← StudyX Devlog

Math Apps That Ask Kids to Write the Answer, and Apps That Ask Them to Choose

Hiroki ObaraHiroki Obara · DeveloperMathCalHow-To
Math Apps That Ask Kids to Write the Answer, and Apps That Ask Them to Choose

Math apps for children usually ask for answers in one of two ways. The app shows 7 + 8, and the child either writes 15, or picks it from a row of choices such as 13, 14, 15 and 16.

In both cases the child answers a question. The difference is whether the correct answer is nowhere on the screen, or sitting there among the options. Memory researchers have compared these two kinds of questions for years, usually as multiple-choice versus short-answer or cued-recall tests. Here are three of those studies.

The basic case for practising by recalling, rather than rereading, is in Is Recalling Better Than Rereading?.

The charge against multiple choice

A 2012 study by Jeri Little and colleagues at UCLA opens with a common criticism of multiple-choice tests. Because the correct answer is shown as one of the alternatives, such tests may engage recognition rather than the productive retrieval known to help later recall.

The researchers tested whether multiple-choice questions could still trigger retrieval if the alternatives were plausible enough.

In the first experiment, 32 undergraduates read two passages of about 1,100 words each, one on Yellowstone Park and one on Saturn. They then took a test on one of the passages, either multiple choice or the same questions with no alternatives. A multiple-choice item might ask for the tallest geyser in Yellowstone National Park, with four of the park’s geysers as the options. A paired question used the same four names but had a different geyser as its correct answer.

On that first test, students answered 70% of multiple-choice questions correctly and 43% of the questions without alternatives. Choosing was easier in the moment.

Plausible options helped with the other answers too

After a five-minute filler task, everyone took a final test on both passages, with no alternatives.

Both formats helped. Questions from the first test were answered better than questions from the passage that had not been tested, whether the first test had been multiple choice or not.

The formats split on the paired questions that had not been asked before. Students who had first answered without alternatives did worse on these paired questions than on questions from the untested passage. Students who had first answered multiple choice showed no such drop.

In the second experiment, 96 undergraduates took part, and half of them saw the correct answer after each question on the first test. Feedback raised later performance for both formats, more so for the questions without alternatives. With feedback, final scores on the repeated questions were about the same for the two formats. On the paired questions, multiple choice came out ahead whether or not feedback was given, and above the untested passage.

The authors conclude that multiple-choice tests, when the alternatives are plausible, can make test takers retrieve why the correct answer is correct and why the others are not.

Writing the answer does not always win

A 2014 study by Megan Smith and Jeffrey Karpicke at Purdue University ran four experiments with 372 undergraduates in total. After reading texts, students practised by typing their own answers, by choosing from alternatives, or by a hybrid: typing an answer first, then answering the same question in multiple-choice form. They were tested a week later.

Every format beat reading alone, on questions taken straight from the text and on questions that needed an inference.

In three of the four experiments, typing an answer or using the hybrid gave little or no advantage over multiple choice. In the fourth, the researchers used shorter texts and chose questions that a pilot group had answered about equally well in both formats (89% multiple choice, 83% short answer), and everyone saw the correct answer after practice. Under those conditions, on a final short-answer test, the hybrid format beat multiple choice and short answer had a marginal advantage over it.

The authors write that the results challenge the simple conclusion that short-answer questions always produce the best learning, and show how much retrieval success matters in retrieval practice.

Second graders: wrong options can stick

Those two studies were with university students. A 2012 study by Elizabeth Marsh and colleagues at Duke University tested 36 second graders.

The children answered general knowledge questions on science, history, geography and similar topics in multiple-choice form, each with the correct answer and three plausible wrong ones. Some children heard the correct answer right after each question. Then all of them answered a mix of old and new questions with no alternatives. Everything was answered aloud.

Questions the children had already seen in multiple-choice form were answered correctly more often than new ones (59% against 25% on average). The gap was larger for the children who had been told the correct answers.

But the children who had not been told gave a wrong multiple-choice option as their answer on 18% of the questions they had seen before, against 3% of new questions. The gap was larger for hard questions. With immediate feedback it was 5% against 4%, so the effect was gone.

The authors write that multiple-choice tests remain a workable option for educators, even with children as young as second graders, and that care should be taken to give corrective feedback so that children do not repeat the wrong options later and take them into what they know.

What the three studies show

In all three, a question answered once, in either format, was answered better later than one that had not been practised.

Multiple choice gives higher scores in the moment. With plausible options it can make learners recall information about the other options as well, which answering without options did not do in Little’s experiments. With children, leaving out the correct answer led some of them to repeat wrong options they had seen, and immediate feedback removed this.

Answering without options came out ahead of multiple choice when learners could answer correctly in the moment and saw the correct answer afterwards.

All three studies used texts or general knowledge questions, not arithmetic. And “answering on your own” here meant typing or speaking, so none of them compares writing by hand as such.

What you can do in MathCal

MathCal is a math app where children write their answers by hand on the screen. I make it.

Every answer is handwritten: the child writes the numbers and symbols on the screen, and the app checks them. In my developer interview, asked why I chose handwriting over number buttons, I said: “Because moving your hand gets the brain working.”

In the Column Arithmetic Drill, the answer goes one digit per box. A box that has not been read yet is tinted, and each digit the app reads appears small above its box. Tapping the space next to it clears that box so the child can write it again. Once every box has been read, the Check answer button turns on and shows the answer it read, as in “Check answer with 123”. On iPad, you can also write with Apple Pencil. Settings also include “Recognizes 5 with 2 strokes” and a “Left-handed interface”.

Today’s Drill, Today’s Challenge and Today’s Lesson on the home screen are free every day, and setting a birthday picks them to suit the child’s grade. When they finish, they can stamp a calendar. Choosing any drill or lesson from the full lists requires a one-time purchase. There are no ads. How to start is in Getting Started with MathCal.

Download on the App Store

MathCalHandwriting game for mental math