On 24/10/15 21:10, Jim Lemon wrote:> Hi Ming, > In fact, the notation lb/1000 is correct, as the values represent the > weight of the cars in pounds (lb) divided by 1000. I am not sure why this > particular transformation of the measured values was used, but I'm sure it > has caused confusion previously.I disagree --- and agree with Ming. The notation is incorrect. Surely "lb/1000" means thousandths of pounds. E.g. 12345 lb/1000 is equal to 12.345 lb. I'm sure that others will come up with all sorts of convoluted lawyerish arguments that the case is otherwise, but as far as I am concerned, any *sane* person would interpret "lb/1000" to mean thousandths of pounds. If in the unlikely event that the documentation for some data set said "Weight (gm/1000)", I'm pretty sure that this would be interpreted to mean milligrams and *not* kilograms! Since the description of the data was presumably taken from that given in the original source ("Motor Trend" magazine) it would probably be inappropriate to "correct" it. However a note/warning should be added to the mtcars help file indicating that Motor Trend got things upside-down. cheers, Rolf -- Technical Editor ANZJS Department of Statistics University of Auckland Phone: +64-9-373-7599 ext. 88276> On Sat, Oct 24, 2015 at 11:59 AM, Ming-Lun Ho <minglho at gmail.com> wrote: > >> Hi, >> I used "?mtcars" to read the documentation for the dataset. I found a >> mistake in how unit is listed, namely, that for the variable "wt," the unit >> should be listed as "1000 lb," not "lb/1000." However, I don't know whom to >> contact exactly for the correction. Please point me to the right place. >> Thanks. >> --Ming
On 24/10/2015 6:07 PM, Rolf Turner wrote:> On 24/10/15 21:10, Jim Lemon wrote: >> Hi Ming, >> In fact, the notation lb/1000 is correct, as the values represent the >> weight of the cars in pounds (lb) divided by 1000. I am not sure why this >> particular transformation of the measured values was used, but I'm sure it >> has caused confusion previously. > > I disagree --- and agree with Ming. The notation is incorrect. Surely > "lb/1000" means thousandths of pounds. E.g. 12345 lb/1000 is equal to > 12.345 lb. > > I'm sure that others will come up with all sorts of convoluted lawyerish > arguments that the case is otherwise, but as far as I am concerned, any > *sane* person would interpret "lb/1000" to mean thousandths of pounds.And we insane ones would read "lb/1000" literally as "pounds divided by one thousand". The problem is that English is ambiguous. In many, many ways. We should rewrite all the help files in Loglan. Duncan Murdoch> If in the unlikely event that the documentation for some data set said > "Weight (gm/1000)", I'm pretty sure that this would be interpreted to > mean milligrams and *not* kilograms! > > Since the description of the data was presumably taken from that given > in the original source ("Motor Trend" magazine) it would probably be > inappropriate to "correct" it. However a note/warning should be added > to the mtcars help file indicating that Motor Trend got things upside-down.
I sanction this discussion. (Google on "auto-antonyms") Cheers, Bert Bert Gunter "Data is not information. Information is not knowledge. And knowledge is certainly not wisdom." -- Clifford Stoll On Sat, Oct 24, 2015 at 4:26 PM, Duncan Murdoch <murdoch.duncan at gmail.com> wrote:> On 24/10/2015 6:07 PM, Rolf Turner wrote: >> On 24/10/15 21:10, Jim Lemon wrote: >>> Hi Ming, >>> In fact, the notation lb/1000 is correct, as the values represent the >>> weight of the cars in pounds (lb) divided by 1000. I am not sure why this >>> particular transformation of the measured values was used, but I'm sure it >>> has caused confusion previously. >> >> I disagree --- and agree with Ming. The notation is incorrect. Surely >> "lb/1000" means thousandths of pounds. E.g. 12345 lb/1000 is equal to >> 12.345 lb. >> >> I'm sure that others will come up with all sorts of convoluted lawyerish >> arguments that the case is otherwise, but as far as I am concerned, any >> *sane* person would interpret "lb/1000" to mean thousandths of pounds. > > And we insane ones would read "lb/1000" literally as "pounds divided by > one thousand". > > The problem is that English is ambiguous. In many, many ways. We > should rewrite all the help files in Loglan. > > Duncan Murdoch > >> If in the unlikely event that the documentation for some data set said >> "Weight (gm/1000)", I'm pretty sure that this would be interpreted to >> mean milligrams and *not* kilograms! >> >> Since the description of the data was presumably taken from that given >> in the original source ("Motor Trend" magazine) it would probably be >> inappropriate to "correct" it. However a note/warning should be added >> to the mtcars help file indicating that Motor Trend got things upside-down. > > ______________________________________________ > R-help at r-project.org mailing list -- To UNSUBSCRIBE and more, see > https://stat.ethz.ch/mailman/listinfo/r-help > PLEASE do read the posting guide http://www.R-project.org/posting-guide.html > and provide commented, minimal, self-contained, reproducible code.
Franklin Bretschneider
2015-Oct-25 22:04 UTC
[R] [FORGED] Re: How to correct documentation?
Dear all, As to stating units in graphs: IMHO it should be as follows: If an axis reads 0 ... 10 ... 20 ... etc and the unit is pounds (lb), the legend should read "weight/lb" (pronounced "weight in pound"). The logic is: 10 lb/lb = 10. In orther words, dividing a dimensioned number by the dimension leaves the bare number, which is what the axis shows. So in the case of the OP the legend should read: "weight/1000lb". Thus, 20000 lb/1000lb = 20, which is what the axis shows. Best, Frank ---- Franklin Bretschneider Dept of Biology Utrecht University bretschr at xs4all.nl
Ming is right. I can't imagine other discipline's standards are substantially different from ours, but e.g. the ACS style manual is very explicit to require ... "Label each axis with the parameter or variable being measured and the units of measure in parentheses." The _units_ are not lb/1000. 1/1000 is a _transformation_ of the value, the unit is lb, or in that case (1000 lb). Writing lb/1000 is just as nonsensical as writing g/k instead of (kg). 1000 is simply a numerical prefix to the unit, like kilo. It gets worse: According to the UK metric association: "The symbol for "per" (meaning "divided by") is ?/? (slash)." Accordingly, "lb/1000" is to be read "pounds per 1000" which is actually wrong by six orders of magnitude. I don't think there is ambiguity here nor occasion for sophistry: as written, the label is wrong. It would be more than appropriate for a community that is passionate about data to correct this. :-) Boris (Good Lord! https://xkcd.com/386/) On Oct 24, 2015, at 6:07 PM, Rolf Turner <r.turner at auckland.ac.nz> wrote:> On 24/10/15 21:10, Jim Lemon wrote: >> Hi Ming, >> In fact, the notation lb/1000 is correct, as the values represent the >> weight of the cars in pounds (lb) divided by 1000. I am not sure why this >> particular transformation of the measured values was used, but I'm sure it >> has caused confusion previously. > > I disagree --- and agree with Ming. The notation is incorrect. Surely > "lb/1000" means thousandths of pounds. E.g. 12345 lb/1000 is equal to > 12.345 lb. > > I'm sure that others will come up with all sorts of convoluted lawyerish arguments that the case is otherwise, but as far as I am concerned, any *sane* person would interpret "lb/1000" to mean thousandths of pounds. > > If in the unlikely event that the documentation for some data set said "Weight (gm/1000)", I'm pretty sure that this would be interpreted to mean milligrams and *not* kilograms! > > Since the description of the data was presumably taken from that given in the original source ("Motor Trend" magazine) it would probably be inappropriate to "correct" it. However a note/warning should be added to the mtcars help file indicating that Motor Trend got things upside-down. > > cheers, > > Rolf > > -- > Technical Editor ANZJS > Department of Statistics > University of Auckland > Phone: +64-9-373-7599 ext. 88276 > >> On Sat, Oct 24, 2015 at 11:59 AM, Ming-Lun Ho <minglho at gmail.com> wrote: >> >>> Hi, >>> I used "?mtcars" to read the documentation for the dataset. I found a >>> mistake in how unit is listed, namely, that for the variable "wt," the unit >>> should be listed as "1000 lb," not "lb/1000." However, I don't know whom to >>> contact exactly for the correction. Please point me to the right place. >>> Thanks. >>> --Ming > > ______________________________________________ > R-help at r-project.org mailing list -- To UNSUBSCRIBE and more, see > https://stat.ethz.ch/mailman/listinfo/r-help > PLEASE do read the posting guide http://www.R-project.org/posting-guide.html > and provide commented, minimal, self-contained, reproducible code.
On Mon, Oct 26, 2015 at 10:44 AM, Boris Steipe <boris.steipe at utoronto.ca> wrote:> Ming is right.... Having started all this trouble, I suppose I should offer a modest explanation. The OP was indeed "right" in the sense that the column heading did not indicate the correct _units_ for the values. I suppose that "kilopounds" would be the correct units if such a unit was acceptable to the relevant standards committee. As Boris noted, lb/1000 is (sort of) the transformation used to get the values. Given the burning interest in this distinction between units (as used to explicitly back transform the values) and and explanatory labels (how did these values come to be?) I should state that the objection I refrained from adding to my original answer was, 'Why didn't they just leave the values in the initial units?" I was reminded of a long past physics lecturer's favorite units of velocity - furlongs per fortnight. Jim [[alternative HTML version deleted]]
On 25/10/2015 7:44 PM, Boris Steipe wrote:> Ming is right. I can't imagine other discipline's standards are substantially different from ours, but e.g. the ACS style manual is very explicit to require ... > > "Label each axis with the parameter or variable being measured and the units of measure in parentheses." > > The _units_ are not lb/1000. 1/1000 is a _transformation_ of the value, the unit is lb, or in that case (1000 lb). Writing lb/1000 is just as nonsensical as writing g/k instead of (kg). 1000 is simply a numerical prefix to the unit, like kilo. > > It gets worse: According to the UK metric association: > "The symbol for "per" (meaning "divided by") is ?/? (slash)."How is that relevant? We aren't trying to represent "per" here, we are trying to represent "divided by". The only valid argument I've heard so far is that we should use what the cited paper used. That was "1000 lbs", so I'll change it. Duncan Murdoch> > Accordingly, "lb/1000" is to be read "pounds per 1000" which is actually wrong by six orders of magnitude. > > I don't think there is ambiguity here nor occasion for sophistry: as written, the label is wrong. It would be more than appropriate for a community that is passionate about data to correct this. > > > :-) > Boris > > (Good Lord! https://xkcd.com/386/) > > On Oct 24, 2015, at 6:07 PM, Rolf Turner <r.turner at auckland.ac.nz> wrote: > >> On 24/10/15 21:10, Jim Lemon wrote: >>> Hi Ming, >>> In fact, the notation lb/1000 is correct, as the values represent the >>> weight of the cars in pounds (lb) divided by 1000. I am not sure why this >>> particular transformation of the measured values was used, but I'm sure it >>> has caused confusion previously. >> >> I disagree --- and agree with Ming. The notation is incorrect. Surely >> "lb/1000" means thousandths of pounds. E.g. 12345 lb/1000 is equal to >> 12.345 lb. >> >> I'm sure that others will come up with all sorts of convoluted lawyerish arguments that the case is otherwise, but as far as I am concerned, any *sane* person would interpret "lb/1000" to mean thousandths of pounds. >> >> If in the unlikely event that the documentation for some data set said "Weight (gm/1000)", I'm pretty sure that this would be interpreted to mean milligrams and *not* kilograms! >> >> Since the description of the data was presumably taken from that given in the original source ("Motor Trend" magazine) it would probably be inappropriate to "correct" it. However a note/warning should be added to the mtcars help file indicating that Motor Trend got things upside-down. >> >> cheers, >> >> Rolf >> >> -- >> Technical Editor ANZJS >> Department of Statistics >> University of Auckland >> Phone: +64-9-373-7599 ext. 88276 >> >>> On Sat, Oct 24, 2015 at 11:59 AM, Ming-Lun Ho <minglho at gmail.com> wrote: >>> >>>> Hi, >>>> I used "?mtcars" to read the documentation for the dataset. I found a >>>> mistake in how unit is listed, namely, that for the variable "wt," the unit >>>> should be listed as "1000 lb," not "lb/1000." However, I don't know whom to >>>> contact exactly for the correction. Please point me to the right place. >>>> Thanks. >>>> --Ming >> >> ______________________________________________ >> R-help at r-project.org mailing list -- To UNSUBSCRIBE and more, see >> https://stat.ethz.ch/mailman/listinfo/r-help >> PLEASE do read the posting guide http://www.R-project.org/posting-guide.html >> and provide commented, minimal, self-contained, reproducible code. > > ______________________________________________ > R-help at r-project.org mailing list -- To UNSUBSCRIBE and more, see > https://stat.ethz.ch/mailman/listinfo/r-help > PLEASE do read the posting guide http://www.R-project.org/posting-guide.html > and provide commented, minimal, self-contained, reproducible code. >