Understanding how to anchor data investigations in real-world phenomena by framing problems before analysis begins.
Before we graph data, calculate averages, or run statistical tests, we need to step back and ask:
What problem are we actually trying to understand?
Framing the problem anchors a data investigation in real-world phenomena. According to Lee et al. (2022), effective data investigations do not begin with tools or techniques; they begin with context, purpose, and questions worth answering. This framing phase shapes every decision that follows, from the research question we pose to the data we choose to analyze.
Importantly, this is not a one-time step. Throughout a data investigation, we return to the framing phase to ensure our analysis still aligns with the real-world problem we care about.
By the end of this module, teachers and students will be able to:
Data investigations begin with real-world phenomena, not datasets.
At this stage, I will ask you to zoom out and identify:
This is where background information comes in. We are not analyzing data yet; we are building contextual understanding.
Suppose we want to investigate electrical fires.
The broader issue might include:
Here, the investigation is grounded in a real system that could be improved, which emphasizes problem-solving in real-world context.
LLMs are especially useful here for brainstorming and contextual exploration.
Example prompt (Zero-shot prompt):
“Act as a science teacher helping students understand a real-world problem.
What are some typical broader issues related to electrical fires in residential neighborhoods?”
At this stage, the LLM helps surface possibilities, not answers. Teachers and students still decide which issues are relevant and meaningful in their context.
Once the broader issue is identified, the next step is to articulate why the investigation is important.
This helps you move beyond “we have data” to “this data helps us understand something meaningful.”
Questions that may be asked here include:
In the electrical fire case:
Example prompt (Zero-shot prompt):
“Why might understanding the causes of electrical fires be important for community safety and engineering decisions?”
This supports sensemaking, but students should still discuss and refine the importance in their own words.
Only after understanding the broader issue do we narrow our focus.
Framing the problem includes determining what focus is most productive. Sometimes, the initial problem needs to be reframed into a question that is more answerable with available data.
Broad issue:
Electrical fires are increasing in residential areas.
Focused investigative question:
“What types of electrical wire materials are most commonly associated with electrical fire incidents?”
This question:
LLMs are particularly useful for question refinement, not question generation alone.
Example prompt (Zero-shot prompt):
“Here is a broad issue: electrical fires in residential buildings.
Suggest three possible data-driven research questions that could be investigated using existing datasets.”
Teachers and students then evaluate:
Once a research question is defined, we ask:
What data would help us answer this question?
This includes:
Research question:
“What electrical wire materials are most commonly used in electrical fire incidents?”
Possible datasets might include:
Here, alignment matters. A dataset that lacks wire material information, even if large, is not useful for this question.
Example prompt:
“Given this research question [paste research question here], what types of data variables would be necessary to investigate it?”
The LLM helps identify what to look for, but teachers and students still evaluate data quality and relevance.
A key point from Lee et al. (2022) is that framing the problem is iterative.
As data is explored, we may realize:
At this point, we return to framing, not as failure, but as scientific practice.
Example Reflection Prompt (Zero-shot prompting):
“Based on the available data, does this research question need to be refined? If so, suggest how.”
This models authentic data science and scientific inquiry.
In this module, I want teachers and students to see that data investigation begins long before analysis. Framing the problem shapes the questions we ask, the data we select, and the conclusions we draw.
LLMs are powerful partners in this phase, but only when used to expand thinking, not replace it.
In the next module, we move from framing to exploring and preparing data, where numbers, variability, and patterns take center stage.
Reference:
Lee, V. R., Wilkerson, M. H., & Lanouette, K. (2022). A call for a reimagined science education data literacy. Journal of Research in Science Teaching, 59(6), 1019-1049.
Here are some more articles you might like to read next: