Summary

This document explores the key principles of creating effective graphs, covering topics from selecting appropriate graph types like histograms and bar graphs, to avoiding common design flaws. It emphasizes the importance of clear axis scaling, informative captions, and avoiding misleading visualisations. The document also examines why pie charts are often a poor visualization choice.

Full Transcript

2/12/25, 11:55 AM OneNote What makes a good graph? Thursday, February 06, 2025 3:12 PM What is your point? Goal: present data in clear and simple way Should your dtat or result b...

2/12/25, 11:55 AM OneNote What makes a good graph? Thursday, February 06, 2025 3:12 PM What is your point? Goal: present data in clear and simple way Should your dtat or result be visualized How many grpahs is too many? Make effective use of figure captions and refer to figure in text Make effective use of colours and symbols to clairty, don’t confuse Choose the right type of graph: What type of variables do you have? ○ Decrete or continous? Ordered categories? Do you want to highlight group differnces in lcoation, disribution, etc.- means plot vs. Box plot Are you interested in a relatioship between continous data variables?- scatter plot Are you showing change across groups?- line graph Does the statistical test you are using have an associated visual Histograms (class a test scores) How big are your bins (values in each bar) Bar Graphs: Proper axis scaling: always include the 0 Avoid cutting off data that exists Shifting the axis scale falsely exaggerates group differences Bar graphs: Too large an axis falsely diminishes group differences https://uvic-my.sharepoint.com/personal/lucybryden_uvic_ca/_layouts/15/Doc.aspx?sourcedoc={9faa0b91-e898-498c-a6e1-c5e15b84c2ef}&action=edit&wd=target… 1/4 2/12/25, 11:55 AM OneNote Line graph: Easy to dimish patterns by including 0 on the y-axis Accommodate data without excessive blank space There are no data between 0-55 Box plots: Similar issue with the axis scale- no need to squash boxes- Purpse is to visulaize and compare distributions Bivariate plot: https://uvic-my.sharepoint.com/personal/lucybryden_uvic_ca/_layouts/15/Doc.aspx?sourcedoc={9faa0b91-e898-498c-a6e1-c5e15b84c2ef}&action=edit&wd=target… 2/4 2/12/25, 11:55 AM OneNote Good captioning: Figure caption should give all infornmation necessary to understand what is being presented May include: Descriptive title+figure # in sequence in text Symbol/colour definitions Sample size info Explanation of graph parts Statistical results Why pie charts are bad (usually): We are bad a differntiaing angels https://uvic-my.sharepoint.com/personal/lucybryden_uvic_ca/_layouts/15/Doc.aspx?sourcedoc={9faa0b91-e898-498c-a6e1-c5e15b84c2ef}&action=edit&wd=target… 3/4 2/12/25, 11:55 AM OneNote https://uvic-my.sharepoint.com/personal/lucybryden_uvic_ca/_layouts/15/Doc.aspx?sourcedoc={9faa0b91-e898-498c-a6e1-c5e15b84c2ef}&action=edit&wd=target… 4/4

Use Quizgecko on...
Browser
Browser