My school has a 50% floor and recently moved from a modified seven point scale to a ten point scale. The biggest difference I see is that slacker D’s and low’s C’s prevent the alarm and intervention that brings a kid to office hours or changes behavior. I used to have kids start attending office hours regularly at the end of a semester to try to pass. They would learn the material, and they would learn that they could master the work with the right interventions, and they would develop a key self-advocacy skill before college (or life). Now they just pass without the intervention.
The difference between our system and yours is that you do not appear to have any process/completion grades. At our school there are process grades, like Do Now grades or participation grades (homework completion grades in my class) that allow a student to pass without mastering any material. That’s a problem.
I am aware of the limits of grades to motivate students, but our 50% floor currently allows students to pass without mastering any no mastery or earn a C by earning an B on only half of the mastery assessments.
Our 9th grade teachers implemented this in 2021 and our students immediately figured out how to game the system and not do anything beyond the bare minimum that would get them above the % they were aiming for - whether that was 90, 80, 70, or 60% (different students had different aspirations).
I teach science. I'd like to imagine that a student passing any subject has mastered more than 60% of the material - particularly for courses that become the foundation for those heading off to medical school or the ones responsible for quality control in engineering, to name a few examples... We'd like to keep our bridges up 100% of the time, as well as our planes.
I’m completely in agreement on a 50 grade floor, as long as students don’t have an unlimited amount of time to turn in work, which is where we are right now. Students at my school receive zeros on missing assignments, but they basically have until the end of the year to turn them in.
I think the bigger problem is averaging exams in the first place, and even bigger is averaging in homework and labs. If you give a 50, or a zero, for a lab that was never turned in, your grade is now about 2 things, mastery of the material, and compliance with the course requirements. Not submitting a lab is not evidence you could not have done the work well if you had chosen to do so. Same with a missed exam. If you're in a class with a comprehensive final, what possible justification is there for giving a grade lower than the final exam score. If the exam is a metric of how much you learned by the end, and that's what grades are supposed to measure... What does the fact that you hadn't mastered the early material at the beginning of the class, if you subsequently master it?
As a someone who taught in a different country and a math teacher, I keep asking why are we plugging in a grade of 50 instead of just changing the scale? Instead of continuing a charade by entering a grade that means zero but isn't zero, we further entrench a system of numerical values that are purely a social construct. 90% literally means nothing bc we're removing the context. 90% is unachievable for major league batters to hit safely but I wouldn't choose elective surgery with only a 90% success rate....but even that depends on the context of the surgery. Clinging to 90-80-70-60 we rob assessments of the opportunity to asking questions that require students to struggle. We never ask them to find their max weight to lift. Change the scales.
Matt does a good job identifying a real underlying issue in how grades are traditionally calculated: a simple weighted average leaves no room to account for improvement, intent, or circumstance. A few early outliers—especially zeros—can shift a student’s average so dramatically that it becomes all but impossible to recover to a passing grade later in the term. That can create exactly the kind of discouragement he describes.
However, this is a problem with the methodology, not the numbers themselves. When a method produces undesirable results, the solution is rarely to alter the data (for example, converting all scores below 50 into 50s). The problem is not the data, it's the methodology used to make it into a final grade.
I would also push back slightly on the framing that the grading scale itself is fundamentally disproportionate. It is true that an “F” covers a 60-point range while other letter grades typically cover 10 points each, but this comparison overlooks the role that the failing grade plays. Rather than viewing the “F” as six times larger than the other ranges, it may be more accurate to interpret the scale as establishing a threshold of competence.
In other words, the real question is: what level of mastery do we consider sufficient to say that a student has learned the material? If a student demonstrates less than 60% of the expected knowledge or skill, many institutions interpret that as insufficient evidence of mastery. From that perspective, the scale is less about distributing letter grades evenly and more about distinguishing between two broad categories: students who have met the standard for competence and those who have not. The remaining grade ranges then provide finer distinctions among students who have met that threshold. In other words, the scale is more about "F" (0-59 points: did not meet threshold to claim they know the material) and "Not F" (60-100 points: met the threshold to say they know the material) - then the other letter grades ("A", "B", etc) are there to give more grandularity on *how well* we think they understood the material.
For that reason, the traditional 100-point scale itself seems largely reasonable. The issue Mark highlights arises not from the scale, but from relying exclusively on a weighted average of all grades to determine a final grade. If the goal is to give students meaningful opportunities to recover from early setbacks, there are other approaches that address this more directly.
For example:
Standards- or mastery-based grading focuses on demonstrated competency rather than accumulating points. These systems often allow students to improve their standing as they demonstrate greater mastery over time, though they can require significantly more instructor effort to implement.
“Best-of” assessment structures can mitigate the impact of a single poor performance. For instance, giving four exams but counting the best three allows an early misstep without eliminating accountability for consistent engagement.
Cumulative replacement assessments can allow students to replace a low score with a later demonstration of understanding. Unlike extra credit, these assessments substitute for an earlier grade and often require a stronger or more comprehensive demonstration of mastery.
The broader point is that if we want students to have realistic paths to improvement, it is often better to rethink the structure of assessment rather than modify the scale itself. Simply replacing zeros with 50s removes important information—whether a student earned a 48, a 12, or never submitted the work at all—and addresses a methodological problem with a blunt adjustment to the data.
A more effective solution is to design grading systems that preserve information while also allowing students meaningful opportunities to demonstrate growth.
The tone and vibe seemed off so I pasted the first ~1/3 of this into pangram AI checker and it came back as 100% AI slop. "That sentence alone is enough to change the temperature in the room" get a grip what is that prose.
AI detectors aren’t reliable — ask almost any teacher who’s had to deal with them, they started "meh" and then reliability dropped off a cliff. But whether or not AI was used to polish a piece isn’t really engaging the argument here. (For the record, the “change the temperature in the room” line was from me after I heard someone use it in a presentation.)
If you disagree with the math about how the 100-point scale behaves, I’m happy to discuss that.
A long, long time ago, I took high school physics from a man who had similar beliefs, but approached the problem slightly differently. His argument: using the standard grading system, he needed to provide 60 points of pretty basic stuff just to make sure people weren’t fighting to pass, leaving him only 40 points to see how the class members actually ranked. His solution: add 100, divide by 2. Now a 20 was passing, and he had 80 points in which to see who excelled, and who was simply getting by. (Seems, in hindsight, that his method was more about finding the standouts, and less about trying to help all students improve… but the people who took that class had already self-selected to some degree, so he had less need for motivating tools and more desire for analytical range.) It’s been more than 4 decades since I took his class, but he still stands out as a powerful teacher. Thanks for reminding me!
If the best evidence that altering low Fs to 50% improves student learning is a vague, unquantifiable feeling experienced by the teacher, then it really seems like a lot of time and effort is being expended to rearrange the deck chairs while the ship goes down. When a student gets a 13% on a test the most pressing problem isn't the impact on future motivation, it's the massive skill or knowledge deficit they have right now. There are better ways to prevent a single screw up from having a disproportionate impact on their term grade (let them drop a score, for instance). Erasing data that shows just how severely a student is struggling is dangerous and a disservice to the student.
Just to clarify, my argument isn’t that a 50% floor improves learning because of a “vague feeling.” The structural point is mathematical. On a 100-point scale, the lower half carries disproportionate weight. That distortion affects how averages behave over time.
If a student earns a 13%, that absolutely signals a serious skill deficit, and it doesn’t happen in isolation - that’s a signal to me that we need to do more work. Nothing about a 50% floor changes the need for intervention, reteaching, or support. The deficit is real and still visible in the gradebook.
What the floor changes is how that one score interacts with later demonstrated mastery. It prevents a single early collapse from mathematically overwhelming subsequent performance.
Dropping a score is one solution. Compressing the lower half of the scale is another. My claim isn’t that this is the only approach — only that the current scale embeds a structural imbalance that’s worth examining. For me, the core question is whether grades are meant to measure cumulative mastery or to amplify early failure.
For example, most driver’s license tests in Australia require a near-perfect score to pass. Make a major error or more than a couple of minor errors and you need to go back and re-learn. This is the easiest sort of test to write. You’re not trying to differentiate. It’s “acceptable standard: Yes or No?”.
Graded assessments are more complex. Broadly, there are three key categories: below standard, at standard, and above standard. Within this, there’s also “how much?”.
Aggregated marks only make sense for “above standard”. It’s as good a tool as any for estimating how well one student is overperforming as compared to another
On the other end of the spectrum it’s flawed. A mix of adequate and inadequate marks doesn’t average out to adequate. The student needs to demonstrate adequacy across all assessable areas.
If the student initially demonstrates inadequacy and later demonstrates adequacy, it makes little sense to penalise their current performance based on their previous failure. The “50% floor” technique doesn’t remove the core problem of using an average to measure adequacy, but does remove the chronological penalty.
Meanwhile, I think any discussion of “slackers” is hiding a lot of assumptions. Leaving aside cheating, does a student who consistently scores 80% actually understand the work better than a student who struggles with interim assessments and then scores perfectly in the final exam?
I’m not saying diligence is not valuable or should not be assessed, but consistency of work and quality of work are not the same thing
So when my school imposed 50% minimums for every assignment, students soon figured out that they could skip many assignments - maybe do 1/3 or 1/4. Because they would still pass! (Do the math.) It’s not like most teachers gave kids 10% or 30% scores anyway; usually crappy work would merit a 50 or 60. And if someone earns a 25% on a test…that is valuable info - not saying you have to keep that score. Zeroes let parents know that the kid turned in NOTHING. Not bad work, nothing.
Thanks for your comment. Implementation absolutely matters.
In my classroom, a 50% floor doesn’t make “do 1/3 of the work and pass” mathematically viable. A student earning nothing but 50s still fails once weights are applied - I tell my physics class that it’s gravity - the mass of 50s gets larger, the more it pulls the rest of the grades towards it. Mastery of major assessments is still required to pass.
I also clearly flag missing work in the gradebook so parents see exactly what was not submitted, and I have conversations with the student to get things back on track. A 50% floor doesn’t erase the record — it prevents a single zero from mathematically overwhelming later demonstrated learning. I think that if students in a system can skip 3/4 of the work and still pass, that’s a weighting and design issue, not a floor issue.
This is fascinating. I've tried a few methods to reward late bloomers and also a modified curve, but I've never felt they were adequate. I don't have many students who fare poorly and then dramatically improve (I teach chemistry which tends to self-limit very poor performance--they leave early). I try to look at individual performance. I'm lucky to have small classes, so I know know pretty quickly who is floundering. But I do reward effort in improving. All that said, most systems are rigid with the 0-100 scale. Its difficult to get around that but I have my ways. Great post, thanks!
This was genuinely hard to read because of all the ChatGPT. Which is a shame, because it’s an interesting topic.
If someone understands less than 60% of a topic, they shouldn’t be passing. It’s appropriate for the failing grade to be a larger proportion than the other grades.
Maybe a sensible policy could be that you can still grade on the hundred point scale, but no grade gets *entered into the grade book* as less than a 50. That is, no great affects the average as being lower than a 50.
The 50% floor is a common "adjustment" to make to a flawed grading scale ... but why not just address the flaw and eliminate it. The easiest way to do this is to pick up the tail of the Bell cure (off of 50% and slide it down to 0%. This gives a grading scales with as many points above average as below, the average now being 50%. This resolves almost all of the problems indicated. This doesn't happen (even though I did it) because teachers want to maintain the conjecture that all students are learning at least half of the material, which we know is a complete fiction.
This scale maps onto testing results perfectly as indicating what fraction of the material studied was learned by the test taker (assuming that they are not ill or otherwise impaired, etc.). Also assumed is that the tests actually address what the students are supposed to learn and that difficulty level is addressed in test construction.
I so appreciate this piece Matt. As a coach and father, I saw kids in high school lose all incentives, get kicked out of sports, and punished irreparably by a missed assignment or illness and lose all hope. As Dr. William Glasser says, if you tell kids they are a failure and make it impossible yo climb out of the ditch, education and school is dropped from their “quality world.”
My school has a 50% floor and recently moved from a modified seven point scale to a ten point scale. The biggest difference I see is that slacker D’s and low’s C’s prevent the alarm and intervention that brings a kid to office hours or changes behavior. I used to have kids start attending office hours regularly at the end of a semester to try to pass. They would learn the material, and they would learn that they could master the work with the right interventions, and they would develop a key self-advocacy skill before college (or life). Now they just pass without the intervention.
The difference between our system and yours is that you do not appear to have any process/completion grades. At our school there are process grades, like Do Now grades or participation grades (homework completion grades in my class) that allow a student to pass without mastering any material. That’s a problem.
I am aware of the limits of grades to motivate students, but our 50% floor currently allows students to pass without mastering any no mastery or earn a C by earning an B on only half of the mastery assessments.
Our 9th grade teachers implemented this in 2021 and our students immediately figured out how to game the system and not do anything beyond the bare minimum that would get them above the % they were aiming for - whether that was 90, 80, 70, or 60% (different students had different aspirations).
I teach science. I'd like to imagine that a student passing any subject has mastered more than 60% of the material - particularly for courses that become the foundation for those heading off to medical school or the ones responsible for quality control in engineering, to name a few examples... We'd like to keep our bridges up 100% of the time, as well as our planes.
I’m completely in agreement on a 50 grade floor, as long as students don’t have an unlimited amount of time to turn in work, which is where we are right now. Students at my school receive zeros on missing assignments, but they basically have until the end of the year to turn them in.
I think the bigger problem is averaging exams in the first place, and even bigger is averaging in homework and labs. If you give a 50, or a zero, for a lab that was never turned in, your grade is now about 2 things, mastery of the material, and compliance with the course requirements. Not submitting a lab is not evidence you could not have done the work well if you had chosen to do so. Same with a missed exam. If you're in a class with a comprehensive final, what possible justification is there for giving a grade lower than the final exam score. If the exam is a metric of how much you learned by the end, and that's what grades are supposed to measure... What does the fact that you hadn't mastered the early material at the beginning of the class, if you subsequently master it?
As a someone who taught in a different country and a math teacher, I keep asking why are we plugging in a grade of 50 instead of just changing the scale? Instead of continuing a charade by entering a grade that means zero but isn't zero, we further entrench a system of numerical values that are purely a social construct. 90% literally means nothing bc we're removing the context. 90% is unachievable for major league batters to hit safely but I wouldn't choose elective surgery with only a 90% success rate....but even that depends on the context of the surgery. Clinging to 90-80-70-60 we rob assessments of the opportunity to asking questions that require students to struggle. We never ask them to find their max weight to lift. Change the scales.
Matt does a good job identifying a real underlying issue in how grades are traditionally calculated: a simple weighted average leaves no room to account for improvement, intent, or circumstance. A few early outliers—especially zeros—can shift a student’s average so dramatically that it becomes all but impossible to recover to a passing grade later in the term. That can create exactly the kind of discouragement he describes.
However, this is a problem with the methodology, not the numbers themselves. When a method produces undesirable results, the solution is rarely to alter the data (for example, converting all scores below 50 into 50s). The problem is not the data, it's the methodology used to make it into a final grade.
I would also push back slightly on the framing that the grading scale itself is fundamentally disproportionate. It is true that an “F” covers a 60-point range while other letter grades typically cover 10 points each, but this comparison overlooks the role that the failing grade plays. Rather than viewing the “F” as six times larger than the other ranges, it may be more accurate to interpret the scale as establishing a threshold of competence.
In other words, the real question is: what level of mastery do we consider sufficient to say that a student has learned the material? If a student demonstrates less than 60% of the expected knowledge or skill, many institutions interpret that as insufficient evidence of mastery. From that perspective, the scale is less about distributing letter grades evenly and more about distinguishing between two broad categories: students who have met the standard for competence and those who have not. The remaining grade ranges then provide finer distinctions among students who have met that threshold. In other words, the scale is more about "F" (0-59 points: did not meet threshold to claim they know the material) and "Not F" (60-100 points: met the threshold to say they know the material) - then the other letter grades ("A", "B", etc) are there to give more grandularity on *how well* we think they understood the material.
For that reason, the traditional 100-point scale itself seems largely reasonable. The issue Mark highlights arises not from the scale, but from relying exclusively on a weighted average of all grades to determine a final grade. If the goal is to give students meaningful opportunities to recover from early setbacks, there are other approaches that address this more directly.
For example:
Standards- or mastery-based grading focuses on demonstrated competency rather than accumulating points. These systems often allow students to improve their standing as they demonstrate greater mastery over time, though they can require significantly more instructor effort to implement.
“Best-of” assessment structures can mitigate the impact of a single poor performance. For instance, giving four exams but counting the best three allows an early misstep without eliminating accountability for consistent engagement.
Cumulative replacement assessments can allow students to replace a low score with a later demonstration of understanding. Unlike extra credit, these assessments substitute for an earlier grade and often require a stronger or more comprehensive demonstration of mastery.
The broader point is that if we want students to have realistic paths to improvement, it is often better to rethink the structure of assessment rather than modify the scale itself. Simply replacing zeros with 50s removes important information—whether a student earned a 48, a 12, or never submitted the work at all—and addresses a methodological problem with a blunt adjustment to the data.
A more effective solution is to design grading systems that preserve information while also allowing students meaningful opportunities to demonstrate growth.
The tone and vibe seemed off so I pasted the first ~1/3 of this into pangram AI checker and it came back as 100% AI slop. "That sentence alone is enough to change the temperature in the room" get a grip what is that prose.
AI detectors aren’t reliable — ask almost any teacher who’s had to deal with them, they started "meh" and then reliability dropped off a cliff. But whether or not AI was used to polish a piece isn’t really engaging the argument here. (For the record, the “change the temperature in the room” line was from me after I heard someone use it in a presentation.)
If you disagree with the math about how the 100-point scale behaves, I’m happy to discuss that.
A long, long time ago, I took high school physics from a man who had similar beliefs, but approached the problem slightly differently. His argument: using the standard grading system, he needed to provide 60 points of pretty basic stuff just to make sure people weren’t fighting to pass, leaving him only 40 points to see how the class members actually ranked. His solution: add 100, divide by 2. Now a 20 was passing, and he had 80 points in which to see who excelled, and who was simply getting by. (Seems, in hindsight, that his method was more about finding the standouts, and less about trying to help all students improve… but the people who took that class had already self-selected to some degree, so he had less need for motivating tools and more desire for analytical range.) It’s been more than 4 decades since I took his class, but he still stands out as a powerful teacher. Thanks for reminding me!
If the best evidence that altering low Fs to 50% improves student learning is a vague, unquantifiable feeling experienced by the teacher, then it really seems like a lot of time and effort is being expended to rearrange the deck chairs while the ship goes down. When a student gets a 13% on a test the most pressing problem isn't the impact on future motivation, it's the massive skill or knowledge deficit they have right now. There are better ways to prevent a single screw up from having a disproportionate impact on their term grade (let them drop a score, for instance). Erasing data that shows just how severely a student is struggling is dangerous and a disservice to the student.
I appreciate the thoughtful pushback.
Just to clarify, my argument isn’t that a 50% floor improves learning because of a “vague feeling.” The structural point is mathematical. On a 100-point scale, the lower half carries disproportionate weight. That distortion affects how averages behave over time.
If a student earns a 13%, that absolutely signals a serious skill deficit, and it doesn’t happen in isolation - that’s a signal to me that we need to do more work. Nothing about a 50% floor changes the need for intervention, reteaching, or support. The deficit is real and still visible in the gradebook.
What the floor changes is how that one score interacts with later demonstrated mastery. It prevents a single early collapse from mathematically overwhelming subsequent performance.
Dropping a score is one solution. Compressing the lower half of the scale is another. My claim isn’t that this is the only approach — only that the current scale embeds a structural imbalance that’s worth examining. For me, the core question is whether grades are meant to measure cumulative mastery or to amplify early failure.
Thanks for writing this. I often had “bad days” which were probably adhd related and I couldn’t dig myself out of the hole.
What is your testing trying to establish?
For example, most driver’s license tests in Australia require a near-perfect score to pass. Make a major error or more than a couple of minor errors and you need to go back and re-learn. This is the easiest sort of test to write. You’re not trying to differentiate. It’s “acceptable standard: Yes or No?”.
Graded assessments are more complex. Broadly, there are three key categories: below standard, at standard, and above standard. Within this, there’s also “how much?”.
Aggregated marks only make sense for “above standard”. It’s as good a tool as any for estimating how well one student is overperforming as compared to another
On the other end of the spectrum it’s flawed. A mix of adequate and inadequate marks doesn’t average out to adequate. The student needs to demonstrate adequacy across all assessable areas.
If the student initially demonstrates inadequacy and later demonstrates adequacy, it makes little sense to penalise their current performance based on their previous failure. The “50% floor” technique doesn’t remove the core problem of using an average to measure adequacy, but does remove the chronological penalty.
Meanwhile, I think any discussion of “slackers” is hiding a lot of assumptions. Leaving aside cheating, does a student who consistently scores 80% actually understand the work better than a student who struggles with interim assessments and then scores perfectly in the final exam?
I’m not saying diligence is not valuable or should not be assessed, but consistency of work and quality of work are not the same thing
So when my school imposed 50% minimums for every assignment, students soon figured out that they could skip many assignments - maybe do 1/3 or 1/4. Because they would still pass! (Do the math.) It’s not like most teachers gave kids 10% or 30% scores anyway; usually crappy work would merit a 50 or 60. And if someone earns a 25% on a test…that is valuable info - not saying you have to keep that score. Zeroes let parents know that the kid turned in NOTHING. Not bad work, nothing.
Thanks for your comment. Implementation absolutely matters.
In my classroom, a 50% floor doesn’t make “do 1/3 of the work and pass” mathematically viable. A student earning nothing but 50s still fails once weights are applied - I tell my physics class that it’s gravity - the mass of 50s gets larger, the more it pulls the rest of the grades towards it. Mastery of major assessments is still required to pass.
I also clearly flag missing work in the gradebook so parents see exactly what was not submitted, and I have conversations with the student to get things back on track. A 50% floor doesn’t erase the record — it prevents a single zero from mathematically overwhelming later demonstrated learning. I think that if students in a system can skip 3/4 of the work and still pass, that’s a weighting and design issue, not a floor issue.
This is fascinating. I've tried a few methods to reward late bloomers and also a modified curve, but I've never felt they were adequate. I don't have many students who fare poorly and then dramatically improve (I teach chemistry which tends to self-limit very poor performance--they leave early). I try to look at individual performance. I'm lucky to have small classes, so I know know pretty quickly who is floundering. But I do reward effort in improving. All that said, most systems are rigid with the 0-100 scale. Its difficult to get around that but I have my ways. Great post, thanks!
This was genuinely hard to read because of all the ChatGPT. Which is a shame, because it’s an interesting topic.
If someone understands less than 60% of a topic, they shouldn’t be passing. It’s appropriate for the failing grade to be a larger proportion than the other grades.
Maybe a sensible policy could be that you can still grade on the hundred point scale, but no grade gets *entered into the grade book* as less than a 50. That is, no great affects the average as being lower than a 50.
The 50% floor is a common "adjustment" to make to a flawed grading scale ... but why not just address the flaw and eliminate it. The easiest way to do this is to pick up the tail of the Bell cure (off of 50% and slide it down to 0%. This gives a grading scales with as many points above average as below, the average now being 50%. This resolves almost all of the problems indicated. This doesn't happen (even though I did it) because teachers want to maintain the conjecture that all students are learning at least half of the material, which we know is a complete fiction.
This scale maps onto testing results perfectly as indicating what fraction of the material studied was learned by the test taker (assuming that they are not ill or otherwise impaired, etc.). Also assumed is that the tests actually address what the students are supposed to learn and that difficulty level is addressed in test construction.
I so appreciate this piece Matt. As a coach and father, I saw kids in high school lose all incentives, get kicked out of sports, and punished irreparably by a missed assignment or illness and lose all hope. As Dr. William Glasser says, if you tell kids they are a failure and make it impossible yo climb out of the ditch, education and school is dropped from their “quality world.”