A Systematic Review of Empirical Research on Graphing Numerical Data in K-12 STEM Education

Graphs are essential representations in the professions and education concerning the science, technology, engineering, and mathematics (STEM) disciplines. Beyond their academic relevance, graphs find extensive utility in everyday scenarios, ranging from news media to educational materials. This underscores the importance of people’s being able to understand graphs. However, the ability to understand graphs is connected to the ability to create graphs. Therefore, in school education, particularly in STEM subjects, not only the understanding but also the skill of constructing graphs from numerical data is emphasized. Although constructing graphs is a skill that most people do not require in their everyday lives and professions, it is a well-established student activity that has been empirically studied several times. Therefore, since a synthesis of the research findings on this topic has not yet been conducted, a summary of the studies investigating graphing via various viewpoints and differing methods could be a valuable contribution. To provide an overview of the empirical literature on this important topic, our systematic review identifies how the construction of convention-based graphical representations of numerical data, referred to as graphing, has been studied in previous research, how effective graphing is, and which types of difficulties are encountered by students. Based on these aspects, we defined inclusion criteria that led to 50 peer-reviewed empirical studies on graphing in K-12 STEM education found in SCOPUS, ERIC, and PsychInfo. Graphing instruction seemed to be beneficial for student learning, not only improving graph construction but also graph interpretation skills. However, the students experienced various difficulties during graphing, both during graph construction and the interpretation and usage of data.

💡 Research Summary

This article presents a systematic review of empirical research on the construction of convention‑based graphical representations of numerical data—referred to as “graphing”—within K‑12 STEM education. The authors searched three major databases (SCOPUS, ERIC, and PsycINFO) for peer‑reviewed studies that (1) involved students actively creating graphs from raw numerical data, (2) were situated in science, technology, engineering, or mathematics curricula at the primary or secondary level, and (3) reported measurable learning outcomes. After applying inclusion and exclusion criteria (e.g., excluding studies that only required interpretation of pre‑made graphs or that dealt with purely mathematical function graphs generated from equations), 50 studies remained for analysis.

The review categorises the methodological landscape of these studies: experimental pre‑post designs with control groups, quasi‑experimental classroom interventions, case studies, and qualitative investigations using interviews or coding of student work. Statistical techniques range from ANOVA and regression analyses to structural equation modelling, while several papers employ content analysis or meta‑analytic synthesis to integrate findings across heterogeneous designs.

Across the corpus, a consistent pattern emerges: graphing instruction improves both the accuracy of students’ graph production and their ability to interpret graphs. The effect is robust across age groups (elementary through high school), subject domains (physics, biology, chemistry, mathematics), and instructional formats (hand‑drawn versus computer‑based tools). Moreover, students who generate graphs demonstrate deeper comprehension of data structure, better alignment of axes and scales with the underlying phenomenon, and enhanced scientific reasoning skills compared with peers who only practice graph reading.

The authors situate these outcomes within two theoretical families. First, the multiple‑representations literature (e.g., Ainsworth’s DeFT framework, the Integrated Model of Text and Picture Comprehension, and Mayer’s Cognitive Theory of Multimedia Learning) predicts that constructing a second, complementary representation (the graph) alongside a primary representation (a table or text) facilitates dual‑coding, reduces cognitive load, and promotes transfer. Second, generative learning theories—including Chi and Wylie’s ICAP model and Wittrock’s generation effect—classify graph creation as a “constructive” activity, which is empirically linked to higher learning gains than passive or merely active tasks. The review confirms that graphing aligns with these theoretical expectations, offering a concrete example of how generative, multimodal tasks can boost learning.

Nevertheless, the synthesis also highlights systematic difficulties that students encounter during graphing. The most frequently reported errors involve (a) incorrect labeling of axes or omission of units, (b) inappropriate selection of scale or interval size, (c) mishandling of data points (e.g., misordering, omission, or inaccurate plotting), and (d) poor choice of graph type relative to the data (e.g., using a line graph for categorical data). These errors are interpreted as manifestations of limited metarepresentational competence—students struggle to translate numeric information into a conventional visual code. The authors argue that explicit instruction on graph conventions, scaffolded practice, and timely feedback are essential to mitigate these pitfalls.

The review identifies several gaps in the existing literature. Comparative studies that directly contrast digital graphing tools (spreadsheets, specialized software) with traditional hand‑drawn methods are scarce, leaving educators without clear guidance on optimal tool selection. Longitudinal investigations of the durability of graphing‑related gains are virtually absent, as most studies assess outcomes immediately after instruction. The relationship between teacher expertise in graph pedagogy and student outcomes remains under‑explored, as does the influence of cultural or linguistic factors on students’ interpretation of graph conventions.

In conclusion, the systematic review provides strong empirical support for incorporating graph‑construction activities into K‑12 STEM curricula. Graphing not only improves students’ ability to produce accurate visual representations but also deepens their data literacy and scientific reasoning. To maximise these benefits, educators should (1) teach graph conventions explicitly, (2) embed graphing within generative, multimodal learning sequences, (3) provide scaffolded feedback targeting common errors, and (4) consider blended use of digital and analog tools. Future research should address the identified gaps, particularly through controlled comparisons of instructional media, longitudinal designs, and investigations of teacher‑mediated factors, thereby refining evidence‑based practices for graphing instruction in K‑12 STEM education.