Application of Statistical Analysis to Analytical Data

Study of Accuracy in Chemical Analysis Using Linear Calibration Curves

“…Once this has been done, the precision and accuracy of the method can be effectively studied by analyzing a series of prepared samples covering the range of concentrations over which the method is applicable. This procedure, which involves the statistical theory of fitting straight lines based on the method of least squares, has been described by Youden (18,19) and applied by Linnig, Mandel, and Peterson (9) and by Lark (8). Essentially, the slope of the fitted straight line can be compared to a value based on stoichiometric or other theoretical considerations; the intercept, to "blank" determinations;…”

Quantitative Determination of Buckwheat Flavonoid Substances by Direct Absorptiometry of Paper Chromatograms.

“…Lark (9) has shown that it is instructive to consider the regression of total error of analytical determination (z=y-x) on the actual amount; the total error, z, includes systematic error and random error. One hypothesis which might be assumed is that systematic error is linearly related to the actual amount.…”

“…Alternatively, the hypothesis may be adlopted that the systematic error is constant (z = a'). In this event the best estimate of the error is the mean of the z values (9). Such mean errors were calculated for the rutin recovery data, and the hypothesis that they are equal to zero was tested through use of the statistic t. For both recovery experiments this hypothesis was accepted at the 5 % level.…”

“…A good idea of the precision of a method may be obtained by inverting the regression equation of y on x and calculating confidence intervals for predictions of actual amounts, x, in new samples (9). Results of such an analysis of the data for pure rutin are presented in …”

Evaluation of Lipophilicity of Piperazine Derivatives by Thin Layer Chromatography and Principal Component Analysis

“…This very interesting and useful conclusion was strongly supported in recent papers. 18,19 Taking into account these results and other disadvantages of the regression method applied in this scope (logarithmic scale, extrapolation) we proposed the use of scores obtained applying principal component analysis as a support for the lipophilicity scale and the "lipophilicity chart" as a new concept. [20][21][22] The purpose of this paper is to investigate the feasibility of the scores, obtained by principal component analysis using RPTLC retention data, as a measure of lipophilicity in correlation with partition coefficient (log P) of some anticonvulsant N-substituted amides of α-piperazine-γ-hydroxybutyric acid.…”