS environment.

Thanks Chris!

Impressive screencast!

Having something that works consistently on different platforms is really my main desire. I'm pretty optimistic that web apps will come along that help out before too long.

I'll give OpenOffice a try though when I get a break between projects. Your screencast will help, thanks!

> It's probably cheaper for them to buy a copy of
> Office than it is to get a copy of RedHat Linux.

Red Hat is an OS, not a productivity suite: OpenOffice is cross-platform, including Windows, Mac and Linux: and downloading OpenOffice is free. It may be more convenient for a student to pirate a copy of MS Office, but that's probably not what you meant.

> by using the kinds of software that they were
> likely to have on their computers and the
> kinds that they are likely to see in the
> business and biotech world when they graduate
> from college.

Productivity apps change regularly: the most recent version of Office has a drastically different UI than older versions (users of these would probably find an easier time migrating to OpenOffice, which has a more conventional interface). Furthermore, you're assuming that Office will retain its dominant position in the market: given the increasing pupularity of OO (particularly in SMBs, government, and emerging markets) this might not last.

There's very little software I used at Uni that's lasted the distance, except command-line OS skills, high-level languages, basic concepts of editing and document preparation etc -- all the stuff that doesn't change much over time.

Knowledge of the intricacies of a particular productivity app doesn't age well.

Have you used Office 2008 on OS X? They've removed the ability to add custom error bars on graphs - something that most, if not all, scientific data requires. I've got no choice but to move away from Office (Mac, but why not all, now that I have to re-train myself?). OpenOffice doesn't have this feature either, so I've decided to learn R.

I've heard that other people are in the same position, but I guess it's not a mass exodus yet. I feel a little bullied into learning a more complex software package, but it'll be much more flexible in the long run. Add that to Endnote not working in Word 2008, and Office 2008 is not an attractive option.

First, how cool is Chris with that screencast? Way to go!

Sandra, your post makes me think of academic silos. Someday I hope to see a post by some teacher about how they were able to work with someone in the CS department in a way which works for both. That is, it would be a great project for CS students to build the software you need as part of their coursework. It could be an opensource project and even (eventually) involve students from anywhere. As a former CS student, I would have found it far more rewarding to be working on something that served an immediate, real need, rather than toy problems.

I agree Trinifar, that is the normal case.

But collaborations between computer scientists and biologies do exist. In fact, our company has a collaboration with a non-profit institute, The HDF Group, to develop open source genotyping software for bioinformatics. I will write about this more in other posts.

Since you wanted to here from instructors too, here goes. I teach in a community college and many of my students would freak out if I told them to dual boot linux, or even install new software. For now I use Excel and Google docs for our work. My goal is to teach them what programs like that can do for them and how it can simplify their lives for some of their tasks. I use it primarily to set up spreadsheets to calculate master mixes for RE digest and PCR reactions. Perhaps graph a standard curve or two. As far as I'm concerned they can use whatever they like as long as it gives them the correct answer. But from my experience they are more comfortable with MS Office products. My advisory committee also likes to see Excel/spreadsheet skills.

There is the ability to work with R in Calc: http://wiki.services.openoffice.org/wiki/R_and_Calc

On the Mac OpenOffice.org aqua final will be released sometime around September or October. Feedback on current development builds is really positive. One particular notable point is that it is faster than NeoOffice.

A few versions back (I think it was 2.2) there was an overhaul of the chart module in Calc, which made it a lot easier to make charts.

If there is something that you can't do, I would advise you to ask on the users@openoffice.org mailing list, or submit a feature request if it isn't available. There may be the possibility that an extension could be created to implement the functionality required.

Developers can only work on improving things, if they know what needs to be implemented. It requires users like you to come up with them. If the features are there and there is no documentation, then you need to speak to the documentation project. http://documentation.openoffice.org/

I thought the entire point of a college education was to teach you how to think and learn. Yes, learning how to use excel 2007 will be helpful for your first job....however, if you can't figure out how to run programs with different UIs you are doomed when you enter the "real world" (or even academia) and programs get switched on you. OO and Excel both work well at what they are for. If you learn on one(graphing, equations, etc), you should really be able to extrapolate those skills to the other. If not, you won't make it anyway....at least not in a capacity where someone like me doesn't end up making your graphs for you.

---note: I'm not the default IT person for my lab because so many people know how to use "specific" software, but anytime there is a slight difference they are lost. I think your students would be best off if they were taught how to use both and learned how to use other stuff on their own.

sorry, ranting is done.