Statistics Rules

Statistics rules are selected from the Document Default Properties or via the CM4D.ini file's [ STATRULES] MODE parameter.

• [STATRULES] MODE= not supplied for default
• [STATRULES] MODE= 0 for Default
• [STATRULES] MODE= 1 for Ford
• [STATRULES] MODE= 2 for Volvo C.
• [STATRULES] MODE= 3 for SAAB

Rule on Supplying Missing Subgroup Values:

1. Default/SAAB Rule: CM4D will not supply missing subgroup values.
2. Ford Rule: CM4D will supply missing subgroup values by substituting the mean of the subgroup for missing values when at least two values in the subgroup are valid and less than three values are missing or invalid and less than five percent of the values for all subgroups in a process are missing or invalid.
   1. Substituted values will be flagged with a "%" indicator when displayed in CM4D's grid.
   2. As masked values will appear as missing, they are treated as missing for the purpose of substitution. Values are masked when they are assigned a cause and the user chooses to mask the assigned value or values.

Rule on Invalid Lower Control Limit:

1. Default/SAAB Rule: CM4D will not calculate or supply zero for control limits for any control limit that can not normally fall below zero and the calculation for that control limit is below zero or cannot be calculated due to insufficient subgroup size. This applies to control limits for range, moving range, and standard deviation. This is a common sense change from prior versions of CM4D because establishing a control limit at zero for values that will never go below zero (such as range) does not make sense.
2. Ford Rule: CM4D will not calculate or supply zero for invalid lower control limits for Range.

Rule of Calculating CP Values (CP, CpK, CR, CpM) and PP Values (PP, PpK, PR):

With the exception of Ford Rules, CP values will be calculated using estimated standard deviation where appropriate when subgroups are used. Otherwise, calculated standard deviation is used which, will make CpK the same as PpK. In all cases, the appropriate equation values must be available for CM4D to calculate CP or PP values.

1. Default Rule: CM4D will calculate all CP and PP values having appropriate equation values present, such as specification limits, etc.
2. Ford Rule: Ford has both Run CP and Process Control CP values.
   1. Run CP values are calculated when:
      1. Subgroup Size > 1
      2. Not Collecting Data
      3. Has control limits
      4. In-control
      5. Check Normality and is normal
      6. Check MSA and MSA passed or acceptable
      7. Has minimum of 30 points
      8. Has minimum of 20 subgroups
      9. Has specification limits (bilateral or unilateral)
      10. Has control limits
   2. Process CP values calculated when:
      1. Has Process Control Statistics (current process or before)
      2. Not Collecting Data
      3. In-Control
      4. Check Normality and is normal
      5. Check MSA and MSA passed or acceptable
3. Volvo C. Rule:
   1. Must be normal distribution of data for Cpk
4. SAAB Rule:
   1. Run CP values calculated when:
      1. Has minimum of 10 points
      2. Has minimum of 3 subgroups

Rule for Out of Control Sequential Count:

1. Default Rule: CM4D will use a run sequence count of seven points in a row to flag an out of control condition for points on one side of the mean or points trending in an up or down direction. The default side of mean sequence count can be changed to 14 or 21 via the Routine or DataSet Properties.
2. Ford Rule: CM4D will use a run sequence count of seven points in a row to flag an out of control condition for points on one side of the mean or points trending in an up or down direction. The default side of mean sequence count can be changed to 14 or 21 via the Routine or DataSet Properties.

Rule for Minimum Number of Subgroups to Calculate Control Limit for XBar-R:

1. Default Rule: 1 or more subgroups are required
2. Ford Rule: 20 subgroups are required
3. SAAB Rule: 3 subgroups are required

Rule for Minimum Number of Points Required to Calculate Control Limit for X-MR:

1. Default Rule: 1 or more points
2. Ford Rule: 30 points
3. SAAB Rule: 10 points

Rule for Minimum Number of Points Required to Calculate Normality:

1. Default Rule: 1 or more points
2. Ford Rule: 30 points

Rule for Minimum Number of Points Required to Calculate CpK:

1. Default Rule: 1 or more points
2. Ford Rule: 30 points
3. Volvo C. Rule: 25 Points
4. SAAB Rule: 10 points

Rule for Minimum Number of Subgroups to Calculate CpK:

1. Default Rule: 1 or more subgroups are required
2. Ford Rule: 20 subgroups are required
3. SAAB Rule: 3 subgroups are required

Rule for Implied Unilateral Values:

This is the value that will indicate a unilateral limit value. When the high or the low limit value is set to the implied unilateral value, then only the other value is valid. CM4D allows for document specific selections for the use and implied value of unilateral values that will override the defaults below. These defaults can be changed via the Document Default properties.

1. Default Rule: The default implied unilateral value is zero
2. Ford Rule: The default implied unilateral value is -99999.99

Rule for the Outer Fence Multiplier to Calculate the Outer Fence of Box and Whiskers Charts:

1. Default Rule: The multiplier is 3
2. Ford Rule: The multiplier is 2

Mid-Third 90% Rule Indicates an Out-of-Control State:

The mid-third 90% rule is violated when more than 90% of points are in the middle third of the control limit region. This applies under all rules. Specific out-of-control markers are used to see where this rule is violated.

Mid-Third 40% Rule Indicates an Out-of-Control State:

The mid-third 40% rule is violated when more than 40% or fewer points are in the upper and lower third of the control limit region. This applies under all rules. Specific out-of-control markers are used to see where this rule is violated.

An optional feature of process baselines may be used with the CM4D database data to track historical changes to engineering data or to set demarcation points in the data. When process baselines are used, CM4D may further divide sets of data (DataSet rows) into processes according to relative time periods associated to the process baselines. In this case, CM4D isolates the data and calculates statistics for each time period or process baseline.

Rule for MSA% When MSA is Out-of-Control and Invalid:

1. Default Rule: The MSA% will not be displayed if the MSA has failed.
2. Ford Rule: The MSA% will be shown even if the MSA has failed.

Can we improve this topic?