How are/should we make the graph?

Data Moves & Topics August 18, 2022

I am often asked a variation of this question by teachers and educators working with learners as young as Kindergarten and as old as undergraduate students. It is a great question, and I think worth more of a strategic approach to how we answer it in our own circumstances.

Additionally making a graph (regardless of how we make it) takes time…which is the most precious and limited resource we have as educators / facilitators of learning so thinking about this more strategically can be beneficial on many fronts.

Let’s explore…

“Big Data, Big Changes? A Survey of K-12 Science Teachers in the United States on Which Data Sources and Tools They Use in the Classroom” article is available for download at https://edarxiv.org/tv4zg/.

FIRST, WHAT ARE WE USING NOW IN OUR CLASSROOMS?

When I ask teachers whom I work with, often the answer is something along the lines of “mostly by hand, but we do some Google Sheets/Excel but that takes SO much time.” The specifics of the responses varies slightly by whether I am talking with elementary, middle, high school, or college educators…but almost always both (by hand and with spreadsheet programs) are included.

Josh Rosenberg and colleagues recently published their findings from a February 2020 poll of 330 elementary-high school science teachers throughout the United States. Their data also indicates that by hand (pen-and-paper) and with spreadsheet or other programs (digital tools) are being used by elementary through high school science classrooms.

Additionally, they also see shifts, similar to those that I hear about on the ground with teachers, in terms of how much of each kind of tool (as well as depending on the kind of data students are looking at). It is this small shift from pen-and-paper to digital tools that I want to explore further to push us to think more strategically and pedagogically about how we are having our students make graphs.

TIME TO UP THE PROGRESSION

As Rosenberg and colleagues (2021) indicated there is a natural shift away from graphing by hand into more digital tools into the upper grade levels of K-12. But we can (and I argue) should be making a wider shift to digital tools. Specifically, our graphing should be mostly by hand and/or with manipulatives in elementary school, and mostly with digital tools in high school. And yes, like many things in our K-12 sequence that puts middle school at the transition or inflection point between those two.

I believe it is important to more actively and strategically make this transition away from graphing by hand in the later years of the K-12 sequence for two primary reasons:

1. Limited dataset size when graphing by hand.

2. Best preparing them for whatever they do in the workforce.

Let’s explore each in turn…

#1 Limited dataset size when graphing by hand.

It physically takes time to graph every data point (or even if they average the data points by group and plot that…which has a whole host of issues with it, but that’s for another time). Therefore, often students are actually graphing somewhere between 2-20 data points/bars when they are making graphs by hand. Don’t get me wrong…I am NOT saying that we should take more time so they can add more data points/bars to their graphs.

But let’s explore this a bit further in terms of how this has unintended consequences on what students are learning about sample size and the conclusions we can draw from data. How many data points we are looking at influences what kinds of conclusions we can draw from the data. Let’s look at an example comparing the relationship among earthquake magnitude and depth using the freely available Devastating Earthquakes dataset from Tuva (I know, I know these are not graphs drawn by hand, but I share them to illustrate the point…so bear with me :)).

There are different things that we would say about the relationship between earthquake magnitude and depth from each of these graphs. Now, this example has been idealized to emphasize the point…but the reality is that whenever we look at smaller datasets we have to always be cautious and think about whether the data is representative of the whole population/phenomenon.

It is worth noting that another aspect of this is that often times the smaller the dataset the less variability (noise, messiness) exists in the dataset. This too can give students a false perception of what relationships in data look like (again that is for another time).

This file comes from Wellcome Images, a website operated by Wellcome Trust, a global charitable foundation based in the United Kingdom. Refer to Wellcome blog post (archive).