RETURN
TO MENU

Online Resources

2003 Annual Meeting -  Hyatt Regency Tamaya, New Mexico

Practical Statistics for Brewing Chemists workshop summary

In opening the session, Dr. Pringle, while asserting that he himself was not a statistician, emphasized that this workshop would not require knowledge of statistics to be pertinent to the audience. He explained that statistics "can be regarded as the last resort to prove a point" or a solid means to get at the truth. The presentation described statistics as a means to provide objective data analysis while minimizing the human element, which, by nature, is "flawed." One example of this "flawed" human nature is the bias often seen in our problem solving from a tendency to look for patterns even when no patterns exist. An analogy was made to the assignment of constellations in the night sky.

The presentation held true to its promise of using only plain English and avoiding "stat-babble." In a further effort to maintain an easy-to-follow format, all statistical techniques were demonstrated in the widely utilized Microsoft Excel. The workshop was divided into four sections, each focused on a different practical analysis tool that brewing chemists would frequently use. These were 1) general data analysis, 2) data comparative methods, 3) data correlation, and 4) quality control charts.

Dr. Cutaia discussed the first two sections, while Dr. Pringle spoke on the last two.

The section on general data analysis discussed how to examine a typical set of data such as a series of pH readings on one brand of beer. Not only were the basic concepts of average and variance covered, coaching was also presented on how to ensure that a series of measurements is valid. Emphasis was placed on verifying that the sample values were indeed randomly obtained and that they are truly representative of the material being examined.

The data comparative methods section demonstrated the t-test and the ANOVA techniques. These useful procedures were shown to be a practical means of determining whether a process change has actually produced significantly different results. The value of always performing a visual assessment of the data by plotting in a graph was stressed through several examples using real production data.

The third section of the presentation was on correlation. Correlation was defined as a measure of relation between two or more variables such as fermenter temperature and ester production. Two parameters, "r" and "R2" were discussed in detail and clearly explained visually with several charts. The term r indicates the extent to which values of two variables are proportional to each other. The similar value, R2, indicates how well a line can be fit to the data. Again, the value of graphing the data was greatly emphasized.

Perhaps the most useful and possibly the most entertaining segment occurred when Dr. Pringle presented several recently published but unnamed examples of correlation data, many of which were faulty. In these examples, Dr. Pringle engaged the audience to evaluate the soundness of the each correlation based on the workshop material that was just presented. The participants, in general, had no difficulty identifying the flaws in each instance.

The final section described the value of quality control charts. Again, visual graphics allowed for each point to be clearly made. The proper construction and evaluation of control charts were detailed. Examples of several control charts were used to demonstrate the seven signs that indicate when a process is out of control. The presentation also pointed out the significance of the variance and distribution in histograms.

Copies of the presentation are available to ASBC members by e-mail request to Dr. Pringle at alastair.pringle@anheuser-busch.com.

404 Object Not Found