On-line statistics

Suggested Representation for Graphical Output

To represent your data for the comparison tests bar charts can be used. This clearly indicates the relative sizes of means or, in the event of non-parametric data, the medians from each sample and should include bars to indicate the error in the sampling.

The vertical bars are 95% confidence intervals (mean ± SE x 1.96). Where the distributions are not normal, standard error is not applicable. In addition to the graphs you should include the critical and calculated values, and the level of probability in your text using the following essentials:

The H_O and H_A

The tests used, including the f-test and K-S where applicable

The calculated value and n

The critical value in the following format: "crit_0.05,(2),14 = 000000". Where crit is the critical value for the test (F-crit for ANOVA, t for t-test etc.), _0.05 is the level of confidence, ₍₂₎ is two-tailed, ₁₄ the degrees of freedom and 000000 the actual value

Always plot a scatter graph before you start analysis.

Note the calculated r-value and the critical values at 5% and 1% confidence levels taken from the product moment tables where degrees of freedom = n - 2.

Again always plot the scatter graph prior to analysis to make an initial assessment.

Note the regression equation and the R² value.

---

Descriptive Stats

Diversity Indices

Comparisons

Correlations

Regression