English term
Germany
US-educated, precise & thorough
Native in: Russian
Works in: English to Russian, Russian to English, Russian
per cent or percentage points?
1.
The survey had a margin of error of 3 percentage points.
2.
VTsIOM, which surveyed 1,600 people in 138 locations nationwide, said the poll's margin of error was plus or minus 3.4 percentage points.
Thank you for your help.
| 4 +5 | percentage points |
Tony M
|
Non-PRO (1): Edith Kelly
When entering new questions, KudoZ askers are given an opportunity* to classify the difficulty of their questions as 'easy' or 'pro'. If you feel a question marked 'easy' should actually be marked 'pro', and if you have earned more than 20 KudoZ points, you can click the "Vote PRO" button to recommend that change.
How to tell the difference between "easy" and "pro" questions:
An easy question is one that any bilingual person would be able to answer correctly. (Or in the case of monolingual questions, an easy question is one that any native speaker of the language would be able to answer correctly.)
A pro question is anything else... in other words, any question that requires knowledge or skills that are specialized (even slightly).
Another way to think of the difficulty levels is this: an easy question is one that deals with everyday conversation. A pro question is anything else.
When deciding between easy and pro, err on the side of pro. Most questions will be pro.
* Note: non-member askers are not given the option of entering 'pro' questions; the only way for their questions to be classified as 'pro' is for a ProZ.com member or members to re-classify it.
Responses
percentage points
--------------------------------------------------
Note added at 12 minutes (2015-06-04 16:21:28 GMT)
--------------------------------------------------
We tend to use 'percent' when referring to some absolute value: "the election turnout was 57%" But we tend to use 'percentage points' when referring to some relative value: "that was 4 percentage points down on last year" In the first instance, it would sound distinctly odd to use 'percentage points' (i'd say arguably downright wrong), whereas in the latter, either sounds fine.
--------------------------------------------------
Note added at 31 minutes (2015-06-04 16:39:59 GMT)
--------------------------------------------------
I don't see why you say this doesn't apply to your context! What's so special or different about your context? I thought I had made it abundantly clear: when talking about an absolute value of percent, it is more natural (and I believe more correct) to use 'percent' — or indeed often the symbol %.
However, where relative percentage points are being talked about (your case exactly!), then I feel percentage points is not only acceptable, but to some extent even preferable.
Where a context contains BOTH absolute AND relative values, I think it is less confusing to simply stick with 'percent' everywhere — unless there is the potential ambiguity as highlighted by Donal.
'Percentage points' are ONLY really used in the rarer case of relative percentages where no absolutes are being cited. And then only usually when describing some trend etc.
--------------------------------------------------
Note added at 36 minutes (2015-06-04 16:45:02 GMT)
--------------------------------------------------
No of course it's not an 'absolute' value — the whole point of a 'margin of error' is that it is a 'relative' value, like a tolerance: "230V ±10%"
If it were an absolute value, it would imply that the MoE was say "10% of the whole sample set"
| I understand that, but my context falls under neither category |
| Actually, you're right: my context falls under the first category. Margin of error is "an absolute value"! |
| Here's why I believe it is an absolute value: http://en.wikipedia.org/wiki/Margin_of_error#Maximum_and_specific_margins_of_error |
| agree |
MPGS
: :-)
11 mins
|
|
Thanks, MPGS!
|
|
| agree |
DLyons
: Yes. // Standard procedure applied :-)
19 mins
|
|
Thanks, Donal! It was the Butler, in the Conservatory, with the Lead Pipe... sssh!
|
|
| agree |
Piyush Ojha
23 mins
|
|
Thanks, Piyush!
|
|
| agree |
Jack Doughty
4 hrs
|
|
Thanks, Jack!
|
|
| agree |
Lindsay Spratt
17 hrs
|
|
Thanks, Lindsay!
|
|
| neutral |
B D Finch
: My Sociology degree studies were a long time ago and jargon may have changed, but I took research methods course units for 3 years and would never have used the term "percentage points" for a margin of error.// Thanks, but I feel somewhat dated. :(
1 day 5 hrs
|
|
Thanks, B! I feel sure your real experience is just as valid today as it ever was; I bow to your superior knowledge.
|
Discussion
If percentage point and % mean one and the same thing in this context, then usage obviously comes down to preference.
And no, I do not understand statistics. I was just making an effort for this question. This is a translators' site, remember?
And it would seem you need to be more specific when using "per cent". Cilian in his example is computing x% of the outcome.
Is "The survey had a margin of error of 3 percentage points" the perfect wording or would it be more correct to say "The survey had a margin of error of 3 per cent."?
As I'm once again suspended from making peer comments, I posted here. Sorry.
https://books.google.ro/books?id=RYtYmMD2ReAC&printsec=front...
And did I miss Cillian's expressing his preference for "per cent"?
Ratios versus differences.
If publishers use "percentage points" to 'spell things out' for their readers instead of the more accurate and shorter "per cent", that's superfluous, isn't it? Of course, I defer to your expertise on this.
Getting back to the point... As the Wikipedia article explains and as I tried to argue, "margin of error" is a characteristic of an entire survey and should not be blindly applied to any particular percentage result, unless it's very close to 50%. That is why "percentage points" may be misleading: it gives the impression that the stated margin of error may be added and subtracted freely. Therefore, when talking about surveys as a whole (as in my question's two original examples), sample of error should, mathematically speaking, be referred to as per cent and not percentage points.
A 'margin of error' is surely by definition relative — to that which is being measured.
"The margin of error for a particular individual percentage will usually be smaller than the maximum margin of error quoted for the survey. This maximum only applies when the observed percentage is 50%, and the margin of error shrinks as the percentage approaches the extremes of 0% or 100%."
http://en.wikipedia.org/wiki/Margin_of_error#Maximum_and_spe...
I conclude that "margin of error" when applied to *the entire survey*--not to a specific individual estimate--is not relative, but absolute.
What do you think?
Suppose a survey polled people asking if they would buy a 12-ounce can of Coke for $1.00. Their estimate is, say, 54% of the total population. Since the sample is only a tiny share of the total population, we calculate a margin of error that tells us how accurate or inaccurate that estimate is. If the survey researchers estimate the margin of error at 3.4%, this means that in all likelihood, roughly between 50.6% and 57.4% of the total population would buy a can of Coke for $1.00. This likelihood is a variable you can tweak; typically it's 95%.
(continued in next post)
This, in my humble opinion, makes a difference.
I have a degree in Mathematics and I would not be asking this question if it related to interest rates.