Statistical significance assesses whether observed results are sufficiently unlikely under a null hypothesis to support rejecting it. A p-value measures how likely the observed results, or more extreme results, would be if the null hypothesis were true. section 4.2.1
Interpret statistical significance with a predetermined threshold and alongside model assumptions:
- Compare the p-value with a chosen significance level; p < 0.05 conventionally counts as statistically significant and leads to rejecting the null hypothesis. section 4.2.1
- Treat 5% as a convention rather than a universal principle; another predetermined level can provide the decision rule. section 4.2.1
- Do not interpret significance as the probability that the null hypothesis is true or false, or as a measure of effect size. section 4.2.1
- Use regression significance to identify influential predictors, while checking collinearity and other model assumptions. section 4.0
Reading a Shapiro–Wilk test
A Python analysis reports p-values of 0.2933 for stage1 and 0.1479 for stage2. These outputs illustrate how a statistical test supplies p-values for comparison with a chosen significance threshold. section 4.2.1
>>> from scipy import stats >>> print(stats.shapiro(mds.stage1)) ShapiroResult(statistic=0.9446673217389622, pvalue=0.29328758849732417) >>> print(stats.shapiro(mds.stage2)) ShapiroResult(statistic=0.9290259439587454, pvalue=0.1478907551379664)
Inspect the paired Shapiro–Wilk test outputs and their p-values. section 4.2.1
