12

This question illustrates the difficulty of a person mastering statistics and probability on their own, in the face of weakly developed resources like Wikipedia.

It occurred to me that consulting statisticians, and there are a few here, may routinely face the challenge of explaining certain concepts and methods to a client. This is the flip side of the pedagogical coin. When one has mastered the concept, it may make sense to conduct a particular avenue of analyses, but one's references may either be inappropriate or difficult to share with a client. So, are there common resources that consulting statisticians like to suggest to their clients? (See update #1 regarding more advanced or specialized topics.)

I can think of a few books that may be useful, but I suspect that a lot of clients will go about searching the web, as Developer did, and will come across rather inane material on Wikipedia. In my answer to Developer, I suggested the NIST Handbook as one such reference that could be used. What else?

Update 1: As Peter Flom has pointed out, for more advanced material or narrower pursuits, it may not be easy to offer a single point of reference. This is correct and I should have worded the question differently for those cases. In such cases, how do consultants find and share accessible references? I believe that many consultants will take the time to write something new in order to explain things to their client, but those aren't references that are found and shared.

Some ideas:

  • Tutorials written by the consultant or others
  • Case studies or analyses from projects that demonstrate the same concepts
  • Excerpts of books (as I'd suggested in my answer to Developer), which describe the concept

What else might be a source or how else do you actually go about finding such references? I realize this is an open ended question, but my answer to Developer shows some of the ways I'd approach this problem. I don't mean to ask of all the ways that one could address this, but in one's own experience, how have you typically provided such explanatory resources?

Community
  • 1
Iterator
  • 2,294
  • 1
  • 15
  • 22
  • Because this question is admittedly open-ended and solicits personal experience, it is a good candidate for CW. – whuber Oct 24 '11 at 18:52
  • 1
    There is an inherent difficulty with this question, though: its interpretation appears to require a subjective understanding of "accessible," which will vary among consultants and their clients. Can it really be answered, then? – whuber Oct 24 '11 at 18:54
  • 3
    It is answerable because accessible isn't the operative term. :) Instead, "how have you typically provided" is one operative phrase. This is open ended, and I agree that CW is perfectly appropriate. Thanks for making it so. – Iterator Oct 24 '11 at 19:51

4 Answers4

7

In my experience, no clients have ever requested additional materials to understand a concept or method; I provide that. About one client in 20 has requested additional materials for pursuing further study or for documentation: many of them have been lawyers or consultants (with scientific/engineering backgrounds). In all these cases they have felt a need for a resource, suitable for self-study, that explains fundamental statistical ideas in a non-mathematical way and provides examples of their application. I typically then have given them a copy of either of the following classic texts, depending on the depth of their interest and the time they might have available:

At one time I was giving away so many copies of How to Lie... (it was only $3.95 in the local university bookstore) that a competitor jokingly characterized this as "distributing [his] bible."

whuber
  • 281,159
  • 54
  • 637
  • 1,101
  • 1
    Great suggestions! By the way, if you order more than 10-20 copies of a book, publishers will often give you a *nice* discount. At around 100 copies of a book, you're doing them a favor and doing their marketing for them. When the book is revised, that's a lot of owners who may want to replace their edition. :) – Iterator Oct 24 '11 at 21:41
3

These example lean more towards the biostatistics realm but for physical references there are many, many, many, many, many, many, many references geared towards our non-statistical collaborators.

And for online references, sometimes a statistics software package will have a really good statistics guide. One example.

  • 1
    I checked all your links and downloaded the last one but to me all look elementary statistics as I categorized them as a. – Developer Oct 22 '11 at 14:19
3

I'm a consulting statistician, mostly to graduate students and researchers in the social, behavioral and medical sciences.

When I need to explain something to a client, they rarely want references for their own edification (although they often want them for papers). Usually I will just explain what I am doing in my own words, using their example to help them understand things. This lets me respond at different levels to different clients.

Once you get past elementary statistics, there's no ONE source that's right - statistics is too broad a field.

But building on what @Mike Wierzbicki said, I also like the documentation in a lot of SAS' statistical PROCs.

Peter Flom
  • 94,055
  • 35
  • 143
  • 276
  • 1
    My observations agree. Beyond introductory material, it is hard to suggest a particular well-known reference. I should update the question to address this. Perhaps I should have asked how one might find & share accessible references for narrower or more advanced topics. – Iterator Oct 23 '11 at 13:33
  • 1
    (+1) There are good points in this answer: `...explain what I am doing in my own words, using their example...`. To my experience in teaching programming languages it is one of good solutions. The third paragraph: `Once you get past elementary statistics, ...statistics is too broad a field.` is also productive and motivational. The language of the comment is also positive. – Developer Oct 25 '11 at 12:49
2

I highly recommend the book "Statistics as Principled Argument" by Robert Abelson. It's a small book, with almost no formulas, suitable for anyone who's had any introduction to statistics at all, and for many people who have had a lot of courses. I reviewed it on my blog here

Peter Flom
  • 94,055
  • 35
  • 143
  • 276