All manner of experts claim to know better than people who do real work. Consultants such as myself have broader exposure to ideas and practices. Executives see the big picture more clearly. And accountants understand the financial realities far better than factory workers, who only know that without a new forklift work will stop when the old one breaks down.

Well, consultants really do have broader exposure, executives do see the big picture more clearly, and accounts do understand the numbers. And we all ask the people who do real work to respect our knowledge, expertise, and perspective.

So why do so few of us return the favor?

We’re going to spend one more week critiquing Paul Strassmann’s thesis that spending on IT hasn’t generated any economic returns. Strassmann, you’ll recall, has amassed a daunting array of financial statistics which he’s sliced and diced more ways than a Vegematic can handle potatoes, all without finding any correlation between financial improvement and IT spending.

I’ve been contending that he’s the kid looking for his lost quarter under the streetlight even though he lost it a block away, because “the light is better here”.

Here’s one place Strassmann hasn’t looked: He hasn’t talked to people who actually use computers to do their jobs. I wonder why not?

People who use computers to do their jobs aren’t misguided children or poor deluded fools. They’re smart people. If they tell you they do their jobs better with a particular technology than without it, they’re more likely to be right than a number-cruncher who’s never talked to them.

Here another problem with Strassmann’s analytical approach: He tries to correlate “IT spending” with various measures of financial return. “IT spending” is an undifferentiated blob. Let’s break it down into its components.

In typical organizations, 70% of IT spending goes to the data center and systems maintenance. These don’t deliver new value. They maintain value you’ve already gained. So by definition only 30% of IT spending has any chance of yielding further financial improvements.

It’s well-known that only 30% of all IS/IT projects are satisfactorily completed. That means while 30% of IT spending tries to create new value, only 9% has the chance. I wonder what conclusions Strassmann would reach if he re-analyzed his data using only the IT spending expected to deliver value.

Because so little spending delivers value in the end, Strassmann claims we’re aiming at the wrong targets, providing technology for technology’s sake. I’d contend the problem is with our technique, not our aim. It’s in the hard work of project management and system implementation that we goof up, not in our ability to link projects to business strategy.

Now here’s something remarkable. I re-analyzed some of Strassmann’s data, converting back to numbers as best I could a graph plotting IT spending per employee against return on equity (ROE) for 20 companies in “the food industry”. As expected, the regression analysis showed no correlation.

Then I tossed out four outliers — data points clearly outside the pack. The new regression showed a strong, statistically significant correlation between IT spending and ROE. For you statisticians, R Square equals .28 at a .033 level of significance. The slope (increase in ROE per $1,000 spent per employee) is about 1.5%.

This new analysis no more proves IT spending does provide value than Strassmann’s analysis disproves it. Since the original analysis lumps supermarkets, agribusiness conglomerates, pet food suppliers and a tobacco company together the whole analysis is just a tad dubious.

But when the inclusion or exclusion of just four companies makes the difference between no correlation and strong correlation, Strassmann’s conclusions must be taken with at least a grain of salt.

Or pepper, if you like spicier stories.