BAD DATA Sun e e Sun Overview Bad

“BAD” DATA Sun e e Sun

Overview Bad Data ä Learning from unexpected results ä Isotherm Analysis ä

Sources of “Bad” Data ä Error in preparation of samples ä ä ä ä mass or volume measurement error contamination improper storage sample substitution sample loss samples with high heterogeneity Apparatus failures ä ä ä leaks incompatible materials inadequate control of an important parameter

Instrument Errors detector malfunction ä below detection limit or above maximum ä interference ä software (instrument or computer) ä hardware (analog to digital converter, power supply, . . . ) ä calibration ä

More sources of “Bad Data” ä Error in data analysis ä ä numerical error (data entry) units (classic errors of factors of 10 and factors of 1000) incorrectly applied theory Error in theory

Bad Data aren’t Bad! ä “Bad” data usually means the results were unexpected ä perhaps unorthodox! ä ä ä Copernicus “Concerning the Revolutions of the Celestial Bodies” 1543 Papal Index of forbidden books until 1835 Bad data for 292 years! ___________ Data do not lie! Data always mean something If you ignore data that you don’t understand you are missing an opportunity to learn

Unexpected Results ä Lack of repeatability (poor precision) ä ä ä scatter for all data outlier systemic error

Unexpected Results ä Inconsistent with theory ä mass balances indicate loss or gain of mass ä inconsistent with previous results ä some “theories” are only hypotheses Sun e e Sun

Responses to Unexpected Results ä ä ä Determine accuracy of technique by analyzing known samples Determine precision of technique by analyzing replicates Evaluate propagation of errors through analysis ä ä are you trying to measure the difference between two large numbers? is the precision of the measurement similar to the magnitude of the estimate? Are you not controlling an important parameter? Is the parameter that you are studying insignificant?

Isotherm Analysis Pointers ä Units ä Express mass of VOC in grams ä Express concentrations as g per m. L ä Remember GC injection volume was 0. 1 m. L Use names to keep track of parameters in spreadsheet ä Build sheet from left to right ä

More Pointers Soil density = 1. 6 g/m. L ä Soil moisture content is 10. 7% ä Soil mass was close to 20 g ä Analyze data sets as sets ä You will get 6 estimates for each parameter. ä Where do all these parameters come from? ä

Proposal for the VOC isotherm lab Change Full Report to Spreadsheet ä Analyze all 6 sets of data (isotherm data summary. xls) ä ä See which parameters are stable ä Calculate all parameters independently (scenarios for each data set? ) ä Extend due date until Friday of next week