Tests & Quizzes Calculated Questions Scientific Notation
Tiffany Stull
Hi all,
I've discovered an issue where the Samigo Tests & Quizzes Calculated Question type can mark correct answers as incorrect. The situation is the following:
- The correct answer is a LARGE number that automatically converts to scientific notation, i.e., 1E11 or bigger.
- The numbers following the decimal point in the scientific notation (in front of the E) would be rounded by Samigo to the decimal place specified in the question's settings.
Here's an example question:
Question Text: {x}^2 = {{y}}
x is any number between 500000 and 1000000, with 2 decimal places
Tolerance is default for the question type: 0.01
Decimal places for answer: 3
One instance of this question that comes up for an example
student is: 871542.68^2 = ___
The answer provided by my TI-83 calculator (which also has a
scientific notation conversion threshold of 1E11) is 7.595866431E11,
but if I type this directly into the answer box, Samigo counts it
incorrect and gives me 7.596E11 for the correct
answer.
Similarly, 759586643061 is counted as incorrect (it SHOULD
be the correct answer), but 759600000000 is marked
correct.
As expected, tolerance applies to the true number Samigo calculates for the answer (ignoring scientific notation).
In the case of the question above, the 0.01 tolerance applies to
759600000000.00. Samigo would mark any value between
759599999999.99 and 759600000000.01 as correct. Thus, for
this question, I would need to set a MASSIVE tolerance (in the
millions) to make Samigo count my calculator's correct answer as
correct. If I did that, a student entering any value in a range
of several million would get the question marked correct.
So the question is: how should Samigo behave when a LARGE number
for the Answer is converted to scientific notation? I
assume it should do the following:
- Calculate the answer.
- Round to the appropriate actual decimal value specified in the question BEFORE converting to scientific notation!
- Apply tolerance on the value calculated in step 2
above to determine the range of acceptable correct answers.
- Convert the answer to scientific notation for displaying it in
the Answer Key (an aside: it might also be nice if the
answer key showed both the number in scientific notation and the
actual value of the number, because either would be correct).
If my assumption is correct, I'll create a bug jira with this information; if not, please let me know what should be the correct behavior!
Thanks,
Tiffany
Shawn Foster
Is what you are seeing related to the rounding issue with Calculated
Questions from SAM-3258?
https://jira.sakaiproject.org/browse/SAM-3258
-Shawn
--
Shawn Foster
eLearning Technology and Application Development
Western Technology Services
Western University
London, Ontario, Canada
On 2018-02-23 3:25 PM, Tiffany Stull wrote:
> Hi all,
>
> Question* type can mark *correct answers as incorrect*. The situation
> is the following:
>
> 1. The *correct answer* is a LARGE number that automatically converts
> to scientific notation, i.e., *1E11 or bigger*.
> 2. The numbers following the decimal point in the scientific notation
> (in front of the *E*) would be *rounded* by Samigo to the decimal
>
> Here's an example question:
>
> *x* is any number between *500000* and *1000000*, with *2* decimal
> places
> *Tolerance is default for the question type:* 0.01
> *Decimal places for answer:* 3
> One instance of this question that comes up for an example student is:
> *
> type this directly into the answer box, Samigo counts it *incorrect* and
> gives me *7.596E11* for the correct answer.
>
> Similarly, *759586643061* is counted as incorrect (it SHOULD be the
> correct answer), but *759600000000* is marked correct.
>
> As expected, *t**olerance* applies to the true number Samigo calculates
>
> In the case of the question above, the 0.01 tolerance applies to
> and 759600000000.01* as correct. Thus, for this question, I would need
> calculator's correct answer as correct. If I did that, a student
> entering any value in a range of several million would get the question
> marked correct.
>
> So the question is: how should Samigo behave when a LARGE number for the
> following:
>
> 1. Calculate the answer.
> 2. Round to the appropriate *actual **de**cimal value* specified in the
> /Answer Key/ (an aside: it might also be nice if the answer key
Matthew Jones
--
You received this message because you are subscribed to the Google Groups "Sakai Users Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sakai-user+...@apereo.org.
To post to this group, send email to sakai...@apereo.org.
Visit this group at https://groups.google.com/a/apereo.org/group/sakai-user/.
Tiffany Stull
Hi Shawn and Matthew,
Unfortunately, I don't think the issues in the jiras you mentioned below will help with this because it looks like those are fixed in 12 and this issue is present everywhere I've tested. I apologize for not including this information in my original message - I tested on the following servers:
I'm also seeing the same issue in UVA's 11.2 instance, on both a server with and a server without backport of SAM-3227 and related follow-on fixes. Without SAM-3227, the issue is more egregious because before SAM-3227, any number greater than 10000 was converted to scientific notation. Thus, without SAM-3227, smaller numbers are ALSO affected by the rounding.
To summarize: the problem only affects LARGE numbers. It affects them because scientific notation conversion takes place BEFORE rounding.
Tiffany