Inferential statistics

Why

To test hypothesis based on a quantitative dataset.

How

First, perform descriptive statistics. Next, formulate a hypothesis that can be expressed in variables of your dataset. Test your hypothesis using an appropriate statistical test. Every test makes certain assumptions about the character of your data; make sure your data complies with these assumptions.

Practice

With the advent of big data, inferential statistics are increasingly important. However, inferential statistics are delicate and companies often use specialists to analyse them. Novices commonly make errors like 'capitalising on chance'. This happens when an analyst tests so many hypothesis that some turn out 'positive' by accident.

Ingredients

• A well-defined dataset and hypotheses
• Statistical software such as SPSS, R, or Excel
• Advanced knowledge about statistics and probability theory
• A keen eye for the difference between signal and noise

Phases

• Analysis
• Realisation

Trade-offs

Read more about the trade-offs: Inspiration or Data / Expertise or Fit / Overview or Certainty.