tag:blogger.com,1999:blog-5841910768079015534.post2579829714454008723..comments2024-03-25T17:14:36.888+00:00Comments on BishopBlog: Why I still use Exceldeevybeehttp://www.blogger.com/profile/15118040887173718391noreply@blogger.comBlogger6125tag:blogger.com,1999:blog-5841910768079015534.post-53438496104503289892016-09-17T07:00:09.165+01:002016-09-17T07:00:09.165+01:00Does Excel have data limits for statistical functi...Does Excel have data limits for statistical functions? I have been told during an Excel call that t-tests, for example, are good for data sets up to N=60. Does that mean if I have a data set of N=120 that the calculation can be in error? Lpalmer@winona.edu<br />Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-5841910768079015534.post-3896900990019146222016-09-04T17:10:00.751+01:002016-09-04T17:10:00.751+01:00SPSS isn't free either, so for reproducibility...SPSS isn't free either, so for reproducibility one would stick to R or something else freely available and open-source. (Without the latter we would be talking about some magic blackbox, not something that everyone can take apart to peek inside, if needed)dimpasehttps://www.blogger.com/profile/14596969929730094920noreply@blogger.comtag:blogger.com,1999:blog-5841910768079015534.post-33804634210571419912016-09-03T22:20:16.006+01:002016-09-03T22:20:16.006+01:00Over the last few years I have slowly come to the ...Over the last few years I have slowly come to the conclusion that using a spreadsheet or anything more complicated than my shopping list was madness. I am now reconsidering my position on shopping lists. <br /><br />In general spreadsheets seem to be often used in extremely inappropriate ways and the actions of a spreadsheet is very opaque to the user which makes auditing a spreadsheet a bit of a problem. <br /><br />For highly experienced and knowledgeable users accustomed to all the vagaries of a spreadsheet, they can be a useful tool particularly for data entry but, as the Genome Biology paper illustrates, too many people don't understand what a spreadsheet does. <br /><br />A possible alternative to a data-entry spreadsheet for some in the Windows world might be EpiData. http://www.epidata.dk/ <br /><br />I think that the best approach for data entry is a well-designed data entry form attached to a database. <br /><br />It seems a pity to teach students and budding young researchers to use an inferior and error prone tool.<br /><br />I totally agree with Thom Baguley on graphs. Using a spreadsheet for graphs is way too much work. <br /><br />Many of the sins of Excel and other spreadsheets are ones that don't appear if one is just using it for data entry but in many cases people will use it for analysis, modeling and heaven knows what else. <br /><br />A problem, not one likely to bother a researcher who is only using a spreadsheet for data entry or a quick bit of EDA is that in Excel and other spreadsheets do not, in general, have strong typing. Without this, you can do some weird things. <br /><br />As an experiment enter your telephone in spreadsheet cell A1 and format it to your usual national format. I am in Canada so my format is usually 555-666-7777. Then in B1 enter the number 2. In C1 enter “=A1*B1”. jrkrideauhttps://www.blogger.com/profile/04869979887929067657noreply@blogger.comtag:blogger.com,1999:blog-5841910768079015534.post-86669069998477952202016-09-02T17:56:53.414+01:002016-09-02T17:56:53.414+01:00I still use Excel a lot. I don't use it for gr...I still use Excel a lot. I don't use it for graphs as I find the time taken to get a graph right exceeds that required to write an R function do plot the data (in most cases).<br /><br />The big advantages are the ubiquity of the package, the utility of sharing csv files and the ability to view the raw data flexibility that is unmatched by any other package I've used. Once you've got data sets that can't be viewed on a single screen this advantage diminishes and I can see a case for other approaches.<br /><br />Two other uses: setting up simple calculators for students and collaborators (showing working etc.) and data checking and cleansing using logic and text functions.<br /><br />One thing I don't generally do is use it for proper stats (except where I can check the results by hand or in a package such as R).thomhttps://www.blogger.com/profile/00392478801981388165noreply@blogger.comtag:blogger.com,1999:blog-5841910768079015534.post-75129267699236849432016-09-01T13:30:16.815+01:002016-09-01T13:30:16.815+01:00I like Excel, and call it Magic Maths Paper.
Howev...I like Excel, and call it Magic Maths Paper.<br />However, it needs to be kept simple, and I find the graphing facilities frustrating. Getting a reasonable graph is easy, getting it just as you want very difficult.<br />No bad for freeware.Anonymoushttps://www.blogger.com/profile/09320614837348759094noreply@blogger.comtag:blogger.com,1999:blog-5841910768079015534.post-2899499769444566372016-09-01T13:29:45.708+01:002016-09-01T13:29:45.708+01:00I like Excel and believe you can do useful things ...I like Excel and believe you can do useful things with it (as long as you are aware of its limitations/auto-adjust behaviour). It is useful for preliminary analyses. However, I don't think it's suited for scientific work. Not so much for the auto-adjust behaviour (to a large extend you should blame the lazy researchers rather than the software) but because it is difficult - for yourself and, especially, for others - to retrace the steps you took. <br />R works with written-down code: everyone (that understands R) can read and replicate exactly what you did. SPSS kind of offers this functionality by having syntax (and the 'paste syntax' button is a good combination of exactly writing down your steps and the easiness of menu-driven software). Excel offers no such thing. (Technically: you can do it using a type of Visual Basic, but almost no-one does that, and by doing so you lose the easiness of menu-driven software and, with that, many of Excel's advantages).Anonymoushttps://www.blogger.com/profile/05364304504311348392noreply@blogger.com