Home / Statistical Tools / MSA / Type 1 Gage Study

Type 1 Gage Study¶

A Type 1 study measures one part, many times, on one gage, by one appraiser. Because the part never changes and the appraiser never changes, everything the readings do is the gage doing it. That makes it the shortest route to two questions: does the gage repeat, and does it read the right number.

Find it at QXL Stat Tools > MSA / Gage R&R > Type I MSA Gage Study (Repeatability Only).

When to use it¶

A Type 1 study is the one you can run on a gage before you commit to anything larger. It needs one part and one afternoon, where a full Crossed study needs several operators, a set of parts and a plan.

It reports repeatability and, when you know the part's true value, bias. It cannot report reproducibility, part to part variation or an operator effect, because the study holds all three constant by design. When you need those, run Crossed, Nested or Extended.

What the study needs¶

You supply	Required	What it buys you
A column of repeated measurements of one part	Yes	Sample size, mean, standard deviation, study variation
A reference value, the part's known true value	No	Bias, the T statistic and its P value, and Cgk
Both specification limits	No	The tolerance width, Cg, Cgk, both percent of tolerance figures, percent of resolution, and the limit lines on the run chart
The gage's resolution	No	Percent of resolution and its acceptance verdict

Two rules are worth knowing before you start.

A tolerance needs both limits. Quantum XL derives the tolerance width as the upper specification limit minus the lower one, and only when you have supplied both. Nothing is substituted for a missing limit, so a study with an upper limit alone reports its summary statistics and its bias and leaves every capability figure as N/A.

Twenty-five measurements is the recommended minimum. Fewer is not refused, but the report prints a note on the sheet saying how many you gave and that the recommended minimum is 25.

The page sequence¶

The dialog has two tabs, Data and Options. Every control is described on Options.

On Data, tick the column or columns holding the repeated measurements.
On Options, type the reference value, the specification limits and the resolution for each column, adjust the study options if the defaults do not suit, and fill in the gage details.
Choose Finish.

The report appears before the dialog does¶

Running the command computes a study from the default options and writes the worksheet first. The dialog then opens on top of the finished report. Every change you commit in the dialog closes it, rewrites the worksheet and reopens the dialog with your settings still in place, so the numbers on the sheet always match the boxes in front of you. A brief flicker is that cycle, not an error, and nothing you entered is lost.

Moving between the Data and Options tabs is not a change, and rewrites nothing.

What you get¶

One worksheet, named Type 1 Gage Study, carrying every study the run produced. Tick two measurement columns and you get two studies on the one sheet, stacked down the page in column order, with the group second when you used GroupBy. The sheet is protected when it is written.

A single study is laid out as a run chart on the left, at columns D to O, and a statistics table to its right, at columns Q and R. Above them, in columns A and B, sit the input echo and the Gage Information block.

The Gage Information block¶

Columns A and B, under the heading Gage Information, carrying the seven details you typed on the Options tab. All seven rows are printed whether or not you filled them in:

Gage Name, Gage No., Gage Type, Part Name, Part No., Date, Performed By.

The statistics table¶

One table per study, headed with the study title, for example Type 1 Gage Study - Diameter, and divided into five labelled sections.

User Supplied Data repeats what you typed, so the sheet stands on its own later: LSL, USL, Tolerance Width, Percentage of Tolerance, # Standard Deviations, Reference Value and Resolution.

Summary Statistics describes the readings themselves: Sample Size, Mean, StDev and the study variation, whose label is built from the multiplier you chose and reads 6*StDev at the default. A Dropped Values row appears only when something in the range could not be read as a number, and is left out entirely when nothing was dropped.

Bias Calculations appears when you gave a reference value: Bias, T and P-Value. The P value is printed in red and bold when the bias is significant.

Capability holds the four figures that need a tolerance: Cg, Cgk, % Repeatability and % Repeatability and Bias.

Resolution holds % Resolution and Resolution Acceptable, the second of which reads Yes, No or N/A.

Any figure the study could not produce is printed as N/A, never left blank and never filled with a placeholder number.

The notes on the sheet¶

Four rules of thumb hover as cell notes on the numbers they judge, in the Capability section:

On Cg: Adequate measurement systems should have Cg values greater than 1.33
On Cgk: Adequate measurement systems should have Cgk values greater than 1.33
On % Repeatability: Adequate measurement systems should have % Repeatability less than 15%
On % Repeatability and Bias: Adequate measurement systems should have % Repeatability and Bias less than 15%

Four more are printed on the sheet as text rather than hovering:

Under Bias Calculations, always: P-values less than 0.05 indicate the bias is significant. The number is the alpha you set.
Under Resolution, always: A resolution of 10 percent of the tolerance or less is acceptable (AIAG rule of tens).
Under Summary Statistics, only below 25 readings: Note: You have only provided 12 measurements, the recommended minimum is 25.
Under User Supplied Data, only when the specification limits need it: Both a lower and an upper specification limit are needed for the reference limit lines on the chart, Cg, Cgk, the two percent of tolerance values and the percent of resolution. when you typed exactly one of them, or The upper specification limit must be greater than the lower specification limit. when you typed two that do not order.

The run chart¶

Titled Run Chart followed by the column name, for example Run Chart - Diameter. The measurements are plotted in worksheet order against Observation on the X axis, as a blue line with round markers, and the Y axis is titled with the column's own name.

Three flat reference lines may be drawn across it:

Series	Where it sits	Colour
Center	the reference value if you gave one, otherwise the mean of the readings	green
Upper	the centre plus half the reference span	red
Lower	the centre minus half the reference span	red

The reference span is the percentage of tolerance applied to the tolerance width, so at the defaults the band is 20% of the tolerance, centred. Upper and Lower are simply not drawn when there is no usable tolerance, rather than drawn at zero, and the legend always reads Center whichever source the centre came from.

The Y axis is scaled to hold both the readings and the band, so a reading outside the band is always visible.

When a study cannot be computed¶

A study the engine refuses prints one section, headed This study could not be computed, followed by a sentence naming the condition that stopped it, and no chart. The other studies in the run are unaffected. Options lists every sentence and its trigger.