Why our reliability ratio is not shown publicly?
Thread poster: Christel Zipfel

Christel Zipfel  Identity Verified
Member (2004)
Italian to German
+ ...
Aug 4, 2006

I always wondered why we can see only our own statistics (if members). It is right that we might make our calculations case by case, dividing points by answers given. But why should we do this, if the result is already there, and in a very elaborated manner?

Maybe this feature is meant only for self-criticism purposes?icon_confused.gif

It is true also that these numbers are expressive only relatively, meaning that there has been no better answer, but I think they are expressive anyway, mainly in a long term view.

I do believe that the reliability ratio (statistics) would be a pretty better indicator than the total points which doesn't say anything.

What do you think?


 

gianfranco  Identity Verified
Brazil
Local time: 14:16
Member (2001)
English to Italian
+ ...
Cumulative effect in KudoZ Aug 4, 2006

Christel Zipfel wrote:

I always wondered why we can see only our own statistics (if members). It is right that we might make our calculations case by case, dividing points by answers given. But why should we do this, if the result is already there, and in a very elaborated manner?

Maybe this feature is meant only for self-criticism purposes?icon_confused.gif


I find very useful to have some personal statistics, but it would be also interesting to have them shown, somehow and somewhere, and in an easy, acceptable format.



It is true also that these numbers are expressive only relatively, meaning that there has been no better answer, but I think they are expressive anyway, mainly in a long term view.


Yes. Their value is significative only above a given threshold, to be decided. For example, after the first 200 answers, or after reaching 500 points.



I do believe that the reliability ratio (statistics) would be a pretty better indicator than the total points which doesn't say anything.

What do you think?



The KudoZ leaders boards (and the directories too) suffer of a cumulative effect. In practical terms, in many communities, the early prozians may have accumulated such an advantage that any newcomer tend to be not very visible for a long period.

Reclassifying both lists (kudoz leaders and directory) not only on the basis of the total number of points, but also, above a certain level, considering how a leading position has been achieved, could be a lot fairer.

Anyone reaching, let's say, 500 points (or for the the first 20 positions at the top, whatever the score) could be listed using a combination of its "total score" and a "scoring ratio", and such combination could be a lot more relevant than the simple score.

A numerical example:

800 points obtained with 400 answers (20 points every 10 answers) is good or excellent

800 points, obtained with 4000 answers (2 point every 10 answers, on average), is not very good at all...

The two performances have not the same value.... the first answerer is very reliable, the second is not, but offering a large number of tentative answers, in 5 years, may appear level or better than other more knowledgeable colleagues.

At the score we could apply a weight, a ratio, a factor (2 in the first case, 0.2 in the second) and obtain a different grading of the top scorers.


bye
Gianfranco


[Edited at 2006-08-05 05:23]


 

Ford Prefect  Identity Verified
Burkina Faso
Local time: 17:16
German to English
+ ...
... Aug 4, 2006

Gianfranco Manca wrote:

I find very useful to have some personal statistics, but it would be also interesting to have them shown, somehow and somewhere, and in an easy, acceptable format.


I'm in favour of having the statistics shown, but I notice mine are not entirely accurate - if I multiply the number of correct answers by 4 (max points per answer) I end up with less points than I actually have. So that bug would have to be fixed.


The KudoZ leaders boards (and the directories too) suffer of a cumulative effect. In practical terms, in many communities, the early prozians may have accumulated such an advantage that any newcomer tend to be not very visible for a long period.


There is a cumulative effect problem with the directory and leaderboard but it is not insurmountable. I for one would benefit considerably from searches being based on reliability rather than total points but as I have vociferiously opposed other recent changes to the directory system I am not going to advocate this one. The directory is very important to some members securing work so any changes should only be made with broad agreement.


 

Charlie Bavington  Identity Verified
Local time: 18:16
French to English
Similar discussion and, er, stuff Aug 4, 2006

See a similar discussion in the last few pages of this thread:
http://www.proz.com/topic/51612


James Visanji DipTrans PhD wrote:

I'm in favour of having the statistics shown, but I notice mine are not entirely accurate - if I multiply the number of correct answers by 4 (max points per answer) I end up with less points than I actually have.


As your statement hints, the reason that correct answers x 4 does not = your points is that you don't always get 4 points per answer. 4 is the max. I quite often see people giving fewer points when the Q is along the lines of "am I right in thinking that xxxx means yyyyy?" and the answer is "yup, you are".

[Edited because a less than sign followed by a greater than sign did not show up as 'not equal to' as I had intended]

[Edited at 2006-08-04 22:57]


 

Uldis Liepkalns  Identity Verified
Latvia
Local time: 20:16
Member (2003)
English to Latvian
+ ...
Yes, besides Aug 4, 2006

new profile design counts only Pro level points, while shows as answered sum of answers both to Pro and Non Pro questions.

Uldis

Charlie Bavington wrote:
I quite often see people giving fewer points when the Q is along the lines of "am I right in thinking that xxxx means yyyyy?" and the answer is "yup, you are".


 

Henry Hinds  Identity Verified
United States
Local time: 11:16
English to Spanish
+ ...
Ratio of answers accepted to questions answered Aug 5, 2006

I would like this.

This might be the best way, indicating not by points (they can be 1 to 4), but by the Ratio of answers accepted to questions answered (for example, 500/1,000, 50%).

It could include all questions, Pro and non-Pro, or if desired separate totals could be devised, but I think just one overall number would be best.

That would be useful, plus even those with a relatively low point total, maybe because they are relatively new to the site, can still distinguish themselves.

In fact someone who answers their first question and gets it right could have 100%!

Of course it would all be looked at in perspective.

Of course I have a lot of points, but then again, how would my ratio stack up? It would be nice to know.


 

Ford Prefect  Identity Verified
Burkina Faso
Local time: 17:16
German to English
+ ...
... Aug 5, 2006

Charlie Bavington wrote:
As your statement hints, the reason that correct answers x 4 does not = your points is that you don't always get 4 points per answer. 4 is the max. I quite often see people giving fewer points when the Q is along the lines of "am I right in thinking that xxxx means yyyyy?" and the answer is "yup, you are".


Charlie, I have 647 points but according to the stats, 160 correct answers out of 310 submitted. So how did I get an average of above four points per correct answer? I must have more than 160 correct answers to have 647 points!


 

Christel Zipfel  Identity Verified
Member (2004)
Italian to German
+ ...
TOPIC STARTER
Sorry! Aug 5, 2006

Charlie Bavington wrote:

See a similar discussion in the last few pages of this thread:
http://www.proz.com/topic/51612




I followed this discussion for quite a while, but then stopped just before Nikki brought up her proposal which I read only today. Indeed, it was thinking about this issue (rubbish answers and how we could get them under control) that I had this idea...

If the reliability ratio is very low, at least the asker becomes suspicious...icon_smile.gif

And, hopefully, those answerers would think twice and wait five minutes before posting other rubbish that would worsen their RRicon_wink.gif

[Bearbeitet am 2006-08-06 08:12]


 

xxxdf49f
France
Local time: 19:16
don't be sorry!! Aug 5, 2006

Christel Zipfel wrote:
Charlie Bavington wrote:
See a similar discussion in the last few pages of this thread:
http://www.proz.com/topic/51612

I followed this discussion for quite a while, but then stopped just before Nikki brought up her proposal which I read only today. Indeed, it was thinking about this issue (rubbish answers and how we could get them under control) that I had this idea...
If the reliability ratio is very low, at least the asker becomes suspicious...icon_smile.gif


Don't be sorry Christel! on the contrary, it was an excellent idea to post a brand new thread on this subject, which ended up gettting lost at the end of Ian's very long thread and deserves further attention and discussion.
dficon_smile.gif


 

Charlie Bavington  Identity Verified
Local time: 18:16
French to English
absolutely Aug 6, 2006

As df said, no need to apologise. I didn't intend to kill the thread, or anything. I just thought it would be useful to point out that a few other people agree with the idea, and there had been some discussion already. Just trying to save a bit of timeicon_smile.gif

While I think that it would be tidier and more logical for all discussion on a given point to be under a single thread, df is quite right, the RR discussion wasn't strictly relevant to the original post title.

And until such time a Proz sorts out the home page so it displays in order of the most recent postings and not simply when the thread was created, I guess we have no choice than to "bump" by creating new threads....


 

Christel Zipfel  Identity Verified
Member (2004)
Italian to German
+ ...
TOPIC STARTER
Where is the staff, really? Aug 8, 2006

As pointed out in another discussion a few days ago (I don't remember which one), we are trying to propose suggestions in order to make ProZ possibly better than it is, but we don't get any feedback from the other end.icon_rolleyes.gif

Should we better keep quiet and not waste our time furthermore?

And while I am here, I have another suggestion/question:
why in the new profile only the last page of answers/questions is displayed? Wouldn't it be more useful to be able to see ALL the answers/questions like in the old profile?

Anyway, happy vacations to everyone!icon_wink.gif


 

Prawi  Identity Verified
Austria
Local time: 19:16
German to Italian
Fully agree Aug 20, 2006

It's an excellent idea. Number of questions answered (pro + non pro)/ reliability ratio. I think it would be fairer to all - even to the newcomers - and moreover it would really improve the quality of given answers. Hope the staff is listening /reading..

Paola


Henry Hinds wrote:

I would like this.

This might be the best way, indicating not by points (they can be 1 to 4), but by the Ratio of answers accepted to questions answered (for example, 500/1,000, 50%).

It could include all questions, Pro and non-Pro, or if desired separate totals could be devised, but I think just one overall number would be best.

That would be useful, plus even those with a relatively low point total, maybe because they are relatively new to the site, can still distinguish themselves.

In fact someone who answers their first question and gets it right could have 100%!

Of course it would all be looked at in perspective.


 


To report site rules violations or get help, contact a site moderator:


You can also contact site staff by submitting a support request »

Why our reliability ratio is not shown publicly?

Advanced search






Anycount & Translation Office 3000
Translation Office 3000

Translation Office 3000 is an advanced accounting tool for freelance translators and small agencies. TO3000 easily and seamlessly integrates with the business life of professional freelance translators.

More info »
BaccS – Business Accounting Software
Modern desktop project management for freelance translators

BaccS makes it easy for translators to manage their projects, schedule tasks, create invoices, and view highly customizable reports. User-friendly, ProZ.com integration, community-driven development – a few reasons BaccS is trusted by translators!

More info »



Forums
  • All of ProZ.com
  • Term search
  • Jobs
  • Forums
  • Multiple search