Lesson 2: Errors

Error

Denotes uncertainty and difference between two natural values.

less representative

Result of a greater error

Types of errors

1. Systematic error

2. Random error

3. Gross error

Systematic error

Result of variation of a measurement of the true value. Affects accuracy and also known as determinate error. Can be detected and eliminated.

• Instrumental

• Method

• Personal

Instrumental - systematic error

Anything that is used and applied in the analysis. Calibration eliminates most systematic errors.

Calibration

Eliminates most systematic errors.

Method - systematic error

Anything related to the method. May be the behavior of the reagent. Is the MOST DIFFICULT to determine and correct.

Personal - systematic error

Something that is inherent on the analyst. Also known as operative errors and may be skill based or a mistake.

CRM

Certified reference materials (uncertainty). Controls or standards do check the quality and metrological traceability of products.

Independent analysis

Asks other chemist to analyze.

Round robin experiment

Different people at different laboratories analyze identical sample.

Blank determination

Contains the reagents and solvents use din determination but no analyte. May be from sample size and can replicate.

• method

• Reagent

• Field

Method - blank determination

Taken through all steps

Reagent - blank determination

Not subjected to all sample

Field - blank determination

Exposed to the sampling site.

Random error

Simply there. No identified cause and gives the ultimate limitation. Affects precision and is indeterminate but can be minimized.

Gross error

Product of human errors. Often large and called outliers, has no alternatives.

Statistics

Way to test or identify outliers

False negative

Indicated as correct but is actually incorrect

False positive

Indicated as incorrect but is actually correct.

Outlier

An observation that lies outside the overall pattern.

Grubb’s test

Compares the deviation of the suspect value from the sample mean. ISO (International Organization for Standardization) recommended.

Dixon’s test

Also known as Q-test. Assess a suspect of measurement by comparison of data. If level of confidence is not given, assume 95%. Outliers are not included in the mean calculation.

Accuracy

Indicates the closeness of measurement to the true or accepted value. Has mean and median under.

Mean

Arithmetic average

Median

Middle value when arranged

Absolute error - expressing accuracy

Difference between measured value and true value. E= xi - xt

Relative error - expressing accuracy

Absolute error divided by the true value. Er = ((xi-xt)/xt) •100%

Precision

Arrangement among several results obtained in the same way.

Types of precision

• Instrument

• Intra-assay

• Indeterminate

• Interlaboratory

Instrument - precision

Injection precision. Reproducibility observed.

Intra-assay - precision

Analyzing aliquots several times a day by the same analyst.

Indeterminate - precision

Ruggedness, same with intra-assay but different people and same laboratory.

Interlaboratory - precision

Most general measure of reproducibility.

Bigger random error

Result when variability is increased

Repeatability

Same all

Reproducibility

Nearness of results

Standard deviation - expressing precision

Measure spread or distribution around the mean of a data set.

Relative standard deviation (RSD) - expressing precision

Standard deviation by mean, often used to compare precision of results.

Coefficient of variation - expressing precision

Relative standard deviation expressed as percentage.

Variance - expressing precision

Square of standard deviation. Useful measure of the scatter particularly for random errors.

Standard deviation of the mean - expressing precision

Referred to as the standard error.

Uncertainty

Measurement without level of certainty.

Absolute uncertainty

Margin of uncertainty associated with a measurement. Nakalimutan q formula omai

Relative uncertainty

Comparing the size of the absolute uncertainty to the size of its associated movement. = (absolute uncertainty/magnitude of measurement) • 100%