Analysis results, total variance

The exponentially decrease of the total variance with increasing sample mass is shown in Fig. 2.7. It can be seen that the uncertainty of sampling, s2samp, decreases and becomes statistically insignificant when the sample amount m exceeds the critical sample mass. Instead of mcnt the proportional critical sample volume vcrit may also be considered, represented, e.g. by a critical microprobe diameter dcnt. Results of homogeneity investigations of alloys, ores, and lamellar eutectics by EPMA (Electron Microprobe Analysis), which correspond to the curve of Fig. 2.7, have been presented by Danzer and Kuchler [1977]. 

Musumarra et al. [43] identified miconazole and other drugs by principal components analysis of standardized thin-layer chromatographic data in four eluent systems. The eluents, ethylacetate methanol 30% ammonium hydroxide (85 10 15), cyclohexane-toluene-diethylamine (65 25 10), ethylacetate chloroform (50 50), and acetone with the plates dipped in potassium hydroxide solution, provided a two-component model that accounts for 73% of the total variance. The scores plot allowed the restriction of the range of inquiry to a few candidates. This result is of great practical significance in analytical toxicology, especially when account is taken of the cost, the time, the analytical instrumentation and the simplicity of the calculations required by the method. 

The total variance in the final result (,v2ita ) is made up of two contributions. One is from variation in the composition of the laboratory samples due to the nature of the bulk material and the sampling procedures used ( ample). The other (Tanalysis) is from the analysis of the sample carried out in the laboratory ... 

The analytical variance can be determined by carrying out replicate analysis of samples that are known to be homogeneous. You can then determine the total variance. To do this, take a minimum of seven laboratory samples and analyse each of them (note that Sample characterizes the uncertainty associated with producing the laboratory sample, whereas sanalysis w h take into account any sample treatment required in the laboratory to obtain the test sample). Calculate the variance of the results obtained. This represents stQtal as it includes the variation in results due to the analytical process, plus any additional variation due to the sampling procedures used to produce the laboratory samples and the distribution of the analyte in the bulk material. 

FIGURE i 6 Diagram indicating that the total variance in the analysis results equals the sum of the method variance and the process variance. A capable measurement system has a method variance that is less than 30% of the design width (difference between upper specification level and lower specification level). The production process is considered to be under full control when the average assay value is centered at Six Sigma values away from the lower and upper specification levels. 

If you are in charge of quality control laboratories in manufacturing companies, it is important to distinguish between the variability of a product and the variability of the analysis. When analyzing tablets on a pharmaceutical production line, variability in the results of an analysis has two contributions from the product itself and from the analytical procedure. Your bosses are interested in the former, and you, the analyst, must understand and control the latter. It is usually desired to use methods of analysis for which the repeatability is much less than the variability of the product, in which case the measured standard deviation can be ascribed entirely to the product. Otherwise, analysis of variance can be used to split the total variance of duplicate results into its components (chapter 2). In the discussion that follows, the emphasis is on measurement variability, but the principle is the same, and the equations and methods can be used directly to obtain information about the product or manufacturing process. 

Segregation variance of increment collection variance caused by nonrandom distribution of ash content or other constituent in the lot (ASTM D-2234). Total variance overall variance resulting from collecting single increments and including division and analysis of the single increments (ASTM D-2013 ASTM D-2234). 

Analysis of variance is a procedure by which the total variance is divided into sources of variations. Depending on the experiment design done, it is possible to separate from the total variance a different number of sources of variations. However, no matter how many sources of variations were selected, they all refer both to those that occur under the influence of systematic variations and to the error resulting from random variations. The aim of applying the analysis of variance method is to answer the question is the difference between the obtained response means for the tested factors a result of the influence of tested factors or has it occurred randomly. 

The exclusive consideration of common factors seems to be promising, especially for such environmental analytical problems, as is shown by the variance splitting of the investigated data material (Tab. 7-2). Errors in the analytical process and feature-specific variances can be separated from the common reduced solution by means of estimation of the communalities. This shows the advantage of the application of FA, rather than principal components analysis, for such data structures. Because the total variance of the data sets has been investigated by principal components analysis, it is difficult to separate specific factors from common factors. Interpretation with regard to environmental analytical problems is, therefore at the very least rendered more difficult, if not even falsified for those analytical results which are relatively strongly affected by errors. 

The G-BASE project has used several statistical packages to perform this nested ANOVA analysis (e.g., Minitab and SAS). It currently uses an MS Excel procedure with a macro based on the equations described by Sinclair (1983) in which the ANOVA is performed on results converted to logio (Johnson, 2002). Ramsey et al. (1992) suggest that the combined analytical and sampling variance should not exceed 20% of the total variance with the analytical variance ideally being <4%. 

As an example, Tables 13.14 and 13.15 show the results of applying principal components analysis to the 10 volatile compounds (methanol, 1-propanol, isobutanol, 2-and 3-methyl-1-butanol, 1-hexanol, cw-3-hexen-l-ol, hexanoic acid, octanoic acid, decanoic acid and ethyl octanoate) analyzed in 16 varietal wines (Pozo-Bay6n et al. 2001), obtained with the STATISTICA program Factor Analysis procedure in Multivariate Exploratory Techniques module, and using Principal Components as Extraction method). The results include the factor loadings matrix for the two first principal components selected q = 2), which explains 70.1% of the total variance (Table 13.14). The first principal component is strongly correlated with d.y-3-hexen-l-ol (-0.888), 1-hexanol (-0.885), 1-propanol (0.870), and... 

Due to the large number of samples in which Fe and Cr were below detection limits it was not possible to include them in the general R-mode analysis. However, a smaller suite of 20 crude oils was selected in which V, Fe, Ni, Co, Mn and Cr were all found at concentrations above detection limits. A separate R-mode analysis was run on these 20 samples and the results of the biquartimin solution are shown in Figure 6.3. The four factors extracted account for nearly 95% of the total variance among the variables. The correlation coefficient between Fe and Co in 38 crude oils in which both elements were above detection limits was 0.42, and the correlation coefficient between Fe and Mn in 36 crude oils in which both elements were above detection limits was 0.18. From these various facts we can state, with some confidence, that Fe is unrelated to either Co or Mn in... 

A principal components analysis was done on a standardised IR emission sp>ectrum data set including all 420 data samples. Results showed that 86.7% of the total variance of the wavelength variables could be explained by the first two principal components. The map of squared correlation coefficients in figure 3 confirms this result. 

The variables (wavelengths) associated with the IR emission spectra were highly correlated. Principal components analysis (PCA), linear and nonlinear PLS showed that at least 86% of the total variance could be explained by the two primary latent dimensions. The forward and reverse modelling results showed that dimensional reduction with a linear model (PLS) produced better models than a nonlinear model (multilayer perceptron neural network trained with the back propagation algorithm) without dimensional reduction. 

Factor analysis and Varimax rotation results revealed that twelve items related to job performance could be clustered into three factors, which together accounted for 65.56 percent of the total variances. These three factors were called achievement at woik, job abihty, and good peer relationship, and their Cronbach alphas were. 90,. 71, and. 63, accordingly. 

The mathematical operation called analysis of variance (ANOVA) provides a mathematical proof that the total variance of the error of a result is the sum of the variances due to each source of error. 

---