When Handwriting Interferes With Assessing

I touched on this in my previous article when I discussed how one of the major improvements of digitizing the screener was no longer having to slog through the interpretation of student handwriting. I thought that point deserved more exploration, because while I am going to be talking about this topic specifically, it is possible that teachers might find similar examples here or in other assessment areas that are simply part of life as an educator. The overarching question is this: What am I really assessing? And is what I am doing actually assessing it?

One of the data points that we gather three times per year is students’ spelling capabilities using the LETRS spelling screener. As a quick recap for those who may not be familiar with how the screener works, in a nutshell there are 25 words, which vary depending on whether you are administering the basic or the advanced version. Those 25 words measure a wide range of phonics rules, including root words, vowel knowledge, and roughly ten other phonics-related features embedded in the words students are asked to spell. In short, the purpose of this assessment is to tell us what a student knows—or does not know—when it comes to these various phonics and spelling rules.

I think any ELA teacher would tell you that this is a perfectly legitimate assessment. That said, adding it on top of all the other assessments we administer—the mandatory assessments, the common assessments, and the “highly recommended” assessments—is part of the reason why simplifying the process by digitizing it felt like such a breakthrough. However, one of the unexpected benefits of digitizing this process was the cleanup of assessment noise that I did not even realize was there, noise that was hiding in student handwriting.

I am not sure what grade you teach, but I teach fifth grade, and I can confidently say that the handwriting of my students spans a traditional distribution curve. On one end, there is handwriting that is graceful, elegant cursive, the kind you might see in one of your grandmother’s letters, bordering on calligraphy. On the other end of the bell curve, the handwriting looks closer to hieroglyphics or vague geometric shapes etched deeply into the paper, torn as if the student thought they had to carve the words out of stone. You all know the kind of handwriting I am referring to—the kind where it looks like the words themselves were struggling to break free from the page they were reluctantly placed on, becoming misshapen and torn in the process.

It is on this side of the bell curve where the noise lives, the noise that can interfere with otherwise good data. The simple fact of the matter is that, assuming I can read this pseudo-alphabet at all, poor penmanship—and occasionally even a student’s unique stylistic letter formation—can make it very difficult to tell exactly what a student meant to write. Is that an e or a z? Now I have to search through the rest of the paper looking for other words with es and zs so I can start comparing. And while for most students I am only doing this occasionally, there are always one or two students per class whose alphabet bears very little resemblance to the letters of our proper Latin alphabet, especially when their letters seem closer to Phoenician than Roman.

On more than one occasion over the years while administering this assessment, I realized that a student had actually spelled the word correctly—if we can forgive the poor penmanship, of course—and that it was my interpretation of the letter that was wrong. The student would assure me that a particular symbol on the page was, in fact, the letter a, and once I recognized that it was an a, the student’s score jumped several points higher than I had originally recorded. This also meant that the phonics groups and spelling interventions I had planned for that student were now entirely irrelevant, because the student did not need help with the word parts I thought they did.

This brings me back to the larger question: What are we really trying to assess? On the spelling screener, what we are theoretically trying to assess is whether a student knows how to spell. That seems like a simple question, but it is often confounded by interfering factors such as poor penmanship, students writing the correct word on the wrong line, writing on the back of the paper, or writing on a separate sheet that they forgot to put their name on. Taken together, these placement errors—and especially the challenges of interpreting student handwriting—were slowing me down while I was already spending several hours manually entering eighty-two data points per student. While I could complete most students in just a few minutes, these particular students often took fifteen or even twenty minutes each.

It was while I was considering building a digital screener that I had the epiphany: I am not trying to assess handwriting or penmanship. That is not the purpose of this assessment. If I wanted to analyze handwriting or penmanship, I would design an entirely separate assessment. This assessment was designed to test students’ knowledge of phonics and spelling rules. But the introduction of handwriting—and the universal acceptance that this was simply how it had always been done—was adding measurement noise on top of an already inefficient process.

While a teacher’s data might be largely correct overall, as I mentioned in my earlier example, even misreading a single letter from a single student can radically change the picture of what that student does or does not know. I have had students who earned a perfect score once they explained what their letters actually represented.

Now, as a teacher, I know that sometimes the obvious answer bears stating anyway, so let me address it directly: How does a digital spelling screener reduce the measurement noise caused by student handwriting? Quite simply—students do not write the words. They type them. Naturally, if they are typing instead of writing, the responses are far easier to read, and from firsthand experience I can tell you that they are.

Not only that, but because students type their responses into a Google Form, I now have a complete digital record of what they wrote. Instead of keeping paper copies stapled to the original assessment, as I had done in the past, I can maintain a digital copy of every screener indefinitely. Did a student type a word on the wrong line? It does not matter—the digital screener places it in the correct location. Is a student’s penmanship poor? Entirely irrelevant, because we are assessing spelling, not handwriting. The digital screener removes the subjectivity—and the time required to manage that subjectivity—improving the overall pedagogy of the assessment and increasing accuracy beyond what was previously achievable, except perhaps in classrooms blessed with uniformly excellent handwriting.

Even if all of your students have great handwriting, the speed and efficiency of a digital screener still saves time, because the data no longer has to be entered manually into a spreadsheet. Being able to see typed responses allows for a quick, clear determination of whether a student does or does not know how to spell a word.

Let’s come full circle and answer the overarching question once more: What is this assessment really trying to assess, and is it actually doing that? In my case, if the goal is to assess spelling, then while this assessment generally succeeds, there were always edge cases. Those edge cases matter, because the students who fall there are often the ones who need the most support. In those cases, the assessment was quietly failing them, creating an unseen inequity where teachers—without any intention—were partially judging spelling ability through the lens of handwriting ability.

From experience, I can tell you that a student with beautiful handwriting can still be a poor speller, and a student whose handwriting looks like every letter had to be carved out of the page can sometimes be your strongest speller. Removing that measurement noise by allowing students to type their responses instead of write them eliminates that unseen bias, improves equity, and gives us what we wanted in the first place: high-quality

phonics data.