Kibbitzer 6

Data: singular or plural?


Students raise this question so frequently that it might be best to try to deal with it in general terms: should data be used as a singular or as a plural noun? The historical answer is clear: the Latin datum is singular and data is its plural. The historical answer, however, may not be a good guide to current usage. The word agenda, for example, was originally plural (from agendum: 'something to be acted on') but is nowadays used only as a singular, and nobody in their right mind would insist that it should be used as a plural.

The following table summarises the results of of an investigation of 750 randomly-selected citations for data from each of the following publications:

The table shows the frequency of data in each publication: the proportion of citations in which data is unambiguously marked as plural or singular: and, of those, the proportion in which it is marked as plural and the proportion in which it is marked as singular:

PublicationFrequencyMarked plur/singPluralSingular
Nature1:563 187 (23.6%)175 (93.6%)12 (6.4%)
New Scientist1:1,970123 (16.4%)84 (69.9%)39 (30.1%)
Guardian1:22,04781 (10.8%)32 (39.5%)49 (60.5%)

The table shows that:

  1. In 'high-level' writing on science (Nature), the use of data as a plural is dominant: in more popular writing on science (New Scientist) its use as a singular form is more frequent, though it is still outnumbered by its use as a plural: and in educated everyday usage as represented by the Guardian newspaper, it is nowadays most often used as a singular.

  2. The difference between the use of data as a plural or singular correlates with the frequency with which the word is used (it is 40 times more frequent in Nature, and over 10 times more frequent in New Scientist, than in the Guardian), and also with the proportion of instances that are clearly marked as plural or singular (the highest proportion being in Nature and the lowest in the Guardian, with New Scientist again occupying an intermediate position).

  3. The choice of plural or singular (and thus publication, frequency, and marking) also correlates with the meaning of the word. As the citations below (all taken from New Scientist) indicate, the (traditional) meaning "evidence used in experimental procedures" is most often plural, while the (more recent) meaning "digital information stored or manipulated by a computer" is most often singular. The best advice I can give is that if you are using the word in the first meaning, you should use it as a plural, but that you are probably quite safe in using it as a singular if it has the second meaning.
 1 orresponding to its importance. The weighted data are then processed to produce a result. The
 2 n of the experiment contained the gene. "Our data suggest that distances from engineered crop
 3 What is perhaps more interesting is that the data appear to be independent of peer group. The
 4 ghts into the way HIV might spread even when data on sexual behaviour are scarce. The first s
 5 mber of animals taken in past seasons. Those data go into an assessment process in which the 
 6  analysis of the registry by the NRPB. These data show a correlation between lifetime exposur
 7 tain about which treatments to use. But when data from large numbers of women are systematica
 8  1989 and 1990, the last two years for which data are published, British discharges of heavy 
 9 sow the benefit wears off over time, but the data are "blind" so no one knows except the inde
10 known as Korner returns.  To cope with these data, hospitals bought large mainframe computers

11 ecord their morning round of checks. All the data is then dumped into the main computer. "If 
12  normal circumstances, only half the digital data in the store is read. If the player is jogg
13 or the Inmarsat A satellites. The compressed data is fed onto the hard disc of a powerful per
14 otates at twice the normal speed so that the data is read off the disc twice as fast. All the 
15 an just charitable feelings," says Dubnick. "Data is being digitised because it's easier to a
16 use, for a given image quality, it used less data than DCT-based systems. `The data rate with 
17 possible for a long time, if at all".  Video data runs at a very high rate, too high for voic
18 nual layers downwards from the surface. "Our data strongly indicates that the last glaciation 
19 ine research, concluding that "the available data provides no basis for testing any experimen
20  nuclear engineers also like to back up this data by examining samples from the pressure vess
Note that when it is used as a plural the word data cannot be preceded by a numeral: so instead of *"There are three data which support this hypothesis" write "There are three sets of data which support this hypothesis" or (depending on the situation) "There are three pieces of evidence which support this hypothesis."

March 1997 footnote. It seems that crtteria (plural of criterion) has begun to journey along the path completed by agenda and on which data is already well under way. Of 1,504 citations from the Guardian for 1994 and Times for 1995, 14 (1%) are marked as being singular:
  1. A criteria which measures the value of shopping, cleaning, cooking, caring for and maintaining oneself and one's intimates in terms of its economic contribution, for example, can find no grounds to contest the replacement of houseworkers and lovers by domestic servants and prostitutes.
  2. Ask the pilots if you want a criteria for failure.
  3. Detailed criteria has been set out as to what is meant by competence.
  4. But optimists should still believe that ``by every criteria - from compassion to self-interest - a reduction in inequality is the sensible choice.''
  5. Our main criteria is to create products based on or inspired by Trust property.
  6. The target they have to meet was reasoned out in 1957, and is known as the Lawson criteria. It is the formula conceived by J D Lawson, a British physicist, for calculating the combination of temperature, plasma density and time needed to sustain a fusion reaction.
  7. Most relied on parents'; geographical proximity to the school as at least one criteria, despite the government's attempts to encourage parents to shop around, according to the study of 3,000 parents across 10 local authorities.
  8. Using 'key negotiators' in each relevant regional office, Uni-key seeks out properties which fit the strict guidelines set out by universities (within a bike ride of the campus is one criteria).
  9. What other criteria is there besides ability?
  10. Unless the criteria is published then the explanation remains a confidence trick.
  11. ''In the odd place where creativity meets commerce, which is Hollywood among other places, their criteria is very simple: to make money.
  12. The system has already tuned in to this criteria.
  13. MHEP fits this criteria.
  14. While safety must always be the ultimate criteria, there is no reason to ignore technological improvements made since Taylor reported.
It will be interesting over the next 10-20 years to see whether criteria continues on the route towards singularity.

Your comments would be welcome!
16th June 1996 and 3rd March 1997 Consultant: Tim Johns
Back to Tim Johns EAP page