T Level Digital: Data and Analysis

Handling data well is mostly a matter of asking one more question than feels necessary, and these ten workplace situations are built around the moments where that extra question changes the answer. A shop has a cell containing 428 and has to say why that is not yet information. A retailer must choose a type for a postcode field, where the real test is not what the value looks like but whether anyone would ever add two of them together. A wholesaler splits a repeated supplier address into its own table and has to say what that actually buys, which is not speed but the impossibility of the same supplier ending up with two different addresses. An analyst sorts when she should have filtered and gets a list that is either cut short or padded. A courier firm quotes a mean of three days when half its parcels take one and half take five, so the average describes almost nobody. And a form rejects a date of birth in the future, which proves the entry is possible and says nothing about whether it is true. Every scenario is invented, so nothing turns on facts about a real company, and each wrong answer is a mistake people genuinely make rather than an obviously silly one. Zestly is an independent study tool. It is not affiliated with any exam board and it is not an exam centre.

• Say what turns data into information, and why a bare figure is neither
• Choose a data type from what will be done with the value rather than from what it looks like
• Choose a primary key that is unique and will never need to change
• Explain what normalising a table prevents, in terms of the inconsistency it makes impossible
• Tell filtering from sorting, and see what taking the top N rows gets wrong
• Match a chart to the question being asked, and read a correlation without assuming a cause
• Recognise how deleting rows with missing values can bias a result, and what a mean hides
• Distinguish validation from verification and state what validation cannot catch

Grimsargh Couriers reports a mean delivery time of three days. In fact about half its parcels arrive in one day and about half in five. What does the mean fail to show? — The spread: two very different groups of customers are being averaged into a figure that describes almost nobody's actual experience.