Variables in research


Variables in research

When doing social research, variables are both important and tricky. Here’s a few words about them.


variable is something that can change, such as ‘gender’ and are typically the focus of a study.
Attributes are sub-values of a variable, such as ‘male’ and ‘female’. An exhaustive list contains all possible answers, for example gender could also include ‘male transgender’ and ‘female transgender’ (and both can be pre- or post-operative).
Mutually exclusive attributes are those that cannot occur at the same time. Thus in a survey a person may be requested to select one answer from a list of alternatives (as opposed to selecting as many that might apply).
Quantitative data is numeric. This is useful for mathematical and statistical analysis that leads to a predictive formula.
Qualitative data is based on human judgement. You can turn qualitative data into quantitative data, for example by counting the proportion of people who hold a particular qualitative viewpoint.
Units are the ways that variables are classified. These include: individuals, groups, social interactions and objects.


Descriptive variables are those that which will be reported on, without relating them to anything in particular.
Categorical variables result from a selection from categories, such as ‘agree’ and ‘disagree’. Nominal and ordinal variables are categorical.
Numeric variables give a number, such as age.
Discrete variables are numeric variables that come from a limited set of numbers. They may result from , answering questions such as ‘how many’, ‘how often’, etc.
Continuous variables are numeric variables that can take any value, such as weight.


An independent variable is one is manipulated by the researcher. It is like the knob on a dial that the researcher turns. In graphs, it is put on the X-axis.
dependent variable is one which changes as a result of the independent variable being changed, and is put on the Y-axis in graphs.
The holy grail for researchers is to be able to determine the relationship between the independent and dependent variables, such that if the independent variable is changed, then the researcher will be able to accurately predict how the dependent variable will change.
Extraneous variables are additional variables which could provide alternative explanations or cast doubt on conclusions.
Variables may have the following characteristics:
  • Period: When it starts and stops.
  • Pattern: Daily, weekly, ad-hoc, etc.
  • Detail: Overview through to ‘in depth’.
  • Latency: Time between measuring dependent and independent variable (some things take time to take effect).


Note that in an experiment there may be many additional variables beyond the manipulated independent variable and the measured dependent variables. It is critical in experiments that these variables do not vary and hence bias or otherwise distort the results. There is a struggle between control vs. authenticity in managing this.


With perfect correlation, the X-Y graph of points (as a scatter diagram) will give a straight line. Whilst this may happen in physics, it seldom happens in social research and a probabilistic relationship is the best that can be determined.
Correlation can be positive (increasing X increases Y), negative (increasing X decreases Y) or non-linear (increasing X makes Y increase or decrease, depending on the value of X).
Correlation can also be partial, that is across only a range of values X. As all possible values of X can seldom be tested, most correlations found are at best partial.


When correlation is determined, a further question is whether varying the independent variable caused the independent variable to change. This adds complexity and debate to the situation.
Sometimes a third variable is the cause, such as when a correlation between ice-cream sales and drowning is actually due to the fact that both are caused by warm weather.