Back to the on-screen lesson ·
identify a value choice that is separate from an evidential claim
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.
You will identify a value choice that is separate from an evidential claim, explaining the inference and its limits in a supplied case.
Evidence can constrain a decision without supplying every priority. Record outcome estimates separately from the rule used to evaluate them.
| Term | What it means |
|---|---|
| Evidential claim | A claim about what observations support. |
| Value judgment | A judgment about what matters or how outcomes should be evaluated. |
| Inductive risk | The risk of error when drawing conclusions beyond direct observations. |
| Sensitivity check | Testing how a result changes when a declared assumption or weight changes. |
Two teams agree that a device issues unnecessary alerts on four trials and misses a target event on one trial. They nevertheless prefer different settings. One team gives high priority to avoiding missed events; the other gives greater weight to avoiding interruptions. Their disagreement need not concern the observed counts. It may concern how the consequences should be valued. Separating these roles makes the disagreement clearer without pretending that either measurements or priorities are unimportant.
An evidential claim describes what the observations support. A value judgment concerns what matters, how outcomes should be weighed, or what ends an inquiry should pursue. A decision often combines both. If you change the observed error rates while holding priorities fixed, a recommendation can change. If you change the priorities while holding the evidence fixed, it can also change. The same conclusion can therefore depend on premises of different kinds.
Write a decision argument in three parts: evidence about likely outcomes, a rule for evaluating those outcomes, and a resulting recommendation. This structure allows another person to challenge a factual premise without rejecting the goal, or challenge the evaluation rule while accepting the data. It also prevents the phrase 'the evidence says we should' from hiding a value assumption. Evidence can strongly constrain reasonable decisions without supplying every priority by itself.
The distinction does not imply that research is completely untouched by values until a final decision. Values can shape which questions are asked, which errors matter, which populations are studied, and how uncertainty is communicated. The task is to make those roles explicit and open to criticism. It is not to conclude that values make every factual claim arbitrary or that evidence can be ignored whenever it conflicts with a preferred outcome.
Another way: Epistemic values and practical priorities
Epistemic values concern the aims and quality of inquiry: accuracy, consistency, explanatory adequacy, and responsiveness to evidence are common examples. Practical or ethical values concern ends such as fairness, well-being, access, or avoiding particular harms. These categories can interact. A commitment to inclusion can improve evidence by revealing that a sample omitted relevant groups. A commitment to accuracy can support a practical goal by reducing preventable errors. The distinction tracks argumentative roles, not necessarily isolated motivations.
Consider choosing between two research designs. One uses a large convenient sample from a single location; the other uses a smaller sample spread across relevant settings. Accuracy about the target population may favor the second design even before an ethical argument about inclusion is added. If the distribution of benefits and burdens also matters, that adds another reason. State each reason separately so that its force can be assessed. Calling a design 'fair' should not conceal an unsupported claim about representativeness, and calling it 'scientific' should not conceal a choice about whose outcomes matter.
Values can influence question selection without changing the truth of the answer. A community may prioritize studying noise near schools rather than noise near warehouses. That priority determines where effort goes; it does not determine what a correctly conducted measurement records. The resulting evidence can still surprise the investigators. An inquiry that allows only a desired answer has replaced evidence-sensitive investigation with advocacy disguised as measurement.
The ideal of objectivity can therefore include transparent methods, independent criticism, disclosed assumptions, and correction procedures. It need not require pretending that researchers have no purposes or priorities. The important question is whether those commitments are handled in a way that preserves accountability to evidence. A transparent value choice can be debated; a hidden one may silently control the conclusion while appearing to be an unavoidable fact.
Another way: Inductive risk and standards of acceptance
Empirical conclusions often involve uncertainty. Accepting a claim too readily can produce a false positive, while requiring too much evidence can produce a false negative or delay. The practical consequences of these errors may differ. Inductive risk concerns the possibility of error when moving beyond directly observed data and the significance of choosing a standard of acceptance under that risk. Philosophers debate how non-epistemic values may appropriately affect such standards.
For a simple illustration, a library's humidity alert can be too sensitive, interrupting work unnecessarily, or insufficiently sensitive, failing to flag conditions that threaten paper. The observed frequency of each error is an evidential matter under a specified test. How much disruption to accept to reduce missed warnings is a decision question. It depends on consequences and priorities as well as the rates. The lesson uses fictional weights to make this combination visible, not to recommend a real preservation policy.
Do not confuse an acceptance threshold with a truth-maker. Deciding to act when the evidence reaches a chosen level does not make the underlying proposition true. A cautious team may temporarily treat a warning as actionable while acknowledging uncertainty. Another may reserve the phrase 'established conclusion' for stronger support. Clear communication distinguishes the current evidence, the action threshold, and the actual condition being investigated.
Different settings can justify different action thresholds while preserving the same factual assessment. A reversible low-cost inspection might be reasonable on weaker evidence than an expensive irreversible intervention. That does not automatically mean there are two different facts. It means decisions incorporate the costs and benefits of acting under uncertainty. An analysis becomes confused when this practical difference is misreported as proof that truth itself varies with convenience.
Another way: Make the weighting rule inspectable
A simple weighted score can expose how priorities affect a recommendation. Suppose a rule assigns one point to each unnecessary alert and five points to each missed event. A setting with four unnecessary alerts and one miss receives four plus five, or nine points. Lower scores are preferred under this declared rule. The arithmetic follows from the weights; the weights do not follow from arithmetic. A different defensible priority scheme could change the ranking while leaving the observed counts untouched.
Such scores are models of a decision, not complete moral accounts. They assume that the selected consequences can be represented and combined in the stated way. They may ignore distribution, rights, qualitative differences, or effects outside the observed period. Use them to understand the logic of a supplied argument, and state their limits. A score's precision can make hidden simplifications look authoritative unless the omitted considerations are also recorded.
Perform a sensitivity check by changing one declared weight while keeping the evidence fixed. If the preferred setting reverses, the recommendation depends materially on that priority. This is useful information for deliberation. It does not prove that either recommendation is irrational. It shows where disagreement about values matters and where further factual measurement alone may fail to resolve the choice. If the ranking stays the same over a reasonable range, the recommendation may be more robust to that particular disagreement.
Keep disagreements about facts separate from disagreements about weights during the check. A team cannot fairly defend its preferred setting by quietly replacing the test counts and the value weights at the same time. Record each change and its reason. Likewise, do not infer that a person disputes the evidence merely because they reject the resulting recommendation. Ask which premise they reject: the outcome estimates, the evaluation rule, the allowed options, or the inference connecting them.
An adequate final explanation contains the evidence, the declared priority rule, the calculated or reasoned recommendation, and an acknowledged limitation. For example, 'Under the supplied weights, setting B has the lower error score, but the score omits who bears the interruptions.' This is both clearer and more honest than calling B simply 'objectively best' without specifying the objective. It preserves the force of evidence while showing where judgment enters.
Another way: Communicating uncertainty is itself consequential
A report can mislead by hiding a limitation even when every printed number is correct. State whether counts come from a small trial, whether a comparison assumes equal conditions, and whether a recommendation depends on a contested weighting rule. This disclosure lets readers evaluate the argument. It is not an invitation to dismiss all results as uncertain. The aim is to communicate exactly how much the evidence supports and which further judgments are being asked of the audience.
Another way: Who bears an error?
Two settings can have the same total weighted score while distributing interruptions differently. One might inconvenience the same volunteer repeatedly, while another spreads the burden. If distribution matters to the decision, add that consideration explicitly rather than pretending the original total already represented it. Equal scores under a simple model establish equivalence only with respect to the consequences that model counts.
A fictional archive tests two alert settings over the same set of independently checked periods. Setting A produces six unnecessary alerts and one missed event. Setting B produces two unnecessary alerts and two missed events. The test conditions and reference labels are stipulated fixed. These four counts are the evidence for the classroom comparison; they should remain unchanged while the teams discuss priorities.
Under a rule assigning one point to an unnecessary alert and five points to a missed event, A scores eleven and B scores twelve. A is preferred under that rule. Under a different rule assigning two points to an unnecessary alert and three points to a miss, A scores fifteen and B scores ten. B is preferred under the second rule. The reversal comes from the weights, not from discovering different test results.
The teams should therefore discuss why each priority scheme is appropriate to the fictional archive. Interruptions might be distributed unevenly among volunteers, and a missed event might have consequences not represented by a single constant weight. These considerations can motivate refining the model. They cannot justify silently altering the measured counts to make a preferred setting win. If the counts themselves are challenged, the teams need a separate audit of the reference procedure and test conditions.
A transparent report presents both calculations and explains the dependence on priorities. It can also ask whether a third setting would improve both error counts, or whether a low-cost secondary check would change the available options. Evidence and values interact throughout that inquiry: priorities help identify a useful research question, and new results can reshape the decision. The role of philosophy is to reveal those connections while keeping observation, evaluation, and recommendation distinguishable.
A weighted score does not discover its own weights. Changing priorities can reverse a recommendation while the factual assessment stays fixed. Conversely, having priorities does not make measured counts optional. A transparent argument identifies both kinds of premise and acknowledges what its simplified decision model leaves out.
Record the evidential counts.
4 unnecessary alerts and 1 missed event
These observed results are held fixed for the decision exercise.
Record the declared weights.
1 point per alert; 5 per miss
The evaluation rule is a supplied value premise rather than an observed frequency.
Score unnecessary alerts.
4 × 1 = 4
Each event receives the weight assigned to its category.
Score missed events.
1 × 5 = 5
The higher weight expresses the stated priority for avoiding misses.
Combine and label the score.
Total = 9 under this rule
The sum evaluates the evidence only relative to the declared model.
Record setting A's counts.
6 alerts and 1 miss
The counts come from the stipulated common trial.
Record setting B's counts.
2 alerts and 2 misses
The same categories and observation period are used for comparison.
Apply weights of one and five to A.
6 × 1 + 1 × 5 = 11
The rule assigns the same category weights to both settings.
Apply the same weights to B.
2 × 1 + 2 × 5 = 12
Comparability requires holding the evaluation rule fixed.
Give the conditional preference.
A has the lower score
The recommendation follows under these weights, not under every possible priority.
Preserve the original evidence.
A: 6 alerts, 1 miss; B: 2 alerts, 2 misses
Sensitivity analysis changes the evaluation premise rather than the observations.
State the new weights.
2 points per alert and 3 per miss
The new rule places relatively more weight on interruptions.
Recalculate A's score.
6 × 2 + 1 × 3 = 15
The same arithmetic structure now uses the new priorities.
Recalculate B's score.
2 × 2 + 2 × 3 = 10
Both settings must be rescored under the identical replacement rule.
Compare the revised totals.
B now has the lower score
The preferred setting changes without a change in the measured counts.
Identify the source of disagreement.
The priority weights materially affect the decision
More repetition of the same counts would not alone resolve the value difference.
A setting has three unnecessary alerts and two misses.
Counts: 3 and 2
The supplied observations remain fixed through the calculation.
The declared rule assigns two points to an alert and four to a miss.
3 × 2 + 2 × 4
The value weights determine how the separate consequences enter the score.
Calculate the conditional evaluation.
A fictional decision model records 4 unnecessary alerts and 2 misses. Its declared rule assigns 1 point per alert and 3 per miss. Compute each contribution and the total.
| Response | |
|---|---|
| Alerts | |
| Misses | |
| Total |
A fictional trial records three alerts and two misses. The declared weights are two per alert and five per miss. Complete the decision score.
Score the alert category.
alerts points
Multiply the observed alert count by the declared alert weight.
Score the missed-event category.
misses points
Use the separate miss weight for the observed missed events.
Add the category contributions.
total points under this rule
The result remains conditional on the supplied priorities and omitted consequences.
Keep the same observed 4 alerts and 2 misses, but use a different declared rule: 3 points per alert and 2 per miss. Compute contributions and total without changing the evidence.
Alert contribution: alerts. Miss contribution: misses. Total under the declared rule: total.
A trial has 5 unnecessary alerts and no misses. The declared weights are 2 per alert and 7 per miss. Compute contributions and total; a high weight does not create an unobserved miss.
Alert contribution: alerts. Miss contribution: misses. Total under the declared rule: total.
Both teams observe 2 false alerts and 3 misses. Team A weights alerts 4 and misses 1; team B weights alerts 1 and misses 4. Calculate A's total, B's total, and number of observed misses shared by both.
| Your reconstruction | |
|---|---|
| A total | |
| B total | |
| Observed misses |
In a fictional archive simulation, one alert setting has 3 unnecessary interruptions and 2 missed humidity events. The supplied teaching rule assigns 2 points per interruption and 5 per missed event. Compute contributions and total under that rule only.
Alert contribution: alerts. Miss contribution: misses. Total under the declared rule: total.
Lesson test: one question per skill, one attempt each, no hints. Your answers are checked when you submit.
Two teams agree on 6 unnecessary alerts and 1 missed event. For the stated comparison use the second team's declared weights: 2 points per alert and 4 per miss. Produce contributions and total; do not treat those weights as additional observations.
Alert contribution: alerts. Miss contribution: misses. Total under the declared rule: total.
Without rereading, explain how to identify a value choice that is separate from an evidential claim. Give a fresh case and identify what would change your analysis.
10. Finish the declared score, step 3
14 points
The total follows from both the evidence and the weighting rule.