Cause analysis: the six conditions

Only one of the six lives inside the person.

Six conditions, and where each of them lives

Gilbert’s contribution was not the list. It was the split. Three of the six conditions are arranged around the performer by somebody else, and three are carried by the performer, and the two halves are bought from different budgets, on different timescales, by different people.

What each one looks like when it is missing

A condition is not recognized by its name. It is recognized by what an organization looks like while it is absent, and those six pictures are worth more in a first conversation than the model is.

The asymmetry the split reveals

Set the six out that way and something becomes hard to unsee. The top row is cheap, fast and owned by someone with a name. The bottom row is slow, expensive and owned by nobody in particular, and the bottom row is what the organization buys.

One thing the split does not claim. It does not say the environment is always the cause. It says the environment is the part you have not looked at, that it is the cheaper half to change, and that the sixth condition is the one everybody proposes before anybody has examined the other five. Number 03 gives the evidence for that ordering; this paper gives the method.

What the model is, and what it is not

Honesty first, because this is the point at which a skeptical reader is entitled to ask what the six conditions actually rest on. The answer is straightforward and worth stating plainly.

Gilbert’s six are not a validated instrument and this series does not present them as one. There is no meta-analysis of the model, and there is unlikely ever to be one, for the reason number 02 sets out: a method whose purpose is to establish that this problem in this organization has this cause does not generate a population of comparable cases for anybody to pool.

What the six are is a diagnostic checklist, and a checklist earns its place differently from a theory. It earns it by being exhaustive enough that nothing important falls between its categories, and ordered well enough that the cheap questions come before the expensive ones. On both counts it holds up, and the individual categories it names have accumulated substantial synthesized evidence around them, built by researchers who had never heard of Gilbert. Section 03 sets that out condition by condition, including where it is thin.

One further consequence follows from the split, and it is the reason the checklist is worth its afternoon. Gilbert drew the line at three and three, but look at who can arrange each. Training can be bought, and capacity can be selected for or designed away, which leaves only motives truly carried in person: five of the six conditions can be arranged around the performer by somebody else. A practitioner who does not have a list will examine the sixth anyway, because the sixth is the one with a face, which is the whole subject of number 04.

Condition by condition, what the research supports

Four decades of work has accumulated around these six categories without ever addressing the model itself. Read across it and the strength of the evidence is markedly uneven, which is itself useful, because it tells you which findings will survive a skeptical room.

And the same, in the numbers themselves

The table says what the research supports. This says it in the form the research reports it, for the reader who wants to check rather than take it on trust.

And the three the person carries

The bottom row is where the money goes, and it is where the evidence is thinnest, not because the conditions are unreal, but because they are far harder to change than the budget implies.

Four questions that identify a binding condition

Mager and Pipe turned this into a decision sequence for practitioners, and the sequence is still the shortest statement of the method (Mager & Pipe, 1997). Four questions, and the order carries most of the value.

Two rules about how they are asked. Ask them of the performer, not about the performer: the person doing the work knows which of the four is true and is rarely asked. And ask the exemplary performer the same four, because the difference between the two answers is the finding, which is the subtraction number 07 sets out.

Rarely one, and rarely all six

Two failure modes sit on either side of this question, and an analysis can walk into either one on the same afternoon.

The first is naming a single cause where three are operating. It happens because the first condition that produces a clear answer is satisfying, and because a single cause makes a tidier report. The correction is arithmetic rather than judgement: for each condition you name, say what share of the gap it plausibly accounts for, and check that the shares are of the right order. Where they add to a fraction of the gap, something is missing and the analysis is not finished.

The second is the over-correction. Having found three conditions, propose something for each, and present the package as thoroughness. It reads well and it is expensive, and it destroys the one thing the analysis was for: after a package, nobody can say which part worked. That is not merely an evaluation problem for later: it is what makes the next analysis harder, because the organization now has a success story with no mechanism in it.

One practical note about ranking. The share a condition accounts for and the cost of reaching it are different axes, and the binding condition is not automatically the one to act on first. A condition that explains a fifth of the gap and can be fixed by changing a report next week will usually be done before one that explains half and needs two directors to agree. Say both numbers and let the person who owns the result decide, which is the limit number 02 sets on the whole discipline.

The dispatchers who all knew the rule

A national maintenance contractor, 62 dispatchers across three centers. The measure was the share of jobs dispatched with a complete parts list. Desired 90 percent; running at 61 over four quarters. The proposal on the table was a refresher on the dispatch procedure.

Question four had established the level: uniform across all three centers and all 62 dispatchers, with nothing concentrated anywhere, and the ceiling test showed that even at the exemplary performer’s rate the unit would reach 90. So the cause was reachable by people, but which condition?

What was changed, and what it cost

Two remedies, both in the top row of the six, and neither of them a program.

The first: the completeness figure was added to the weekly report that already existed, per dispatcher, next to call handling time. That is a report change, and it took an analyst two days. The second: the team meeting agenda was changed so that the two figures were discussed together rather than one of them alone, which is a five-minute decision by a center manager, and it was by far the harder of the two to get.

Two notes about the ending. The remaining six points were the catalogue: the third question’s finding, quantified afterwards at roughly the right size, and fixed the following year when the catalogue was replaced for unrelated reasons. And the analysis had to say out loud that handling time would rise, before the change, because a remedy that quietly moves a second number and does not warn anybody is how a good finding loses its credibility in the following quarter.

What this does not claim

This is the paper in the series with the weakest formal foundation, and saying so is the condition for the rest being usable.

• The six conditions are a checklist, not a validated model. There is no synthesized evidence for the model itself and there is unlikely to be. What the six have is exhaustiveness, a useful order, and convergent support from literatures built independently of them.
• The supporting evidence is about categories, not about the model. Wisniewski and colleagues studied feedback in education, Kluger and DeNisi studied it at work and in the laboratory, and Knight and Parker studied work redesign. None of them studied Gilbert’s conditions. Reading those syntheses as validation of the model is the error this paper is trying not to make; what they establish is that the category is real and that its effect is conditional.
• Nothing here establishes how often each condition binds. No study reports that distribution, for the same reason set out in number 08. The claim is about the order of examination, which holds whatever the distribution is.
• The sixth condition is real. Sometimes people genuinely cannot do the work and instruction is exactly the right answer. This paper argues that it is the last of six to examine and the first of six to be proposed, which is a claim about sequence and not about worth.

Six conditions, on the gap in front of you

Half a day, with the people doing the work rather than with the people describing it. Line one is answered in the room; the rest are answered where the work happens.

Two problems, one method, sixteen papers

Two models carry the method. Gilbert’s six conditions ask what kind of cause; Rummler and Brache’s three levels ask where it sits. Both are sound and this series uses both, and it leads with Gilbert, because his unit is the accomplishment, which is what an organization already counts, and because his six conditions are the categories an organization can actually change.

Where the method comes from

Two further pieces close the set and carry no number, because each points at all the numbered papers. Before the guide sits Arets L&D Business Models, the destination piece for the people who receive the requests: four models a learning function can run, and the route from HR-driven to business-driven. Last sits Solutions that cannot work, the field guide, ordered by the sentence that lands on your desk: seventeen forms a performance problem arrives in, each with an address in one of the numbered papers. Three orders in all: the destination you choose, the question you are on, and the form that landed on your desk.

Sixteen papers, six parts

Number 01 is the map and number 02 is the foundation. From number 05 on, every number is depth on exactly one step of the method. The filled block is the paper you are holding; the two unnumbered pieces close the set.

Where every claim in this paper comes from

The three books named earlier in this paper are listed in full in number 02, so that they do not recur in every reference list.