CoolData blog

15 February 2010

I’m a driver, not a mechanic

Filed under: Training / Professional Development, Uncategorized — Tags: — kevinmacdonell @ 1:33 pm

Kevin MacDonell at work - not exactly as shown. (Photo used by Creative Commons license. Click on photo for source.)

It’s clear from the number of daily page views this blog gets, and the amount of feedback I receive personally, that there is a strong interest in data mining and predictive modeling for fundraising. I can only surmise that demand for expert opinion on the subject is outstripping the supply, because let me be clear: I’m no expert.

That is, I’m not an expert in the sense that I’ve figured it all out, that my models all do what they’re supposed to do, that I understand what “degrees of freedom” means. (I’m beginning to suspect that no one can explain what degrees of freedom means. One statistician I read will say only that it’s an “elusive concept”. I get that much.)

I’m not an expert – I’m an explorer. I recognize that this predictive modeling stuff can be a powerful tool. Others have shown that certain simple techniques can be learned by anyone looking to harness some of that power for the benefit of their institution, and I’ve tried to learn those as best I can.

I also recognize that there is no end of learning. Where’s the summit of this mountain? I’ve no idea, but I sometimes catch a glimpse when I post a question to this data mining listserv. As the responses roll in from people who have been doing this a long time, I realize I’m at Base Camp 1 without enough oxygen.

Now, there’s nothing wrong with Base Camp 1. The view is nice. A person can do quite a lot of good from here, including trying to help others reach that level.

Allow me a metaphor shift …

I drive a car. It gets me places, allows me to do things I otherwise could not do. I understand the vague outlines of the workings of the machinery inside. I know the general idea of how the propulsive thrust is generated and transmitted to the wheels. Keeping an eye on the gas gauge and speedometer, I know enough to drive safely, and to recognize trouble.

But I am not a mechanic!

It’s the same with regression analysis. I know the interior workings only in outline. For the rest, I simply try to define my destination as accurately as I can, and then keep my eye on the indicators I see on the dashboard: the p-values, R-squared and so on. They keep my on the road, and tell me if I’m getting into trouble. (When I do get into trouble, I consult a mechanic. They’re out there, and they are helpful.)

I don’t choose to be ignorant about my car. Learning more about how it works will never be a waste of time, and it might save me someday. But being a non-expert is no reason to avoid getting behind the wheel. The number of us who are coming to appreciate data and analytics via our careers in fundraising is growing every year: Wherever you find yourself stranded on this road, someone will be by to help!

P.S. — Startled to discover this evening (28 June 2010) that Jacob Cohen et al. used exactly this automotive metaphor in the opening pages of the landmark book, “Applied Multiple Regression/Correlation Analysis for the Behavioral Sciences.” They write: “We note that to drive a car, one does not need to be a physicist, nor an automotive engineer, nor even a highly skilled auto mechanic, although some of the latter’s skills are useful when one is stuck on the highway.” I didn’t (knowingly) lift this idea!



  1. I’ve been lurking around your blog for a couple weeks now on the recommendation of Peter Wylie. The blog and your conversational writing style make the topic very approachable. Unfortunately, I’m not a mechanic either…heck, I’m not even a driver! I’m in the passenger seat trying to figure out how to get my seatbelt latched.

    Do you have some books or articles that you’ve found helpful over your trip down Predictive-Modeling Lane? As a layperson at this, finding resources that aren’t completely offputting is difficult.

    Thanks for sharing your successes and trials with us!

    Comment by Sarah — 16 February 2010 @ 4:39 pm

    • Hi Sarah – Thanks for that. And I do relate to not even being in the driver’s seat. Sometimes I’m out behind the vehicle, pushing! It is quite true that publications for the layman are really in short supply. I will admit, too, that even though I try to keep the tech-talk as accessible as possible, the blog format is too disorganized and discontinuous for anyone trying to work their way through building their first model. As you’ve been speaking with Peter, you’ll already be aware of his book, “Data Mining for Fundraisers”. That book has really been the foundation document for me. (I took it down off the shelf today and re-read some sections. It’s worth owning your own copy for reviewing now and again.) I would supplement that with a few of Peter’s white papers (in my “books and resources” page I’ve got quite a few links – scroll down a bit). After that, my top recommendations aren’t books at all: A good conference, and one-on-one training. A few good sessions at the right conference can really light a fire under your data mining efforts. (I’m not sure what your profession is, but the Association of Professional Researchers for Advancement – APRA – occasionally runs a very good symposium on the subject.) As for training, I probably would not have gotten anywhere without it. Feel free to email me ( if you want discuss further. I don’t offer training myself, but might have some advice about finding it. Best of luck, and thank you again for coming out of lurk mode!

      Comment by kevinmacdonell — 16 February 2010 @ 9:50 pm

RSS feed for comments on this post. TrackBack URI

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

Blog at

%d bloggers like this: