stats 2

These flashcards cover essential vocabulary and concepts related to p-values, hypothesis testing, errors in statistical analysis, and the use of parametric and non-parametric tests.

p-value

The probability of obtaining results at least as extreme as the observed results, assuming the null hypothesis is true.

Significance of p-value < 0.05

Indicates a significant difference, suggesting you are roughly 95% sure there is a genuine effect, allowing you to reject the null hypothesis.

Interpretation of 'Sig.' values in SPSS normality tests

If p > 0.05: Data is normally distributed; If p < 0.05: Data is NOT normally distributed.

Type 1 Error

A False Positive; it occurs when we believe there is a genuine effect in the population when there actually is not.

Type 2 Error

A False Negative; it occurs when we believe there is no effect in the population when a genuine effect actually does exist.

Two-tailed test

Use for a non-directional hypothesis (e.g., 'x may or may not predict y').

One-tailed test

Use for a directional hypothesis (e.g., 'x will predict y').

Tests for normally distributed data (2 groups)

Dependent t-test for dependent groups; Independent t-test for independent groups (requires checking Levene’s test for equal variance).

Non-parametric tests for non-normal data

Mann-Whitney test: For 2 groups; Kruskal-Wallis test: For 3 or more groups.

Caution with P-values

Do not rely solely on the p-value; consider if the experiment was well-designed, if the population was sampled appropriately, and if any bias exists.

Decision Tree Logic

Your choice of test always starts by checking distribution via the Kolmogorov-Smirnov (K-S) or Shapiro-Wilk (S-W) tests.

Parametric tests

Normal distribution tests (e.g., t-tests, ANOVA).

Non-parametric tests

Non-normal distribution tests (e.g., Mann-Whitney, Kruskal-Wallis).

Courtroom Analogy for Type 1 and Type 2 Errors

Type 1 error is like convicting an innocent person; Type 2 error is like letting a guilty person go free.