24761 Chap.6 Interpreting Data: Signals, Correlation and Causation
Interpreting Data: Signals, Correlation and Causation
Three depths of reading
Week 4 treats interpretation as giving data meaning so that it answers a question and supports a decision.
The lecture separates descriptive reading (what happened), exploratory reading (what patterns exist) and causal reading (why it happened), and only the last needs a test showing that one thing produces another.
Signals before insights
Patterns, trends and relationships are the signals the lecture asks you to name first.
They move you from describing data towards interpreting it, but each still needs an explanation, and descriptive statistics such as the median and interquartile range help you state them precisely.
Correlation versus causation
The chapter takes apart a supermarket claim that a loyalty card makes shoppers spend 66% more, showing how selection and reversed direction create a gap that is not an effect, and how a random comparison would test it.
The habits here protect the first report from overclaiming and prepare the causal questions behind the primary research you recommend later. Practice items test the same reasoning on new claims.
What this chapter covers
- 01
Descriptive, exploratory and causal reading
- 02
Patterns, trends and relationships as signals
- 03
Positive, negative and zero correlation
- 04
Mean, median, mode and interquartile range
- 05
Selection effects and reversed direction
- 06
Designing a test before claiming an effect
Worked example · free
Classify three claims about loyalty spending
- 1Statement 1 is descriptive: it reports what happened, and the chart supports it directly.
- 1Statement 2 is exploratory: it names a pattern across groups, and the chart supports it because members spent $63 before joining, $22 above non-members.
- 1Statement 3 is causal and unsupported: the $27 compares different people, the within-shopper change is only $5, and even that needs a controlled comparison before it can be called an effect.
Key terms
- Descriptive Research
- Observation that summarises what is happening, without changing any variable, and often serves as the first step of a study.
- Exploratory Research
- Examining and comparing sources with curiosity to uncover patterns and signals before conclusions are drawn.
- Causal Research
- Testing whether a change in one variable directly produces a change in another, usually through a controlled experiment.
- Correlation
- Measures how strongly and in which direction two variables move together, without showing that either causes the other.
- Data Trend
- The general long-run direction, upward or downward, in which a measure moves over time.
Interpreting Data: Signals, Correlation and Causation FAQ
What is the difference between correlation and causation in marketing?
Correlation means two measures move together, such as card membership and basket size. Causation means changing one directly changes the other. Marketing data is full of correlations created by who chooses what, so a causal claim needs a test, ideally a random comparison between people who did and did not receive the change.
What are descriptive, exploratory and causal research?
Descriptive research states what happened, for example average spend per visit. Exploratory research looks for patterns, trends and associations across sources. Causal research tests why something happened by showing that one factor produces another. The Week 4 lecture uses the three as increasing depths of reading the same data.
What counts as a pattern, a trend and a relationship?
A pattern is something that recurs, such as the same complaint across many reviews. A trend is the overall direction a measure takes over time, up or down. A relationship is two or more variables changing together. The lecture calls all three signals: they still need explaining before they become insights.
When should I report the median instead of the mean?
Use the median when a few extreme values would drag the mean, as with spend or income, because the middle value is barely affected by outliers. The mean suits measured data without heavy skew, and the mode is the only option for categories such as store format or payment type.
How can I tell if a marketing result is just a selection effect?
Ask whether people chose their group. Members, subscribers and email openers often differed before the programme reached them. Compare the same people before and after, or groups that are similar in every other way, and treat any remaining gap as a lead to test rather than as proof.
Assessment move
Collect three charts or statistics from sources you might use and write one sentence for each at every depth: what it shows, what pattern it suggests, and what causal claim someone might be tempted to make. Then list the alternative explanations for each causal claim, especially who chose to be in each group and whether the direction could run backwards.
Practise computing the median and interquartile range on a small invented set of spend figures, and finish each exercise by naming the test that would turn your association into evidence of cause. Keep these notes for the second report. Review your notes weekly.
Working through Interpreting Data: Signals, Correlation and Causation in 24761? Sia is AskSia’s AI Marketing tutor — ask any 24761 Interpreting Data: Signals, Correlation and Causation question and get a clear, step-by-step explanation grounded in how 24761 is taught and assessed. Read this chapter free, then take your hardest questions to Sia.