Stratification in quality control in simple way
Stratification is the separation of data into categories. This tool is usually used during the diagnostic process to identify which categories contribute to the problem. We can separate data into different categories so that the data of that category can be observed to solve the problem.
Stratification is a system of formation of categories, classes or layers. Data collected by check sheet need to be meaningfully classified. Such categorization help in understanding the data so that further analysis can be done to get the proper result. This tool separates the data so that patterns can be seen.
A category is defined by a specific combination or range of variables. In a category there is more than one variable. Stratification by itself does not point out the root cause of the problem. It helps to sort out which areas need further investigation. Stratification is actually the base of other tools like Pareto analysis, scatter diagram etc.
What is the use of stratification?
- To categorize data from different sources like machines, operators, dates, suppliers.
- Stratifying customer complaints by type of complaint or type of customer can help in identifying the important problems.
- It is useful in focusing the investigation on the most serious symptoms and giving an indication of where to look for the root cause.
- Stratification help in testing theories in case of cause and effect diagram.
- It is useful to stratify data on the new performance level by the variable used during the diagnostic process.
What are the precautions in the use of stratification?
- Small differences between data classes should not be taken into consideration.
- It should not consider that the abnormal category is the cause of the problem.
- When there is a need of collecting new data, there must make an effort to collect as much identifying information as required for stratification.
How to stratify?
Step 1: Select the stratification variables: When new data are collected, make sure that all potential stratification variables are included like a machine, operator, date etc.
Step 2: Establish categories for each stratification variable: A category is a value or range of value of a stratification variable. For example dates of the month are 1 to 31.
Step 3: Sort observations into categories of one of the stratification variables.
Step 4: Calculate the phenomenon measured for each category. Collect the data date-wise.
Step 5: Display the result. Usually graphic displays are the most effective.
Step 6: Prepare and display the results for additional stratification variables.
Step 7: Plan for additional confirmation.
How to interpret stratification?
It is usually best to present the stratification result in graphic form. If the stratification results are in bar graph form. It will be easy to see the categories of the variables to see whether one or more of the categories stand out.
- Does one machine has more defect rate?
- Does one operator have more defect rate?
- Does the shift or specific date is the reason for the high defect rate?
After stratification, if the result provides a clear indication of the likely source of the phenomenon being studied. Then the team has to validate their initial result. If we do not get any useful results then see the other variables with each category.