Data analysis and evaluating research (accuracy, precision, validity, errors, outliers): VCE Psychology Units 3 and 4

What this dot point is asking

Every VCE Psychology exam includes data: a table, a graph or a described study. The key science skills ask you to process data (percentages, percentage change, mean, median, mode, and an understanding of standard deviation), present it appropriately, evaluate its quality (accuracy, precision, repeatability, reproducibility, validity, errors, uncertainty, outliers), and draw conclusions that the evidence actually supports. The Unit 4 Area of Study 3 key knowledge applies the same criteria to your own investigation and poster.

The study design defines these terms precisely, and VCAA marks against those definitions, so learn them in its words.

The answer

Types of data

• Primary data are collected by the researcher for their own investigation; secondary data were collected by someone else (for example, a literature review uses secondary data).
• Quantitative data are numerical (reaction times, ratings on a scale, counts); qualitative data are descriptive, in words or images (interview responses, diary descriptions).
• A questionnaire with a rating scale produces quantitative data even though it asks about feelings. Such self-report data are subjective, while physiological measures such as EEG readings are objective.

Processing quantitative data

• Percentage: the part divided by the whole, multiplied by 100.
• Percentage change measures the size of a change relative to where it started:

A negative result is a decrease.

• Mean: the sum of scores divided by the number of scores. It uses every score but is pulled by extreme scores.
• Median: the middle score when the scores are ordered (the average of the two middle scores if there is an even number). It is not affected by one extreme score.
• Mode: the most frequent score.
• Standard deviation: a measure of variability, showing how spread out the scores are around the mean. VCE requires an understanding, not a calculation: a larger standard deviation means scores are more spread out; a smaller one means they cluster closely around the mean.

Presenting data

• The IV goes on the horizontal (x) axis and the DV on the vertical (y) axis.
• Use a bar chart when the IV is categorical (groups or conditions), with bars that do not touch. Use a line graph when the IV is continuous (time, amount).
• Give every graph a title, labelled axes with units, and an appropriate scale. The 2024 examiners noted that many students drew a line graph for categorical data, which was not appropriate.

Evaluating the quality of data

Internal validity: an investigation is internally valid if it investigates what it sets out or claims to investigate. It depends on the design, sampling and allocation, and on whether extraneous and confounding variables were controlled. If a study is not internally valid, its external validity is irrelevant.

External validity: the results can be applied to similar individuals in a different setting. It is increased by broad inclusion criteria and sampling techniques that produce a sample resembling the wider population.

Errors, uncertainty and outliers

• Random errors are unpredictable variations that produce a spread of readings, affecting precision. Reduce them by repeating measurements and calculating a new mean, increasing sample size, or refining the method.
• Systematic errors shift every reading from the true value by a consistent amount or proportion, affecting accuracy. Repeating the measurement does not fix them; better calibration and correct use of instruments does.
• Personal errors are mistakes, miscalculations and observer errors. They are not reported or analysed as error; the experiment is repeated correctly.
• Uncertainty is the lack of exact knowledge of the value being measured. It is present in all measurements and is often higher in psychology because many variables are psychological constructs. VCE requires only a qualitative treatment of uncertainty.
• Outliers are readings that lie a long way from the others. They must be analysed and accounted for, not automatically removed; repeating readings can help decide whether they reflect an error or a genuine result.
• Contradictory data (incorrect data) and incomplete data (missing answers or observations) should be identified, along with possible sources of bias.

From results to conclusions

• A conclusion is a statement based on the results of this particular study, and it says whether the evidence supports or refutes the hypothesis. The 2025 examiners distinguished a conclusion from an implication, which is a possible consequence or broader impact of the findings.
• Only a controlled experiment with good internal validity can support a cause-and-effect conclusion.
• Generalise only to the population the sample represents.
• Distinguish evidence (systematically collected data) from opinion (a personal view) and anecdote (an individual story).