Episode 2: The Ishikawa Diagram, Mapping Every Possible Cause

Theo Ashworth

Last episode we followed a single chain of whys, one question stacked on the last, until we hit something we could fix. And we admitted that the method gets awkward when a problem has several causes tangled together rather than one tidy line.

Nell Harding

Right, and accidents and incidents can have several causes tangled together. Today's tool helps us map those possibilities. It can also be used in other areas, including quality improvement, but our focus is investigation. It's the Ishikawa diagram, also called the fishbone or the cause and effect diagram. Theo, where does it come from?

Theo Ashworth

Kaoru Ishikawa, 1943, working with engineers at Kawasaki Steelworks. He was a chemistry professor at the University of Tokyo, and later a key figure in Japan's postwar quality control circle movement. The idea was a visual, shared picture that frontline workers and engineers could build together to pull apart variability in a process. It ended up as one of the Seven Basic Quality Tools.

Nell Harding

Okay, so what does it actually look like? Because I always picture a fish and then lose track of which part is which.

Theo Ashworth

Four parts. The head, on the far right, is the problem statement, the effect. The spine is a horizontal arrow pointing at that head. Then the major bones are the categories, branching off the spine diagonally. And the minor bones, the ribs, are specific causes and sub causes hanging off those categories. You generate those by asking "why?" down each line.

Nell Harding

So that's where the 5 Whys comes back in. Each bone becomes its own little chain of whys.

Theo Ashworth

Exactly. The fishbone is the wide map, and the whys are the drilling. One guide I read, from Kaizen, puts it neatly: list the causes on the diagram first, then apply 5 Whys to each one.

Nell Harding

Now the head. You said a clearly defined problem statement. Is that really worth stressing? It feels like the boring bit.

Theo Ashworth

It's the bit that can wreck an investigation. Take a hypothetical case: "someone was injured" leaves too much open. "A worker injured their hand while clearing a jam on a conveyor" gives the team a more specific event to investigate. Stick to what is known, and don't build an assumed cause into the statement. If half the room is investigating a different event from the other half, the diagram just becomes a tidy picture of an argument.

Nell Harding

Fair. So, the categories. Everyone knows the 6Ms.

Theo Ashworth

For manufacturing, yes. Machine, Method, Material, Measurement, Manpower, and Mother Nature, meaning the environment. Some people add a seventh M, Money. In an incident investigation, those bones can prompt questions about equipment, procedures, materials, measurements, people and conditions. They are starting points, not a list of proven causes.

Nell Harding

But that's a factory vocabulary. What if we're investigating an incident on a hospital ward or in an office?

Theo Ashworth

Choose categories that fit the incident. You might consider equipment, work procedures, the environment, training, supervision and safeguards. The categories are prompts against tunnel vision, not dogma, and you can change them as the investigation develops. The same diagram can also be adapted for quality, service and other kinds of problems.

Nell Harding

And the shape of the thing tells you something, doesn't it? If a bone's empty.

Theo Ashworth

That's one of my favourite bits. A barren bone usually doesn't mean that area is fine. It means nobody in the room knows it well, so you go and find the person who does. And the opposite: if one bone is overcrowded, split it into sub categories. If two bones have almost nothing, merge them.

Nell Harding

Here's what worries me, though. You get ten people with sticky notes, and an hour later there are sixty causes on the wall. It looks thorough. But what do you do with it?

Theo Ashworth

That's the heart of it, and I'll say it plainly: the fishbone doesn't tell you which cause is real. It's an idea generation and hypothesis mapping tool. Every note on that wall is a suspect, not a verdict.

Nell Harding

Which sounds like an easy trap. You finish the diagram, everyone feels productive, and the meeting ends.

Theo Ashworth

It's the classic failure. Treat the finished diagram as the investigation and you either freeze, paralysis by analysis, or you try to fix dozens of speculative causes at once and fix nothing. So the next step is narrowing. Dot voting, sometimes called multi voting: everyone gets a few dots, and you're looking for the top two or three suspected root causes.

Nell Harding

Votes are still opinions, though. Ten people agreeing doesn't make something true.

Theo Ashworth

No, and that's the point. Voting can help decide what to examine first, but it doesn't prove or rule out a cause. Then go to the actual place where the incident happened, when it is safe to do so, and collect evidence. Look at records, observations and the physical evidence, and speak with people who know the work.

Nell Harding

And Pareto comes in there?

Theo Ashworth

It can. A Pareto chart can help a team examine patterns across a set of incidents and decide where to look more closely. But it doesn't establish what caused an individual incident. We'll explore that method next time.

Nell Harding

Can we walk through an incident example? How would the diagram help without turning guesses into facts?

Theo Ashworth

Take that hypothetical hand injury during conveyor jam clearing. The team might ask about the equipment and any safeguards, the procedure for clearing jams, the working environment, and how people were trained and supervised. A possible issue could appear on more than one branch. None of that tells us what happened yet. Each suggestion points to evidence the team needs to collect and test.

Nell Harding

So the diagram's result was, in effect, "don't guess yet".

Theo Ashworth

Which is a perfectly good result. There's also a healthcare paper by Kumah and colleagues, 2024, describing Nyaho Medical Centre. A quality improvement project there used a fishbone to identify possible causes of needlestick injuries, and the injuries went from 11 cases in 2018 to 2 in 2021. I'd be careful, though. The paper reports the fall after the project; I wouldn't claim the diagram alone did it.

Nell Harding

That's a sensible caveat. So if someone's running one tomorrow, what's the short version?

Theo Ashworth

Get people who know the work in the room. Agree one specific incident statement based on what is known. Pick categories that fit, and change them if they don't. Ask why down every bone. Tidy the sprawl, then decide what to investigate and test the possibilities against evidence. Don't mistake a vote or a completed diagram for proof.

Nell Harding

The map first, then the evidence. Next time, we look at Pareto analysis and how it can help us explore patterns across incidents.