Home / Statistical Tools / Control Charts / Laney P' Chart
Laney P' Chart¶
From Excel click...
QXL Stat Tools Tab > Control Charts > Attribute Charts > P, NP, and Laney's P'
Want to build one first?
Laney P' Chart How-To builds the same twenty subgroups twice, once as a p chart and once as a Laney P' chart, from sample data you copy off the page. The p chart marks seven points out of control and the Laney P' chart marks none.
The Laney P' chart, read "P-prime", plots exactly what a p Chart plots and draws its limits differently. It multiplies the binomial standard deviation by an estimate of how much the subgroups really scatter, so a process with genuine subgroup-to-subgroup variation gets limits that describe it rather than limits that flag most of it.
| Panel title on the worksheet | Laney P' Chart, one panel only |
| What it plots | defectives divided by sample size, the same as a p chart |
| Y axis | Proportion Defective |
| X axis | Subgroup |
| Distribution behind the lines | binomial, with the spread rescaled by \(\sigma_Z\) |
One ribbon entry opens three charts. The button reads P, NP, and Laney's P', and the Variant: radio row at the top of the dialog selects which one you get. Choosing Laney P' retitles the window Laney P' Chart. The worksheet it writes is named "Laney P Chart", without the apostrophe, because Excel strips characters it does not allow in a sheet name.
When to use it¶
- You would draw a p Chart, but its \(\sigma_Z\) diagnostic is well above 1, so the points scatter more than binomial variation alone explains and the p chart's limits are too tight.
- Large subgroups make this more likely, not less. Binomial limits tighten as the sample grows, so a few thousand items per subgroup plus any real lot-to-lot variation gives a p chart that flags almost everything.
- The counterpart for defect counts rather than defective items is the Laney U' Chart.
How it works¶
The plotted statistic and the center line are the p chart's, unchanged: \(\bar{p}\) pooled across the baseline subgroups. Only the spread differs.
The factor is \(\sigma_Z\). Each subgroup is standardized, turned into how many binomial standard deviations it sits from the center line. If the binomial model were right, consecutive standardized values would move about as much as consecutive values from a standard normal distribution. \(\sigma_Z\) measures how much they actually move, as an average moving range divided by the usual constant, and the ratio is the correction:
- \(\sigma_Z\) near 1: the model fits, and a Laney P' chart is barely distinguishable from a p chart.
- \(\sigma_Z\) above 1: overdispersion. The Laney limits are wider than the p chart's by exactly that factor.
- \(\sigma_Z\) below 1: underdispersion. The Laney limits are narrower.
Every subgroup's standard deviation is multiplied by the one \(\sigma_Z\) estimated for the chart, then the limits are placed three of those either side of the center line by default, floored at zero and capped at one. The center line never moves, because \(\sigma_Z\) scales spread and not level. Because the correction is a single multiplier, the band stays the same shape: it still steps with the sample size, just further out.
\(\sigma_Z\) is reported on the plain p and np charts too, where it is a diagnostic only. On this chart it is the arithmetic.
The exact formulas are on the Math Details page, and \(\sigma_Z\) is defined in Laney Overdispersion.
Output¶
The analysis writes one worksheet named Laney P Chart. Under GroupBy there is one worksheet per group, each named Laney P Chart - {GroupName}.
Worksheets are divided by GroupBy group, not by chart. Every chart belonging to a group is stacked down that one worksheet, one below the other, so selecting three measurement columns without GroupBy produces three charts on a single sheet rather than three sheets. The User Input block is written once at the top of each worksheet; everything after it repeats for each chart on the sheet.
The sheet carries a User Input block, the Laney P' Chart panel, a summary table listing p-bar, UCL, LCL, Total subgroups, OOC points and Outliers marked, and an Overdispersion Interpretation block headed by the computed \(\sigma_Z\) and then a three-row legend for reading it, which is a guide rather than a verdict on yours. Every row of that table is read at the LAST plotted point, so on a chart with varying sample sizes read the steps off the chart. A split chart adds a Per-Split Breakdown block with a \(\sigma_Z\) per phase, and the pooled \(\sigma_Z\) above the legend is then not printed at all. There is no capability block and no histogram.
When the analysis finishes, the Control Chart task pane opens on the right. See Task Pane.