1. What causal reasoning tests

A cause is a condition, event or action that contributes to producing an outcome. The outcome is its effect. In reasoning questions, the task is not simply to find two related events but to identify the direction and strength of the relationship justified by the evidence. For example, a ruptured supply pipe may cause reduced water pressure. Reduced pressure, by itself, does not establish that a pipe has ruptured: pump failure or unusually high demand could produce the same observation.

Cause-and-effect reasoning supports several CSAT tasks: drawing inferences from passages, assessing arguments, identifying assumptions and interpreting trends. A question may ask which event is the cause, which explanation best accounts for an observation, or which additional fact strengthens a causal claim. The required standard comes from the question stem. A probable explanation need not be certain, while a conclusion that must follow requires much stronger support.

Some practice materials use paired statements with categories such as I causes II, II causes I, independent causes, effects of independent causes and effects of a common cause. Treat this as a practice convention rather than a guaranteed UPSC format. Read the actual answer choices carefully because their categories and wording can differ.

Separate the observed facts from the explanation proposed for them.
Identify whether the task asks for possibility, probability, support or certainty.
Use the passage's evidence first; ordinary background knowledge should not become a substitute for missing facts.

2. Recognising the main relationships

Direct causation occurs when one stated event produces the other through a reasonably clear mechanism. If an inspection confirms that a fallen tree severed a feeder line and electricity failed immediately afterwards, the tree fall explains the outage. Reverse causation is a competing possibility when the assumed direction is wrong: an association between public complaints and inspection visits may arise because complaints trigger inspections, not because inspections create complaints.

A common cause produces two separate effects. Heavy rainfall may cause both waterlogging and disruption of outdoor work. Neither effect must cause the other. A confounder is a factor that influences both the proposed cause and the outcome, potentially distorting their apparent relationship. For instance, unusually hot weather may increase both electricity consumption and heat-related illness; higher electricity use need not explain the illness.

Independent causes and effects of independent causes are different categories. In the former, the statements describe separate causal events; in the latter, they describe outcomes generated by distinct antecedents. Yet independence should not be asserted merely because no connection is mentioned. If a question provides too little evidence to distinguish these possibilities and offers a cannot-be-determined option, that option deserves serious consideration.

Causation can also operate through a chain or feedback loop. Rainfall may damage a road, which delays supplies and raises local prices. Separately, low income and poor health may reinforce each other over time. In such cases, identify the specific stage or period under examination rather than forcing a complex relationship into a single timeless direction.

Common cause: one antecedent branches into two outcomes.
Causal chain: one event produces an intermediate event, which produces a later outcome.
Feedback: an effect subsequently influences the original causal factor.

Cause-and-effect decision process

1. Identify the observations and the proposed explanation.
2. Check chronology and the precise scope of the claim.
3. Look for a credible causal mechanism.
4. Test reverse causation and common causes.
5. Examine comparison evidence and measurement changes.
6. Select the conclusion requiring the fewest unsupported assumptions.

3. A systematic solving method

Begin by reducing the statements to neutral events: A happened and B happened. Note dates, time lags, locations and affected groups. A proposed cause must precede the particular effect being explained. Statement order is irrelevant: Statement II can describe an earlier event than Statement I. Temporal priority helps reject impossible directions, but it does not prove the remaining direction.

Next, identify the mechanism connecting the events. Ask how A could produce B using the facts supplied. A specified mechanism is stronger than an unexplained association. Then test alternatives: could B influence A, could another factor influence both, or could the pattern result from chance, selection or changed measurement? Avoid adding elaborate possibilities, but do not ignore a straightforward rival explanation.

A useful counterfactual question is: would the outcome probably have occurred if the proposed cause had been absent? An appropriate comparison group helps answer this. If attendance rises after a school programme, the increase is more persuasive evidence of impact when comparable schools without the programme do not show a similar rise. Differences between the groups and simultaneous policy changes still require attention.

Finally, match the strength of the answer to the strength of the evidence. An option stating that a measure contributed to an outcome may be justified even when an option calling it the sole cause is not. For strengthening and weakening questions, choose the fact that most directly changes confidence in the proposed causal link, rather than one that merely discusses the same subject.

Check sequence, mechanism, competing explanations and comparison evidence.
Treat words such as only, always, necessarily and entirely as demanding claims.
If two choices remain, identify which needs fewer unsupported assumptions.

Quick classification of causal relationships
Relationship	Illustration	Key check
Direct cause	A confirmed pipeline rupture reduces water pressure.	Does the stated mechanism connect the events?
Reverse causation	Complaints prompt inspections rather than inspections prompting complaints.	Which event initiates the relationship?
Common cause	Heavy rainfall causes waterlogging and outdoor-work disruption.	Is there a shared antecedent?
Separate causes	A power fault stops one pump; scheduled maintenance stops another.	Are distinct antecedents explicitly established?
Association only	Two indicators rise during the same period.	What evidence supports causation beyond co-movement?

4. Frequent errors and conceptual distinctions

The post hoc error assumes that because B followed A, A caused B. A fall in accidents after a publicity campaign may instead reflect reduced traffic, changed reporting or another safety intervention. Correlation likewise describes association, not necessarily causation. Even a strong association can reflect reverse causation or a shared underlying factor.

Necessary and sufficient conditions must be distinguished. A necessary condition must be present for an outcome; a sufficient condition guarantees the outcome within the stated model. Having a valid admit card may be necessary for entry to an examination centre, but possession alone is not sufficient if other mandatory requirements are unmet. Many real-world causes are contributory: they change the likelihood of an outcome without guaranteeing it.

Another trap is affirming the consequent. From 'if the specified sensor fails, the alarm activates' and 'the alarm activated', it does not follow that this sensor failed; other conditions may activate it. Similarly, the absence of one proposed cause need not imply the absence of the effect. Read whether the rule is one-way or explicitly says if and only if.

Selection bias and measurement changes can create misleading patterns. Coaching attendees may perform better partly because more motivated students choose coaching. Recorded crime may rise after easier complaint registration even without an equivalent increase in underlying crime. Also watch regression to the mean: an unusually poor result may be followed by a more typical result even without an effective intervention.

Do not equate an administrative response with the original cause of the problem.
Distinguish actual changes from changes in reporting, classification or detection.
A plausible story explains how causation could occur; evidence is needed to show that it did occur.

5. Applying causal reasoning to public-policy passages

Policy passages often connect an intervention with outcomes such as learning, employment, health or road safety. Separate inputs, implementation, immediate outputs and final outcomes. Installing water-treatment equipment is an input; functioning treatment is implementation; safer supplied water is an intermediate result; reduced waterborne illness is an outcome. Failure at an intermediate stage can break the proposed causal chain.

Evidence becomes more informative when it establishes timing, verifies the mechanism and compares otherwise similar groups. Random assignment can reduce systematic differences between treated and untreated groups. Natural experiments and carefully designed observational studies can also help, although their conclusions depend on assumptions. CSAT generally requires understanding this logic rather than knowing technical statistical procedures.

Under examination conditions, avoid trying to establish scientific certainty where the question asks only for the most reasonable inference. Equally, do not accept an absolute claim from a short before-and-after comparison. State mentally what the evidence warrants: association, a plausible causal contribution, or a stronger causal conclusion. This habit improves accuracy in both reasoning items and comprehension passages.

For strengthening questions, look for verified mechanisms or evidence ruling out a major alternative.
For weakening questions, look for reverse causation, confounding or unreliable measurement.
For inference questions, remain within the scope of the supplied evidence.

Real-world case studies

John Snow and cholera in London, 1854

John Snow investigated cholera deaths around the Broad Street pump in Soho and argued that contaminated water transmitted the disease. His wider comparison of households supplied by different water companies strengthened the waterborne explanation. The lesson is not merely that deaths occurred near a pump: location, exposure patterns and comparison evidence supported a mechanism. The outbreak was already declining before the pump handle was removed, so the later decline alone should not be treated as decisive proof of that intervention's effect.

Randomised remedial education in India

Evaluations of Pratham's Balsakhi remedial education programme in Mumbai and Vadodara used randomised assignment to assess learning effects. Comparing programme and comparison groups offered stronger evidence than observing score improvements among participants alone. The reasoning lesson is that a credible counterfactual helps separate programme effects from ordinary learning over time or differences in the students selected.

Previous year questions

No source-linked PYQ is available here yet. Use the practice questions below.

Practice questions

Practice MCQ 1 · Easy

Consider the following statements: I. Water supply to a residential colony stopped at 10:15 a.m. II. At 10:10 a.m., an excavator ruptured the colony's only supply pipe. Inspection confirmed that the rupture completely blocked water delivery; no other supply failure occurred. Which of the following correctly describes the relationship?

A. Statement I describes the cause of the event in Statement II.
B. Statement II describes the cause of the event in Statement I.
C. Both statements describe effects of a power failure.
D. The statements establish only a coincidence.

Practice MCQ 2 · Easy

An intense heatwave struck a town. A municipal study found that high temperatures independently increased demand for cooling appliances and caused more cases of heat exhaustion. It found no causal connection between appliance purchases and heat exhaustion. Consider: I. Purchases of cooling appliances increased. II. Cases of heat exhaustion increased. Which relationship is supported?

A. I caused II.
B. II caused I.
C. I and II arose from unrelated causes.
D. I and II were effects of a common cause.

Practice MCQ 3 · Easy

A survey finds that districts with more public libraries also have higher average household incomes. An official concludes that constructing libraries necessarily raises household incomes. Which is the most appropriate assessment?

A. The association alone does not establish whether libraries raise incomes, higher incomes support more libraries, or another factor influences both.
B. The survey establishes that higher incomes are caused entirely by libraries.
C. The survey establishes that libraries cannot influence incomes.
D. The official's conclusion follows because both variables were measured in the same survey.

Practice MCQ 4 · Easy

Wards receiving more sanitation inspections also record more complaints about uncollected waste. A councillor argues that inspections generate complaints. However, municipal rules require an inspection within two days whenever a ward crosses a specified complaint threshold, and records show that complaints generally rose before inspections increased. Which explanation best fits the evidence?

A. Inspections necessarily worsen waste collection.
B. Complaints and inspections must be unrelated.
C. Increased complaints trigger additional inspections, reversing the councillor's proposed causal direction.
D. Both complaints and inspections must have been caused by population decline.

Practice MCQ 5 · Easy

A city introduced a night-time restriction on heavy vehicles on 1 July. Its recorded night-time accidents had already fallen sharply between 1 April and 31 May. No announcement, trial or anticipatory change connected with the restriction occurred before July. Which conclusion necessarily follows?

A. The restriction increased accidents after July.
B. The restriction's operation cannot explain the accident decline during April–May.
C. The earlier decline proves that the restriction could never reduce accidents.
D. The earlier decline must have resulted from improved street lighting.

Practice MCQ 6 · Moderate

Students who voluntarily joined a free morning coaching programme scored higher in a later examination than students who did not join. The organiser attributes the entire difference to coaching. Which additional finding most seriously weakens this attribution?

A. The programme covered topics appearing in the examination.
B. Several participants attended every coaching session.
C. Participants already had substantially higher scores before the programme, and the average improvement was the same in both groups.
D. The examination was held after the programme ended.

Practice MCQ 7 · Moderate

A municipality replaced defective water meters and subsequently reported a fall in household water consumption. It claims that meter replacement caused households to use less water. Which additional evidence would most strongly support this specific claim?

A. Replacement was randomly phased across comparable neighbourhoods, and separate calibrated inlet meters showed a decline only in neighbourhoods receiving replacements, with no concurrent differential changes.
B. Bills generated by the replacement meters were lower than bills generated by the old meters.
C. Most residents said that wasting water was undesirable.
D. Water consumption declined throughout the region during the same unusually wet season.

Practice MCQ 8 · Moderate

A district's total registered thefts fell from 1,000 to 700 after online complaint filing was introduced. The superintendent argues: 'The fall in registered thefts shows that actual thefts declined.' Which assumption is required for treating the fall in registrations as evidence of a fall in actual thefts?

A. Online filing eliminates every incentive to commit theft.
B. Every theft was registered before online filing began.
C. No district without online filing experienced a decline in theft.
D. Changes in reporting, registration or classification did not themselves account for the fall in registered thefts.

Practice MCQ 9 · Moderate

In a town, unreliable bus services initially encouraged commuters to buy motorcycles. As motorcycle traffic increased, congestion slowed buses further. These longer delays then encouraged additional commuters to abandon buses. Which interpretation best captures the causal relationship?

A. Motorcycle purchases were the sole original cause of unreliable bus services.
B. Initial bus unreliability encouraged motorcycle use, which subsequently reinforced bus unreliability through congestion.
C. Bus delays and motorcycle purchases must be independent effects of population growth.
D. Because the two variables influence each other, neither can be a cause at any stage.

Practice MCQ 10 · Moderate

A transport department proposes the chain: a bypass reduces freight traffic on the old road; lower freight traffic reduces roadside dust; lower dust reduces respiratory clinic visits. After the bypass opens, freight traffic on the old road falls sharply and clinic visits decline. However, reliable monitors show that roadside dust exposure remains unchanged because nearby construction generates additional dust. Which conclusion is best supported?

A. The decline in clinic visits cannot be explained by the proposed dust-reduction mechanism on the evidence given.
B. The bypass definitely had no effect on any health outcome.
C. Fewer clinic visits prove that actual dust exposure fell despite the reliable monitoring results.
D. Construction activity must have directly reduced respiratory illness.

Practice MCQ 11 · Difficult

Two comparable road corridors, P and Q, are observed for equal periods before and after speed cameras are installed only on P. P: Before—120 crashes during 60,000 vehicle journeys; After—84 crashes during 70,000 vehicle journeys. Q: Before—90 crashes during 50,000 vehicle journeys; After—80 crashes during 50,000 vehicle journeys. Assume that, without cameras, P's crash rate per 1,000 journeys would have fallen by the same absolute amount as Q's rate. The cameras did not affect journey numbers, and no other differential changes occurred. How many crashes on P during the after-period are estimated to have been prevented by the cameras?

A. 28
B. 36
C. 56
D. 42

Practice MCQ 12 · Difficult

An employment-support programme reports the following outcomes: Participants: 23 of 25 applicants with prior work experience obtained jobs; 45 of 75 applicants without prior experience obtained jobs. Non-participants: 72 of 90 applicants with prior experience obtained jobs; 4 of 10 applicants without prior experience obtained jobs. An evaluator compares the groups after giving applicants with and without prior experience equal weight within each group. Participation was voluntary, and no other background differences were measured. Which assessment is correct?

A. Participants' unadjusted employment rate is 8 percentage points lower, proving that the programme reduced employment.
B. Participants' equally weighted employment rate is 16 percentage points higher, but this adjusted association alone does not establish causation.
C. Participants' equally weighted employment rate is 8 percentage points higher, proving that the programme caused the difference.
D. Participants' equally weighted employment rate is 36 percentage points higher, but only among applicants with prior experience.

Practice MCQ 13 · Difficult

During a particular storm, both river overflow and hillside runoff will occur. The following causal rules are exhaustive: • River overflow closes the hospital's access bridge unless action E protects the bridge. • River overflow contaminates the hospital's regular water supply unless action T treats that water. E does not protect the water intake. • Hillside runoff also contaminates the regular water supply unless action D diverts the runoff or action T treats the water. • The hospital suspends services if the bridge closes, or if its regular water is contaminated and no emergency clean-water reserve is available. • Action S provides an adequate emergency clean-water reserve but does not remove contamination from the regular supply. Initially, none of these actions is in place. Which option lists all two-action combinations sufficient to prevent suspension?

A. E + D and E + T only
B. E + S and D + T only
C. E + T and E + S only
D. E + D, E + T and E + S only

Practice MCQ 14 · Difficult

A labour regulation is introduced only in district P. Average monthly earnings equal the hourly wage multiplied by monthly paid hours; all workers within each district have the stated wage and hours. P: Before—₹100 per hour and 200 hours; After—₹120 per hour and 150 hours. Comparable district Q, without the regulation: Before—₹80 per hour and 200 hours; After—₹88 per hour and 180 hours. Assume that, without the regulation, P would have experienced the same percentage change in wages and the same percentage change in paid hours as Q. No other differential changes occurred. What is the estimated effect of the regulation on a P worker's monthly earnings in the after-period?

A. A decrease of ₹1,800
B. A decrease of ₹2,000
C. An increase of ₹1,800
D. A decrease of ₹3,600

Practice MCQ 15 · Difficult

Eligible jobseekers are randomly assigned to four equally sized groups. Their employment rates after six months are: Neither training nor counselling: 60%. Training only: 66%. Counselling only: 65%. Both training and counselling: 80%. For this question, treat these rates as the exact causal outcomes of the assignments, with full compliance and no spillovers. Define the 'additional joint effect' as the improvement from both services over neither, minus the sum of the improvements from each service alone over neither. Consider the following statements: 1. The additional joint effect is 9 percentage points, equal to 45% of the total improvement from providing both services. 2. The incremental effect of training is 6 percentage points without counselling but 15 percentage points with counselling. 3. Providing both services improves employment by exactly the sum of their separate effects. Which of the above statements are correct?

A. 1 only
B. 2 and 3 only
C. 1 and 3 only
D. 1 and 2 only

Cause and effect