A time series is a sequence of data collected or recorded chronologically, usually at uniform intervals. We encounter this type of data in various fields, including economics/finance (e.g., daily closing stock prices, monthly price indices, quarterly unemployment rates), meteorology (e.g., hourly temperature/precipitation/wind speed), and engineering (e.g., sound waves, electric signals). By examining those sequential measurements over time, researchers can identify trends, patterns, and relationships that might not be apparent from a single snapshot or sample. In addition, time series analysis can be a powerful tool in communication research, for example, to illustrate how legacy/social media’s attention to specific issues fluctuates over time and to demonstrate how changes in media content influence public opinion or political/commercial campaigns over time. Particularly, in case the time series is characterized by a linear relationship between current and past values (so-called autocorrelation), ARIMA (Autoregressive Integrated Moving Average) models can be suitable for understanding and predicting the time-dependent media phenomenon and effects. In this talk, I will review fundamental concepts in time series analysis and discuss ARIMA applications in communication research based on two real examples selected from recent communication publications.

Bu Hyoung Lee is a statistician and an associate professor of statistics at Loyola University Maryland in the United States. He earned an M.A. degree in communication from Seoul National University in 2006 and a Ph.D. degree in statistics from Temple University in 2016. His research interests include spatial/temporal data analysis and machine learning with complicated/incomplete data.