Psychological Statistics and Scientific Rigour

Vocabulary-style flashcards covering t-tests, statistical assumptions, non-parametric alternatives, correlation, and scientific rigour based on lecture notes.

t-test

A test used to compare the means of two groups to see if the differences between them are because of chance.

Repeated Measures t-test

A test used when the same people are in both groups, meaning each person acts as their own control group.

t-test Formula

The difference between the means divided by the variability or error.

Practice effect

A limitation of repeated testing where people might do better the second time simply because they have practiced the task.

Fatigue effect

A problem in testing where participants might do worse on a subsequent task because they have become tired.

Counterbalancing

A technique to reduce bias where half the participants do the tasks in one order and the other half do them in the opposite order.

Significant difference threshold

The point at which a result is considered significant, typically when $p < 0.05$.

One-Sample t-test

A test that compares the mean of a group to a known meaning of a population.

Non-significant result

A result that does not provide enough evidence of a difference, though it does not necessarily mean the groups are identical.

t-test Assumptions

The requirements that data be normal, interval or ratio level, free of outliers, and have equal variances.

Non-parametric tests

Tests that use ranked data instead of raw scores, used when standard t-test assumptions are not met.

Mann-Whitney U Test

A non-parametric test for two groups that converts scores into ranks and compares the rank totals.

Wilcoxon Signed-Rank Test

A non-parametric test used for two related groups that calculates and ranks difference scores.

Correlation

Standarised covariance that tells us both the direction and strength of the connection between two continuous variables.

Covariance

A measure that tells us the direction of a relationship and is dependent on the scale of measurement.

Scientific rigour

The practice of doing research ethically and responsibly, following best practices to avoid fraud or data manipulation.

Reproducibility and replication

Efforts and history of attempts to repeat research to verify if results can be found again.

Statistical Power

The ability to detect an effect that exists, which is often tied to the size of the sample.

Effect Size vs. Sample Size

If an effect is small, more participants will be needed; if the effect is large, fewer participants are needed.

Standard Power Level

A target usually set at $80\%$, meaning there is an $80\%$ chance of finding an effect that exists.

Psychological effect size

Most effects in psychology are considered small, requiring large numbers of participants, such as $200$, to detect.

Interval or Ratio level

The specific levels of measurement required of data to meet the assumptions of a standard t-test.

Ranked data

The type of data used by non-parametric tests like the Mann-Whitney U or Wilcoxon tests.

Data peeking

An analysis decision that can affect final results and inflate the likelihood of achieving a significant result.

Informed Consumer

A person who evaluates science by looking at sample size, replication efforts, and data availability.