Featuring Star Bars

An elegant way to visualize significant differences in your data.

Annotate your statistics with Graphmatik

With version 0.1.0 of Graphmatik, whenever you run a statistical analysis on row or column charts, star bars will automatically be plotted when you tab over to the chart workspace.

Star bars are a great way to visually represent significant differences in your data. Each individual bar depicts a comparison between 2 groups and the number of stars indicates the corresponding p-value.

1. * represents a p-value <= 0.05
2. ** represents a p-value <= 0.01
3. *** represents a p-value <= 0.001
4. **** represents a p-value <= 0.0001

Star bars settings

After running a statistical analysis, you can customize your star bars by accessing the settings menu.

Inside the settings menu you can:

• toggle star bars on/off.
• switch between the fast or optimize stacking algorithm.
• set the p-value cutoff to 0.1, 0.05, 0.01, 0.001

How can I show insignifcant results?

Typically any comparison between groups that doesn't show a * is considered not significant. That said, there are times you may want to highlight insignificance.

• If you have a large effect size, but low statistical power. You may want to describe a "trend".
• If you have an effect that looks significant, but is in fact not.

If you select a p-value cutoff of 0.1 you will have the option to show insignifance as:

• "ns" --> not significant
• "0.061" --> as a number rounded to 3 decimal places

Be careful with "trends". A P value does not define a study's worth nor prestige. Using words like "trends towards significance" or "nearly significant" can introduce bias & subjectivity in scientific reporting. Rather discuss observed differences in terms of their effect size, variability, & limitations (eg. statistical power).