User talk:Thopper

Welcome!

Hello, Thopper, and welcome to Wikipedia! Thank you for your contributions. I hope you like the place and decide to stay. Here are a few good links for newcomers:

I hope you enjoy editing here and being a Wikipedian! Please sign your name on talk pages using four tildes (~~~~); this will automatically produce your name and the date. If you have any questions, check out Wikipedia:Where to ask a question or ask me on my talk page. Again, welcome! Vsmith 05:00, 8 January 2006 (UTC)[reply]

Your recent edits to Shewhart individuals control chart

I'm looking at Wikipedia:Stub and wondering what criteria this article meets that warrants its classification as a stub... -- DanielPenfield (talk) 16:27, 19 January 2010 (UTC)[reply]

It lacked detail, lacked organization and contained inaccurate information. Your edits help, some; perhaps it should now be marked for cleanup. Tom Hopper 16:31, 21 January 2010 (UTC)

Ok, none of those concerns translate to criteria that appear in Wikipedia:Stub. But now you've piqued my curiosity: What details did you not see that you expected to see? Which information is inaccurate? -- DanielPenfield (talk) 17:00, 21 January 2010 (UTC)[reply]

Taking them in order, there was insufficient detail to enable a reader to create an ImR chart on their own. For instance, the explanation of how to calculate the limits on either the individuals or the moving range chart is, at best, confusing. There were also no examples given to further illuminate either the technique or the results, so the average reader could only be left with a vague impression that some numbers are calculated and, perhaps, plotted (here's where your edits helped). I would also argue that some history of the chart should be included, at least by reference to Control Charts.

The article is organized as a single block of text, even though there is some organization to the information, such as overall description, application and calculation. The explanation of the individuals and moving range charts are unclear and confusing, as the author does not use clear language and shifts between one chart and the other from one sentence to the next. In addition, the calculations given should more clearly state the dependent variable and to which chart they apply.

With regard to inaccuracies, incorrect information is provided in the statement that "the normal distribution is the basis for the charts," and the following three bulletted assumptions are likewise incorrect. They should be deleted and, preferably, replaced with accurate information regarding the underlying assumptions and potential weaknesses of the ImR chart (e.g. skewness increases the false-positive rate on the chart; for processes with infrequent sampling and a need to rapidly detect small mean shifts the CUSUM or EWMA chart will provide greater sensitivity).

So, with regards to stub status, the article is, in my opinion, grossly incomplete and inaccurate, but offers a fair start at describing a subject that can and should be treated in much greater length. In my opinion, therefore, the article meets the criteria for a stub as described both at Wikipedia:Stub and Croughton-London rule.

Tom Hopper 10:23, 26 January 2010 (UTC)

Note on assumptions: data is assumed to be mostly homogeneous. i.e. Data comes from the same process, and note from "identical" parts coming off of parallel production lines. However, the chart is intended to detect nonhomogeneity, so the data doesn't have to be completely homogeneous. Tom Hopper 19:06, 26 January 2010 (UTC)

After reading your response, I believe that you really are more concerned with Wikipedia:Make technical articles accessible and/or Wikipedia:Please clarify (a view that I share, BTW). There is no way the version of the article that you tagged met the "not so short as to provide no useful information" criterion of WP:Stub. As for what you feel are inaccuracies:

Your objection	My reaction
"With regard to inaccuracies, incorrect information is provided in the statement that "the normal distribution is the basis for the charts,"	I'm looking at Montgomery page 237 in the section on the Shewhart Control Chart for Individual Measurements: "...we conclude that if the process shows evidence of even moderate departure from normality, the control limits given here may be entirely inappropriate...". See also [1], "d2 is a correction factor with the ASSUMPTION of a normal distribution. We estimate a value for the standard deviation of the individuals by dividing the average range value by d2."
"and the following three bulletted assumptions are likewise incorrect.	So you're saying that you can use these control charts on serially-correlated process observations? Or that you can use these control charts for differently (vs. identically) distributed process observations? Or that you can change your inspection procedure at whim and still expect to reliably detect out-of-control conditions?
"(e.g. skewness increases the false-positive rate on the chart; for processes with infrequent sampling and a need to rapidly detect small mean shifts the CUSUM or EWMA chart will provide greater sensitivity)	Those concerns are not unique to the individuals/moving range chart--they're true for the Xbar and R chart and the Xbar and s chart as well and they're dealt with to some degree by Control chart#Types of charts. Should these be repeated in every control chart-specific article?
"However, the chart is intended to detect nonhomogeneity, so the data doesn't have to be completely homogeneous.	There's a subtle distinction here that probably should be explicitly spelled out in the article on control charts: We can't control the data directly--we can only influence it by adjusting the process inputs or the steps in the process. The best we can do is use quality-improvement techniques (e.g., Design of experiments) and knowledge of the process to eliminate as many sources of variation as we can, run the process, and plot the observations to determine whether we have evidence that the process is in control or not. The requirement absolutely is that at the point we establish control limits, we've got a process in a state of statistical control (the process observations behave like independent, identically-distributed random variables (same distribution, same values for parameters)). In other words, we absolutely have to have statistical evidence that the control limits are calculated from "homogenous data". To summarize: The data are what they are, whether we're in the control chart setup phase or monitoring phase The "homogenous data" requirement applies only to those process observations we decide to use for the control limit calculations during control chart setup If the data don't pass the statistical tests that indicate they're "homogenous", we shouldn't be using them to calculate control limits for any control chart (instead, we should be investigating and eliminating any sources of special-cause variation)

-- DanielPenfield (talk) 21:22, 26 January 2010 (UTC)[reply]

Excellent discussion. You raise some good points that have made me think carefully about the process behavior chart; thank you! I agree, too, that Wikipedia:Make technical articles accessible and/or Wikipedia:Please clarify would be better categorizations than Wikipedia:Stub.

If the intent of the statement that the normal distribution forms the basis of the charts is simply to point out that the three-sigma limits are calculated based on a symmetric, normal distribution, then this should be stated as such. As it is currently, it could be misinterpreted to mean that the data must be normally distributed, which is demonstrably false. Control charts work with skewed and one-sided data, and in fact the three-sigma limits were chosen because of their robustness to different distributions rather than because of the characteristics of the normal distribution. In my opinion, Montgommery's presentation confuses this point somewhat, though I do not recall him stating explicitly that data must be normally distributed..

You have good points about independent observations and homogeneous data, however, the bullet points are stated as required assumptions, implying that all data on the chart must meet these three requirements (normally distributed, independent and from the same measurement process). Since process behavior charts are intended to detect nonhomogeneity, these assumptions must be violated for some data points if the charts are to be of any value. With this in mind, I think that the subtle distinction that you discuss would be good to add. Probably in the article on control charts there should be some greater emphasis on the need to work toward homogeneous data (i.e. collecting data in the same way, from the same process, and working to bring the process under statistical control).

I would, however, caution against categorical statements that to create a chart, the data must pass statistical tests for homogeneity. There is, after all, no statistical test that will tell you if the data is homogeneous; you can only fail tests of homogeneity, and testing for homogeneity is the core purpose of the process behavior chart. If data could be made homogeneous, then we would not need the process behavior chart. The real test for homogeneity that satisfies the requirements of a process behavior chart is a careful process analysis with a skeptical eye turned toward ensuring that there are no controllable differences from one data point to the next in either the process that produces the measured product or the measurement process that produces the plotted data. The use of some form of standardized work and the application of continuous improvement, rather than any statistical tests, are probably the key requirements for successful use of control charts.

Regarding the issues that are dealt with in the general control chart article but not in the Individuals article (impact of skewness, etc.), this may be a question of editorial style. In my opinion, the article should stand reasonably well on its own, at least pointing to such considerations in the other article.

Tom Hopper 14:55, 27 January 2010 (UTC)

Proposed deletion of Mind42

The article Mind42 has been proposed for deletion because of the following concern:

Article has no third party reliable sources that establish notability.

While all contributions to Wikipedia are appreciated, content or articles may be deleted for any of several reasons.

You may prevent the proposed deletion by removing the {{dated prod}} notice, but please explain why in your edit summary or on the article's talk page.

Please consider improving the article to address the issues raised. Removing {{dated prod}} will stop the proposed deletion process, but other deletion processes exist. The speedy deletion process can result in deletion without discussion, and articles for deletion allows discussion to reach consensus for deletion. MrOllie (talk) 16:01, 21 September 2010 (UTC)[reply]

Mind42 has been around for some time, and there are multiple reviews of the product online. I have attempted to add several of the more notable ones.

In addition, please note that I am a user of Mind42 and not in any way associated with the development; I added Mind42 to Wikipedia because, as a user, I think that it's notable.

Thanks for your continued efforts; I appreciate the prodding to improve my edits. Tom Hopper (talk) 08:00, 23 September 2010 (UTC)[reply]