Note: DO NOT take this debate if you are unfamiliar with the terms used in my case. If you are unclear on the topic but are interested, please ask in the comments, and then take it after I answer. Also, being from Australia there might be small differences with the USA education system (or systems) .
Resolved: Point deduction is a better test marking system than the one currently in place
Often people question things in education like standardised tests, what should be evaluated etc. One thing that is not properly thought of is how things (ie. Tests) should be evaluated. Since evaluation of students (tests or no tests as an example) and the method behind each form of evaluation (ie. The 2 systems in this debate) are critical to assessing what students have learnt, it is key to examine how tests are scored.
3. Pro (myself) , is allowed to expand upon the point deduction system to refute criticisms from Con (ie. Con asks about working marks for Maths, I clarify as to how that would work under point deduction)
The 2 Systems
To give an idea of the 2 systems, we’ll start with the standard system. Let’s say there’s 100 1 mark questions for a 100% score of 100. Let’s also say that the students gets 50 correct and 50 wrong – with all the wrong answers being incorrect answers (as in to say they answered) .
50 correct questions = 50 marks
50 wrong answers = 0
Total = 50/10
Note that if the person did not answer the questions they got wrong that the score would not change – ie. ‘marks are not deducted for incorrect answers’ , so we can change these 50 non correct answers into any combination of incorrect and not attempted questions and get the same score – guessing is rewarded.
Now, here’s how my proposed point deductions system works:
1. Each question in the test is assigned a rank (lower numbers indicate tougher questions when ranks are varied) , similar to the current system (in the scenario described above, 1 mark for each question) , which is used to score the test
2. Students start with perfect scores (this can be changed, but it’s complicated)
3. If a student gets a question right, he/she loses no marks
4. If a student does not attempt a question, he/she loses marks equal to the rank
5. If a student attempts a question and gets it wrong, he/she loses marks equal to 2.5 times the questions ranks (hence, lower numbers being tougher questions – less penalties for being wrong because of higher question difficulty)
So let’s say we take the same test under point deduction. The student starts with a perfect score of 250 (maximum possible loss of marks, assuming all questions are rank 1). Like before he gets 50 questions correct, so his score stays on 250. Now, however comes the key point. The student’s score, depending upon the no. of incorrect answers vs. non-attempts, will vary between 125 and 200.
For example, let’s say the student got 25 wrong and had 25 non-attempts, his score will be 162.5. However, let’s say the student gets 15 wrong and has 35 non-attempts, his score will be 177.5. Therefore, this system creates marks based upon 3 categories – correct, incorrect and non-attempts, vs. the current system with only correct and incorrect.
Problems with guessing
Okay, as one might be able to deduce, the major problem with the current system is guessing. Now, what is wrong with this? What is wrong is that tests are meant to determine if a student has learnt what they have been taught, and this is done through forms of questioning (ie. Multiple choice) , and these forms of questioning can produce incorrect results, due to guessing.
As an example, let’s say we have 8 multiple choice questions, with answers A, B, C, D and our poor friend Billy has no clue as to the answer of these questions whatsoever or which answers are more likely. Clearly, what a test marker wants to know is does Billy know the answer to these questions; to which the answer is no. However, Billy knows that guessing wildly is not punished, so he answers A on all the questions. On average, Billy will get 2 questions right.
Now, this will result in those who wished to garner what Billy has learnt being given an inflated view of his ability’s.
Conversely, if Billy knows that there’s a strong penalty for incorrect answers (ie. 2 incorrect answers = 5 non-attempts), it’s not in his best interests to wildly guess and he won’t guess, leaving the questions blank and giving the examiner a better understanding of Billy’s knowledge. Of course, if Billy does decide to guess, instead of there being an incentive to guess, there is now a disincentive to guessing.
Let’s show briefly why guessing is a bad policy to be encouraging, teacher feedback aside.
Scenario: David wants to buy some shares. Unfortunately, he does not know some key financial terms and is only 2/3 (arbitrary) sure that he is making a sound financial investment. Also, this investment would constitute a substantial investment for David, ie. Unlike probability where if he just does this sort of thing 3 times, he’ll likely do well, David can only afford to make one major investment.
Now, would David invest based on an educated guess? Would David do better to actually know something about what he was investing in? Undoubtedly, David would do far better to know what he was investing in and key points about it.
While the primary reason behind testing is for feedback to teachers and such, students also get feedback from their test results. These results tell them how they did, what they knew/ didn’t, what they did wrong and how to fix it etc. Now, when a student gets back a test paper under the current system, if he’s interested in his results, he’ll likely look at the questions he got wrong, but not the ones he got right.
However, since guesses would give the student extra correct questions, under the current system the student will be looking at less problems and faults in their knowledge – a student will not check correct answers vs. incorrect answers in all likelihood. This means that the student will not realise (and later perhaps improve on) their faults and not learn it to a sufficient degree.
These reasons are enough to provide strong grounds for a change of test scoring system. With this, I turn it over to Con.
These reasons are enough to provide strong grounds for a change of test scoring system. With this, I turn it over to Con.
Resolved: Point deduction is a better test marking system than the one currently in place
Pro proposes a scheme where the value of questions is 2.5* rank if correct, -2.5*rank if attempted and are wrong and -rank if just left out. There is a very simple illustration as to why this system is extremely imbalanced.
A1) Let us make a test with 10 questions, all of equal rank.
The student looking at the test sees two questions that are both easy to answer and they are certain of the answer. They thus answer those two questions and the spend the rest of the time wondering when will the Reptoids will contact them so they can ascend to the Illuminati. Here is the amazing part, the percentage score they get is 68% and they pass (in the comments Pro clarified that 65% is the pass mark) with only two out of ten questions, or 20% of the testing material answered.
If clarification is required, their percentage mark is
(2.5 * rank * 10 - 8 * rank)/(2.5 * rank * 10)
which reduces to
1-8/25 = 0.68 and 68%
Or to look at it another way, they start off with 25 marks, they leave out eight questions and thus they score (25-8)/25 = 68%. Now if Pro attempts to counter with, oh wait what if they get one of them wrong, well yes they fail, but this is just mean to illustrate just how imbalanced the system is in terms of what it will produce and this is because the reward for getting an answer right is so much higher than just leaving it out.
If one wants a more general refutation then just consider the following; this is a list of percentages for answering the questions right from one question to ten and leaving the rest out - 64, 68, 72, 76, 80, 84, 88, 92, 96, 100. Note the score starts off at 64% for just one question right! This means if the test was only nine questions then you would actually pass with just one question answered and the rest left out.
Now to complete this with more rigor you would want to use the general formula for expected gain (summation of expected gain * probability of gain) but even without this it should b obvious that they proposed system is extremely imbalanced and leads to scenarios which are even on the surface just plain odd and even Michael Tsarion would consider it too unbelievable to support.
If this is not convincing enough to refute the resolution then the additional two arguments can be made :
A2) It is very difficult for a student to determine what to do in limited time.
In the real world the student is often in a situation where they are deciding which questions to attempt as they are pressed for time and as the hour draws near they are thinking should they recheck or continue with new questions. With the current marking system this determination is not overly complicated but nearly impossible to figure out what to do for the student during a test. Imagine if you had to start figuring out what your best path is if the questions have different ranks and you think you know one question maybe 60% certain but it has rank 5 and you know another one of rank 9 but only 40% certainty and you know only almost perfect at 95% certainty but it is of rank 1. Trying to figure out what to do here is almost worst than trying to answer the questions.
A3) Marking the tests is much more complicated
One of the deciding factors in how much testing to do is actually the amount of time to mark, grade and tabulate scores, more so when there are grading curves. The above system makes it much more demanding on teachers as they now have to multiply ranks, take into account attempts vs blanks and of course when they are making the tests they have to look at question ranks and decide and adjust to make a fair test. This is a lot more work and as the above shows it produces an imbalanced system anyway.
To close with a quote, David Icke was asked for commentary on the resolution that such a scheme would offer a significant advantage :
"What, that is insane - no one would believe that, stop bothering me with trivial questions. Wait, did you just blink with a second eyelid - REPTOID!"
In short :
-the system is extremely complicated for the student
-it puts more work on the teachers
-it is severely imbalanced
The resolution is negated.
Firstly, many thanks to Cliff for taking the debate. He has provided strong criticisms so far, especially in important details I had not necessarily thought of. Let us return to the debate.
Con makes 3 points in his R1 about point deduction:
It is a complicated system for the student
I will address these points while furthering my case where possible below.
Complications for the student
On this point, Con gives us an example of involving certainties and ranks. He then proceeds to say that students will have difficulty weighing up various scenarios. There are a few problems with this point:
1 – Students already evaluate such cases. There is only a marginal difference in point deduction.
Tests already include such situations. In fact, different mark questions (ie. 1,2,3,4) are essentially the equivalent of ‘ranks’ for question difficulty. In both systems, the student is faced with such complications, if they are calculating this.
The only difference for the student in point deduction is that they must consider whether or not to answer a question. However, this is A, not overly complicated and B, is actually included for other benefits – a primary reason for point deduction is to prevent guessing, which skews test results.
2 – Racing against the clock is only a problem for some.
Obviously, if a student finishes a test before the time is done (a reasonable amount of time) , then it is likely then have not guessed. The only students that could be affected by these asserted ‘complications’ (see 1) are those underprepared and who are guessing haphazardly. However, this is the wanted effect – point deduction is used primarily to try and discourage random guessing, a thing the current system does not, and cannot prevent.
3- It is doubtful that worse performing students actually calculate such percentages as thought by Con
I ask readers, do you truly think that a guessing person who is likely a person who doesn’t try in class and such actually calculates the likelihood of their getting questions correct and then proceeds to answer? I think not. And if the students who are meant to face this ‘additional complication’ already don’t do this in identical circumstances (see 1) , then why would then do this under point deduction?
Firstly, I’m going to concede that the marking point for the pass/fail score is not ideal. It wasn’t something I’d thought of, so Con wins that point. However, I believe that this is not a major problem (it could be rectified very easily) , and the general idea of the system is still strong. Essentially, a fixing of the pass/fail marks would be no problem for large organisations (private sector, government etc.). Also, given that different tests have different pass mark percentages, this issue would be dealt with adequately at some point and is minor because under both the current system and a point deduction system, pass/fail marks are not uniform. Below are some other points on this subject (marking imbalance):
1 – Pass marks are not everything to a test.
For example, let’s say we have a 20 question test and everyone in the class gets in the range of 16-20. The pass mark is 10. Also, for convenience, assume that there are some people who nearly always fail tests. Has the class suddenly skyrocketed in performance? Perhaps, but it is more likely to be an easy test.
Tests are meant to assess outcomes. In the case above, it is clear the outcomes assessed are of a lower scale than a standard test. Also (assuming this class represents a reasonable range), everyone would pass this test. This makes a pass mark meaningless. Also, test results are often assessed and compared student vs. student – ie. Any score outside the range of scores (in this case 16-20) is not important. Effectively, pass marks are preset, arbitrary results meant to show certain levels of achievement, not that they necessarily are themselves levels of achievement.
2- Student answers will still provide feedback.
Let’s say this Illuminati believing student answers 2 of the 10 questions. While the student will get a reasonable percentage, it illustrates to the teacher that the student does not know 80% of the content he was taught. Regardless of the percentage scored, the teachers get the same feedback, except that the point deduction system discourages guessing.
In addition, the fact remains that a higher no. of non-correct answers will likely provoke greater student reflection and feedback for the students.
3- Tests are also assessed based on the range of scores of students
Although I also addressed this in (1), a change in the percentages of a student does not matter by itself. So long as the test marking system can adequately distinguish between students, assess students and provide adequate feedback, the test will provide the needed information. Both systems accomplish this, except point deduction discourages guessing.
4- Just dodging questions isn’t smart in point deduction either.
Just because a person can get an okay percentage with ease isn’t reason to not try harder to get a much better percentage. If anything, a system which attacks guessing should provide greater critical thinking and thought during the test – if you want points you might have got from guessing, you’ll have to try harder.
Even in point deduction one gets marks for working, henceforth, just looking at 8 questions and have slight fears is not reason to avoid putting down working.
The first point I would like to mention about this subject is that under point deduction there would likely be a much higher number of blank answers. Blank answers require less effort than incorrect answers to mark (eg, checking for working, faulty working, incorrect answers, parts of a question etc.). Therefore, point deduction systems actually can help lower the teacher workload in marking.
Secondly, Con states that teachers will have to ‘look at question ranks and adjust to make a fair test’. What he doesn’t note is that teachers already weigh questions based upon difficulty and their effect on the test and then give a maximum score obtainable on that question. As I’ve said this round, both systems do this.
Furthermore, teachers actually wouldn’t have to multiply ranks. The 2.5 doesn’t apply because a student either gets nothing (wrong) , nothing happens (right) and a loss of 1*rank in the case of blanks. Teachers can multiply by 1. Also, even if this was a problem, the system might be adjusted slightly at a later stage (as mentioned before) to a better, refined version. This version might for example include whole numbers for ease (ie. – 2*rank)
Finally, the system can be made into a positive scoring system with great ease. Imagine a rank 2 question:
Right: 0 (currently)
A positive scoring system:
This would further simplify any difficulties that might arise from the point deduction system.
I’d like to note that Con did not attack any of my benefits of point deduction. Con did not disagree with the fact that my system discourages guessing, which is not advised to be used on serious, real life situations (R1 example). Con also did not attack the point about feedback.
While the actual pass mark and exact scoring system may not be perfect, it is a minor concern. Given the ease with which these issues can be fixed, and the benefits of point deduction, point deduction is a better system.
Rome wasn’t built in a day. Neither are entire marking systems. Fortunately, the key foundations are built very quickly.
Pro offers the following in refutation :
1) It is not much more complicated to evaluate with the proposed scenario
2) Not all students have a problem with time
3) Students who would need to can not calculate the percentages anyway
In response, as both a teacher and student I would offer the following :
In a normal marking scheme all a student has to do is consider the mark of a question and multiply that by the chance they think they are right and this is the expected return. Example, you look at a question which has a mark of 10, you estimate you have a 60% shot at being correct and thus it is worth 10*60% or 6 marks to attempt. You then think how much time it will take and decide if you want to try for those 6 marks.
Now under the proposed scheme by Pro the student has to do the following calculation; 2.5*rank*certainty-2.5*rank*(certainty-1) and they have to see if this expected value is greater than -rank which is what happens if they just leave it out. Thus instead of one simple multiplication they are doing six operations and then a comparison.
With an actual example; in the same scenario as noted in the above but using the marking scheme proposed by Pro the student would calculate 2.5*10*60%-2.5*10*40% which comes out to a mark of 5 (out of a possible 25) which is greater than -10 (if they left it out) so they should attempt the question if the time return is sensible.
Note however there are two further issues which come out of this calculation.
Issue #1) There are some scenarios where the expected value is negative, i.e., the average score the student will get is a loss and the are better off not attempting it. The demarcation point is actually 30%, if the student has less certainty that this they should not in general answer the question. Just think of what that means in general and is it a good thing to teach students that 30% certainty is useless and even more so consider than anything above 30% but less than 100 is *seriously* devalued.
Issue #2) Because of the imbalance of the marking scheme it does very odd things to the expected value. As an example, if the student knew 60% of the material and attempted all the questions they would get a score of 20% on the exam. It is hardly the case that they could know which 60% they will get right on and leave the rest out. To show the full imbalance of Pro's system, in order to achieve the passing grade you have to be on average 85% or more certain on the questions!
As for students not doing the calculations, Pro is right, due to the complexity few will do it unless they learn a few shortcuts (for example if you play with the equations a little you will realize that you need a 30% probability of being certain to break even, if you are less certain than that you should not [in general] answer the question). Thus students are being encouraged not to make careful decisions but instead just make wild guesses about what they should or should not do.
A1) Marking Imbalance
Pro returns with :
1) Pass marks are not everything to a test.
2) Student answers will still provide feedback.
3) Tests are also assessed based on the range of scores of students
4) Just dodging questions isn't smart in point deduction either.
Now here is where I must cry foul and cry as long and hard as askbob when he spots a multi/sock account.
In the comments section before I accepted this debate I asked specific questions to clarify the resolution. Among other clarifications Pro noted that the passing grade was 65%, clarified the total score and how it was calculated. When in rebuttal he was shown that this leads in very imbalanced testing Pro then attempts to change the resolution to something much more general akin to "Point deductions should be used on multiple choice questions."
Again I cry foul! Pro has to defend the exact marking scheme proposed or concede the debate, they can not simply change the resolution. If they want Pro can defend a general claim of attempt penalty vs zero for incorrect answers but that has to be another debate - here they have to defend the resolution given.
However, if Pro is going to propose such a grading scheme to transform the marks, which is possible, it is not sufficient simply to say it could be done, but rather the translation formula should be given. This should be able to deal with for example the one student who attempts all questions but gets them all wrong vs the one student who just leaves everything blank. If you have the second student get zero the first student actually gets a negative mark.
A3) Teacher workload
Pro makes the following arguments :
1) there would likely be a much higher number of blank answers.
2) What he doesn't note is that teachers already weigh questions based upon difficulty
3) Furthermore, teachers actually wouldn't have to multiply ranks.
Pro fails to realize that developing a fair and balanced test is more demanding that marking it and in the developing of the test the teacher has to look at each question and then perform the same rank calculation that the student does and determine a fair mark based on the total score of the test. With even a trivial consideration it should be obvious this is not easy as while you are developing the test the total score is not known which means that the teacher could be constantly revising past ranks.
Note further that Pro argues that his method reduces marking as teachers do not have to add up marks for correct answers because his scheme starts off with full marks and only subtracts, this of course can be done with the current marking scheme as well and in fact anyone who marks a lot of papers does just that - they just record all the negative marks and add them up and subtract off the total score.
Pro makes the following arguments :
1) my system discourages guessing, which is not advised to be used on serious, real life situations (R1 example).
2) Con also did not attack the point about feedback.
As noted, the scheme is severely imbalanced I was going to let the point stand on its own but added two more commentaries as it looked like Pro was attempting to formulate a marking scheme and I just wished to share my experience as a teacher. If it is wished to refute the latter two points then that can be done as well.
First, guessing is a part of life as nothing can be known with absolute certainty and thus it is always the case that you are using expected gain calculations. In the case that Pro raised, many investments are worth investing if you only have a slight chance. If for example you have a 20% chance to win a million dollars and it only costs you $10 do you do it? Of course. It would be a horrible message to send to students that guessing is a bad thing and you should only act when certainty is 100%, they would be unable to function in the real world.
Second, if Pro wanted to encourage feedback then they can do two simple things to change marking schemes. Stop taking off marks, grade all questions only positively, reward all attempts, even reward multiple attempts. When I was grading I would encourage students to try multiple approaches if they were uncertain because in the real world it is not the case that you either know something for certain or you are useless, it is often the case that you have to attempt several times and succeeding after a few attempts obviously has value.
Even for multiple choice, I would encourage students if they were not certain to give multiple answers because it is obviously worth something if a student knew that it was either one of the two answers but unsure which one vs not knowing anything. I used a gradual marking scheme based on the number of attempts used. I would have to revisit my notes but it was something like (total marks/# attempts) per question and then then entire class was normalized based on the inflation.
Thanks again to Cliff for the reply. He has made some good points as before. Let’s return to the debate.
I answer based on Con’s numbers from before (which follow on from my points)
1) Okay, I agreed that is was somewhat more complex under point deduction. I should note while Con says it is 6 times more complex, this is wrong. Merely, Con eliminates a variety of steps through things like 10*1 being simplified to 10. Now, naturally *2.5 is tougher than *1, but the difference is required for other things (the whole differentiation between incorrect and no-attempts for example). Also, if 2.5 was just simplified to 2, this would be no problem. However this is a separate issue and I’ll leave it to voters as to whether modifications can be made. Also important is whether it is my specific proposed system vs. the current system, or that mine should be implemented despite small problems, because those problems can be fixed nearly instantly. I’ll let voters decided and understand that Con believes I should not change here.
Con raises 2 issues with this point (replied to as a and b)
a) I would ask that readers reread my example about guessing in R1 to get the context correct here. On the 30% point, I think it’s reasonable that guesses below that percent do not really constitute firm knowledge. I’ll address this in greater depth later in this round.
b) Con contradicts himself somewhat here. In his previous round he claimed one need only know 20% of the content and now says one needs to know 85% of the content. Con can not hold on to both examples together. I understand Con’s point, but until this is resolved I see no need to answer this.
2) Unanswered by itself. It ties in with some themes of 3 though.
3) Okay, this is a key point. While my opponent may be a teacher, I must kindly state that he is teaching students who, if they do have problems, are at least genuinely trying and such, being a post secondary teacher. Point deduction is particularly geared towards eliminating gains from guessing, which is more likely at lower levels of testing and lower levels of student effort, both of which are less common with post secondary teaching.
The point is not that point deduction may be more complex (see 1) , but that under neither system the students with lower achievement will test probabilities, and therefore, this is a non-issue (issue – students testing probabilities in only 1 system) . Meanwhile, point deduction attacks unjustly gained marks.
I addressed this partially above (1) and so state that I understand my opponents stance and such with arguing about the ease of change to the system for overall benefit. I’ll let voters decide what is fair. I have to admit my aim was more based upon a general debate at first, but that is irrelevant. I would also note that Con doesn’t address those 4 points I made, so I see no need to expand on them.
Basically, the main argument here is whether it should be allowed to argue for change, with the point that the system could be changed easily or whether I must strictly stick to the system proposed.
Also, my opponent says that if a student answers every question wrong they get a negative mark, which is incorrect. The total mark is total rank (ie. Ranks added together) * 2.5 . The total losable marks = 2.5 * total rank.
Firstly, I don’t believe that Con rebutted my point 1) here, so I don’t need to speak on that this round. Let’s address Con’s points.
Con states that more effort is put into developing balanced tests than marking them. Fair enough, and I don’t have the knowledge to argue against this, except to merely question the assertion. I also agree that in point deduction one may have to revise ranks as they continue the creation of a test. However, this is no different to the current system! Teachers simply revise the ‘marks’ given in a question vs. ‘ranks’. If a teacher finds a 2 mark question from the start of a test at the end of a test’s creation and deems it only worth 1 mark in hindsight (due to other questions and such) , they will change it! Or if they don’t, there is no reason they would change it in point deduction instead!
Con’s second paragraph just puts both systems on par (excluding 1, which was not addressed), which is fine.
This is perhaps the key point of the entire debate. So I’ll address this a little more deeply than the other points.
Con’s counter example is the one with the million dollars. Con claims that I support only doing things with 100% certainty. This is not true. Rather, I advocate improving one’s chances where I can, instead of acting upon lower percentage chances. In this specific example, it can easily be seen that the rewards from a potential win and the percentages involved far outweigh a loss. Also, it is likely that a person would not be crippled by the loss of $10 and would be smart to retake the offer again if they could.
This is different from my example in that my example works with larger amounts of money to be lost, given the same income. Also, in my example, it is quite likely that the odds could be improved through doing things other than random guessing (just like a test, you could take a good guess, or through better preparation, know it for sure). Also, assuming one could retake Con’s offer they probably would. In my example, the amount mentioned could not be attempted a second time. These are key differences. Let me clarify these below.
My main point above is simple. While David has x chance based upon a guess with his current knowledge, through research and such, David can improve upon x chance and make it more likely that he will make a profitable investment. Obviously, it would be ludicrous to say that one has to be 100% sure of something to do it. Merely, in situations that can have a profound impact (ie. Impact assets), it is better for a person to have a stricter mindset towards guessing.
Since point deduction is anti-guessing, it follows that point deduction has an advantage in the preferred mindset towards serious decisions. Also, importantly, guessing becomes more allowable when the consequences of a wrong decision are lessened. For example, if David were to instead only invest $100 (forget about brokerage fees and such here), there would be little reason for him to research because the time spent outweighs the likely gain from extra research.
What point deduction attempts to do is create a mindset where guessing is discouraged in major decisions. Instead, knowing is valued far higher and guessing, discouraged.
Also, although I’ve already said it, a key thing about my scenario is that it cannot be repeated due to lack of funds. Con said in response to the below issue that guessing multiple times is preferable. 1, he is not defending such a system and 2, does not apply in this case.
Con’s system definitely has a variety of strengths and such, and I consider it a marked improvement on the current system. Unfortunately, Con is not defending such a system, although I believe such a system would also do to be implemented if point deduction was not. Nevertheless, since a debate is about learning, I urge voters to read his points, as they are pertinent to test scoring.
Also, I believe my point about additional feedback (through less correct answers) remains firm (except for indirect attacks, like guessing).
While Con has attacked the specifics of my scheme, I urge voters to read my defence of that case and decide. Also, having defended my other points and expanded where needed, and mentioned unaddressed points; I believe I have upheld my case.
"I should note while Con says it is 6 times more complex, this is wrong."
My point here was simple, under the current system a student can look at a question and do a trivial calculation to determine if it is worth attempting a question if they are uncertain. However, in the proposed system they have to do multiple calculations and then attempt a comparison.
During a test, with time pressing, these trivial calculations may not be such and if a student makes a mistake they may end up attempting a question they should not. Paradoxical to the intent, a lot of students will end up not doing these calculations and just guessing if they should attempt the question, this is the very thing the system is supposed to prevent.
"I think it's reasonable that guesses below that percent do not really constitute firm knowledge."
I would agree, but I would ask the simple question - if a student was 30% certain of the material in a course what mark should they get - is there no reward for anything which is not firm knowledge?
Under the proposed replacement system if they attempted the questions they would most likely get them wrong and end up with zero, what message does this send to students - if you can not learn something to certainty then do not learn it at all and leave out the question as you will get a much better mark.
Is that really the message a teacher should send to students?
"Con contradicts himself somewhat here. In his previous round he claimed one need only know 20% of the content and now says one needs to know 85% of the content."
Those were two different assertions, the were comparing what is needed to pass if all questions were attempted vs the minimum amount of right answers needed with everything else left blank. My point was that this is insanely imbalanced system no matter which way it is examined and the fact that you are given two extremely different numbers is more proof of this imbalance.
"Point deduction is particularly geared towards eliminating gains from guessing, which is more likely at lower levels of testing and lower levels of student effort, both of which are less common with post secondary teaching."
In secondary schools there are a lot of courses taken just as prerequisites and often students are just looking for a pass/credit. In first year physics for example students will usually ignore torque unless they are intending to do physics as a major/minor as it is the most demanding concept and thus they won't even attempt to learn it and just guess at the answers, similar examples can be given for other courses.
"Also, my opponent says that if a student answers every question wrong they get a negative mark, which is incorrect. The total mark is total rank (ie. Ranks added together) * 2.5 . The total losable marks = 2.5 * total rank."
To clarify, this was in response to the comment about balancing the system. The current scheme has a student get 60% if they just leave out every question and with the pass mark is 65% this is obviously a problem. But it is not trivial to devise a formula which will "curve" the marks and not generate someone getting negative marks if they attempt all questions, Pro certaintly did not provide one only imply that "the details could be worked out".
"However, this is no different to the current system! Teachers simply revise the ‘marks' given in a question vs. ‘ranks'."
Yes, it is simply more complicated due to the calculations required. Again, look at the current system and what a teacher has to do to decide if a question should effect the pass fail point on an exam, which any committed teacher should do. Then look at the machinations they have to do under the current system, it is far more complicated and sample calculations have been provided in the above.
"Rather, I advocate improving one's chances where I can, instead of acting upon lower percentage chances."
Ok, but here is the thing, you are in a test, you don't have any ability at that point to improve on anything - you just have the chance to show what you know. It is not as if I would argue students should not learn, but simply that a test should not discourage students from proving all the information they do learn so a full assessment can be given.
"Con's system definitely has a variety of strengths and such, and I consider it a marked improvement on the current system. Unfortunately, Con is not defending such a system, although I believe such a system would also do to be implemented if point deduction was not."
Comments were made to discuss alternates to provide more a more in depth argument as to why point deduction is in general a bad thing as it discourages students from sharing information. In any classes I have ever taught one of the things I made sure to do was produce an environment where students were free and confident to express any ideas they had, on a test or not, this meant that I had the fullest understanding of their strengths and weakness and could teach accordingly. When I had the maximum understanding of what they knew I was at maximum effect as a teacher.
I was extremely demanding, workloads were high consistently, the last courses I taught were calculus based introductory physics and I had two tests every week, one random, two assignments every week, a lab every week - and with all of this student feedback was consistently high. I achieved this level of interaction and corresponding performance by creating an atmosphere where students knew that even the attempt at a solution would be rewarded even if, in the words of one student, "the answer is completely borked". My perspective which comes from my view as an experimental physicist is that all attempts, even failures, are critical to learning and everything I did as a teacher was to foster learning.
In short, the opening refutations have not been refuted and I have given a proposed counter scheme which will maximize feedback to the instructor and maximize rewards to the students.
(no reptoids were harmed during this debate)
Thanks again to Cliff for the reply, it has been a good debate so far and has taught me much about how tests should be scored. Also, because of some issues with the debate.org editor, some of the spacing and formatting is not as should be, I ask readers to bear with this and follow things in a logical manner. Time for the final round.
Con here attacked 3 of my statements from where I had ‘student complications’ as a heading. I will address them in order.
Basically, I addressed this point last round – the number of calculations is equal, just regular systems times by 1 vs. 2.5. Also, as said before, assuming a simple change (if allowed) to say 2, multiplication problems become a non-issue because of simplicity (like times by 1). Also, the calculations we are talking about here are less likely to be attempted by students who are guessing (people who are likely to not be trying so hard at school; lower achievers) , so this issue is not key to some of the main targets of point deduction.
I addressed this point in my guessing section, and which Con also talks about later. Basically, the system rewards people with higher percentage certainty. Ie. The test does not want to send a message that guessing at levels such as 30% is a good idea in real life. Instead, the person who is 30% sure in real life should improve their certainty through things such as research. Obviously, certainty as the only time for rewards would be absurd. I’ll address this a bit more later.
Con doesn’t actually resolve his contradiction here, except states that the system can produce these results, given opposite student approaches. Well, that’s the entire reason for point deduction – differentiation between non-attempts and incorrect answers! If there was no mark differentiation, there’d be no reason to not guess, would there? It would be like the current system.
I’d like to make a point here. Con gives an example of a lack of interest in learning at post secondary. However, I did not claim that such things did not occur at post secondary, but that they were less common. Obviously, since those at post secondary are typically more academically gifted and interested in learning, it makes sense that if they can show a lack of interest, it is likely to be more common among people with a lesser interest in learning.
Con – ‘and not generate someone getting negative marks if they attempt all questions...’
I already disproved this myth conception. Even the quote shows this. On the point of ‘the details could be worked out’, I see no serious difficulties for major institutions like government solving such problems. I would have attempted to show this myself, however, Con specifically stated that I could not change the details midway through, ie. ‘Pro has to defend the exact marking scheme...’. This would have been poor conduct to do so. If Con wishes to hold me to his own belief of what I should do, then he should not ask for information that he claims should not be allowed.
As with all other headings, I am basing this on what section of mine Con responded to (through my statements).
As addressed earlier this round, this is purely because of a 2.5* issue. If this became 2 (again, if allowed), then this is a non-issue. Also, since teachers typically add up scores and then do a multiplier (ie. Times 1) , this should be no problem – only times by 2.5 at the end.
This is a key issue, and con only attacks with ‘you can’t improve during a test’. This is true. However, the test is not just there to assess you. Tests can also teach the students through simply doing the tests. Ie. Make mistakes, improve; learn better how to write an extended response next time etc.
In this case, the lesson to be learned is about guessing. The test can teach a person about the value of not guessing, and rather, being strongly backed by knowledge in life, where guessing can be hazardous and knowledge is rewarded.
As stated before, I think Con’s system is a marked improvement on the current system. I also admire a good teacher, and student feedback, while not always perfect, is, especially at higher levels, a good indicator of teaching, so I commend Con here. Ideally, every teacher would not be stuck to a set curriculum and there would be an intellectual to and fro between teacher and students. Sadly, this is not a common occurrence.
However, with all due respect, I must note that Con is not arguing for his personal system, he is arguing for the current system. While I believe that it is definitely a thing worth reading for readers, due to the improvements made in it, it is not what the issue at hand is, and so it can be dismissed.
While Con has attacked specifics of my scheme (ie. 2.5 etc.), I have defended the merits of my system and successfully argued many points, sometimes without response. Also to consider is whether the system should be considered based on it’s exact form, or the positives when able to be changed ever so slightly (ie. *2) for a superior system. Finally, I ask that voters do not solely vote on their personal opinions. Also, I thank Con again for a fine debate.
"Basically, I addressed this point last round – the number of calculations is equal, just regular systems times by 1 vs. 2.5."
This is not the case, as a direct requote :
"Example, you look at a question which has a mark of 10, you estimate you have a 60% shot at being correct and thus it is worth 10*60% or 6 marks to attempt. You then think how much time it will take and decide if you want to try for those 6 marks.
Now under the proposed scheme by Pro the student has to do the following calculation; 2.5*rank*certainty-2.5*rank*(certainty-1) and they have to see if this expected value is greater than -rank which is what happens if they just leave it out. Thus instead of one simple multiplication they are doing six operations and then a comparison."
To spell this out in detail :
Scenario 1 - the student asks himself one question, it will take me five minutes and my expected value is 10*0.6 or six marks, is this worth it.
Scenario 2 - the students asks himself, it will take me five minutes and my expected value is 2.5*10*0.6-2.5*10*(1-0.6), is this worth it compared to just leaving it out which has the expected value of -10.
If there are multiple questions and the student has to see which ones to attempt in the time, i.e., which questions produce the most marks per time then the second scenario becomes even more of a tangle.
"Also, the calculations we are talking about here are less likely to be attempted by students who are guessing (people who are likely to not be trying so hard at school; lower achievers) , so this issue is not key to some of the main targets of point deduction."
Again, ironically as noted see the result, students will not do it, that is the point, they will instead move from making a determined calculation to just guessing how to proceed - it has the exact opposite result as to what is intended.
"Instead, the person who is 30% sure in real life should improve their certainty through things such as research."
Yes, just like you should study before you attempt a test - the problem is that few students get 100% consistently and thus do you want to reward students who attempt problems when they are uncertain - or instead punish them?
"Con doesn't actually resolve his contradiction here, except states that the system can produce these results, given opposite student approaches."
There is no contradiction, if you attempt all questions you need to know 85% to pass, if you just attempt the minimum amount you only need to know 20% to pass.
The point here is clear, this system encourages students to take a very small part of the course material and learn it very well and ignore the rest because the penalty for leaving something out is minute compared to a wrong attempt.
"I already disproved this myth conception."
This was never disproven, it was simply stated to be false. It would have been nice to see the transformation function g(f(x) which would have done so because it is impossible under standard normative transforms (bell curve, etc.).
"As addressed earlier this round, this is purely because of a 2.5* issue. If this became 2 (again, if allowed), then this is a non-issue."
Again, this is just stated. Note that I have constantly showed very specific equations and math to support claims, at most Pro has just stated there is no issue without actually showing the math to support them. The position may very well be true, but details would be appreciated.
"Tests can also teach the students through simply doing the tests. Ie. Make mistakes, improve; learn better how to write an extended response next time etc."
Yes, but the point what that does not help you *during* the test.
"However, with all due respect, I must note that Con is not arguing for his personal system, he is arguing for the current system."
Care needs to be taken here, let us return to the OP :
"These reasons are enough to provide strong grounds for a change of test scoring system."
The point in presenting an alternate is to show that if a change is to be made to the system there are far better methods than point deduction and an alternate method was so exhibited, that is all.
In short, the resolution stands refuted for the reasons noted. Which are :
-The system is imbalanced, a student can leave out all questions and be 5% short of passing, or answer two questions and pass, or know 80% of the material and fail. That is an obvious and gross problem which produces literally insane results.
-If a student is unsure of a question, they have to do a nontrivial calculation to see if they should attempt it. If the student has to try to figure out between multiple questions as they are pressed for time (which are the exact students who are likely to be guessing) they will not be able to decide intelligently due to the calculations required. In short they will be forced to guess more not less.
-It increases the workloads of teachers, especially in creating tests.
-It sends the message that attempting a problem if you are unsure is a bad thing. It this stops all creativity during a test and encourages simply regurgitate.
-It limits the ability of a teacher to learn about the students as there will be more blank answers which tell nothing, vs incorrect answers which can provide insight into where the lack of understanding is situated.
I do appreciate the effort of Pro to consider an alternative marking scheme and encourage them to review the most well known one which is simply; one mark is right, -one mark if wrong, zero if blank. This is well used in various fields and has well documented strengths and weaknesses.
I would also encourage them to do the math required to see what happens to the pass and fail percentages and the effects on the grading curve for various systems as this is critical to the results. It is very easy to check all of this out in Excel with a few simple formulas.
|Agreed with before the debate:||-||-||0 points|
|Agreed with after the debate:||-||-||0 points|
|Who had better conduct:||-||-||1 point|
|Had better spelling and grammar:||-||-||1 point|
|Made more convincing arguments:||-||-||3 points|
|Used the most reliable sources:||-||-||2 points|
|Total points awarded:||0||3|