To choose the appropriate capability analysis, determine which type of data you have. The two main types of data for capability analysis are continuous data and attribute data. Minitab offers normal and nonnormal analyses for continuous data and binomial and Poisson analyses for attribute data.
If you can choose to collect either continuous data or attribute data, try to collect continuous data because they typically provide more information and are more objective. Attribute data are easier to collect and thus are often used when continuous measurements are difficult to obtain.
Continuous data measures a characteristic of a part or process, such as length, weight, or temperature. The data often include decimal values. For example, a food manufacturer wants to investigate whether the weight of a cereal product is consistent over time. To collect data, a quality analyst records the weights from a sample of cereal boxes.
Continuous data from industrial processes often follow a normal distribution. Continuous data that are not normally distributed may follow a specific type of nonnormal distribution, such as a Weibull distribution or an exponential distribution. Sometimes, you can transform nonnormal data to fit a normal distribution.
Attribute data typically count the presence of a characteristic or condition, such as a physical trait, a type of defect, or a rating, such as pass/fail. Attribute data usually depend on a subjective assessment, and are thus subject to rater interpretation. There are two main types of attribute data: counts of defects, which are nonconformities, and counts of defectives, which are nonconforming items.
A defect refers to a specific quality characteristic for an item, such as a tear, scratch, or discoloration. Each item can have more than one defect and a defect may not always result in the item being unusable. For example, analysts at a textile company inspect towels for tears, pulls, or improper stitching and record the number of defects for each batch of 25 towels. Each towel can have more than 1 defect, such as 1 tear and 1 pull. When you monitor defects, you collect Poisson data.
A defective refers to whether the overall status for an entire item is acceptable or not. Therefore, the data is often of the form yes/no, pass/fail, or defective/nondefective. Because an item may have many quality characteristics, it may have many defects, but the item itself is either defective or not defective. For example, an analyst inspects a sample of bulbs from a supplier and counts the number of broken bulbs in each sample. When you monitor defectives, you collect binomial data.