Statistics for #Research - L1 - Introduction to Statistical Variables

The session lays the groundwork for research statistics by defining what “statistical data” and “statistical variables” actually mean, then sorting variables into practical types researchers use when building analyses. The core takeaway is that statistical data is numerical information organized by cases (records) and that statistical variables are the measurable properties attached to those cases—properties that can vary across individuals, times, or contexts. Without that clarity, later choices about analysis methods (and how to code questionnaire responses) become guesswork.

Statistical data is described as a collection of numbers located within cases or records tied to entities such as people. A case might be a respondent’s answer, a country-year observation like GDP per year, or a record for a specific unit such as a hospital or a region. The same dataset can include cases at multiple levels—for example, one dataset for individual patients, another for medical emergency teams, and another for hospitals—allowing comparisons across groups or even across cities.

A statistical variable is then defined as a special kind of mathematical variable that represents a conceptual space within a broader set of concepts. That conceptual space can be abstract (like personality traits such as introversion/extroversion) or physical (like height or weight). Two properties distinguish a variable: (1) it holds a measurement value for each individual case, and (2) across cases, it can take one or more values. If a measurement never changes—staying fixed for a respondent, event, or unit—it behaves like a constant rather than a variable.

The session also connects variables to how researchers organize observations about real-world concepts. Variables can capture straightforward demographics such as gender, but they can also represent complex constructs like attitudes, job commitment, turnover intention, or perceptions of an organization’s social responsibility.

Finally, the session categorizes variables into types that drive analysis choices. Categorical variables assign each possible value to a distinct category—examples include gender, job rank, education level, driver’s license status (provisional/open/no license), and city of residence (e.g., Peshawar, London, Lahore, Manchester, New York). Continuous variables, by contrast, fall along a spectrum—age, height, and weight are given as examples.

A key nuance is that some responses are technically categorical but behave like continuous measures in practice. Likert-style “liquid scale” responses (the transcript uses this phrasing) such as “How much do you enjoy statistics?” range from “greatly enjoy” to “greatly dislike.” Even though these are categories, researchers often treat them as metric/numerical responses by mathematically manipulating them as continuous variables, which enables common statistical techniques for explaining variance in dependent variables and assessing how predictors influence outcomes. The session positions this variable-type decision as essential preparation for the later R and SPSS implementation that will follow in the series.