Getting Messy With Data

Abstract

Analyzing and interpreting data is essential to the practice of scientists and is also an essential science and engineering practice for science teaching and learning. Although working with data has benefits for student learning, it is also challenging, particularly with respect to aspects of work with data that are not yet very common in schools, such as developing quantitative models, understanding variation in data, and using larger, complex data sources. In this article, we aim to describe tools for engaging students in work with data in your class as well as three general strategies, including understanding data collection, experiencing the transformation of messy data sets in preparation for analysis, and modeling the data to answer a question. We show how these strategies can be employed using the freely-available, browser-based Common Online Data Analysis Platform, and outline connections to curricular standards.

Full text

GETTING MESSY WITH DATA:
Tools and Strategies to Help Students Analyze
and Interpret Complex Data Sources
Joshua Rosenberg1, Alex Edwards2, and Bodong Chen3
1University of Tennessee, Knoxville
2Tate’s School Knoxville
3University of Minnesota
This is a preprint.
APA Citation: Rosenberg, J., Edwards, A., & Chen, B. (2020).
Getting messy with data: Tools and strategies to help students
analyze and interpret complex data sources. The Science Teacher,
87(5), 30–34.
Abstract
Analyzing and interpreting data is essential to the practice of scientists
and is also an essential science and engineering practice for science teaching
and learning. Although working with data has benefits for student learn-
ing, it is also challenging, particularly with respect to aspects of work with
data that are not yet very common in schools, such as developing quanti-
tative models, understanding variation in data, and using larger, complex
data sources. In this article, we aim to describe tools for engaging students
to work with data in your class as well as three general strategies, includ-
ing understanding how data is collected, experiencing the transformation
of messy data sets in preparation for analysis, and modeling the data to
answer a question. We show how these strategies can be employed using
the freely-available, browser-based Common Online Data Analysis Platform,
and outline connections to curricular standards.
1

GETTING MESSY WITH DATA The Science Teacher, 87(5)
K ;A< J
1Introduction
Analyzing and interpreting datamay present challenges to teach-
ers and students because the Next Generation Science Standards em-
phasizes data analysis–related capabilities that are not often studied in
the classroom, such as developing quantitative models (Kastens 2015). Even mak-
ing ”simple” observations, such as the height of the school’s flagpole, requires
knowing what, how, and how many times students should measure and record
observations in light of variation in the data (Lehrer, Kim, and Schauble 2007).
Due to the shift called for in the NGSS from knowing about scientific theories
and ideas to figuring out how the world works (Schwarz, Passmore and Reiser
2017), students can now learn about data not only as interpreters of quantitative
models, but also as creators of those models themselves (Lehrer, Kim, and Jones
2011).
In addition, analyzing and interpreting newer sources of data, such as the
“big” data sets collected and created by scientists and engineers, presents ad-
ditional opportunities and challenges for science educators (Finzer, Busey, and
Kochevar 2018; Kastens, Krumhansl, and Baker 2015; Lee and Wilkerson 2018).
Traditionally, as part of the data-modeling approach, students use the data they
have collected themselves. In the context of larger sources of data, students often
use data originally collected for some other purpose, such as data from the city
in which they lived (Wilkerson and Laina 2018). Facing large data sources, stu-
dents also need to deal with analytical challenges with large data sources, such
as the importance of structuring hierarchical data (Konold et al. 2017) and using
technological tools (Finzer 2013).
In addition to what the educational tool can do, it is also essential to consider
how it aligns with particular pedagogical aims, content area, and context (Mishra
and Koehler 2006). Thus, we selected tools that we think exhibit some of the
characteristics of effective data analysis platforms for learners (see McNamara
2015). Additionally, the tools for working with data we identify are those that are
freely available (and do not require purchasing a license), browser-based (and so
can easily be used across computer operating systems), and relatively easy to use,
especially for students. This article also describes three strategies for analyzing
and interpreting complex data using the Common Online Data Analysis Platform
(CODAP).
K K;9NAB8
2
F

2Tools for working with data
K ;A< J
2Tools for working with data
2.1Desmos
Mathematics and science teachers commonly use Desmos (see
“On the web”) with, or in replacement of, graphing calculators. Like
graphing calculators, Desmos works well with functions that do not
require a data table, such as the use of the function f(x) = sin(x) to display the
form of that function. It also works well with datasets. Data can be typed directly
into Desmos or can be copied from Google Sheets or other spreadsheet software.
Then, functions, such as a sin, linear, or quadratic function, can be estimated
based on the data and added to a graph. Even for those who are not accustomed
to writing an equation can easily write complex functions.
2.2Google Sheets
Google Sheets (see “On the web”) is widely used by science teachers and students,
especially in school districts using Google Suite. A benefit of Google Sheets is
that it bears similarities to other, widely used tools, namely Microsoft Excel. This
may make it easy for students to begin to use this tool. Unlike Excel, Google
Sheets is browser-based, making it is easy for students to collaborate through
a single Google Sheet. While many high school students may be familiar with
Google Sheets, its advanced functionality, such as writing commands to populate
cells with values that rely on other cells (i.e., to create the mean of multiple
variables) or fitting functions to data, likely requires additional support. Finally,
while easy to use, sometimes Google Sheets can make it so easy to create a figure
that students may not have the opportunity to think carefully about what each
part of the figure represents.
2.3JASP and R
JASP is a statistical software program, based on R (R Core Team, 2019), that
students can use for data analysis (see “On the web”). R is a programming lan-
guage designed for data analysis. Unlike R, JASP has a point-and-click interface,
through which it is possible to perform a wide array of statistical tests. JASP may
8CAM:<J J
F3

GETTING MESSY WITH DATA The Science Teacher, 87(5)
K ;A< J
be most useful for teachers who want students to conduct complicated analy-
ses, such as t-tests for how two means (or averages) differ or multiple regression
analyses. In addition, R is most commonly used via R Studio software, which ex-
ecutes R and provides additional functionality for enhanced data-analysis work-
flows. While challenging to use, in some advanced applications—such as for the
use of methodologies for analyzing phylogenetic data—it may be useful to turn
to R. JASP has both a desktop version and a browser-based version. There is a
browser-based version of R Studio available, known as R Studio Cloud.
2.4The Common Online Data Analysis Platform (CODAP)
The Common Online Data Analysis Platform (CODAP) provides a distinctive
interface to view, transform, and analyze data and create and interpret graphs.
Developed by the Concord Consortium, CODAP (see “On the web”) draws upon
past research and development of TinkerPlots and Fathom statistical software.
One distinctive feature, related to how both data and graphs can be viewed to-
gether, is that elements of graphs, such as dots on a scatterplot, can be clicked on
to view the data to which they correspond (Figure 1). Another distinctive feature
of CODAP is its drag-and-drop interface. For example, to create a graph, columns
from a data table can be dragged to the x- or y-axes or the grid of the graph, to
color points based on the values in the column. It is also easy to load data (as
long as you can save the data as a .CSV file, which can be done in Google Sheets
or other software). A .CSV file can be dragged into the window to load the file as
a table. In addition to being easy to use, CODAP has more advanced functional-
ity, such as the ability to fit quantitative models (i.e., linear models or models for
simple linear regression). Additional resources include tutorials, example data
sets, and activities.
3CODAP strategies for analyzing and interpreting
complex data sources
We have been engaged in research about how tools such as those de-
scribed above—and in particular CODAP, because of its distinctive fea-
tures—can be used to support student learning in the context of the
NGSS. We have identified several research-based strategies that align with past
K K;9NAB8
4
F

3CODAP strategies for analyzing and interpreting complex data sources
K ;A< J
Figure 1:A screenshot of the freely-available, browser-based CODAP software.
research, and that can be used as a part of longer investigations across a lesson
sequence, over a unit, or as a part of a single lesson or class.
While we focus on how these strategies can be employed using CODAP, each
could also be employed using other tools described in this article or other than
those described here. Finally, while the strategies can be considered on their own,
they may best be considered as a part of a cycle, where students first explore how
the data were collected, then prepare a data set for analysis, and finally model
the data in order to answer a driving question.
3.1Strategy 1: Explore how the data were collected or created
Creating or collecting data is an essential step in the data analysis process (Han-
cock, Kaput, and Goldsmith 1992). This step can also serve as an introduction
to working with data, particularly for students who are familiar with the prac-
tice. When students record observations themselves, they have the opportunity
to consider how the data gets created and may be more confident when analyz-
ing it later on. When students use already-collected data or secondary data, there
8CAM:<J J
F5

GETTING MESSY WITH DATA The Science Teacher, 87(5)
K ;A< J
are still benefits to considering how the data came to be. Thus, when students
are analyzing already-collected data it is still important for students to have the
chance to think about how the data were originally collected or created. These
discussions may lead students to question how and why the data were collected
and to consider sources of bias (deliberate or unintentional) that change the na-
ture of the data, which can be seen as an example of critical data literacy (Hautea,
Dasgupta, and Hill 2017).
Figure 2:Which mammal eats meat and lives on both land and
water? Corresponding data points and their place in figures.
To help students ex-
plore how the data
were collected, start
with data that repre-
sents a single case. Of-
ten, the data that stu-
dents are analyzing
are aggregates of in-
dividual cases of data,
such as when a data
set includes a column
representing the mean
of a measurement col-
lected multiple times.
In CODAP, this is sup-
ported by the con-
nections between the
data points and the
figure (see Figure 2).
Another way is to
talk through, with stu-
dents, what the data
collection process was
like, or what it could
have been like, as fa-
cilitated through a dis-
cussion of a description of a study associated with the data, a codebook describ-
ing what the variables are, or a data collection instrument (or a description of
one).
K K;9NAB8
6
F

3CODAP strategies for analyzing and interpreting complex data sources
K ;A< J
3.2Strategy 2: Analyze complex data
Working with clean, tidy data makes it easier for students to reach conclusions;
however, particularly with complex sources of data, the need to think about
and work through the messier parts of the process—such as renaming and se-
lecting variables and joining together multiple datasets—can also be beneficial
(Kjelvik and Schultheis 2019; Konold, Finzer, and Kreetong 2017; Schultheis and
Kjelvik 2015; Wilkerson-Jerde et al. 2017). In CODAP, it is easy to include data
sets that are hierarchically structured or to create nested data structures. In this
way, students can see and explore connections between data at multiple levels.
Figure 3depicts how all of the observations associated with one elephant seal
are grouped.
Figure 3:Analyzing hierarchical data in CODAP.
Another way to
engage students in
the messier parts
of data analysis is
even more simple:
allow some time
for students to ex-
plore the data and
to generate their
own ideas about
the data. This can
be an especially
useful way to ex-
pose students to
raw, messy data,
akin to the kinds
that scientists cre-
ate and use, but
which may also re-
quire greater time and effort than is required in typical data analyses (see Data
Nuggets in “On the web” for structured activities that involve students in ana-
lyzing complex data from scientists).
8CAM:<J J
F7

GETTING MESSY WITH DATA The Science Teacher, 87(5)
K ;A< J
3.3Strategy 3: Model and explain variability in the data to an-
swer a question and solve a problem
Finally, a central goal of statistical models—and statistics—is to understand what
is going on in light of variation in the data (Aridor and Ben-Zvi 2019; Lehrer, Kim,
and Schauble 2007). Importantly, explaining variability does not need to involve
highly complex models: even a mean or a median can be an important summary.
A key part of using this strategy is recognizing that it is not essential for students
to learn about the mean or the median; it is important that students have the
opportunity to use statistics that are useful for determining what is going on
with something concrete: a phenomenon.
Figure 4:Adding a regression (linear model) in CODAP.
When using this
strategy, it is im-
portant to push
students to reach
and to defend their
conclusions in light
of variability to
answer an authen-
tic question, such
as a driving ques-
tion that allows
students the op-
portunity to an-
swer the question
in multiple ways,
as well as share and revise their answers. In CODAP, modeling and explaining
variability is easy to do by clicking on an already-created figure, as demonstrated
in Figure 4. In addition to adding a model, such as the linear model depicted in
Figure 4, students can also add statistics, such as the mean and median to a graph
(and to groups depicted within a graph). Students can also represent how spread
out a variable is through the inclusion of statistics, such as the standard deviation
or the range and through adding graphical representations of these statistics to a
graph.
K K;9NAB8
8
F

5References
K ;A< J
4Conclusion
Working with data is essential to the practice of scientists as well as
to science teaching and learning. As you consider these tools and strate-
gies, we encourage you to think creatively: data do not have to be about
something separate from students’ investigations of the world. In many cases,
data can come directly from students’ experiences in your classroom—or their
lives. We encourage you to not only seek out complex data sources that help
your students demonstrate a performance expectation, or standard, and that also
provide chances for students to investigate, critique, and share what they find
about topics that are of interest, personal investment, or relevance to them. Do-
ing so can support a shift away from students learning about the world toward
figuring out how and why the world works in the ways it does.
On the Web
CODAP: https://codap.concord.org/releases/latest/static/dg/en/cert/index.html
https://codap.concord.org/for-educators/
Data Nuggets: http://datanuggets.org Desmos: www.desmos.com/calculator
Google Sheets: www.google.com/sheets
Fathom: https://fathom.concord.org
JASP: https://jasp-stats.org www.rollapp.com/app/jasp
R Studio: https://rstudio.cloud
Tinkerplots: www.tinkerplots.com
5References
Aridor, K., and D. Ben-Zvi. 2019. Students’ aggregate reasoning with covariation.
In Topics and Trends in Current Statistics Education Research, 71–94. New
York: Springer.
Finzer, W. 2013. The data science education dilemma. Technology Innovations in
Statistics Education 7(2): 1-9.
8CAM:<J J
F9

GETTING MESSY WITH DATA The Science Teacher, 87(5)
K ;A< J
Finzer, W., A. Busey, and R. Kochevar. 2018. Data-driven inquiry in the PBL class-
room. The Science Teacher 86 (1): 28–34.
Hancock, C., J.J. Kaput, and L.T. Goldsmith. 1992. Authentic inquiry with data:
Critical barriers to classroom implementation. Educational Psychologist 27(3):
337–364.
Hautea, S., S. Dasgupta, and B.M. Hill. 2017. Youth perspectives on critical data
literacies. Proceedings of the 2017 CHI Conference on Human Factors in Com-
puting Systems, 919–930. https://doi.org/10.1145/3025453.3025823
Kastens, Kim. (2015, May). Data Use in the Next Generation Science Standards
(revised edition) [White paper]. Waltham, MA: Oceans of Data Institute, Edu-
cation Development Center, Inc. Retrieved from http://oceansofdata.edc.org/our-
work/data-next-generation-science-standards
Kjelvik, M.K., and E.H. Schultheis. 2019. Getting messy with authentic data: Ex-
ploring the potential of using data from scientific research to support student
data literacy. CBE—Life Sciences Education 18 (2): 1–8.
Konold, C., W. Finzer, and K. Kreetong. 2017. Modeling as a core component of
structuring data. Statistics Education Research Journal 16 (2): 191–212.
Lee, V. R., and M. Wilkerson. 2018. Data use by middle and secondary students
in the digital age: A status report and future prospects. Commissioned paper
for the National Academies of Sciences, Engineering, and Medicine, Board
on Science Education, Committee on Science Investigations and Engineering
Design for Grades 6-12. Washington, DC.
Lehrer, R., M.J. Kim, and L. Schauble. 2007. Supporting the development of con-
ceptions of statistics by engaging students in measuring and modeling vari-
ability. International Journal of Computers for Mathematical Learning 12 (3):
195–216.
Lehrer, R., M.J. Kim, and R.S. Jones. 2011. Developing conceptions of statistics by
designing measures of distribution. ZDM 43 (5): 723–736.
McNamara, A. 2015. Bridging the gap between tools for learning and for doing
statistics [doctoral dissertation]. Retrieved from
https://cloudfront.escholarship.org/dist/prd/content/qt1mm9303x/qt1mm9303x.pdf
Mishra, P., and M.J. Koehler. 2006. Technological pedagogical content knowl-
edge: A framework for teacher knowledge. Teachers College Record 108 (6):
K K;9NAB8
10
F

5References
K ;A< J
1017–1054.
National Research Council. (2012). A framework for K–12 science education:
Practices, crosscutting concepts, and core ideas. Washington, DC: National
Academies Press.
NGSS Lead States. 2013. Next generation science standards: For states, by states.
Washington, DC: National Academies Press
R Core Team (2019). R: A language and environment for statistical computing. R
Foundation for Statistical Computing, Vienna, Austria. URL https://www.R-
project.org/.
Schultheis, E.H., and M.K. Kjelvik. 2015. Data nuggets: Bringing real data into the
classroom to unearth students’ quantitative & inquiry skills. The American
Biology Teacher 77 (1): 19–26.
Wilkerson, M.H., and V. Laina. 2018. Middle school students’ reasoning about
data and context through storytelling with repurposed local data. ZDM 50
(7): 1223–1235.
Wilkerson, M.H., K.A. Lanouette, R.L. Shareff, T. Erickson, N. Bulalacao, J. Heller,
and F. Reichsman. 2018. Data moves: Restructuring data for inquiry in a
simulation and data analysis environment. In Making the Learning Sciences
Count, 13th International Conference of the Learning Sciences, eds. J. Kay
and R. Luckin, 1383-1384. London, UK: International Society of the Learning
Sciences.
8CAM:<J J
F11