Work sampling calculator

It is a study made of snap observations: at random times the analyst records what each person or machine is doing at that moment. The share of times an activity shows up estimates the share of time it takes. Use it when the work is not repetitive (maintenance, stockroom, small-batch assembly), when you want to measure waiting and downtime, or when you want to observe many people at once; for a short, repetitive cycle, a stopwatch time study is more precise.

They are different things with similar names. Attribute acceptance sampling (AQL, ANSI/ASQ Z1.4 / ISO 2859-1) decides how many parts of a lot to inspect to accept or reject the lot. Work sampling measures how people’s and machines’ time splits between work, waiting, setups and downtime. For lot inspection, use the attribute sampling calculator.

With relative precision: n = z² × (1 − p) ÷ (s² × p). In the example, with 75% working time, ±5% relative and 95% confidence (z = 1.96), that is 513 observations. With ±5 points of absolute precision, n = z² × p × (1 − p) ÷ e² gives 289. At 99% confidence and ±5% relative it rises to 885. The smaller the activity you want to measure, the more observations you need.

Spread the observations over the days and the people seen per round. In the example, 513 observations over 5 days with 8 people per round give 13 rounds a day (520 observations). The times are random and change from day to day; the calculator draws them from a seed so you can print the plan and the field sheet. Spread the days over the week and the shifts you want to represent.

Out of 520 observations, 390 found people working: 75.0%, with an interval of 71.3% to 78.7% at 95% confidence. The precision achieved is ±4.96% relative, within the ±5% requested. On the p chart, all 5 days fell between 73.1% and 76.9%, inside the 62.3% to 87.7% limits: the process was stable and the days can be pooled. Waiting for material is the largest loss, 8.1% of the time, about 26 hours in the period.

Standard time = hours observed × working share × performance rating ÷ output × (1 + allowances). In the example: 320 h × 75% × 95% ÷ 4,800 parts = 2.85 min of normal time; with 12% allowances, 3.19 min per part, or 18.8 parts per hour. Output must come from the same group and days observed, and the rating should be assessed during the rounds.

Tell the crew and explain that the study measures the process; hidden observation tends to breed distrust and worse data. Categories must not overlap and must be defined before you start. Cover every weekday and shift you want to represent. The confidence interval uses the normal approximation, which breaks down when a category shows up fewer than 5 times. Where a union represents the employees, time studies and standards are usually covered by the contract.