I found the first one (Excel errors) very interesting, in particular, because much of the coverage highlighted that a spreadsheet application is the wrong tool and that researchers should ditch it in favor of R or Python.
This makes sense.
However, it is rarely as easy as that much repeated sentence makes it sound. There are reasons why scientists use spreadsheets, and they may be ignored. Some ideas:
- Excel is well known. An unfamiliar tool may do even more damage. - Import and Export are rather easy (preview) in in Excel or Calc. This concerns usecases like cleaning data, sharing data, quickly eyeballing… - Along the lines: Eyeballing! An underestimated topic said Tuckey, too, quite a while ago! (Exploratory Statistics). If you know the syntax, you can easily create diagrams in R. But a GUI is easier and quicker to use and leads probably to fewer errors. Totally absent from (most) statistical packages is conditional highlighting of cells. All in all: Exploratory statistics are important but much undeserved by (open source) software.
In my opinion, yes, let's ditch Spreadsheets! But only after we can support the needs that is satisfies.
Jan
PS.: Some years ago, I did some minor qualitative research on use of statistical programs. If there is interest, I dig it up again. PPS.: For the eyeballing, Polestar is great ( https://github.com/vega/polestar ) and open source.
2016-08-28 21:02 GMT+02:00 Pine W wiki.pine@gmail.com:
Two articles of potential interest to medical and Wikimedia researchers:
http://www.nytimes.com/2016/08/28/opinion/sunday/do-you- believe-in-god-or-is-that-a-software-glitch.html
https://www.washingtonpost.com/news/wonk/wp/2016/08/26/ an-alarming-number-of-scientific-papers-contain-excel-errors/
Pine
Wiki-research-l mailing list Wiki-research-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wiki-research-l