VARIABLES IN RESEARCH

Research variables are the “language” that lets scholars communicate what they measure, how they measure it, and how different factors relate. The core takeaway is that a study only becomes analyzable once its variables are identified and defined in measurable terms—then classified correctly (independent, dependent, confounding, moderating, extraneous, and controlled) and matched to the right scale of measurement. That chain matters because the variable choices shape the research objectives, questions, hypotheses, literature review, conceptual framework, and ultimately the statistical tools used.

The lesson begins by grounding variables in real proposal writing. Using a sample topic—“influence of institutional capacity on performance of parastatals (Kenya Meat Commission modernization project)”—institutional capacity is treated as the independent variable and performance of parastatals as the dependent variable. The instructor emphasizes that titles and objectives must reflect these variables clearly; wording like “influencing/effects/effectiveness” signals a design that typically supports quantitative approaches (often surveys or mixed methods), while other phrasing can restrict or expand design options. Variables also drive what appears in Chapter Two: literature review sections and how the variables relate.

A variable is defined as a measurable characteristic that assumes different values across subjects or observations—variation is the defining feature. If everyone in a class shares the same characteristic (e.g., all are female, all are married), that characteristic becomes a constant rather than a variable. Variables must also be operationally defined: the meaning used in the study must be tied to how it will be measured. “Performance,” for example, is not treated as a dictionary concept; it is operationalized as what Kenya Meat Commission reports (e.g., money made in a specific financial year) and compared against what was proposed.

Next comes variable types. The foundational pair is independent variable (IV) and dependent variable (DV): the IV is the presumed driver of change in the DV, and the DV is expected to change with the presence/absence or magnitude of the IV. The instructor stresses that causality claims require appropriate design and control of other factors.

When other variables interfere with the IV–DV relationship, confounding variables enter. Intervening variables are hard to see, measure, or manipulate directly; they are inferred from outcomes (e.g., students’ underlying IQ inferred from performance). Extraneous variables must be seen and controlled by removing their effects—such as restricting a study to schools within the same social-economic setting to neutralize parental background differences.

A moderating variable is introduced as a second independent variable that changes the strength or direction of the IV–DV relationship. The instructor illustrates moderation as a “pacifying/modifying” factor that alters how the IV affects the DV (e.g., a house help improving household dynamics rather than a disruptive third party). Controlled variables are those held constant in experimental or quasi-experimental work to isolate the IV’s effect.

Finally, the lesson links variable classification to measurement scales. Variables can be categorical (nominal, ordinal) or continuous (interval, ratio). Nominal scales name categories without quantitative meaning (e.g., gender). Ordinal scales rank categories without equal spacing (e.g., grades). Interval scales have equal differences but no true zero (e.g., temperature, attitude scores). Ratio scales include a true zero (e.g., weight, income). These scale levels determine which statistical summaries and tests are appropriate—mean/median choices and tools like chi-square for categorical data, and Pearson vs Spearman correlations depending on whether the data are interval/ratio or ordinal.

The session closes by reinforcing that variables and scales must be decided before writing and analyzing: questionnaire construction, data analysis, and the logic of the conceptual framework all depend on getting variable identification and measurement right.