Qonvert

Defining Accuracy and Precision

Accuracy refers to how close a measured value is to the true or accepted value. If you weigh an object that is exactly 500 g and your scale reads 499 g, that is a fairly accurate measurement. If it reads 350 g, it is inaccurate — far from the truth.

Precision, on the other hand, refers to how consistent or repeatable a series of measurements is. If you weigh the same 500 g object five times and always get 478 g, your measurements are highly precise — they agree with each other — but they are not accurate because they consistently miss the true value.

The Bullseye Analogy

The classic way to visualize accuracy vs. precision is with a target and arrows:

• Accurate and precise: All arrows cluster tightly around the bullseye.
• Precise but not accurate: All arrows cluster tightly together, but away from the bullseye (a systematic error).
• Accurate but not precise: Arrows are scattered around the bullseye — some hit it, others don't (random errors cancel out on average).
• Neither accurate nor precise: Arrows are scattered randomly and far from the bullseye.

This analogy makes it immediately clear that precision alone is not enough — you can be very consistent at being consistently wrong.

Types of Measurement Error

Understanding why measurements miss the true value requires knowing the two fundamental categories of error:

Systematic Error

Systematic errors push every measurement in the same direction — always too high or always too low. They are reproducible and predictable. Common sources include:

• A miscalibrated instrument (e.g., a scale that always reads 5 g too heavy)
• Environmental factors affecting the instrument consistently (e.g., temperature causing a thermometer to expand its fluid too much)
• Parallax error when a reader always views a gauge from the same angle
• Zero error — where an instrument doesn't start at zero

Systematic errors affect accuracy. They can be corrected once identified — through calibration, better technique, or adjusting the instrument.

Random Error

Random errors are unpredictable fluctuations that vary from measurement to measurement with no consistent pattern. Sources include:

• Electrical noise in sensitive instruments
• Slight variations in how a person reads a scale or stopwatch
• Environmental vibrations or air currents affecting a balance
• Inherent variability in the quantity being measured (e.g., the roughness of a surface)

Random errors affect precision. They cannot be eliminated, but their impact can be reduced by taking multiple measurements and averaging them.

Significant Figures and Rounding

Significant figures (sig figs) are the meaningful digits in a measurement — the ones that carry real information about precision. The number of significant figures you report communicates how precise your measurement is.

Rules for Counting Significant Figures

• All non-zero digits are significant: 4.56 has 3 sig figs.
• Zeros between non-zero digits are significant: 4.056 has 4 sig figs.
• Leading zeros are never significant: 0.0045 has 2 sig figs.
• Trailing zeros after a decimal point are significant: 4.500 has 4 sig figs.
• Trailing zeros in a whole number are ambiguous without a decimal point: 4500 may have 2, 3, or 4 sig figs.

Reporting too many significant figures implies a false level of precision. If your ruler can only measure to the nearest millimeter, reporting a length as 145.37 mm is dishonest — you should report 145 mm.

Calculating and Expressing Error

Once you have a measurement and know the true value, you can quantify how far off you were. The most common way is percentage error:

For example, if you measure the boiling point of water at 99.2°C when the true value is 100.0°C:

A 0.8% error is reasonably small. In precision engineering, errors below 0.01% may be required; in cooking, 5% might be perfectly acceptable.

Expressing Uncertainty

Professional measurements always include an uncertainty — a statement of how much the true value might differ from the reported value. This is written as:

This tells the reader that the true value likely falls between 45.1 and 45.5 cm. Uncertainty can be estimated from instrument resolution, repeated measurements, or stated manufacturer specifications.

Calibration: Restoring Accuracy

Calibration is the process of comparing an instrument's output to a known standard and adjusting it so that its readings are accurate. It is the primary tool for fighting systematic error.

For example, a laboratory balance is calibrated using certified reference weights. A thermometer might be calibrated at the freezing and boiling points of water (0°C and 100°C at sea level). Once calibrated, any offset is corrected — either electronically, mechanically, or by applying a correction factor to readings.

Calibration establishes a chain of traceability: your instrument is compared to a local standard, which was compared to a national standard, which was compared to international standards maintained by laboratories like NIST (USA) or the BIPM (international). This chain ensures your measurement ultimately connects to the SI system.

Why This Matters: Real-World Consequences

The difference between accuracy and precision is not merely academic. In many fields, getting it wrong has serious consequences:

Medicine and Pharmacy

Drug doses must be measured accurately. A patient prescribed 10 mg of a medication cannot receive 8 mg or 12 mg consistently — underdosing is ineffective, overdosing can be fatal. Medical balances are calibrated daily and checked with reference weights before use.

Engineering and Manufacturing

Machine parts must fit together with tolerances often measured in micrometers (millionths of a meter). A car engine piston must be the right diameter — precise and accurate — or the engine won't run. Imprecise parts cause vibration, wear, and failure.

Science and Research

Experimental results that are imprecise cannot be replicated. Science depends on reproducibility. A measurement that varies wildly (low precision) cannot support a conclusion, even if it happens to be near the true value on average.

Cooking and Food Production

Baking is sensitive to ingredient ratios. Too much baking powder (inaccurate measurement) ruins a cake. In large-scale food production, mass dosing of ingredients must be both accurate and precise to ensure consistent product quality and comply with nutrition labeling regulations.

Improving Both Accuracy and Precision

Here are practical strategies for getting better measurements in any context: