Case Study: Precision Comparison

Download: ArgentometricChloride_PrecisionComparison_init.xlsx

Introduction

In this screencast we will apply what we have learned to compare the measurement precision of the two endpoint detection methods of the argentometric titration data. In particular we are interested in comparing two types of precision:

• the variability of measurements of a single student using both techniques (within-student precision)
• the variability of measurements across students (between-student precision)

We will use a number of tools in our toolbox in this analysis.

Learning Goals

Nothing new is introduced here but the case study should give you practice in applying what you have learned. The non-parametric nature of bootstrapping and the greater versatility in choosing comparison statistics when bootstrapping are highlighted.

Tasks

• Create some named filters: error_faj, error_pot, n_faj, n_pot, s_faj, s_pot. Also create fajans and pot data block filters. These will be used as inputs involving ANOVA1S.
• Parametric comparison of within-student precision: pool the standard deviations and compare with an F-test
• Perform a non-parametric F-test by calculating the same CI with a bootstrap. Then use a robust alternative statistic to compare within-group precision.
• Set up outlier removals from Fajans and Potentiometric data blocks using MAD criteria applied to the errors
• Parametric comparison of between-student precision: perform ANOVA on the data and apply an F-test to compare the $s_B$ estimates. Do this for both raw data and data with outliers removed.
• Perform non-parametric versions of those same tests using bootstrapping

By the end of the screencast your two worksheets should look like the images below.