Comparing results, examining relationships and making decisions based on data.
What does it mean?
Hypothesis testing helps us assess whether an observed difference or relationship could be due to random variation in the data. For example, we can investigate whether two machines produce different results or whether an improvement has changed a process.
A test result does not make a decision for us. We must also assess the quality of the data, the assumptions of the method and the practical importance of the result for the process being studied.
Aim of the training
Participants learn to state a null and an alternative hypothesis, choose a test suited to the type of data, determine the required sample size and interpret the p-value, confidence interval and size of the observed difference.
They complete exercises in Statystyka, using examples from quality management and process improvement. Particular attention is given to the distinction between statistical significance and practical importance.
Who is the training for?
The training is intended for:
- quality department staff,
- continuous improvement staff,
- process engineers and technologists,
- people carrying out Lean Six Sigma projects,
- people analyzing trials, measurements and implemented changes.
Training organization
The training lasts one or two days, depending on the desired level of detail and the number of analyses participants will perform. The one-day version focuses on the foundations of statistical inference and selected tests. The two-day version provides time to practice a wider range of methods, check their assumptions more thoroughly and discuss the interpretation of results.
Example program for the two-day training
Day 1 - foundations and comparison of results
- The null and alternative hypotheses.
- The Central Limit Theorem - drawing conclusions about a population from a sample.
- Type I and Type II errors, significance level, p-value and confidence interval.
- Determining sample size before collecting data.
- Types of data and selection of an appropriate test.
- Normality test, examination of the distribution and variance test.
- Two-sample t-test for independent groups.
- Paired t-test for measurements before and after a change.
- ANOVA for several groups and residual analysis.
Day 2 - event data and relationships between variables
- Two-proportions test, for example to compare the percentages of defective products.
- Two-sample Poisson rate test, for example to compare defect counts relative to production volume.
- Chi-square test for categorical data.
- Correlation and testing for the absence of a linear relationship.
- Regression - coefficient significance, model assessment and residual analysis.
- Statistical significance versus practical importance.
- Final exercise - selecting a method independently, analyzing data and writing a conclusion about the process.
Training cost
The price of dedicated training is agreed individually, depending on its duration and the scope of exercises. Use of Statystyka is free.