Testing multiple

Often, analysts will want to run special short-term tests with the operating unit in order to identify the cause of the trouble being experienced By the unit. Operators are naturally leery of running tests outside their normal operating experience because their primaiy focus is the stable control of the unit, and tests outside their experience may result in loss of control. Multiple tests with few results may decrease their cooperation. 

If the object of debottlenecking is to run heavier feeds, multiple test runs may be needed with heavy feed added in stages. 

Standard deviation s = 2.8 (determined by multiple test with one of the enzyme mixtures). 

A maximum latency of 50 seconds was allowed to prevent tissue damage and to enable multiple testing. Drugs were given by cumulative IV injections at 42-minute intervals up to a maximum dose one unit below the dose which produced rotarod failure. 

Multiple Comparisons and Multiple Tests Using SAS Text and Workbook Set (books in this set also sold separately) by Peter H. Westfall, Randall D. Tobias,... 

Likewise, the method was able to measure changes in abundance as low as 1.5-fold with confidence (p < 0.05), following multiple testing adjustment using the Benjamini-Hohcberg method (Fig. 20.7). 

Given the above observations it was essential in the present study that multiple test methods be used, representing evaluation of the effects of coatings on ignitability, flame spread, heat release, ease of extinction, and smoke. Samples should be commercially prepared and representative of materials commonly used in business machine applications. 

The FDA, USDA, and EPA have established methods and standards for detecting food- and water-borne pathogens these methods are available on the agencies websites [4—81. Many of the procedures for microbe identification rely on culturing the sample for subsequent identification by multiple tests such as colony characteristics, selective growth conditions, and biochemical assays for metabolites. Such traditional methods take upward of 24 h and remain the standard against which new... 

In summary, GC adjusts for population stratification without the assumption or estimation of parameters such as the number of subpopulations involved in the study. It provides control of false-positive results caused by population structure as well as by multiple testing. One possible drawback of this method is that the correction of the test statistic is constant across the genome. As a result, GC may have less power in certain situations. 

Benjamini, Y., and Hochberg, Y. (1995) Controlling false discovery rate a practical and powerful approach to multiple testing. J R Statist Soc Br. 57, 289-300. 

When the laboratory uses different methodologies or instruments, or performs testing at multiple testing sites, a system is to be in place that evaluates and verifies the comparability between these test results. For example, correlation studies must ensure that manual and automated methods of immunostaining within a laboratory are in agreement. This must be documented biannually. In addition, any reference laboratories utilized must be CLIA-88-certified, and the lab director must monitor the quality of test results received from these outside sources. A mechanism must be in place to evaluate immunohistochemical results that are inconsistent with clinicopathologic studies. This evaluation should be performed and recorded by a laboratory physician. 

O Brien PC, Pleming TR. A multiple testing procedure for clinical trials. Biometrics 1979 35 549-56. 

Westfall, P. H. and Young, S. S. (1993) Resampling-based multiple testing. Wiley, New York, NY. 

Also, the analysis plan should identify the statistical methods that will be used and how hypotheses will be tested (e.g., a p value cutoff or a confidence interval for the difference that excludes 0). And the plan should prespecify whether interim analyses are planned, indicate how issues of multiple testing will be addressed, and predefine any subgroup analyses that will be conducted. Finally, the analysis plan should include the results of power and sample size calculations. 

In some circumstances it may be necessary to combine several events/endpoints to produce a combined or composite endpoint. The main purpose in doing this is to again avoid multiple testing and more will be said about this in Chapter 10. In addition, combining endpoints/events will increase the absolute numbers of events observed and this can increase sensitivity for the detection of treatment effects. 

For the first sample in the computer simulation the 99 per cent confidence interval is (78.64, 81.80). This is a wider interval than the 95 per cent interval the more confidence we require the more we have to hedge our bets. It is fairly standard to use 95 per cent confidence intervals and this links with the conventional use of 0.05 (or 5 per cent) for the cut-off for statistical significance. Under some circumstances we also use 90 per cent confidence intervals and we will mention one such situation later. In multiple testing it is also sometimes the case that we use confidence coefficients larger than 95 per cent, again we will discuss the circumstances where this might happen in a later chapter. 

The problem with this so-called multiplicity or multiple testing arises when we make a claim on the basis of a positive result which has been generated simply because we have undertake lots of comparisons. Inflation of the type I error rate in this way is of great concern to the regulatory authorities they do not want to be registering treatments that do not work. It is necessary therefore to control this inflation. The majority of this chapter is concerned with ways in which the potential problem can be controlled, but firstly we will explore ways in which it can arise. 

As mentioned in the previous section, multiplicity can lead to adjustment of the significance level. There are, however, some situations when adjustment is not needed although these situations tend to have restrictions in other ways. We will focus this discussion in relation to multiple primary endpoints and in subsequent sections use similar arguments to deal with other aspects of multiple testing. 

---