Tag Archives: partitioning

Grouping is different from partitioning [TDI 5]

After last week’s pizza-slicing interlude, we are back on task for the closing half of the Decimal Institute.

This week, I want to invite discussion of the question, How much are decimals like whole numbers?

In case you are this far into things and cannot guess my answer (and in case you haven’t read this week’s title!), I offer the following clue.

From a purely abstract and logical perspective, decimals are exactly like whole numbers. No matter what place you are considering, the place to the left is worth 10 times as much, and the place to the right is worth \frac{1}{10} as much.

But there are many important ideas that this logical analysis ignores. And people do not always find abstract logical arguments compelling. So we’ll dig deeper than that.

I have four major ideas for us to consider. You, class, will surely have more.

1. Grouping groups is different from grouping units. Thanheiser (2009) demonstrated that some preservice elementary teachers could work competently with two-digit numbers yet make important errors with three-digit numbers. These teachers could explain the grouping inherent in writing a number like 23, but did not extend this reasoning to numbers such as 235. If decimals are really just like whole numbers, we should expect that all whole numbers are the same for learners. Thanheiser has demonstrated that they are not.

2a. Grouping patterns and partitioning patterns are often mismatched. The metric system was established by the scientific community for ease of working with our base-10 numeration system. It was developed intentionally at a moment in time when correspondences between numeration and measurement were of increasing importance.

Other measurement systems probably reflect the informal and natural ways people have of working with measurement. The Imperial system, for instance, is probably based on how people naturally view quantities.

In that case, consider the inch. Inches are grouped in twelves. They are partitioned in twos and powers of two.

The teaspoon is grouped in threes (making tablespoons) and partitioned in twos and fours.

Cups? Those are partitioned in twos, threes and fours. But they are grouped only in twos.

Time and again, the size of the grouping is not related to the number of partitions. Perhaps this is because partitioning and grouping are not closely related processes in people’s minds.

2b. This is borne out in my own work with preservice teachers. Go read my post titled, Measurement explored for full details. My experience in having students develop length-measurement systems includes these observations:

  1. Students nearly always partition in 4ths, 8ths and 16ths.
  2. Students almost never partition into 10ths.
  3. Students may group in threes or sixes, but they never ever partition this way.
  4. Students rarely think to group the same way they partition. That is, if they made 8ths, they might very well group in sixes. The convenience that would be afforded by consistency does not tend to occur to them in advance.

The comments on that post are thought-provoking and we should feel free to pick up threads of those comments in this week’s discussion.

3. Place value understanding does not seem to cross the decimal point easily. I do alternate place value work with my preservice teachers. Bear with me on this if you’re not familiar. In a base-5 system, we count 1, 2, 3, 4, 10. We make groups of this many: ***** instead of this many: **********; the latter is what underlies our usual base-10 system.

This means we write 10_{five} for our usual five and 100_{five} for our usual twenty-five. After mastering grouping with fives instead of tens, we move to partitioning. If decimals are just like whole numbers, this should present no difficulty.

But it presents tremendous difficulty. Even my strongest students have a common struggle, which is this: They view the whole and the part of a decimal number separately and treat them equivalently.

Here is what this means. Consider the base-10 number 20.20. This is “twenty and twenty hundredths”. My students tend to correctly interpret whole number part of this. Twenty is four groups of five so they write 40_{five}. But then they do the same thing with the decimal part, writing .40_{five}, so that 20.20_{ten}=40.40_{five}.

But this is not right. The decimal part represents 20 hundredths. But if we have changed bases, then the values of the decimal places change too. The first place is fifths; the second is twenty-fifths; and so on.

Through the use of grids and activities paralleling those from the Rational Number Project (Cramer, et al., 2009), they come to understand that 20.20_{ten}=40.1_{five}

The underlying difficulty seems to be that…

4. The unit changes when we add digits to the right of the decimal point. When you read whole numbers aloud, the unit is always the same—one. Thirty-two means thirty-two ones. 562 means 562 ones. Yes, the 6 has a value, and this value changes depending on its place. But no matter the number of digits, the number counts ones.

This is not true with decimals. 0.32 means thirty-two hundredths. 0.562 means 562 thousandths. Thousandths are different units from hundredths. The unit changes to the right of the decimal point in way that it does not for whole numbers.

To summarize, our question this week is: How much are decimals like whole numbers? My answer is that they are not very much alike at all. I outlined four reasons: (1) Even whole number place value is more challenging than logic suggests, (2) Our experiences with grouping and with partitioning tend not to parallel each other, (3) We tend to think of whole number parts and decimal parts as separate things, and (4) The units we count are different to the right of the decimal point, depending on how many digits there are.

How say you, class?


Cramer, K.A., Monson, D.S., Wyberg, T., Leavitt, S. & Whitney, S.B. (2009). Models for initial decimal ideas. Teaching children mathematics, 16, 2, 106—117.

Thanheiser, E. (2009). Preservice elementary school teachers’ conceptions of multidigit whole numbers, Journal for research in mathematics education, 40, (3), 251–281.

Task design

For the last two years, I have been working with the four surviving original authors of Connected Mathematics on a revision that is responsive to the Common Core State Standards. My task has been top-to-bottom revision of three of the four rational number units, Bits and Pieces I, Bits and Pieces II and Comparing and Scaling.

The process for this work is unusual for commercially available U.S. curriculum materials, so I want to share a few observations from the inside. They will trickle out over the next few months, and they’ll get filed under “Curriculum”.

The task

In CMP2, we focused the initial fractions unit on careful introduction of the number line. The premise was that children had lots of elementary school experience with area models for fractions, and that we wanted to introduce the more sophisticated linear models.

We introduced the licorice lace problem.

In this problem, a group of four kids is going on a hike. They have a 48-inch licorice lace and they want to share it equally among themselves. Sid (the protagonist in our narrative) carefully marks the places where he will make the cuts.

Just before he actually cuts, two more kids show up. Now they need to make new marks on the already marked-up lace. The cycle is repeated a couple of times. At each phase, we ask students to name the part of a licorice lace each hiker receives.

If you try this yourself, you will notice that it’s pretty hard to locate the marks for sixths when there are already fourths marked. Not impossible, but hard.

Ideally, some students in class will try, and some students will go to twelfths; others will go to twenty-fourths. Then when it’s time to name the fractions we have sixths, twelfths and twenty-fourths on the table and we can talk about equivalence and partitioning linear things.

From classroom feedback and my own experience working the problem with adults (both in professional development and college courses), it was clear that the problem needed a redesign. The set up was wordy, using one and a half pages of text to work through a small set of tasks. The marks before cutting were slightly implausible. The sharing and re-sharing was too complex for simple problem-posing.

The redesign

Two years ago, I took on the task of redesigning this problem.

I knew we needed something that was (1) linear, (2) shareable, and (3) already marked.

Linear and shareable are properties of licorice laces (these are not Twizzlers, each of which-while shareable-is not plausibly shareable among four children). That third criterion was new. If I could find something whose pieces were already marked, I could get rid of the complicated storyline and a tremendous amount of text.

“Marked pieces” is important because this is a problem about partitioning and repartitioning. We want kids to have pieces that they need to cut up further, and to have to think about names for these new pieces.

Skittles would not work.

These, while delicious, are mathematically unproductive for our purposes.

A bag of Skittles is composed of the original unit, one Skittle.

I needed something where we partition the original unit. It is perhaps shameful how many hours of thought went into this. But I eventually found it.

It’s perfect. Linear, shareable and already marked. You want to share equally? Each person gets \frac{1}{2}, sure. But can’t you see that each gets \frac{4\frac{1}{2}}{9}? Or that it could also be \frac{9}{18}?

It is helpful that most people don’t know the standard partitioning of a Tootsie Roll. (Did you know it was ninths? Be honest, now. It hasn’t always been; it used to be sevenths.) If you don’t know the standard partitioning of a Tootsie Roll, then we can make as many pieces as we like to start each new task. No more marking and re-marking. We just give a new  Tootsie Roll and a new number of people.

We know from research that sharing is a productive context for understanding fractions. We’re sharing something that is already partitioned, so we need to repartition when the number of sharers is not a factor of the number of pieces.

Feedback from classrooms and my experience working with adults (again-professional development and college courses) suggests that we get more mathematics with a lot less effort setting up than we did with the previous version.


Not every problem in Connected Math has gotten this level of attention, of course. But a lot of them have. This is a curriculum that takes context seriously as a basis for mathematical activity and abstraction.

Once we have committed to a particular mathematical development (e.g. partitioning in linear situations in order to move to the number line), we seek a problem in which the right mathematical activity naturally results. I am proud to have been a part of that.

There are unshareable sizes?

Enough about me. Let’s get back to overthinking our teaching, shall we?

This came in the mail today.

Seems innocuous enough. No goofy combination discounts. No bait and switch. Just a straight-up buy one, get one free offer.

But wait. Buy one what?

One “shareable-sized flatbread pizza”.

Contrary to what we have been teaching kids in middle school classrooms, Cosi is suggesting that there are sizes too small to share.

There are units of which we cannot take a fraction.

The rational numbers are not dense.

Who knew?