ANALYSIS OF VALIDATION INSPECTION DATA 



Data from the two on-site inspections were analyzed to determine 

 the required versus actual sample size as well as the suitability of the 

 data to make predictions of structural capacity and maintenance require- 

 ments. A complete listing of the data from these two inspections is 

 presented in Appendixes C and D. 



The following sample statistics were calculated from the data 

 obtained during the inspection of the piers at St. Helena: 



1 . Pile Measurements (Variable Sampling) 



• Mean diameters 



• Coefficient of variation for cross-sectional area 



• Standard deviation 



• Standard error of the mean 



• Confidence interval at 95 percent 



2. Proportions (Attribute Sampling) 



• Standard error of proportion 



• Confidence intervals 



• Maintenance criteria 



• Good condition 



• Wrap 



• Concrete jacket 



• Replacement 



• Proportion of borer activity 



3. Other Statistical Measures 



• Chi-square "goodness of fit" to a normal distribution 



• Multiple regression analysis to estimate the time required 

 to move from bent to bent, time to perform different 

 inspection levels, and time for other related activities 



Table 4 summarizes the analysis of the data collected from the 

 variable and proportional sampling plans. Each pier has two columns of 

 data. The first column is data from the sample set of 30. The second 

 column is data based upon the larger sample size required by the FPO-1 

 contract. These columns are labeled NCEL and FP01, respectively. The 

 estimates of the population parameters differ very little between the 

 two sample sizes for measurement statistics based upon mean pile dia- 

 meter. The major difference between the two columns is the predicted 

 proportion of borer activity and the width of the confidence interval 

 for mean diameters. Borer activity is an attribute of the pier popula- 

 tion; hence, it requires a larger number of samples to obtain an esti- 

 mate. Pier 11 was the only pier which was specifically sampled for this 

 attribute. 



32 



