A Blog Around The Clock

MS Excel is the tool of the Devil!

There is a good reason why scientists in general despise MS Excel. It is cumbersome, non-common-sensical, and the stats cannot be trusted. The graphs are ugly. I am sure it took a lot of hard work to design Excel (and Word), but if I were Charles Simonyi, I would hide the authorship of those two programs as much as possible. Charles went to the Space Station, after all, paying for the ticket out of his own pocket, so there is something much more exciting (and safe) to brag about (not to mention dating Martha Stewart).

There are so many good pieces of software out there, many capable of doing some very complex statistics. For quick simple tests, nothing beats the free online GraphPad QuickCalcs. A few seconds of pasting in the data, click on “Calculate” and the numbers are all there, ready to copy and paste into the manuscript.

I still use CricketGraph (the ancient version 3.0, not anything newer) for drawing graphs as nothing beats its simplicity and the crisp clarity of the graphs. For the stuff in my field, the open source program Circadia, small enough to fit on the old big soft floppies, not updated since 1982, is still a golden standard that no newer software package can begin to match in its ease of use, clarity and ability to do everything a chronobiologist wants to do with data in a matter of few minutes. I wish I did not have to keep an old Mac around just for those two programs (new MacOS cannot read them).

But, darned Excel is the corporate standard. So, I spent the last few days cussing and cursing, being mean to the dog, and generally having a bad time, because I had to use Excel. And I started out all wrong – it was a common-sensical way to do it, but, hey, common sense does not operate here. I consulted people better versed with it to see if I was doing it right. When they said No, I had to start from scratch. And I am still doing it (a day overdue now). And it appears I’ll be doing it all day or longer…. Yuck!


  1. #1 Tegumai Bopsulai, FCD
    August 27, 2007


  2. #2 Gerard Harbison
    August 27, 2007

    There is a good reason why scientists in general despise MS Excel

    They don’t. I use Excel, frequently. So do most of my colleagues. When were you elected spokesman for ‘scientists in general’?

  3. #3 lunartalks
    August 27, 2007

    Welcome to the world of MSR. MicroSoft Rage. Truly, it sucks a fat one.

  4. #4 afarensis
    August 27, 2007

    Excel denialist!

  5. #5 coturnix
    August 27, 2007

    I have yet to meet in person a scientist who likes Excel. Perhaps some corporate type with a bureaucratic mind, perhaps a Republican… 😉

  6. #6 CRM-114
    August 27, 2007

    Try Perl, the scripting language. Easy to learn, extremely flexible. Scripts can digest data in any format (because you tell it how) and output in any format (again, because …). Then all you need is a graphics package that runs on your OS.

  7. #7 Hank Roberts
    August 27, 2007

    Recommended reading on spreadsheet errors (yes, it’s science!)

  8. #8 Rosie Redfield
    August 27, 2007

    Ah, CricketGraph – my first love! I guess you must have to run it in Classic mode…

    Why oh why did the creators of Excel think that all graphs should have grey backgrounds, and all graph symbols and bars should have shadows? And why can’t these defaults be changed?

  9. #9 afarensis
    August 27, 2007

    No, no. I have a book, by a scientist, on applied statistics in Excel. It says, and I quote, “Once the appropriate technique has been selected, all that’s needed is guidance using the software – nothing more complicated than point and click instructions”. Did you try pointing and clicking?

  10. #10 coturnix
    August 27, 2007

    Yes, Classic mode. Abel PharmBoy got me an old OS installed on a new Mac so it works.

    And one can change the defaults, but, even you did not figure out how! And you are smart! Students showed me how a long time ago but I forgot now and of course it is not in any obvious way.

  11. #11 Warren
    August 27, 2007

    Have you looked into the OpenOffice counter-MS apps? No idea if they’d work better for you or not, but they might be worth a shot.

  12. #12 coturnix
    August 27, 2007

    OpenOffice sounds really good. I’ll try to download it to my home computer to test it. I don’t think the company laptop will let me download new programs.

  13. #13 Rob Knop
    August 27, 2007

    OpenOffice calc is probably no better than Excel. All the same problems exist….

    A spreadsheet can be very useful for a lot of things in science. I’ve kept gradebooks in it. I’ve used it to put together telescope exposure time calculators. I’ve used it to make quick-n-dirty plots of things like blackbody curves for classes.

    But if you’re doing serious analysis or statistics, a spreadsheet is the wrong way to do it. Likewise, the graphs and plots are designed for businessfolk, not scientifolk. You can get them to do sort of what you want, but it’s not the natural way those programs work, and they think you’re looking at data a weird way.


  14. #14 riptide
    August 27, 2007

    Just so we’re clear, your company (which is PLoS One, n’est pas?), which champions “open source” biology, has locked down your laptop so you’re unable to install programs? The cognitive dissonance is making my mind shudder…

  15. #15 coturnix
    August 27, 2007

    I can, but it’s a friggin’ complicated method to bypass various security features and I don’t want to do it until I know for sure I need and want this.

  16. #16 Eric Lund
    August 27, 2007

    I find M$ Excel to be useful for the things it was designed to do. It works remarkably well for things like proposal budgets and other routine calculations involving simple arithmetic. If you’re trying to do medium-duty data analysis, Excel is the wrong tool. I also find it annoying that Excel thinks it’s smarter than I am, e.g., by flagging “inconsistent formulas” (defined as a formula which differs from the formulas in neighboring cells) as errors.

    That said, I find Excel to be the least evil component of M$ Office. Word is bloated with feeping creatures; I try to avoid generating documents in Word. Outlook/Entourage has long been a major security risk. And let’s not forget the most evil Office component of them all, PowerPoint: it was designed to shut discussion down, not facilitate it as we (usually) would like our scientific presentations to do.

  17. #17 Gerard Harbison
    August 27, 2007

    I have yet to meet in person a scientist who likes Excel. Perhaps some corporate type with a bureaucratic mind, perhaps a Republican… 😉

    How odd. I work in a university chemistry department, and I’d say pretty much everyone uses it; nor have I heard any notable complaints about it. It’s actually very good at what it’s meant to do; adding, tabulating, and sorting data, and graphing it. You can do perfectly good graphs with Excel as long as you don’t rely on the default formats; I have shown students how to propagate differential equations using it.

    If you’re going to do real statistics, of course, you probably want a real statistical package. And if you’re doing math, you use Mathematica or equivalent. But Excel is a good tool for most standard number hacking and display applications.

    I think we have a lot of poor workmen on this thread blaming their tools. Maybe that’s a Democrat thing? 🙂

  18. #18 coturnix
    August 27, 2007

    And tongues in cheeks are not readily visible online 😉 so some people take everything waaaay too seriously – Republicans perhaps 😉 ?

  19. #19 jeffk
    August 27, 2007

    In my physics/astronomy lab, we all use it from time to time. If you have a small amount of hand-taken, unofficial data and you just want to get a quick look at what you’re dealing with, it’s really not all that painful. If you truly want pain, try dealing with gnuplot on linux.

  20. #20 Davej
    August 27, 2007

    I don’t think Excel is so bad. Try old Lotus 123 if you want a headache. I find Excel usable and Word far more annoying, especially that sadistic little paperclip man who comes to taunt you.

  21. #21 mark
    August 27, 2007

    I use Sigma Plot and am generally satisfied, but the price of upgrades has gotten extremely high so I may just use ProStat if my current version of SigmaPlot has any problems in the future. I characterize a number of plotting/spreadsheet applications as being designed for shoe salesmen.

  22. #22 Lori
    August 27, 2007

    Actually, I would take exception only to your title. As far as I’m aware, the devil likes things to be beautiful, graceful, and easy to use. He is way into the aesthetics. It is the puritans, however, who always wanted things clunky, totally devoid of aesthetic qualities, and who believed that a little pain and suffering was good for your immortal soul . . . perhaps you have aligned yourself with the wrong side? 🙂

  23. #23 AE
    August 28, 2007

    I’ve always wondered why Excel uses such abysmally appalling default colours. I mean, I have a pretty dreadful eye for colour and style myself, so if it offends me it must be particularly dire.

  24. #24 SLC
    August 28, 2007

    I would suggest trying Deltagraph. I mainly like it because it allows one to do curve fitting with user designed functions.

  25. #25 kate
    August 28, 2007

    i also use sigmaplot, and also like it. for my aims, it’s pretty straightforward and user-friendly. you should give it a whirl.

    for statistics, try using SAS. i’m no programmer and quickly develop a horrible headache whenever i have to program/run something new. it’s just awful. but, hoping to avoid another learning curve (and perhaps b/c i’m stubborn), i won’t spend the time to learn anything new.

  26. #26 PhysioProf
    September 1, 2007

    For graphing, we use Microcal Origin. It is *extremely* flexible, and it exports high quality PDFs, which can be opened in Adobe Illustrator and tweaked to your heart’s content.

    For stats, SPSS.

New comments have been disabled.