An introduction to data types

What is data?

Data is a broad and subjective designation. The National Institutes of Health’s National Library of Medicine states that “The term “data” does not have one clear definition and can be interpreted differently depending on the context, such as a researcher’s field of study¹.” Because data is so widely and contextually defined, and because of the complexity of how human-centered projects work in the wider world, we recommend delineating your data into three types. Using a mix of data types in your work, and documenting your decisions at each step in your process, will help others to engage with them effectively. Imagine measuring impact as composing a piece of music. Each data type serves as an instrument, each with its own unique instrumental voice and role. By coordinating these different ‘data voices,’ we can create a cohesive and harmonic understanding of the intervention’s impact. Let’s explore the different data types before we consider how the diversity of data types together contribute to telling a more comprehensive impact story.

Data types

There are three broad data types to use in measuring human-centered projects:

• Quantitative
• Qualitative
• Historical

Using qualitative and quantitative data together “…provides a tradeoff between breadth and depth and between generalizability…²”, while historical data provides a means by which we can derive value from data-gathering efforts that preceded our work.

What is quantitative data?

Quantitative data “…are data represented numerically, including anything that can be counted, measured, or given a numerical value.”³ It captures a lot of data points, but those data points aren’t very nuanced. This data type can be thought of as hard facts and figures that help us understand the world around us in a precise way. It’s definitionally backwards-facing: something has to have happened for quantitative data to be accrued.

Quantitative data is frequently perceived as being objective and most useful in evaluations. However, any dataset is subjective in terms of significance, method(s) of gathering, interpretations, and use. Instead of “objective,” consider quantitative data for exactly what it is: a foundational part of your measurement strategy; one to be used in concert with other types of data. Examples of quantitative data:

• Visitor and attendance numbers
• Completion times
• Budget numbers

A nuance from data and evaluation science

In data and evaluation science, any data that is captured using numbers is called “quantitative data.” Examples of this include:

• Customer sentiments such as feelings, opinions, and attitudes towards a product or service, frequently packaged as Likert (i.e., rating scale) surveys
• Surveys containing multiple choice answers where the answers are subjective

The inputs of these surveys are subjective, i.e., people’s feelings and perceptions, not the objective data you might think of when you think of quantitative data. But because participants assign numbers to their feelings, or quantify them, this data is quantitative. We recommend following this standard in your own work as well.

Advantages

The main advantage of quantitative data is that it can be big, and it can be objective. Counting the number of births, business licenses applied for, or revenue earned are all examples of really useful quantitative data. It can also be used to mathematically scale insights across large populations and into the future for forecasting. Quantitative data is often characterized as broad and shallow in nature, since it can tell you a little but at a large quantity.

What is qualitative data?

Qualitative data “are data representing information and concepts that are not represented by numbers.”⁴ It’s typically thought of as narrow and deep, or even as “thick.”⁵ This means that information can be in-depth, nuanced, and layered, but in a very focused area. This can be an objective description of a physical situation, as in the statement “Rainwater is cold,” or a human’s perspective on that situation, such as “Rain is pretty.” In contrast to quantitative data, qualitative data can give us information about desired or feared future states, as in the statement, “Rainwater is cold, so even though it’s pretty, I don’t want to bathe in it.”

Other types of qualitative data aren’t initially represented by numbers, such as:

• Observational data
• Notes from interviews with individuals or groups to hear about their positive and negative experiences
• Answers from open-ended questions in a survey
• Current photographs, or audio/video recordings that capture someone’s experience, emotions, or perceptions

This non-numeric data can be represented numerically, which we’ll do when using it in concert with quantitative data as part of a compiled indicator. For details on how to do this, please see the qualitative data scoring section of the HCD measurement operations guide.

Advantages

Qualitative data’s main advantage is that it can help us understand people’s perspectives, desired future states, or desired past states. It is often a good complement to round out quantitative data.

For a deeper dive on how to gather qualitative and quantitative data, please see the HCD discovery operations guide. Link Needed

What is historical data?

In the public sector, interventions can run for years, decades, and even generations. Because of this long-term context, public sector workers often inherit interventions upon starting a new job, instead of building interventions from primary research, through design, delivery, and into measurement. This long-term context means that it’s important to be able to use the data captured in previous data-gathering efforts and/or accumulated prior to our landing in our jobs, to help us understand the effects interventions are having. In this guide, we call this historical data.

Historical data helps us understand how we got to our current situation, but it’s problematic in that we can’t measure the past directly, nor can we control how past data was gathered and stored. This means that historical data is almost always partial, not clean, and can be biased. To deal with these challenges, historians rely heavily on data triangulation to verify different facets of a situation. This means looking at historical datasets in relationship to each other to understand the situation your intervention created, and was a part of, in the past.

Advantages

Historical data helps us understand how we got here. It allows us to understand how and why we are the way we are, and do the things we do. Without looking into the historical record, however flawed it might be, we cannot accurately interpret our present context. This is especially true in government, where the relationship with the customer, the public, is longer than lifetimes; it is intergenerational, and the impact our work has on the public is life-changing. Indeed, in the words of Horst Rittel, the public sector worker “has no right to be wrong,”⁶ even in the midst of imperfect data. We must know what came before in order to understand how best to move forward.

Conclusion

Using mixed data types means compositing data does not share a unit of measure, or a format, or even a common logic model. While it can be tempting to try to compare “apples to apples” and only use one type of data, that simple approach will not allow you to measure complex interventions.

In the next section, we’ll talk about what your data mix can look like, and the goal you should have in assembling it.

Up Next