Living on the Data Fringe: Big Data, Small Data, Thick Data, Oh My …

Posted in: #CC/Academic List, #Public/K-12 List, All Members, Data Science

---

As I have been watching the COVID-19 daily numbers of cases, hospitalizations, and deaths unfold, I have also been noticing the increase in the number of COVID ‘stories’ being shared across many news and media outlets. Even the hospitals are using a more qualitative approach to COVID-19 by having nurses and doctors tell their COVID-19 experience stories about the challenges of caring for patients, and their concerns about infecting their families, as they plead for people to wear masks. Although quantitative data (numbers and statistics) and qualitative data (words, stories and images) are very different, used together they each contribute to drawing a more holistic picture of our current and dire situation. One data approach is not better than another; in reality they support and enhance each other.

Big data are numerical or quantitative data (Example John Hopkins University COVID website). Big data analysis involves very large datasets, either structured or unstructured, that are analyzed by specialized software and requires advanced data skill to clean, manipulate, and synthesize the data. The NNLM Data Thesaurus is a great resource to learn more about big data. Big data can also be large textual datasets analyzed using computational processes like text mining, natural language processing, machine learning, and artificial intelligence methods (Ex. Medical record data). On the other hand, smaller and more manageable datasets of numerical data are called small data. It is data that you can manually analyzed in Excel, for example. An example of a small data website would be the CDC Places: Local Data for Better Health. Built on a larger national dataset, data are organized so you can easily visualize data on health outcomes, prevention, and unhealthy behaviors and then download small subsets of data for further analysis. Another example of small data is using library data such as gate counts, collection usage data, or instructional statistics to take action or make decisions about library work.

Textual or qualitative data that is analyzed in a more manual process is called thick data. Examples of qualitative data are  interviews or focus group transcripts, observation or field notes, and open-ended survey questions. Social media text can also be analyzed using qualitative methods such as thematic or sentiment analysis. For example related to Covid-19, a database of oral histories from the Voces of a Pandemic Collection at the University of Texas Austin, presents Latinx Covid-19 experiences and transcripts of these COVID stories could be analyzed for patterns and themes using qualitative thematic analysis.

Qualitative data analysis can provide a rich description of the quantitative data findings if the two types of data are used together. Quantitative data can be used to explore what is happening, and qualitative data  can be used to get at the why and how of what is happening. This mixed method research design (using both quantitative and qualitative data collection methods together) is becoming a more common method of data analysis for improving business organizations, exploring health science or medical topics, doing assessment and evaluation, designing products, and studying innovation practices. Tricia Wang, a technology ethnographer, makes a case for why big data needs thick data. Want to read more about how ethnography? The Association for Medical Education in Europe (AMEE) has created an informative guide on ethnography and how it is being used in medical education.

Image Sources: Numbers & Letters