Published on: **Mar 4, 2016**

- 1. Presenting Data in Charts,Graphs and TablesMDLB 5100 – Principles of Research DesignCatherine Loc-Carrillo, Ph.D.
- 2. Learning Objectives identify the types of charts and graphs and when the use of each is appropriate
- 3. Purpose of displaying data Figures and graphs help the reader visualize the story Can make information easier to understand and remember
- 4. General rules for displayingdata K.I.S.S. Use clear descriptive titles and labels Provide a narrative description of the highlights
- 5. A good graph Includes: axes, labels, scales, symbols and legend Should draw attention to data not itself Data points should be easy to read Labels are easy to read and distinguish The scales used should match the range of the data The legend is clear and concise The data deserved to be graphed Visually balanced axes (i.e. ratio of 1.0 to 1.3)
- 6. Line graphs &scattergrams Use horizontal and vertical axis Illustrates the relationship between 2 or more variables Independent variable (plotted on x axis) can be manipulated or changed by the investigator Dependent variable (plotted on y axis) depends on the value of the independent variable Use for continuous variables
- 7. Scattergrams When displaying data that is distributed over an area, bar graphs can have limitations The ‘mean’ and SD, or median and 95% CI can be easier to display with a scattergram Source: Annesley (2010) ClinChem 56(8): 1394-1400
- 8. Example 1 Line graph Symbols representing doses are clear and easy to distinguish Connecting lines are clear The scale is proportional to the range of values Minimal wasted space throughout graph Legend is concise and easy to understand Source: Annesley (2010) ClinChem 56(8): 1229-1233
- 9. Exercise 1: How can thegraph be improved? Symbols Lines Text size Use of decimal point Wasted space The ratio of axis Source: Annesley (2010) ClinChem 56(8): 1229-1233
- 10. Bars graphs and pies charts Used when variables are discontinuous Bar graphs are useful for visual comparisons of data
- 11. Bar graphs Used to plot a set of discontinuous independent variables versus continuous dependent variables Consider: The space between the bars should be narrower than the width of the bars Use shading or patterns that help distinguish bars from one another Avoid use of a suppressed-zero scale Y-axis represents frequency X-axis may represent time or different classes
- 12. Exercise 2: Which graph isbest and why? Source: Annesley (2010) ClinChem 56(8): 1394-1400
- 13. Pie charts A circular drawing divided into segments Each segment reflects proportion of total area Pie charts are most understandable if the number of categories is limited to <6 These charts are most accurate when all available data or outcomes are included Source: Annesley (2010) ClinChem 56(8): 1394-1400
- 14. Tables A rectangular arrangement of data in which the data are positioned in rows and columns. Each row and column should be labelled. Rows and columns with totals should be shown in the last row or in the right-hand column. Region Adults and adolescents Children <15 Total ≥ 15 years years 1 14 800 200 15 000 2 400 000 20 000 420 000 3 997 000 3 000 1 000 000 4 985 000 15 000 1 000 000 5 1 460 000 40 000 1 500 000 6 465 000 35 000 500 000 7 940 000 10 000 950 000 8 380 000 220 000 600 000 9 900 000 600 000 1 500 000 10 545 000 5 000 550 000 Total 7 086 800 948 200 8 035 000
- 15. Making friends with yourdata Statistical results should be clear and suitable for the typical reader The statistics conducted should be appropriate There isn’t a ‘right’ one The data should be analyzed in several ways Do the results convey meaningful information Don’t be afraid to learn new statistical procedures Don’t try to impress the readers with statistical complexities
- 16. Presenting the results A 2nd year undergraduate should be able to read the results section and know what the main findings are Don’t report the same stats more than once ’Statistical significance’ Is a result where chance is an unlikely explanation i.e. p> 0.05 When p> 0.001, report the actual number, but when p< 0.001 then just report as is Don’t report p=0
- 17. Take home message Researchers spend months/years collecting their data then throw it into a computer and try to analyze in minutes The data deserves better! Quick and reckless approach to data analysis often fails to identify important aspects of the data