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.
60 Procedures and Guidelines for Validating Contractor Test Data 2. REFERENCED DOCUMENTS 2.1 AASHTO Standards and Publications ï§ R 9 Standard Recommended Practice for Acceptance Sampling Plans for Highway Construction - Verification, Statistical Tests, Process Verification Practices, F- and t- tests, Dispute Resolution, Dealing with Outlier Observations, Resampling and Retesting. ï§ R 10 Standard Recommended Practice Definitions of Terms Related to Quality and Statistics as Used in Highway Construction - Multiple Related Definitions. ï§ R 18 Standard Recommended Practice for Establishing and Implementing a Quality Management System for Construction Material Testing Labs - Dispute Resolution, Appeals. ï§ R 42 Standard Recommended Practice for Development of a Quality Assurance Plan for Hot Mix Asphalt - Dispute Resolution. ï§ R 44 Standard Recommended Practice for Independent Assurance (IA) - Comparison of Test Results, Split Samples, Disadvantage of D2S, Use of Paired t-test. 2.2 ASTM Standard. ï§ E 178 Standard Practice for Dealing with Outlying Observations. 2.3 Others. ï§ NCHRP Research Report 946: Procedures and Guidelines for Validating Contractor Test Data. ï§ FHWA-RD-02-095 Optimal Procedures for Quality Assurance Specifications - Quality Assurance Specifications and Risk Assessment. ï§ FHWA-HRT-04-046 Evaluation Procedures for Quality Assurance Specifications - Quality Assurance Considerations. ï§ State of Maine Department of Transportation Standard Specifications, 2014. 3. TERMINOLOGY 3.1 acceptance planâalso called âacceptance sampling planâ or âstatistical acceptance planâ. An agreed-upon process for evaluating the acceptability of a lot of material. It includes defining lot size, frequency of sampling and testing, sampling methods, quality measures, acceptance limit(s), validation procedures, and pay adjustment provisions.
Proposed Practice for Validating Contractor Test Data 61 3.2 acceptance sampling and testingâsampling and testing performed to evaluate acceptability of the final product. 3.3 acceptance quality characteristic (AQC)âa quality characteristic that is measured and used to determine acceptability of a material or a constructed pay item. 3.4 accredited laboratoriesâlaboratories that are recognized by a formal accreditation body as meeting quality system requirements including demonstrated competence to perform standard test procedures. 3.5 accuracyâthe degree to which a measurement, or the mean of a distribution of measurements, tends to coincide with the true population mean. When the true population mean is unknown, the degree of agreement between the observed measurements and an accepted reference value may be used to quantify the accuracy of the measurements. 3.6 averageâa measure of central value that usually refers to the arithmetic mean ( ). The mean, median, and mode are equal for a normal distribution. As a distribution becomes more skewed, the mean, median, and mode will differ more and more. 3.7 certified technicianâa technician certified by some organization as being proficient in performing certain duties. A certified technician is considered to be qualified. A qualified technician may or may not be certified. See qualified technician. 3.8 characteristicâa measurable property of a material, product, or item of construction. 3.9 coefficient of variation (COV)âthe ratio of the standard deviation to the mean expressed as a percentage. It provides a measure of dispersion or spread relative to the mean. 3.10 composite pay factorâalso called âcombined pay factorâ or âoverall pay factor.â A multiplication factor, often expressed as a percentage that considers two or more quality characteristics and is used to determine the Contractorâs final payment for a unit of work. 3.11 confidence intervalâan interval in which the estimated parameter will lie for a predetermined probability (called the âconfidence levelâ). 3.12 confidence levelâif a large number of confidence intervals are constructed, the proportion of time that the estimated parameter will lie within the interval. A confidence level is usually expressed as a percentage, typically ranging from 90 to 99 percent. Confidence level = 1 â level of significance, Î±. See level of significance. 3.13 confidence limitsâthe end points of a confidence interval. 3.14 conflict resolutionâsee dispute resolution. 3.15 degrees of freedomâthe number of independent observations in a data set minus the number of population parameters to be estimated from the data set.
62 Procedures and Guidelines for Validating Contractor Test Data 3.16 dispute resolutionâalso called âconflict resolutionâ and âreferee testing.â A procedure used to decide in favor of one side or the other when discrepancies occur between a SHAâs results and Contractorâs results that are of sufficient magnitude to impact acceptance/payment. The procedure may include the testing of independent or split samples as an initial step and third-party arbitration as a final step. 3.17 F-test (F-statistic)âa hypothesis test that involves the comparison of the variances of two sets of data. The null hypothesis is that variance of one population equals the variance of another population. 3.18 Independent Assurance (IA)âactivities that provide an independent verification of the reliability of the acceptance data obtained by the SHA and the Contractor. The results of IA testing or inspection are used for quality system management and are not as a basis of acceptance. 3.19 independent sampleâa sample taken without regard to any other sample that may also have been taken to represent the material in question. An independent sample is sometimes taken to verify an acceptance decision because unlike those from split samples, these data contain independent information reflecting all sources of variability, i.e., sampling, testing, materials, and construction. 3.20 independent t-testâa hypothesis test that involves the comparison of the means of two independently obtained sets of data. The null hypothesis for this test is that the mean of one normal population is equal to the mean of another normal population. 3.21 level of significance (Î±)âthe probability of rejecting the null hypothesis when it is true. The level of significance is also referred to as the probability of a Type I error. 3.22 lotâa specific quantity of material from a single source that is assumed to be produced or placed by the same controlled process. 3.23 meanâthe arithmetic average. â â denotes the arithmetic average of a sample. â â denotes the arithmetic average of a population. 3.24 medianâthe midpoint of a set of values that have been ordered from the smallest to the largest, or the largest to the smallest. There are as many values above the median as below it in the data array. 3.25 Monte Carlo simulationâa simulation technique (usually performed by a computer) that uses random numbers to sample from probability distributions to produce hundreds or thousands of scenarios (called âiterations,â âtrials,â or ârunsâ). A complete Monte Carlo simulation thus uses each result from each individual iteration. 3.26 normalityâthe state of being normally distributed or having a normal distribution. 3.27 paired t-testâalso called ât-test for paired measurements (samples).â A statistical test used to compare the test values for multiple pairs of data. The test uses the differences
Proposed Practice for Validating Contractor Test Data 63 between the pairs of test values and determines whether the average difference is statistically significant from zero. 3.28 pay adjustmentâthe actual amount, either in dollars or in dollars per area/weight/volume, that is to be added to or subtracted from the Contractorâs bid price or unit bid price. 3.29 pay adjustment schedule (for quality)âalso called âprice adjustment scheduleâ or âadjusted pay schedule.â A pre-established schedule, in either tabular or equation form, for assigning pay factors associated with estimated quality levels of a given characteristic. Pay factors are usually expressed as percentages of the Contractorâs bid price per unit of work but may also be given as direct dollar amounts. 3.30 pay adjustment system (also called âprice adjustment systemâ or âadjusted pay determine the overall pay factor for a submitted lot of material or construction. 3.31 pay factor (PF)âa factor based on a single AQC, often expressed as a percentage that is multiplied by the Contractorâs bid price for an item to determine the payment for a unit of work. 3.32 percent defective (PD)âthe percentage of the lot falling outside specification limits. [PD may refer to either the population value or the sample estimate of the population value.] 3.33 percent within limits (PWL)âthe percentage of the lot falling above a lower specification limit, beneath an upper specification limit, or between upper and lower specification limits. PWL = 100 â PD. PWL may refer to either the population value or the sample estimate of the population value. The term âpercent conformingâ is not a synonym for PWL. 3.34 qualified technicianâa technician who has been determined to be qualified (i.e., meeting some minimum standard) to perform specific duties. A qualified technician may or may not be certified. SHAs recognize qualification based on proficiency demonstrated by combinations of training, written and/or practical exams administered by SHAs, regional or national programs. See certified technician. 3.35 quality characteristicâa product attribute that is measured either for quality control (QC) purposes or for conformance with acceptance requirements. Quality characteristics are specific material properties evaluated by QC and acceptance sampling and testing. Quality characteristics used in acceptance specifications are typically selected because they: (a) relate to initial and long-term performance; (b) are quantifiable or measurable; and (c) can be measured with good repeatability. 3.36 resamplingâthe process of obtaining a new sample, if warranted, to replace the original sample for an individual sublot. [See retesting.] systemâ)âAll pay adjustment schedules, equations, and/or algorithms used to
64 Procedures and Guidelines for Validating Contractor Test Data 3.37 retestingâthe process of performing another test to confirm the initial test. Retesting is performed using a second test portion or a specimen prepared from the same original sample. [See resampling.] 3.38 sampleâalso called materials sample when it provides (1) a small physical quantity of material or a measurement obtained in some manner so that the portion is representative of the whole, or (2) a quantity of material fabricated in a lab on which future tests can be run. Also called statistical sample when (1) all of the individual samples obtained from a lot provide information that may be used to quantify the quality of the entire lot, or (2) the integer number of random material samples obtained from a sublot or lot. The context in which the word âsampleâ is used determines its meaning. For example, âobtain a sample hereâ could mean either obtain a physical quantity of material at this location or take a test (obtain a measurement) at this location, while âthe sample size equaled 9â would mean that a total of 9 individual material samples were obtained in a random manner and thus comprised a statistical sample of size = 9. 3.39 sample size ( ) (referring to a statistical sample)âthe number of random individual observations (i.e., data points, test results, measurements) under consideration or comprising a sublot or lot. The sample size (i.e., number of random observations) has important implications on how well other statistical measures represent the population. The greater the sample size ( ), the greater the degree of confidence in inferences made about the population. 3.40 split sampleâa sample that has been divided into two or more portions representing the same material. Split samples may be used for comparison of results from two or more operators or laboratories. The variability calculated from differences in split sample test results is attributed solely to testing variability. 3.41 standard deviation ( )âa biased measure of the dispersion of a series of results around their mean, expressed as the square root of the quantity obtained by summing the squares of the deviations from the mean of the results and dividing by the number of observations minus one. 3.42 standard normal distributionâa mathematical construct of a continuous probability distribution having a symmetrical, asymptotic bell-shaped curve that is fully defined by Âµ and Ï, where Âµ = 0 and Ï = 1, and for which the mean, median, and mode are all equal. All distribution curves having a similar shape may be modeled by the standard normal distribution via the z-score transformation. 3.43 statisticsâa branch of mathematics that deals with the collection, analysis, interpretation, and presentation of masses of numerical data. Statistics use mathematical theories of probability to impose order and regularity on the analysis of data.