Variables, Hypotheses and Controls
Variables in Psychological Research
Before running a study, a psychologist must decide exactly what they are testing. The independent variable (IV) is the variable the researcher deliberately changes or manipulates. The dependent variable (DV) is the variable the researcher measures — it is the outcome that is expected to change because of the IV.
For example, in a memory experiment a researcher might change how information is presented (IV: word list vs picture list) and measure how many items participants recall (DV: number of items correctly recalled). Only the IV should be deliberately changed between conditions — everything else should stay as similar as possible so that any change in the DV can be traced back to the IV.
The IV is the variable the researcher manipulates; the DV is the variable the researcher measures. If a study finds a difference in the DV, it should be caused by the IV — not by something else the researcher forgot to control.
Extraneous Variables
Sometimes something other than the IV affects the DV. Any variable, other than the IV, that could affect the DV is called an extraneous variable. If extraneous variables are not controlled, a researcher cannot be sure the IV caused the change in the DV, which weakens the validity of the whole study. Extraneous variables are usually divided into two types: situational and participant variables.
Situational variables come from features of the research situation itself, rather than from the people taking part. Examples include noise levels, lighting, the time of day the study runs, or the specific instructions given by the researcher. If one group is tested in a noisy room and another in a quiet room, any difference in the DV might be caused by the noise rather than the IV.
Participant variables come from differences between the individual people taking part, rather than the situation. Examples include age, memory ability, mood, motivation and prior experience. If one group happens to contain more experienced participants, that difference — not the IV — could explain a difference in the DV.
Situational variables come from the research environment (noise, timing, instructions); participant variables come from differences between individuals (age, mood, ability). Both types must be controlled so results can be trusted.
Controlling Extraneous Variables
Standardised procedures mean every participant experiences exactly the same conditions, instructions and environment. This controls situational variables — if every group hears the same instructions in the same room at the same time of day, any difference in the DV cannot be blamed on inconsistent procedure.
Counterbalancing controls order effects when participants complete more than one condition. Half the participants complete Condition A first then Condition B; the other half complete Condition B first then Condition A. This spreads practice or fatigue effects evenly across both conditions rather than favouring one.
Randomisation uses chance — such as picking names from a hat or a random number generator — to decide the order of conditions or how participants are allocated to groups. This reduces the researcher's own bias in decision-making and helps spread participant variables evenly across conditions.
In a single-blind technique, participants are not told which condition they are in or the true aim of the study. This stops participants changing their behaviour simply because they know what is being tested, which would itself become an extraneous variable.
In a double-blind technique, neither the participants nor the researcher running the session know which condition is being tested. This also removes the researcher's own expectations from influencing how they treat participants, which can happen even unintentionally.
Four ways to control extraneous variables: standardised procedures (same conditions for everyone), counterbalancing (balances order effects), randomisation (removes bias from order or allocation), and single/double-blind techniques (removes the influence of expectations).
Writing Hypotheses: Null and Alternative
Before collecting data, a researcher states two hypotheses. The null hypothesis predicts that there is no difference or no relationship between the variables being tested, and that any difference found in the results is simply due to chance.
The alternative hypothesis predicts that there will be a difference or a relationship between the variables — this is what the researcher usually expects to find, based on theory or previous research.
For example, in the memory experiment above: the null hypothesis might state “There will be no significant difference in recall between the word-list and picture-list conditions.” The alternative hypothesis might state “Participants shown a picture list will recall significantly more items than participants shown a word list.” After the study, statistical testing decides whether the null hypothesis can be rejected in favour of the alternative hypothesis.
The null hypothesis says any difference is due to chance; the alternative hypothesis says there IS a real difference or relationship. Both must refer to the specific IV and DV being tested in the study.