Online Statistical Resources for Direct Marketers
Fortunately, the great World Wide Web has it all, if you know where to look. Analytical help and a deepening of your statistical sophistication are only a few mouse clicks away.
General sites to educate yourself. For the beginner, there's a terrific site, Statistics Every Writer Should Know, at http://nilesonline.com/stats/. The work of Robert Niles of Lakewood, CO; it's well written, funny and accessible. It covers basic concepts like standard deviation, margin of error (confidence interval) and sample sizes (or, as Niles puts it, "So how come a survey of 1,600 people can tell me what 250 million are thinking?"). For someone trying to get oriented, or as a refresher course, this site is a complete delight. Niles also includes a handy link, Finding Data on the Internet, that acts as a portal to a host of useful data sites.
As a mid-level site -- not too easy, not too hard -- there's the Statistics Homepage site at http://www.statsoft.com/textbook/stathome.html. Organized like a textbook, it takes you from variables and statistical significance right through to the basis of some routine statistical analyses -- such as cluster, discriminant and factor analysis -- that you might see in business, science and economics.
For the professional or near-professional number cruncher, there's Rainer's Home Page for Statisticians, http://ourworld.compuserve.com/homepages/Rainer_Wuerlaender/stathome.htm#top. This won't be everyone's cup of tea, and it even provides a link for German-speaking Web surfers (nothing like a few Gesellschaften to add that enviable patina of scholarship) but there's undeniably as much depth here as you could possibly ask for. It's certainly where I like to spend my Saturday nights.
Online statistical tools. If you are looking for a few practical tools online where you can enter your numbers and get answers, here are a few sites you might want to bookmark. Interestingly enough, these sites are rarely, if ever, designed for business purposes. They're often academic sites meant for students and teachers of the physical or social sciences, but completely adaptable to our nefarious purposes as marketers.
My personal favorite is offered by the linguistics department of Georgetown University. Called the Web Chi-Square Calculator, http://www.georgetown.edu/cball/webtools/web_chi.html, this is a handy tool to validate the significance of observed differences in a mail test. Use the columns and rows of the calculator to set up the results you observe in your mail matrix, and let the Web do its magic.
What I particularly like about this site is that it will find out whether or not your observable differences are significant in a brief or verbose manner. That is, you can say, "Just give me the answer," or "Show me how you got that answer" to see the full set of calculations. It's a dream site for this application, and I use it all the time.
Two suites of tools are available at the next two sites. In a site designed by the Physics Department at the College of St. Benedict and St. John's University in Minnesota, called Statistics to Use, http://www.physics.csbsju.edu/stats/, you can find a host of calculators to be applied in a manner very similar to the way you run the Web Chi-Square Calculator. Mean, standard deviation, contingency tables, ordinary least squares and more are all there for the taking. Figure out the statistical test you need for your business, plug in your numbers and go.
More directly applicable to our needs as direct marketers is Moore's Direct Mail Calculators Page, http://www.moore.com/solutions/integratedsvcs/dirmailcalc.html. Here's where you'll find a nice little input device to determine if your test beat your control package, or another to see how large a variance in response you can expect when you move from test to rollout. You run these online calculators at the level of statistical confidence that makes you feel comfortable. These calculators Moore has developed return your answers quickly in plain language. Personally, I don't like that you can't see what's going on under the hood, but that's just me. As direct marketers, we should all be grateful for the existence of this page -- practical to use and refreshingly jargon free.
There's also a set of online statistical calculators at the UCLA site, at http://www.stat.ucla.edu/calculators/, but I find most of these applications rather specialized -- calculating power, running t-tests, generating random permutations and the like -- and so I don't use the site very often for business purposes. Messing around there, on the other hand, can be a blast.
Other stat sites of interest in cyberspace. If you're looking for an online course in Business Statistics (at the graduate level), try out Prof. Arsham's http://ubmail.ubalt.edu/~harsham/Business-stat/opre504.htm. This course has the virtue of being both fairly comprehensive yet served up in rather digestible chunks. It also offers a variety of helpful graphs, tables and flow charts. Nevertheless, many visitors will probably find all the mathematics here tough sledding.
Many statistical clearinghouse pages are also available on the Web, but here are two of my favorites:
At http://www-vdc.fas.harvard.edu/hdc/other_links.shtml, a joint effort of Harvard and MIT despite MIT's absence from the URL, you'll find scores of links to statistical software sites, government sites (e.g., Bureau of Labor Statistics) and sources of economic stats (Federal Reserve, etc.). This is a first-rate portal to the statistical wealth of the Web.
Similarly, visit Statistics on the Web at http://www.execpc.com/~helberg/statistics.html#software. Maintained by Clay Helberg, it's kept reasonably up to date and includes statistical professional organizations, educational resources, publications and more.
If you're looking for links to statistical journals and other periodicals on the Web, check out http://www.minitab.com/resources/Journal/index.html. You'll find links to nearly two-dozen publications, from The Annals of Statistics to, yes, The Journal of Multivariate Analysis. Also noteworthy are all the wonderful white papers you can find at the SPSS site (http://www.spss.com/cool/papers/index.html). These are a singular education in and of themselves covering a broad variety of topics.
Finally, there's Dr. B's Wide World of Web Data. I'm not sure who Dr. B is, but, heavens, what a site. Created at Arizona State University in 1994, http://seamonkey.ed.asu.edu/~behrens/siip/webdata/index.html provides links to stats of every possible variety on the Web. A lot of fine work went into this site. If you want to drill down into, say, Japanese baseball's pitching leaders -- or just about anything else you can imagine -- this is the place to do it.
Most people find statistics boring or frightening or both, drenched in jargon and obscure-sounding stuff -- F-ratios, regression, ANOVA, degrees of freedom, the null hypothesis, Type I and Type II errors and Poisson distributions (which have nothing to do with fish) -- the list goes on and on. But to really do our jobs well as direct marketers, we ultimately must ask the simple yet compelling question: How confident can we be that the difference we're seeing is genuinely a difference? It's comforting to know that there are a host of offerings on the Web to help you both get familiar with that way of thinking as well as answer the questions posed by your results.
And of course, the Web is also a great source of jokes. Q: What's the definition of a statistician? A: Someone who stands in a bucket of ice water, sticks his head in an oven, and says, "On average, I feel fine."