Part 2. The FLOAT Method
2.1 What is the FLOAT Method?
What is the FLOAT Method?
The FLOAT Method is a five-step process that facilitates your ability to outline your project by (1) conceiving of a research question, (2) explaining how you locate, (3) organize, and (4) analyze a given data source, and finally, (5) transforming your findings into a visualization.
Finding Pearls with the FLOAT Method
In Storytelling with Data, Cole Nussbaumer Knaflic brings attention to two types of analyses that relate to the visualization process: exploratory and explanatory analysis. Exploratory analysis is what you do to understand the data and figure out what might be noteworthy or interesting to highlight. Explanatory analysis is when you settle on a specific finding you want to explain – or, a specific story you want to tell.
She writes, “We might have to open 100 oysters (test 100 different hypotheses or look at the data in 100 different ways) to find perhaps two pearls.” She goes on to note people too often overwhelm audiences by trying to show 100 oysters, that is, in other words too much. The more ideal situation is this: “Concentrate on the pearls, the information your audience needs to know.”
The FLOAT process enables you to focus on pearls and communicate insightful findings to your audience. Data analysis can be part of a study, which involves the development of research design, data collection, and interpretation. The process of exploring data brings about different possibilities for deriving insight about a particular finding.
The FLOAT Method (Extended)
F – Formulate (a good research question)
- The first step is FORMULATING an insightful and feasible research question, which will define the parameters of your project and guide your analysis.
- Exploratory Question: This initial question is broad in scope and tends to lack specificity. How difficult would it be to locate the data should be taken into consideration during this phase.
- Explanatory Question: Researchers usually revise this question after analyzing the data source in-depth. This revised question is more targeted in scope and reflects the limits of a given data source.
L – Locate (a sustainable data source)
- The second step involves LOCATING a suitable data source to answer your research question.
O – Organize (your data source)
- The third step is ORGANIZING your data to make sure it is cleaned and in a suitable format to be manipulated using data-analysis and visualization software.
A – Analyze (your data source)
- The fourth step is ANALYZING data in order to identify trends and other patterns that help to answer your overall research question(s).
T – Tell (a data-driven story)
- The final step, TELLING a story, involves composing a narrative to effectively convey your findings. To tell the story, you will need to choose a suitable visualization medium. The explanatory research questions facilitate the creation of useful visualizations.
Setting Expectations
To be comfortable within the FLOAT method, researchers must be able to do two things: (1) set and reset expectations at the start (and as they work on a project) and (2) determine whether they need to cycle or recycle through steps (often called iteration). This is a metacognitive process where a researcher thinks about things purposefully.
It can help to begin with general questions ranging from “How feasible is this research?” and “Will this analysis be useful in some way?” to practical questions, such as, “How much effort is involved and will there be costs?” These initial questions assists a researcher in making reasonable assessments about the feasibility of a project, and determining whether they are making adequate progress. As the process occurs, confusion around what is next will signify that prior steps need to be revisited in order to strengthen decisions made in order to move forward.
A good process of setting expectations is similar to your high school science class where you learn about how science is the process of testing a hypothesis, or an educated guess. In the case of data analysis, before you actually test a research question, you must first set a clear “hypothesis” for the analysis itself. This involves gathering information to (1) determine if the analysis will be useful, (2) determine what information is out there, and (3) determine if the data you have are suitable.
For example, a college student may want to go on a date with a potential partner on a Saturday night. There is information that the college student may need to collect prior to asking the question of the other party. First, do they have enough money to go on a date? They can research this by looking at the cost of a few potential restaurants’ meals, movie tickets, and other potential date costs.
They may also need to check their bank account or determine the next time they will be paid – especially, if they plan to pay for any or all parts of the date. This person may also need to interact with the other party ahead of time to determine if there is chemistry and interest. They may also try to determine if the other party is un-partnered and available on a potential date and time.
Some of these activities may involve some effort, such as working extra hours at a part-time job to have more spending money. After exploratory research has been performed, then the college student is able to then ask the question, “Will you go on a date with me on Saturday?”
For data analysis, the same process occurs, but the difference is in the information sources. Researchers are expected to search through the literature to see how similar topics have been researched. You may also need to see (1) what data is available, (2) what software and technological resources are at your disposal, and (3) what time and financial costs are needed to complete this project.
This process makes it easy to FLOAT from beginning-to-end of any given data driven project.
Iteration through the FLOAT Method
As a researcher set expectations before and during data analysis. There may be times that this leads to reverting to previous points in the FLOAT method. Let us discuss how this works in practice: Suppose a researcher is interested in how Black artists are recognized in the music industry. A researcher would first start by searching to see if others have researched this or similar topics already.
Imagine this: a researcher finds the recent work of others looking at the number of awards Black artists has received in comparison with artists of other races. While these projects are a good start, a researcher may want to look at a different facet of the data. So, it may be time to go back to a new research question by returning to FORMULATE. Are there other ways to investigate this topic?
Perhaps this research could look at just a particular artist or subset of artists who have reached widespread acclaim, coupling theories of breaking past Blackness into pop superstardom, with entertainers like Beyoncé or Rihanna. Could data be collected to determine how these “breakthrough” Black artists are recognized? What data is available about these artists? And, how would one define “recognition?”
Perhaps, the researcher determines, for their case, that recognition will be defined as number of awards from mainstream institutions (MTV Awards, Grammy Awards, etc.), and traditionally Black awards (BET Awards, NAACP icon awards, etc.). As the researcher moves on to LOCATE and works to collect data, there may be issues with how information about award shows is provided. Is each award category for each award show placed on its own webpage? Do you would have to take a massive effort to gather the data? Or, are the data downloadable or available in an API? Are some important components of the data missing or unavailable?
If so, the researcher may have to re-formulate the question using data that is available. Or, if the time and resources are available, they would need to spend a significant effort during the ORGANIZE step accessing and structuring the data for their project. As the researcher continues to ANALYZE, they may find that a crucial variable is missing. This may require returning to LOCATE more data or additional ORGANIZE efforts before returning to the analysis steps.
Finally, when analysis is complete, as the researcher is attempting to TELL the story about a particular insight, there are often additional questions that they think of or that come from others looking at their preliminary work. Have these rates changed over time? Are Beyoncé and Rihanna seeing the same treatment as other Black artists? Are there exceptions or trends not originally anticipated? This requires the researcher to go back and ANALYZE a slightly different component of their data. The various iterations of the FLOAT method ensure that a researcher can identify a specific finding and create an effective data visualization story in the end.
By Peace Ossom-Williamson & Kenton Rambsy
(See bibliography for sources)