EDPS0007 Chap.6 Statistical Controls and Normative Questions
Statistical Controls and Normative Questions
Statistical Controls and Normative Questions introduces the statistical reasoning the module expects you to read critically, without heavy calculation. Lecture 5 shows a raw comparison in which British-Pakistani pupils averaged 10% below White British pupils at GCSE, then 5% above once household income is controlled, reversing a deficit story some politicians had told.
Controlling for a variable means removing its influence so the effect of the variable of interest can be isolated. The summer-born effect illustrates probabilistic causation: pupils with summer birthdays trail older classmates on average, by 25% at age 7 and 6% at 16, yet many individual summer-born children do well.
The chapter then separates normative questions about what should be, answered by reason and judgement, from empirical questions with measurable criteria, and shows how strong arguments use evidence to inform values without pretending the evidence settles them. Confidence intervals and significance follow in Week 9.
What this chapter covers
- 01
Raw comparisons and hidden third variables
- 02
Controlling for household income
- 03
The summer-born effect
- 04
Probabilistic causation and phrasing on average
- 05
Treatment and measured variables
- 06
Normative and empirical questions
- 07
Statistics as the voices of many people
Worked example · free
Describe the summer-born gap at four ages without overclaiming
- 1Trend: the shortfall falls at every age, from 25% at 7 to 6% at 16.
- 1Narrowing: 25 - 6 = 19 percentage points between ages 7 and 16.
- 1Persistence: the gap never closes completely, which fits the lecture's account of starting behind and never quite catching up.
- 1Probabilistic sentence: on average, summer-born pupils score below older classmates, though many individual summer-born pupils do not.
Key terms
- Statistical control
- A technique that removes the influence of chosen variables to isolate the effect of one variable.
- Probabilistic causation
- A cause that raises or lowers the likelihood of an outcome on average rather than in every case.
- Normative question
- A question about how things should be, answered through reasoning and judgement.
- Empirical question
- A question with specific, measurable criteria, answered by collecting and analysing data.
- Treatment variable
- A variable a researcher deliberately manipulates to see its effect on an outcome.
Statistical Controls and Normative Questions FAQ
What does controlling for income actually change?
It removes income's influence so that the remaining difference reflects the variable of interest. In the GCSE example, controlling for household income turned a 10% shortfall for British-Pakistani pupils into a 5% advantage, overturning a deficit explanation.
If summer babies do worse, why do some do very well?
Because the effect is probabilistic. The finding describes averages across many children, so individual exceptions are expected and do not disprove it. Adding "on average" or "typically" makes such statements accurate.
Is "Should university be free?" an empirical question?
No, it is normative, answered through values and reasoning. It has empirical counterparts, such as whether free tuition increases the number of graduates, and good essays use those findings to inform, not replace, the judgement.
Why does a control not prove causation?
Unmeasured variables may still drive whatever difference remains. A control only removes the influence of the variables held constant, so the careful conclusion is that the leftover gap is linked to the variable of interest, not that it is wholly caused by it.
Exam move
Learn this chapter through examples rather than terms. For the GCSE case, be able to state the raw gap, the controlled gap and the variable controlled in one sentence. For the summer-born effect, memorise the four ages and figures and practise reporting them with "on average".
Then build a two-column list of normative and empirical questions from across the module, such as who should pay for university against what share of students graduate in debt, and pair each normative question with an empirical counterpart. When you read a study for an essay, run the five-step routine: comparison, third variables, control, probabilistic phrasing and the point where judgement takes over.
This chapter supplies tools for all three exam questions, so plan to use at least one controlled comparison or probabilistic phrase in each essay. Finally, rewrite three sentences from your own drafts that state a group difference, adding the comparison group, the control if any and the phrase on average, and notice how much more defensible each becomes.
Working through Statistical Controls and Normative Questions in EDPS0007? Sia is AskSia’s AI Education tutor — ask any EDPS0007 Statistical Controls and Normative Questions question and get a clear, step-by-step explanation grounded in how EDPS0007 is taught and assessed. Read this chapter free, then take your hardest questions to Sia.