Because they are not careless. The mistakes you are talking about, the dropped minus sign, the 7 that became a 1 two lines down, the you solved for and then reported as , happen on operations you have practiced hundreds of times, and that is not a coincidence. It is the cause.
Human-error researchers have a name for this category, and understanding it is the difference between fixing these mistakes and spending another year being told to "just be more careful", which you have already tried, and which has already failed. Not because you did not mean it. Because it targets the wrong mechanism.
Slips happen to experts, on autopilot, by design
The classic research on human error, built by the psychologist James Reason and colleagues, splits mistakes into fundamentally different species. A mistake in the technical sense is choosing the wrong plan: you used the wrong method because you misunderstood the problem. A slip is executing a correct plan incorrectly, and the research on slips reaches a conclusion students find almost insulting: slips occur precisely during skilled, automatic performance, and they need two conditions: a routine task familiar enough to run without supervision, and attention captured by something else at the wrong moment.
Read those conditions against your own math. Distributing, carrying, copying a line: all automated years ago, all running without conscious attention while your mind is on the next step, the clock, or the phone that just lit up. Slips are not evidence that you are bad at math. They are a tax on being good enough at it that your hands work unsupervised. Which is also why "concentrate harder" fails as a fix: the whole point of automation is that it runs without concentration, and no one can consciously supervise every carry for an hour. The realistic fixes work with automaticity, and they come after one more piece of theory.
The capture error: your strongest habit fires instead
The most predictable slip has its own name in the literature: capture, where a more practiced routine takes over from the less practiced one you intended. Math is full of it, and once you see the pattern you will find it all over your own quizzes:
- becoming . The years-old routine "apply the operation to each term" captures the newer skill of squaring a binomial, which actually needs .
- read as 9. The instinct to square the 3 captures the order of operations, which says the answer is , since only is 9. If exponent-and-sign combinations are a recurring bite, drill exactly that pattern on the exponent calculator until the correct reading is the automatic one.
- becoming . The plain distributing routine runs, and the sign flip on the second term, the part needing attention, gets skipped: the correct result is .
Notice what these have in common: the wrong answer is not random noise. It is the fingerprint of a specific stronger habit firing in place of a weaker one. That is why your slips repeat, and why the next section works.
Your slips cluster. Find your top two.
In years of tutoring I have never met a student who makes "careless mistakes" in general. I have met sign-flippers, miscopiers, exponent-droppers, and question-misreaders, each convinced they were randomly sloppy. Pull your last three quizzes and tally every slip by type. Almost always, two categories cover most of the damage, and that changes everything, because "be careful about everything forever" is impossible while "guard the sign when a negative enters" is a Tuesday's work. The tally is the same taxonomy I use in the 20-minute redo, where slips get separated from method errors precisely because their fixes are different. The redo finds your pattern; this article is about installing the guardrail once you know it.
Guardrails that work with autopilot, not against it
Since slips strike unsupervised routines, every effective countermeasure does one of two things: it puts a physical mark where attention must return, or it removes the conditions slips need. None of them involve trying harder.
- One operation per line. Compressing three moves into one line is running three unsupervised routines at once; the slip hides in the compression. Writing each move gives every routine its own checkpoint, and as a bonus it is exactly what graders pay for, as I have argued from the rubric side in what teachers are actually grading.
- Parentheses around every negative that moves. Substituting as with parentheses makes the sign physically present at the moment the routine runs. Most sign slips are the negative existing in your head but not on the paper.
- Slow only the danger moments. You cannot supervise everything, so supervise the two spots your tally named. A sign-flipper who pauses one beat whenever a negative crosses an equals sign has patched their actual leak at a cost of seconds per problem.
- Recopy the problem, then verify the copy. A surprising share of "math errors" are transcription errors in line one, after which the math is flawlessly performed on the wrong problem. Five seconds of comparing your copy to the original kills the whole species.
- Remove the capture source. The research's second condition for slips is attention captured elsewhere. A phone face-up next to homework is a slip generator by definition. So is doing the last five problems at double speed to be done; if the tally shows your slips living at the bottom of every page, the fix is scheduling, not mathematics.
Why tests multiply your slips
Students always report the same asymmetry: homework is mostly clean, tests are a slip festival, and they conclude test-day carelessness is a character defect that arrives with the proctor. The mechanism is more forgiving. Whatever supervision your automatic routines do get comes from working memory, and a timed test is working memory's worst day: the clock occupies some of it, and for anxious students the worry itself occupies more, which is measurable in the research and is the same resource-theft that makes people blank on material they knew. I unpacked that mechanism, and how to tell anxiety from a knowledge gap, in why you get it in class and blank on the test.
The practical consequence: your guardrails matter more under time pressure, not less, because they are the supervision that keeps working when attention is spread thin. Parentheses written on homework out of habit are parentheses that still appear on the test when nothing else is available to guard the sign. This is why the guardrails have to be installed as habits during low-stakes practice. A checklist you only attempt on test day is itself a task competing for the working memory you do not have.
Catching the ones that get through
Guardrails cut slips; nothing eliminates them, in students or in surgeons, which is why the research tradition treats checking as part of the work rather than an insult to it. Give every answer the three-second plausibility glance, and give the problems your tally flags a real line-by-line check at the end. The point of all of this is not perfection. It is that "careless" was never the right diagnosis, and once you replace a character flaw you cannot fix with a mechanism you can, the mistakes stop being a mystery and start being a maintenance schedule.