Blog
Understanding Cross-Section, Time Series, and Panel Data Structures in Research
For those of you currently conducting research, I believe it’s important to have a solid understanding of data structure before starting. This is crucial because the structure of your data will determine the appropriate analytical tools to use when analyzing your research results.
In general, when conducting research, we can classify data into three types: cross-section data, time series data, and panel data. Perhaps you are using one of these types in your study. Have you truly understood the differences among them?
Here, I will try to share my perspective on the differences between cross-section, time series, and panel data structures. Through understanding these data structures, I hope you’ll be able to select the right statistical analysis methods more effectively.
Cross-Section Data Structure
Let’s begin by discussing cross-section data. This type of data is often referred to as “snapshot data” and is usually collected through field surveys.
By definition, cross-section data is collected at one point in time across multiple subjects or variables. The key thing to remember is that this data is gathered during a single period.
This period can be a year, a semester, a month, a week, or even a single day.
To make this easier to understand, let’s look at an example. Suppose a researcher collects data on product sales from 50 companies in region ABC in the year 2024. This type of data is considered cross-section data.
Let’s look at another example. Suppose a researcher conducts a study on the profit of rice farming businesses using 200 sample respondents from regions X, Y, and Z in 2023. The data collected here is also categorized as cross-section data.
Hopefully, these two examples help you better understand what cross-section data is.
In essence, if we conduct a survey and collect data within a single time period, then that data is referred to as cross-section data. The structure typically consists of observational units such as respondents, companies, consumers, farmers, livestock breeders, and so on.
Now, let’s move on to the next data structure.
Time Series Data Structure
Unlike the previous data type, what we’ll discuss now is called time series data. In our language, it can be translated as sequential or time-ordered data, meaning the data is collected in a continuous time sequence.
From a definitional standpoint, time series data is gathered from a single subject over multiple time periods. These time periods may be annual, monthly, weekly, or even daily.
To make it clearer, let’s go through an example. Suppose a researcher observes household food consumption in District ABC from the year 2000 to 2024. This annual data on household food consumption is classified as time series data.
Here’s another example. Suppose a researcher observes daily egg prices in region XYZ from January 1, 2024, to July 17, 2025. This kind of daily price data is also referred to as time series data.
Hopefully, this helps you better understand how time series data is structured. Now let’s move on to the final type, which is panel data.
Panel Data Structure
You may have heard of or even read journal articles that use panel data. But what exactly is panel data?
By definition, panel data is a combination of cross-section and time series data. Based on this, panel data refers to information collected from multiple subjects, each observed over multiple time periods.
To help explain this further, let’s use an example. Suppose a researcher wants to study the operational costs of 30 store branches from 2020 to 2024. In this case, the operational costs of each of the 30 branches are observed annually—2020, 2021, 2022, up to 2024. This is what we call panel data.
Let’s look at another example. A different researcher may observe the Gross Regional Domestic Product (GRDP) of 27 provinces from 2015 to 2020. This would also be considered panel data.
Conclusion
Based on the article I’ve written, I hope this helps deepen your understanding of the differences between cross-section, time series, and panel data structures.
I also hope this article provides value and new insights for those who need it. Thank you and stay tuned for the next update from Kanda Data.