|Statistical Application Software
|FREE STATISTICAL SOFTWARE is available below (titles are in large red font)
FREE webinar demonstrations of the software are available upon request -- send request to JOHNZORICH@YAHOO.COM
FREE ARTICLES on a variety of statistical subjects are available at the bottom of this webpage.
We offer powerful MS Excel spreadsheets designed to simplify activities in Mfg., QA, and R&D.
Each commercial spreadsheet is formally validated; all of them are "GMP" and "Part 11" compliant.
VALIDATION PROTOCOLS, EXECUTED AND SIGNED BY JOHN ZORICH AS VALIDATION REPORTS,
ARE PROVIDED FREE TO COMMERCIAL CUSTOMERS. Our Validation Protocols/Reports have successfully
withstood the scrutiny of auditors from FDA, FDB, TUV, BSI, KEMA/DEKRA, and NSAI
WEBINAR TRAINING IN HOW TO USE THE SOFTWARE IS PROVIDED FREE.
More than 260 of our validated spreadsheets have been sold to more than 110 clients
(mostly American companies governed by FDA regulations). (Click here for LIST OF PRICES)
REVISED May 2019 to allow sample sizes up to 25 and up to 400 samples.
XbarR, XbarS, XmR, & other types of Statistical Process Control Charts are automaticly generated, along with a histogram chart, raw-data-points
chart, and capability indices (& their lower 1-sided confidence limits). SPC charts automaticly identify out-of-control points.
New Version July 2018: Now with nP as well as P and U statistical process control charts and a Pareto chart that are automaticly generated.
Out-of-control points, trends, and series are automaticly identified. New version allows up to 400 samples and is easier to use.
New version, March 2020 -- This program is for calculation of component reliability (i..e., % in-specification), confidence levels, Normal
K-factors, or related sample size (with a correction for off-center process/sample averages), as well as stress/strength interference,
non-normal tolerance limits, & Confidence Limits on Cpk/Ppk. Calculations are performed automaticly, with easy-to-use interfaces and with a
range of inputs that is much wider than is available in textbooks. Calculation is simplified for confidence limits on MTTF, MTBF (mean time to
failure, & mean time between failures) as well as for their corresponding failure rates and % reliabilities. Probability Plots (to assess whether data is
"normally" distributed or "exponentially" distributed) are created automatically from inputted data (after the user chooses a transformation to
normality for inputted raw data, the program provides a table of the transformed raw data). For the most accurate %-in-specification, a 2-sided
"Normal reliability plot" is now included (this is a simplified version of "Reliability & Distribution Plotting" that is described below).
REVISED: File outputs include Sample Size, two types of AOQL, two types of OC curves, Sample Size curve, and two AOQ curves. Instead of
relying on a sample-size table, this program calculates the exact absolutely smallest sample size that gives the user-chosen protection level for a
given exact size lot. Now for lot sizes up to 10,000.
REVISED to simplify entry of non-numeric options choices --- Four types of power curves are provided (as opposed to the one type provided in
most other commercial stat-packages): Power vs. Sample Size, Power vs. Hypothesized Difference, Power vs. Alpha, and Power vs. Population
Standard Deviation. Simple user interface.
New Version April 2018: Powerful tools for quantifying measurement uncertainty. Methods include Gage R&R (up to 3 persons, 3 gages, 3
replicates, and 20 parts, using either the ANOVA method or the XbarR method; includes interaction plots), Gage Correlation (up to 3 gages),
Gage Bias, Gage Linearity, Specification/Inaccuracy Ratios, and Specification Guardbanding. An essential tool for setting product specification
limits that take into consideration the uncertainty in the measurement process. Automatically generated customizable reports and charts.
Examine OC curves for your own custom sampling plans. Use either binomial or hypergeometric calculations. Now be able to explain the
"valid statistical rationale" of sampling plans you already use.
UPDATED (bug-fix) January 26, 2018 --- Provides an analysis and planning tool for sample sizes in situations where lots undergo sequential
inspections (e.g., 1st by Manufacturing, 2nd by QC, and finally by QA).
New version, October 2019
(1) Reliability is calculated simultaneously for 2-sided specifications, rather than only for 1-sided specs, with up to 2000 data points
(2) data placed into the "exact" entry table no longer needs to be in numerical order
(3) data can be sorted while in the Data Entry Table, and can also now be moved around between tables by "cut&paste" and "drag&drop"
(4) data can include zeros or negative numbers (prior versions allowed only positive numbers).
This program is for calculation of component % in-specification (i..e., process capability or % reliability) at a given % confidence, using "Reliability
plotting", which is also known as "probability plotting", "rectification", etc. Pre-formatted customizable reports are automatically created.
This may be the only method possible for calculating a reasonable level of process capability when there are either small sample
sizes, many replicate data values, unfinished experiments, or data that cannot be transformed into Normality. Reliability plotting can
also be used to confirm data as being "normally distributed" or can be used to identify what transformation to Normality is needed. Some of the
other distributions that can be explored include: Fatigue Life, Weibull, Largest Extreme Value, Exponential, Logistic, Cauchy, Student's t, etc.
Two detailed articles about Reliability Plotting are provided at the bottom of this page; also, mouse-click this link: RELIABILITY PLOTTING.
File provides sampling plans that were described in Annex IV, Section 6, of the version of the MDD that had been in place up thru 2009 (i.e., a
valid plan, but one that is no longer mandatory). This program outputs the smallest sample size that has no more than a 5% chance of passing
QC (as required by that version of the MDD). Automatically created OC curve and SampleSize vs. LotSize curve. For lot sizes up to 10,000.
FREE, STATISTICS-RELATED ARTICLES (see below) by John Zorich (mouse-click title to download):
Reasons for Teaching & Using the Signed Coefficient of Determination instead of the Correlation Coefficient
(published in 2018, in the Journal of the American Mathematical Association TYC (a.k.a., MathAMATYC Educator)
This article explains how the Signed Coefficient of Determination is a much better indicator of correlation strength than is the Correlation Coefficient.
Four Formulas for Teaching the Meaning of the Correlation Coefficient
(published in 2017, in the Journal of the American Mathematical Association TYC (a.k.a., MathAMATYC Educator)
This article provides easy ways to explain a difficult topic.
Better Alternatives to Sampling Plans
based upon hour-long seminars given by John Zorich at each of the following venues:
-- Ft. Worth TX ASQ's Annual CowTown Quality Roundup (April 19, 2013)
-- ASQ's Annual World Conference on Quality and Improvement (May 5, 2014)
-- Houston TX ASQ's Annual Regional Quality Conference (November 13, 2015)
Reasonable Confidence Limits for Binomial Proportions (published in 2010, in MD&DI Magazine)
This article introduces a new type of confidence interval for proportions (that is, percentages);
the new interval is called a "Reasonable Confidence Interval".
Better AOQ (and AOQL) Formulas
This article provides formulas necessary for calculating exact rather than approximate AOQ and AOQL values when sampling is "Type-A" (that is,
when sampling seeks to evaluate each lot for quality rather than to evaluate a stream of lots for process quality). It also provides Type-B formulas
that instructors of this subject may find easier for students to comprehend. Different formulas are provided that apply to different IQC practices
(e.g., are rejected parts replaced or simply discarded?). Risk-management recommendations are given at the end of the article.
The Pre-history of Probability (seminar given in 2000 by John Zorich to the Silicon Valley ASQ Statistics Group)
Literature Reference Justifications for Transformations to Normality (list assembled 2012)
Gage R+R, using Excel (this explains how to use Microsoft Excel to perform Gage R&R, simply!)
Reliability Plotting Explained (this provides a step-by-step instruction for this valuable tool)
Why Does Reliability Plotting Typically Provide a Higher Process Capability Result (i.e., %Reliability) Than Does a K-Table?
END OF PAGE
2018: This program provides an easy-to-use way to determine if your data is not normally distributed. If the program indicates that your data is
non-normal, it provides 14 different "transformations" to normality for you to consider. The program includes 10 different "tests of normality",
including Anderson-Darling's A2*, Cramer-von Mises' W2*, Shapiro-Francia's W' (a large-sample-size extension of the Wilks-Shapiro test),
Watson's U2*, Gan-Koehler's K2, Skewness, Kurtosis, Kuiper's V*, Lilliefors' Kolmogorov-Smirnov D*, and Chi-square. Histograms and probability
plots (with 95% confidence intervals [ a.k.a., "prediction bands"] & correlation coefficients) are automatically generated for all 14 transformations. A
formal customizable printable test report is automatically generated, including test results, raw data, and transformed data.
Several literature-reference justifications for normality transformations can be downloaded by mouse-clicking this sentence.
Provides a way to objectively decide if the complaint rate currently observed for your product is significantly larger than the historical complaint
rate for that product, using either the "confidence limit" or "significance test" methods.