Back to the on-screen lesson ·
Test whether evidence supports a proposed decision, accounting for costs, selection, bottlenecks and the difference between a metric and a goal.
Paper packet. Every task here also exists on screen, where it is checked automatically; answers written on paper are not assessed by Nydus. When you are back at a device, enter your answers there.
Test whether evidence supports a proposed decision, accounting for costs, selection, bottlenecks and the difference between a metric and a goal.
Test whether evidence supports a proposed decision, accounting for costs, selection, bottlenecks and the difference between a metric and a goal.
| Term | What it means |
|---|---|
| Proxy metric | An observable measure used to represent a different outcome of interest. |
| Incremental contribution | The additional revenue less additional variable costs associated with a decision. |
| Counterfactual | The relevant outcome expected if the proposed action were not taken. |
| Selection effect | A difference attributable to how cases enter or remain in a comparison group. |
A proposal usually aims at an outcome such as total profit, processing time or reliable service. Identify that outcome before accepting a supporting statistic. A higher margin per unit does not by itself establish higher total profit. Faster completion of one stage does not guarantee a shorter end-to-end journey. Fewer recorded incidents do not necessarily mean fewer actual incidents. These distinctions require no specialized business knowledge: the passage supplies the relevant quantities and relationships. Your task is to keep the stated goal separate from a convenient proxy and then decide whether the offered evidence connects the two.
Another way: steps
Goal to evidence to missing link: identify the promised business outcome, label what was actually measured, and ask what must connect them. For a plan, compare incremental benefits and costs against the no-change alternative before assessing feasibility.
Local improvements can be offset elsewhere. A smaller delivery vehicle may use less fuel per kilometer but require more trips for the same workload. A faster machine can leave finished output unchanged when another required stage remains at capacity. A component with a lower purchase price can raise operating cost if it needs frequent replacement and every replacement stops production. Draw a short chain from intervention to final objective. At each link, ask what must remain constant or what additional condition is needed. An evaluation question is useful when its answers would reveal whether one of these links holds, rather than merely adding background information about the product or organization.
A resource can be idle on average while fully occupied at the time demand must be served. Off-peak availability does not resolve a peak-hour shortage unless work can be shifted in time. A branch's overall score can also reflect its mix of easy and difficult cases rather than staff performance on comparable work. Compare like with like before interpreting an aggregate. This does not mean averages are always misleading. It means a conclusion about a particular constraint or category needs evidence at that level. Ask whether the measured unit matches the decision: hour, trip, customer, service type, completed problem or total operating period.
People who choose training, customers who renew and devices eligible for an update can differ from the comparison group before the intervention. Their later performance therefore does not automatically isolate the intervention's effect. Locate the selection mechanism and ask what it may be associated with. Motivation can influence both training participation and sales. Device age can influence update eligibility and baseline reliability. A representative comparison that addresses the relevant preexisting differences is more useful than simply enlarging the same biased groups. The key causal question is what would have happened to comparable cases without the intervention, not whether the treated group looks successful in isolation.
A policy can alter reporting behavior or the incentive to optimize a statistic. Staff evaluated by calls per hour may end calls faster while leaving issues unresolved, increasing repeat contacts. Requiring signed safety reports may reduce reports because employees fear blame. In both examples, the recorded metric can move in the intended direction without the underlying goal improving. A weakening answer may identify this divergence without proving that actual service or safety worsened. Read the strength of the conclusion carefully: evidence sufficient to challenge a claim that the entire change was caused by an intervention may be insufficient to establish that the intervention had no effect at all.
Only members receive a discount means discount implies membership, not membership implies discount. A guarantee issued only when materials arrive by Tuesday makes timely arrival necessary; it may not make it sufficient. These logical distinctions also matter in plans: customer familiarity may be necessary for conversion without guaranteeing that conversion revenue covers every trial's cost. Compare the best remaining alternatives by naming the specific gap each one addresses. A strengthen answer should make the missing link more credible; a weaken answer should introduce a relevant competing account or obstacle. An evaluate answer should have contrasting possible answers that change your judgment. Do not favor a dramatic statistic if both the proposed plan and its main alternative would predict it equally well.
Business arguments often measure a goal indirectly. Calls handled per hour may represent productivity, but it can rise when agents end calls before resolving problems. If repeat contacts increase, total service cost may rise despite the improved local metric. Identify the final outcome the conclusion promises and the intermediate measure offered as evidence. A strong answer addresses the link between them. Evidence that agents like the new procedure may be relevant to morale but insufficient to establish durable resolution. Conversely, evidence that repeat contacts declined helps connect faster handling to the intended result. Do not assume every proxy is invalid. A metric can be useful if the argument supplies a reason it tracks the outcome in this setting. The task is to assess that specific bridge, including whether incentives or measurement procedures changed when the policy was introduced.
A proposal's effect depends on what would happen without it. If a firm offers a discount, additional sales revenue cannot be treated as pure gain: some buyers would have purchased at the original price, and producing the extra units may incur costs. Distinguish fixed costs that remain unchanged from costs that vary with the decision. Suppose a service upgrade costs two thousand credits per month and is expected to retain twenty customers who each contribute one hundred and fifty credits after variable costs. The projected incremental contribution is three thousand, leaving one thousand before any additional fixed costs. This calculation supports the plan only if those customers would otherwise leave and the contribution estimate applies to them. An answer about total customer count can distract from that counterfactual. An evaluation question about likely retention without the upgrade directly tests the claimed benefit and can change the decision in opposite directions.
A comparison between successful adopters and nonadopters may reflect selection. Firms choosing a new inventory system could already have better managers, more capital or faster growth. Evidence of a preexisting difference weakens the inference that the system caused the entire performance gap. It does not show the system caused failure or had no benefit. Likewise, excluding firms that closed can make surviving adopters appear unusually successful. Ask who entered the sample, who remained observable, and whether outcomes were measured consistently. For before-and-after comparisons, consider seasonality and broader market changes. The most useful new evidence distinguishes the proposed explanation from a concrete alternative. A fact that every growing firm also sells products is too general to discriminate. A matched comparison showing similar growth without the new system is more directly relevant, though its force still depends on the similarity of the groups and the strength of the conclusion.
When reviewing a decision argument, make a three-column note: intended outcome, supporting metric, missing link. For a total-cost proposal, the missing link may be replacement frequency or an indirect cost. For a service improvement, it may be downstream capacity or durability of resolution. For a causal claim, it may be a comparable untreated group or stable measurement. Then choose a new problem with a different surface setting but the same structure. In timed work, this compact diagnosis helps you discard answers that are true yet irrelevant. It also prevents outside business experience from supplying assumptions the question never gave: evaluate the stated model before imagining a more elaborate organization.
A warehouse can pack 90 orders per hour, while its mandatory inspection station can clear only 60. Raising packing capacity to 120 does not by itself increase completed orders beyond 60 per hour. The proposal might still reduce packing delays at particular times, but that narrower benefit differs from the promised increase in total throughput. An evaluation question about whether inspection capacity will also increase is therefore diagnostic. A fact about the color or purchase date of the new packing equipment would not establish the missing operational link.
Identify the proposed objective
A plant proposes substituting a component costing 60 credits for one costing 100. The goal is lower annual operating cost.
The target is not the price of one component. It includes the consequences of keeping the line operating over an equal period.
Add the supplied service-life information
The old component lasts six months; the new one lasts three months. Each replacement adds 40 credits of shutdown and labor cost.
The cheaper component must be bought and installed twice as often to supply a full year of service.
Compare annual totals
Old: 2(100+40)=280. New: 4(60+40)=400.
A common horizon reveals a higher total cost despite the lower purchase price. The model uses only the costs and lifetimes stated.
Limit the inference
Under these assumptions, the proposed substitution does not meet its cost objective.
This conclusion is local to the supplied figures. It does not establish that cheaper components are always worse or that the old design is universally preferable.
Separate the metric and goal
A support team is rewarded for tickets closed each day. The firm's goal is to resolve customer problems.
Closing a record and resolving its underlying problem are related but distinct events; the policy may alter that relationship.
Read the competing observations
Closed tickets rise after the reward begins, but reopened tickets about the same issues also rise.
Both observations can coexist if agents close work prematurely. The second observation tests what the first statistic actually represents.
Choose relevant additional evidence
The share of cases remaining resolved after a week falls.
This measure is closer to the stated durable-resolution goal and therefore challenges the declaration of improved service.
Avoid an excessive claim
A rise in closed tickets alone is insufficient to establish better resolution.
The evidence need not imply that every agent behaved badly or that every customer received worse service. The argument's metric-to-goal link is what fails.
Separate local and final outcomes
Compare the measured activity with completed service rather than assuming the two always move together.
A policy can improve an intermediate statistic while shifting work or cost to another stage that determines the promised outcome.
Identify the observed association
Staff volunteering for a negotiation workshop later win more contracts than nonparticipants.
This is a comparison between selected groups. The observation does not by itself explain why their outcomes differ.
Look for an earlier difference
Volunteers had higher contract-win rates before the workshop as well.
Baseline performance supports the possibility that motivation, experience or assignments account for some of the later difference.
Improve the causal test
Compare changes in performance among initially similar staff with different workshop exposure.
A better comparison addresses the known baseline difference, though it does not automatically remove every possible source of confounding.
Keep the forecast bounded
The original group difference cannot simply be assigned to every future compulsory participant.
That forecast assumes both causation and a uniform effect across people. Neither follows from the original selected-group comparison.
Examine the comparison population
Check whether future compulsory participants resemble the volunteers whose results were observed.
A forecast transfers evidence to a new population and therefore needs support for the similarity relevant to the treatment's effect.
Test an alternative prediction
Ask whether the preexisting performance gap would predict the later difference without training.
If it would, the observation fits a rival account; that weakens attribution without proving that training has no effect.
Map the process
An express registration desk doubles its processing capacity, but every registrant must next use an unchanged identity-check station.
Both stages are mandatory, so final throughput depends on more than the first desk.
Identify the decisive question
Was identity checking already operating at capacity?
If it was, faster registration can enlarge a queue instead of increasing completed processing.
State the qualified conclusion
Which information is most important in evaluating the projected fuel saving?
A delivery service proposes replacing large vans with smaller vehicles to reduce fuel expense. Each small vehicle uses less fuel per kilometer, but it can carry only half as many packages.
Which fact most weakens the recommendation?
A retailer's premium product has a higher profit margin per unit than its standard product. The director recommends dropping the standard product to increase total profit.
What is the strongest criticism of this inference?
A call center began evaluating staff by calls completed per hour. That measure rose, while repeat calls about unresolved issues also rose. Management declares service quality improved because the first measure increased.
The conclusion overlooks that
A subscription firm surveys only customers who renewed and finds high satisfaction. It concludes that dissatisfaction cannot explain the recent increase in cancellations.
Which explanation best reconciles these facts?
A factory installed a faster cutting machine, but finished output did not rise. Every cut part must pass through a coating line that was already operating at full capacity.
Which assumption is most questionable?
A firm compares employees who volunteered for training with those who did not. Volunteers subsequently closed more sales. The firm concludes that mandatory participation would increase every employee's sales by the observed difference.
Which conclusion is best supported?
A bank plans to reduce abandoned applications by shortening its online form. An internal audit finds that most abandonment occurs only after the completed form displays an unexpectedly high fee.
Which finding would most weaken that explanation?
An office reports fewer recorded safety incidents after replacing a confidential reporting system with reports signed by employees. Managers attribute the decrease to safer working conditions.
Which fact most strengthens the plan's financial rationale?
A manager claims that offering customers a free trial will increase paid subscriptions. Trials create familiarity, but the plan succeeds only if enough trial users later pay to cover the cost of all trials.
Which conclusion follows?
A store offers a discount only to members, but not every member receives one. Customer N received a discount.
Which question most helps evaluate whether the proposal reaches its goal?
An airport wants to reduce total passenger processing time. Faster kiosks reduce check-in time, but each kiosk user must then visit a document-check desk. The number of such desks will remain fixed.
Three fresh reserved problems. Choose your method without hints; explanations follow your response.
The purchaser's reasoning confuses
A contractor guarantees delivery by Friday only when materials arrive by Tuesday. Materials arrived on Tuesday, so a purchaser concludes that Friday delivery is guaranteed.
Which calculation best evaluates the proposed saving?
A company plans to replace a costly component with a cheaper one. The new component lasts half as long, while replacement requires shutting down the entire production line.
Which judgment is most defensible?
A retailer credits its new loyalty program for a rise in repeat purchases. The program began at the start of the annual holiday season, when repeat purchases have increased by a similar amount in each of the preceding five years.
Explain the decisive restriction or inference in your own notes, then answer three new reserved questions without hints. Review and classify errors before trying a different problem after a study gap.
17. Locate a downstream constraint, step 3
A local speed improvement alone does not establish a shorter total processing time.
This identifies a missing link without proving the proposal has no other benefit.