The Seven Basic Quality Tools

ASQ's steps for making one:

1. "Decide what event or problem will be observed. Develop operational definitions."
2. "Decide when data will be collected and for how long."
3. Design the form so that data "can be recorded simply by making check marks or X's or similar symbols and so that data do not have to be recopied for analysis."
4. "Label all spaces on the form."
5. "Test the check sheet for a short trial period to be sure it collects the appropriate data and is easy to use."
6. "Each time the targeted event or problem occurs, record data on the check sheet."

Worked example: the help desk's check sheet

During release 2.0's first registration week, ExamReg's help desk kept a check sheet: one tally mark for each complaint, under its type and its day. The six days end on the last day before the late fee begins. The program prints the sheet as the help desk kept it, then stratifies the same data by day.

# the help desk's check sheet for release 2.0's first registration week (FINDINGS 5.11):
# complaints by type, one count per day; day 6 is the last day before the late fee starts
sheet = {"payment taken, form not submitted": [3, 4, 2, 5, 6, 14],
         "session expired while filling form": [2, 1, 3, 2, 3, 8],
         "account locked after wrong passwords": [1, 2, 1, 1, 2, 5],
         "wrong fee shown": [1, 0, 1, 1, 0, 2],
         "other": [1, 1, 0, 2, 1, 3]}

def tally(n):
    """Tally marks in groups of five, as a check sheet records them."""
    return " ".join(["|||||"] * (n // 5) + (["|" * (n % 5)] if n % 5 else []))

print("the check sheet, as kept for the week")
for complaint, counts in sheet.items():
    print(f"   {complaint:<38}{tally(sum(counts)):<42}{sum(counts):>3}")

print("the same data stratified by day")
print(f"   {'':<38}" + "".join(f"{d:>4}" for d in range(1, 7)))
for complaint, counts in sheet.items():
    print(f"   {complaint:<38}" + "".join(f"{n:>4}" for n in counts))
days = [sum(day) for day in zip(*sheet.values())]
print(f"   {'all complaints':<38}" + "".join(f"{n:>4}" for n in days))
print(f"day 6 alone: {days[-1]} of {sum(days)} complaints ({days[-1] / sum(days):.0%})")

the check sheet, as kept for the week
   payment taken, form not submitted     ||||| ||||| ||||| ||||| ||||| ||||| ||||   34
   session expired while filling form    ||||| ||||| ||||| ||||                     19
   account locked after wrong passwords  ||||| ||||| ||                             12
   wrong fee shown                       |||||                                       5
   other                                 ||||| |||                                   8
the same data stratified by day
                                            1   2   3   4   5   6
   payment taken, form not submitted        3   4   2   5   6  14
   session expired while filling form       2   1   3   2   3   8
   account locked after wrong passwords     1   2   1   1   2   5
   wrong fee shown                          1   0   1   1   0   2
   other                                    1   1   0   2   1   3
   all complaints                           8   8   7  11  12  32
day 6 alone: 32 of 78 complaints (41%)

What the check sheet shows. In a week, 78 complaints. One kind dominates: payment taken, form not submitted, 34 of them, more than the next two kinds together. That is a Pareto question, and Chapter One Hundred Four, on Pareto diagrams, asks it properly. It is also the problem most worth a cause-and-effect diagram, which Chapter One Hundred Five, on cause-effect diagrams, draws for failed fee payments.

What stratification adds. Split by day, the complaints are steady at 7 to 12 a day for five days and then jump to 32 on day 6, 41 per cent of the week's total, with every kind of complaint rising. Two readings are possible, and the data alone do not decide between them: the portal may fail more under the last day's load, or more students may simply use it that day. The next tools separate them: a scatter diagram of complaints against daily logins (Chapter One Hundred Six, on scatter diagrams) and a run chart of complaints over the season (Chapter One Hundred Seven, on run charts).

Why the check sheet came first. Without it, the help desk's impression might have been students keep getting logged out, the most irritating complaint to handle. The sheet's counts put the payment problem first and showed the deadline effect, which is what ASQ's first step, deciding exactly what to observe and defining it, makes possible.

The tools together

The tools form a natural sequence for a quality problem, and this module's last chapters follow it.

1. Count with a check sheet: what happens, how often, when.
2. Rank with a Pareto chart: which problems matter most.
3. Explore causes with a cause-and-effect diagram: what could produce the top problem.
4. Test a suspected cause with a scatter diagram or stratification: is it related, and in which groups?
5. Watch the process with a run chart or control chart: did the fix work, and does it stay fixed?

What it does not mean

A check sheet is not a checklist. A checklist says what to check; a check sheet counts what happened.

Basic does not mean weak. The tools are simple so that everyone can use them; most quality problems need nothing more.

The list is not fixed. ASQ notes that newer lists include the flowchart instead of stratification, and some include the run chart.

A tool does not supply the cause. The check sheet showed the deadline effect; deciding what causes it needs further data and judgement.

Quick revision

• The seven basic quality tools (ASQ; first highlighted in Ishikawa's Guide to Quality Control): cause-and-effect diagram, check sheet, control chart, histogram, Pareto chart, scatter diagram, stratification (or flowchart; some lists add the run chart).
• Each answers one question: how often (check sheet), how spread (histogram), which few (Pareto), what causes (cause-and-effect), related (scatter), stable (control chart), different by group (stratification).
• Check sheet (ASQ): a prepared form for collecting data by tally marks; not a checklist; steps: define what to observe, decide when and how long, design and label the form, trial it, record each event.
• Together: count, rank, explore causes, test a cause, watch the process.
• Worked example: 78 complaints in the first week; payment taken but form not submitted 34; day 6 alone 32 (41 per cent), every kind of complaint rising.

Test yourself

1. Name the seven basic quality tools and say what each is for. The check sheet records how often problems occur; the histogram shows how a measurement is distributed; the Pareto chart ranks problems to find the few that matter most; the cause-and-effect diagram lists possible causes of a problem; the scatter diagram shows whether two variables are related; the control chart shows whether a process is stable; stratification splits data into groups to see whether the picture differs. Some lists use a flowchart or a run chart instead of stratification.

2. What is a check sheet, and how is one made? A structured, prepared form for collecting data by marking tallies as events occur. To make one: decide exactly what will be observed and define it; decide when and for how long to collect; design and label the form so that marks need no recopying; test it for a short period; then record every occurrence.

3. How does a check sheet differ from a checklist? A checklist lists items to check or steps to follow, such as the questions in a review; a check sheet records how many times each kind of event happened, producing data for analysis.

4. What is stratification, and what did it show in the worked example? Breaking data down into categories to make sense of them. Split by day, the help desk's complaints were steady at 7 to 12 a day for five days and jumped to 32 on the last day before the late fee, 41 per cent of the week, a deadline effect the weekly totals hid.

5. In what order are the basic tools typically used on a problem? Count occurrences with a check sheet; rank them with a Pareto chart; explore possible causes of the top problem with a cause-and-effect diagram; test a suspected cause with a scatter diagram or stratification; and watch the process after a change with a run chart or control chart.

6. Why are the seven tools called basic, and why did Ishikawa promote them? Because they need only counting and drawing, so anyone can use them without statistical training. Ishikawa promoted them so that quality work could be done by everyone in the organisation, not only by specialists, which is the idea of company-wide quality control.