Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Proposed Practice for Validating Contractor Test Data 85 assessment. In the case that CVL 1 is not validated then lot 1 is considered not validated, and the next CVL should consider results from lot 2, lot 3, and lot 4. A.2.7 Restarting the CVL â When a CVL is not validated through Primary Validation, Secondary Validation, and Dispute Resolution, the next CVL must start with a new lot. For example, if CVL 1 is not validated, then CVL 2 would start with lot 4 and include lots 5 and 6. ANNEX B â PROCEDURE FOR DETERMINING STATISTICAL OUTLIERS SHA and Contractor data should be evaluated for potential outlier observation using an established valid method such as ASTM E 178, Standard Practice for Dealing with Outlying Observations. This annex describes the procedure developed by the Maine Department of Transportation basedon ASTM E 178. Scope: This procedure deals with identifying outlying observations in sets of at least three results. Definition: An outlying observation, or âoutlier,â is one that appears to deviate markedly from other test values from the same population or lot. When considering outliers, two conditions may exist: (1) the value may be an extreme value of the population or excessive variability of the population; in either case the value should not be discarded or (2) it may be the result of gross deviation from prescribed sampling and or test procedures, errors in calculations, or errors in recording of numerical values; in this case the value should be discarded. This procedure provides steps to determine if the value is not an outlier and should be retained or the value is an outlier and should be discarded. Procedure: Step 1. Determine if a physical reason is known for the outlier. Possible reasons include sample mishandling prior to testing, test equipment malfunctioning, or a computation error. If a computation error is found, the value may be corrected and used as the test result. Step 2. If no reason is found for the outlier, the following calculation procedure should be used. Calculation Procedure - This procedure is based on a âtwo-tail t-testâ with level of significance (Î± ) of 0.05. The two-tail test means that the outlier may be either on the high or low side of the average. The level of significance means that there is a 5% chance that the identified value is not an outlier.
86 Procedures and Guidelines for Validating Contractor Test Data 1. Calculate the sample average ( ) and standard deviation ( ) of the results in the sample set (e.g., lot). 2. Find the critical t-value âtcritâ from Table B.1 using the total number of samples ( ) in the sample set. 3. Determine D, the total allowable deviation on either side of , by multiplying âtcritâ by . 4. Establish values for MAX and MIN by adding and subtracting D to and from . 5. Any result greater than MAX or less than MIN is determined to be an outlier. Example 1. The following eight (8) density test results were obtained. Is one an outlier? Sample Relative Density, % 1 89.5 2 94.0 3 93.3 4 93.3 5 92.8 6 92.6 7 93.5 8 94.3 Calculations n = 8 = 92.9 = 1.49 tcrit = 2.126 (from Table B.1) D = tcrit Ã = 2.126 Ã 1.49 = 3.17 MAX = + D = 92.9 + 3.17 = 96.07% MIN = â D = 92.9 â 3.17 = 89.73% Because Sample 1 density is 89.5% which is less than the MIN of 89.73%, this sample result is identified as an outlier and should be investigated.
Proposed Practice for Validating Contractor Test Data 87 Example 2. The following three air-void results were obtained. Is one an outlier? Sample Air Voids, % 1 5.2 2 5.2 3 6.6 Calculations n = 3 = 5.7 = 0.81 tcrit = 1.155 (from Table B.1) D = tcrit Ã = 1.155 Ã 0.81 = 0.94 MAX = + D = 5.7 + 0.94 = 6.64% MIN = - D = 5.7 â 0.94 = 4.76% Because all results lie between the MAX of 6.64% and the MIN of 4.76%; none of the sample results is an outlier; all results should be used in further calculations.
88 Procedures and Guidelines for Validating Contractor Test Data TABLE B.1 tcrit values for a 5% Significance Level * tcrit 3 1.155 4 1.481 5 1.715 6 1.887 7 2.020 8 2.126 9 2.215 10 2.290 11 2.355 12 2.412 13 2.462 14 2.507 15 2.549 16 2.585 17 2.620 18 2.651 19 2.681 20 2.709 21 2.733 22 2.758 23 2.781 24 2.802 25 2.822 26 2.841 27 2.859 28 2.876 29 2.893 30 2.908 *Table adopted from the procedure developed by the Maine Department of Transportation.