Space Science Workshop '98 Newsletter

Top Stories

- HOA Activities

- Binocular Visions

- Statistical Savvy

- Amazing AXAF

Wright Center Workshops

- Art and Science

- Oceanology

- Frontiers in Science

Other News

- Wright Center

- AAVSO

- HOA

How to Win Friends and Influence People with Statistics

Statistics a term that is often feared by many but unsuspectingly used in every day life. For example four out of five dentists recommend....88.3% of the population is in favor of funding a new school....are all examples of where statistics enter the media on a daily basis. However the interpretation and explanation of statistics is important in understanding what is actually happening in the experiment being tested or the phenomena being observed.

Grant Foster demonstrates why one should be careful to read between the mathematical lines.

You do the math - Grant Foster demonstrates why one should be careful to read between the mathematical lines.

Grant Foster, mathematician for the American Association of Variable Star Observers, gave a presentation titled “How to Win Friends and Influence People with Statistics” which made statistics interesting and easy to understand. One of the major points he relayed to the workshop participants was that there are many areas that appear in the statistical analysis of data that either by deliberate choice or honest error provide erroneous information to the general public. This, according to Grant, even occurs within the scientific world.

What is a “statistically significant” result and at what level of confidence can one state that the null hypothesis has been rejected were questions answered through this presentation. A concrete example that was given when testing a null hypothesis was related to ice cream. If everyone in the world believes that this is vanilla ice cream, a statistical test determine if what everyone believes to be true is actually or if it only happened by accident. After obtaining a significant number of samples, one can safely accept or reject the null hypothesis. If the testing reveals that there is a 5% or less chance that this is vanilla ice cream the researcher can safely reject the null hypothesis. However, it is not safe to say that the sample is chocolate ice cream. It could be something else. The only thing that we are sure of is that it isn’t vanilla.

In conclusion, Grant explained other areas in which statistics may be misleading which was done purposely to help further a point or by accident in a true mistake. One thing was sure at the end of the presentation, the workshop participants walked a way with a better understanding that when something appears to be statistically significant, it may be or it may not be. More investigation needs to be done to ensure that we are getting vanilla when we think we are getting vanilla.