Data, plural for the Latin word datum.
I find it useful to distinguish between data and phenomena.
Phenomena: recurrent features of the world.
Data: “public records produced by measurement and experiment that serve as evidence for the existence or features of phenomena” (Woodward 2011, p. 166)
Even in well-designed experiments, the data will reflect the influence of many other causal factors that have nothing to do with the phenomena of interest.
But there’s more.
When we think about “data” we generally think about standardized data—i.e., symbols that are arranged in a way that’s convenient for computer processing. We think about symbols that are stable, manipulable, and transportable—or what Bruno Latour (1987) calls “immutable and combinable mobiles.”
Note. Accepting a distinction between data and phenomena means that we can have different explanations for the phenomena (scientific models) and for the observed data (statistical models). The most obvious examples of this distinction are statistical models whose purpose is to account for some form of measurement error or to impute missing data. All variables are measured with error. The distinction becomes less clear when we start thinking about “bespoke statistical models” (McElreath 2020, chap. 16).
References
Latour, Bruno. 1987. Science in Action: How to Follow Scientists and Engineers Through Society. Harvard University Press.
McElreath, Richard. 2020. Statistical Rethinking: A Bayesian Course with Examples in r and Stan. CRC press.
Woodward, James F. 2011. “Data and Phenomena: A Restatement and Defense.” Synthese 182: 165179.