My slides from my Podcamp 6 Presentation

Published on: **Mar 4, 2016**

Published in:
Technology Business

- 1. Or You Can Lie With Statistics but it’s a Lot Easier with Words
- 2. Everything is Numbers Statistics are used to estimate & describe patterns in nature that aren’t easy to see with the naked eye Sports-Earned Run Average, Slugging Percentage, QB Rating, Goals Against Average Economics-Gross Domestic Product, Unemployment, Inflation Medicine-Heart Rate, % Body Fat, T-Cell Counts Education-IQ Scores, SAT scores, Dropout Rates As long the statistic is from a source of data that is verifiable, it’s hard to lie using it.
- 3. Types of Statistics Measures of Central Tendency (aka Averages) Continuous-Number can take any value. Mean (sum of all data divided by the number of data points) Median (midpoint of all data when it is ranked from highest to lowest) Mode (most frequently occurring data value) Discrete-Value can only take certain values eg. 0 or 1, true or false. Proportion-sum of values taking a certain value for a given variables divided by the maximum value for that variable.
- 4. Types of Statistics (cont.) Measures of Spread Range-highest data value-lowest data value Variance-Average squared deviation from the mean Standard Deviation-square root of the variance Probability Used to measure the chance of events Also used to make a statement about the relationship between a sample and a population that it’s taken from eg. margin of error.
- 5. But a Summary Statistic can NeverTell You the Whole StoryGraph with States Graph without States
- 6. Graph TypesBar Graph-Good visually but not Line Graph-Better for showinggood for trends trends over time 6time 4 5 4time 3 Product A 3 Product A Product B 2 Product Btime 2 Product C Product C 1 time 1 0 time time time time 0 10 20 1 2 3 4
- 7. Graph Types (Cont)This is the first pie chart created byFlorence Nightingale to show thenumber of British soldiers in theCrimean War who died due toinfection rather than combatinjuries.
- 8. Graph Types (cont.)Mapping using GeographicalInformation Systems (GIS) is agood way to represent data byregion. In this graph I showedwhich areas of the city have thehighest number of crimes bycensus tract in the city for 2005.
- 9. Posting Graphs on the Web Line, Bar, Pie, & other Graphs can be created using Microsoft Excel, SPSS, SAS, ArcGIS, R, & other Packages If that data package will allow you to save that graph as a .jpg, .gif, or .png file you can easily add it to your blog. Microsoft Excel requires a visual basic command to save graphs as image files.
- 10. Statistical Packages Microsoft Excel-Most readily available but not really built for statistical analysis. OK to make basic graphs. SPSS-Better for more advanced analysis and graphics but less accessible due to cost. User friendly. R-Free software package that can be downloaded from the web. Can do many types of analyses. BUT it is syntax driven. Can save graphics as image files using syntax.
- 11. Cutting Edge Graphics The Gapminder institute provides great interactive graphics for free that can be seen in the documentary the Joy of Stats. URL: www.gapminder.org Joy of Stats Clip: http://csiwodeadbodies.blogspot.com/2010/12/income-and- life-expectancy-what-does-it.html The website Fractracker uses advanced graphics and mapping techniques to monitor the impacts of Marcellus Shale drilling in Pennsylvania and New York. URL: http://www.fractracker.org/
- 12. Poor Statistical Reasoning Example The blog The Audacious Epigone posted an analysis of the IQ’s of a sample of McCain & Obama voters which can be seen at http://anepigone.blogspot.com/2011/05/iq-wars- mccains-voters-win.html
- 13. Some Good Statistical Blogs FiveThirtyEight-Nate Silver’s blog which forecasts elections, the Oscars, and other sporting events. http://fivethirtyeight.blogs.nytimes.com/ Data Visualisation-Has more examples of cutting edge graphics. http://www.datavis.ca/ The Incidental Economist-Good Analysis of health care data. http://theincidentaleconomist.com/wordpress/ CSI without Dead Bodies-My own website http://csiwodeadbodies.blogspot.com
- 14. Sources of Data on the Web Many websites, such as The Census Bureau’s provide data for download with which to do your own analysis. Example-Small Area Health Insurance Estimates (SAHIE) makes state and county level estimates for the whole US from 2005-2007 (2008 and 2009 estimates are forthcoming) http://www.census.gov/did/www/sahie/index.html Other sites provide data that can be copied and pasted into a data file. Example-CNN makes it’s poll reports available as PDF’s but not the raw data
- 15. Summary When analyzing data leave no stones unturned or if that is impossible turn over as many as possible and acknowledge that you couldn’t turn all of them over. When interpreting an analysis ask yourself if they have turned over the important stones and or accounted for the ones that they couldn’t turnover.