# Talk:Numerical question type

## Additional content

Lesson Module said: This type of question requires a number as the answer. In it's simplest form it requires just one answer to be specified. For example "What is 2 plus 2?" with the answer 4 given a forward jump. However, it is better to specify a range because the internal rounding of numerical values can make single numeric comparisons rather hit or miss. Thus, if the question were "What is 10 divided by 3" it would be necessary to give the answer as Minimum:Maximum, that is two values separated by a colon. Thus if 3.33:3.34 is given as the acceptable range for the answer, then the answers 3.33, 3.333, 3.3333... would all be taken as correct answers. "Wrong" answers would include 3.3 (less than the minimum) and 3.4 (greater than the maximum).

More than one correct answer is allowed and the answers can be either single or pair of values. Note that the order in which the answers are tested is Answer 1, Answer 2... so some care needs to taken if the desired response is to appear. For example the question "When was Larkin born?" could have the single value of 1922, the exact answer, and the pair of values 1920:1929, the 20's, as the less exact answer.The order in which these values should be tested is, obviously, 1922 then 1920:1929. The first answer might have the response "That's exactly right" while the other answer's response might be "That's close, you've got the right decade, it is was actually 1922."

Wrong answers can be given but depending on their actual range, care should be taken to place them after the correct answers. For example in adding the wrong answer 3:4 to the "10 divided by 3" question it needs to come after the correct answer. That is the answers are ordered 3.33:3.34 (the "correct" answer) then 3:4 (the "wrong" answer, but not wildly wrong answer!).

Unfortunately this question differs from the numerical quiz question and the numerical embedded question (Cloze) in a couple of ways. 3:4 in those questions means "3 plus or minus 4", in other words from -1 to 7. The embedded question doesn't support interval boundaries. The numerical question, if imported in GIFT format, can use "3..4" as the interval from 3 to 4. Another difference is that those questions accept , as decimal in student answers, but the lesson numerical question doesn't.

- Please move text above to Numerical question type as appropriate. --Helen Foster 05:34, 11 September 2006 (CDT)

## Further possible additional content

(Prior to Moodle 1.7, you can only add one correct answer in the user interface. If you import the question with a GIFT format file you can specifiy multiple answer(intervals) with accompanying feedback and point-percentage. This is done in a similar way to the CLOZE Numerical format. There is no units support in the Cloze type.) It is possible, though not simple, to get support for several answer intervals **and** unit support if you create the question in the numerical interface and export it in Moodle XML format. Than you can duplicate the <answer> segment and put in another answer interval and the feedback and grading factor you want for that interval. Than import it again. You will not be able to edit the question in the normal numerical interface though.

Units can be specified and work to a degree. Unfortunately if the student answers with the right number but no unit he can get full points. And if he thinks of another unit and has the right number and no unit, he gets no differentiated feedback, just wrong. You must also give the conversion factor . So if your main answer was **5500** with unit **W** and you wanted to allow the unit **kW** you would have to specify the factor **0.001**. If you wanted to allow **Watt** you would use the factor **1**.

- Text above moved from Numerical question type as unsure whether it is still relevant in Moodle 1.7. Please move back if appropriate. --Helen Foster 09:45, 21 February 2007 (CST)

## Numerical CLOZE questions

From the student perspective, a numerical CLOZE question looks just like a short-answer Cloze question. (CLOZE can also be called *embedded* or *fill in the blanks*.)

The difference is that numerical answers are allowed to have an accepted error. This allows a continuous range of answers to be set. You can also express your answer in some different numerical formats. 23.4 23,4 (some countries use , as a decimal separator) and 2.34E+1 (meaning 2.34*10^1) would be interpretted as the same.

More examples: 0.5 accepts .5 0.5 ,5 0,5 0.500 5e-1 5E-1 but not 1/2 50% 50% accepts 50% 50.0% 5E1% 50/100 even50/1000 50but not 500/1000 0.5 1/2 accepts 1/21/3 1twentybut not 2/4 0.5 0,5 3/6 50% ½ ½ accepts ½ HALF doesn't even accept HALF (maybe0?) If you want to accept several variants you can have them in the same {} butbe careful, notice the "false positives" in bold!

The writing of a NUMERICAL CLOZE question is about the same as the other CLOZE questions and they can be mixed in the same question.

You write your question/incomplete text and where the student is to enter the numerical answer you write (Preferably in the source code mode, the RTF editor can insert linebreaks that make the question not function. So the linebreak is for readability in the example box below. A problem with these questions is the readability of the code! ):

{2:NUMERICAL:=23.8:0.1#Feedback for correct answer 23.8 ~%50%23.8:2#Feedback for ½credit near correct answer}.

In this example

- 2: is the question point weight which would say that this question has twice the weight in the final point(s) for this question as other partial answers with weight 1 (or no declared weight you can start {: for weight 1) in the same question.
- NUMERICAL: says what kind of question it is. It must be in CAPS.
- =23.8:0.1 = or %100% means correct if the answer is 23.8 with an accepted error of 0.1, then any number between 23.7 and 23.9 will be accepted as correct. (In the GIFT numerical question one can express an interval like this 13..15 or 14:1 but in CLOZE only 14:1 works.)
- #Feedback for correct answer 23.8 is preceded by #
- ~%50%23.8:2 ~ is the separator for answer alternatives %50% means this answer would get 50% of the score that the more precise answer had gotten. Because the tolerans here is 2, 21.8 to 25.8 would get this point and feedback

The feedback (which is seen as OverLib popup windows when the user hovers over the answer space) is formatable with HTML tags. For example if you want an exponent, surround it with superscript tags: <sup> </sup>. You can even include pictures in the feedback popup but you must clean out all " and save while still in code mode (not RTF). This works in feedback popup:

#See this picture:<br><img src=Something.gif />}

but not

#See this picture:<br><img src="Something.gif" />}

(ALGEBRA and TEX filters don't work in the feedback popups, but they can be very useful in the question writing for math/science expressions).

If you want to give feedback for any answer that didn't fit the intervals you already have specified feedback for, add some BIG general intervals, like for positive answers (if they aren't bigger than 20000 you could add:

~%0%10000.0001:10000#Feedback for unspecified not_right answers}

This would give feedback for anything from 0.0001 to 20000.0001 (that hadn't already gotten feedback). I didn't want to include 0 since that special case as well as negative ought to have specific reactions.

~%0%0#Hey! It can't be zero ~%0%-10000.0001:10000#We just want the size here, so a negative value is not what we want}

- Text above moved from Numerical question type as unsure whether it should be on a different page. Please move back if appropriate. --Helen Foster 09:45, 21 February 2007 (CST)