A survey reports 71% favorable. Before HR celebrates, ask: favorable about what, among whom, out of how many eligible employees, and compared with which earlier group? The number is useful only after its denominator and limitations are clear.
Employee Engagement accounts for 17% of the current PHR outline. The domain includes employee-lifecycle measures and performance-management activities, not merely running surveys. This lesson uses an original fictional dataset to practice analysis, action selection and follow-through. It is not an industry benchmark, a validated engagement instrument or a claim about any real employer. HRCI's current PHR outline provides the official domain description.
Distinguish three questions
| Area | Core question | Suitable evidence | Common mistake |
|---|---|---|---|
| Engagement and employee experience | What conditions support people's connection, participation and ability to contribute? | Carefully designed feedback, listening and workforce patterns | Treating one survey score as a complete diagnosis |
| Performance management | What outcomes and behaviors are expected, and what support or accountability is needed? | Job standards, observed work, feedback and agreed goals | Treating dissatisfaction as proof of poor performance |
| Employee relations | Is there a specific complaint, conflict, policy concern or protected right to address? | Case-specific facts and an appropriate process | Routing an urgent allegation into a future survey action plan |
These areas interact, but their evidence is not interchangeable. A productive employee can be dissatisfied. A satisfied employee can need coaching. A serious harassment or safety concern needs appropriate case handling even if the team's overall score is high.
The fictional Northline survey
Northline invites all 100 employees to a short survey during one defined week. Each person may submit once. The three items are scored on a five-point agreement scale; ratings of 4 or 5 count as favorable. Every respondent answers all three items, so each item has the same denominator within a team. The remaining responses may be neutral or unfavorable; this table does not distinguish them.
The items are:
- Clarity: “I understand the work priorities for my next shift or workday.”
- Resources: “I have the tools and information needed to do my work.”
- Speaking up: “I can raise a work concern without expecting unfair treatment.”
| Team | Eligible/invited | Respondents | Clarity favorable | Resources favorable | Speaking-up favorable |
|---|---|---|---|---|---|
| Day operations | 40 | 32 | 20 | 24 | 22 |
| Night operations | 30 | 12 | 3 | 6 | 4 |
| Office | 30 | 24 | 18 | 18 | 20 |
| Total | 100 | 68 | 41 | 48 | 46 |
These deliberately small, synthetic numbers make the arithmetic visible. Real reporting should apply a privacy and disclosure-risk process before displaying small groups or comments.
Step 1: calculate participation before interpreting sentiment
Response rate = respondents ÷ eligible invitees × 100.
- Day operations: 32 ÷ 40 = 80%.
- Night operations: 12 ÷ 30 = 40%.
- Office: 24 ÷ 30 = 80%.
- Overall: 68 ÷ 100 = 68%.
The overall rate hides a coverage concern. Night employees make up 30% of the eligible workforce but only 12 ÷ 68 = 17.6% of respondents.
HR should check access before interpreting silence: Was there paid time and an accessible device? Did the invitation reach the shift? Were language or accessibility needs met? Did employees understand who could see responses? A low rate does not tell HR which explanation is true.
Nonresponse bias becomes possible when people who respond differ systematically from those who do not. Increasing participation helps coverage, but a high rate alone does not prove an unbiased result. The Census Bureau's response-rate explanation describes this general methodological concern. Its survey-specific procedures are not employer-survey requirements.
Step 2: calculate item favorability with the right denominator
Item favorability = favorable valid responses ÷ all valid responses to that item × 100.
| Team | Clarity | Resources | Speaking up |
|---|---|---|---|
| Day operations | 20/32 = 62.5% | 24/32 = 75% | 22/32 = 68.8% |
| Night operations | 3/12 = 25% | 6/12 = 50% | 4/12 = 33.3% |
| Office | 18/24 = 75% | 18/24 = 75% | 20/24 = 83.3% |
| All respondents | 41/68 = 60.3% | 48/68 = 70.6% | 46/68 = 67.6% |
Interpretation: Among respondents, clarity is the least favorable item overall and on the night shift. The night-shift results warrant careful follow-up, but they do not establish what every night employee believes or why the pattern exists.
Avoid these errors:
- Calling 70.6% the company's “engagement score.” It is the resources item's favorability among respondents, not a validated composite.
- Calling the remaining 29.4% “disengaged.” Nonfavorable includes neutral responses in this scoring scheme.
- Dividing each item's favorable responses by all 100 invitees and labeling the result item favorability. That answers a different question.
- Using 68 for an item that some respondents skipped. In a real survey, disclose each item's valid-response denominator and treatment of “not applicable.”
Step 3: do not average team percentages casually
The simple average of the resources percentages is:
(75% + 50% + 75%) ÷ 3 = 66.7%.
But the actual respondent-level result is:
(24 + 6 + 18) ÷ (32 + 12 + 24) = 48 ÷ 68 = 70.6%.
The first number weights each team equally. The second weights each respondent equally. Use the second if the claim concerns the responding employees collectively. Neither automatically estimates all employees' views without assumptions about the missing responses.
One could weight team scores by the eligible workforce: 40% × 75% + 30% × 50% + 30% × 75% = 67.5%. That calculation assumes each team's respondents adequately represent its nonrespondents. The assumption is especially uncertain for the night shift. Weighting is not a substitute for investigating missingness.
A useful boundary instead of false precision
There are 48 favorable resources responses and 32 nonrespondents. If none of those 32 would respond favorably, the all-employee favorable proportion would be 48/100 = 48%. If all would, it would be 80/100 = 80%.
This 48% to 80% range is a logical bound under the stated data, not a confidence interval or an estimate of the likely result. It shows why “70.6% of all employees feel well equipped” overstates what was observed.
Step 4: watch the composition of follow-up responses
Imagine a follow-up survey with unchanged resources favorability within every team:
| Team | Follow-up respondents | Follow-up favorable | Team favorability |
|---|---|---|---|
| Day operations | 32 | 24 | 75% |
| Night operations | 6 | 3 | 50% |
| Office | 24 | 18 | 75% |
| Total | 62 | 45 | 72.6% |
The overall result rises from 70.6% to 72.6%, roughly 2.0 percentage points, even though no team's favorable percentage improved. Fewer responses came from the lower-scoring team.
Best interpretation: The apparent overall improvement may be entirely compositional. Compare team rates, response coverage, wording, survey timing and eligible populations before attributing a change to an intervention.
All numbers here are fictional. In a real report, a six-response group may require suppression or combination with another appropriate group. Set a reporting threshold before collection, then assess reidentification risk from job roles, demographics, linked filters and free-text comments. A minimum group size alone does not guarantee anonymity. Do not promise anonymity if the system retains identifying links; explain actual confidentiality and access controls.
Step 5: connect feedback to turnover and retention carefully
Use a separate fictional monthly workforce ledger. This example is not asserted to be the same survey population over time.
- Beginning headcount: 100.
- Hires during the month: 18.
- Separations during the month: 8, comprising six voluntary quits and two employer-initiated separations.
- Ending headcount: 110, because 100 + 18 − 8 = 110.
- Of the eight leavers, seven were present at the beginning; one was hired during the month.
Monthly turnover
For this organization's stated convention, average headcount = (beginning + ending) ÷ 2.
(100 + 110) ÷ 2 = 105.
Total monthly turnover = 8 ÷ 105 × 100 = 7.62%.
Voluntary turnover = 6 ÷ 105 × 100 = 5.71%.
Employer-initiated turnover = 2 ÷ 105 × 100 = 1.90%.
Rounded component percentages may not sum exactly to the rounded total. The two-point average is a convenient convention, not the only method. Daily or payroll-period averages may better represent a rapidly changing workforce. State the method and period, then use it consistently.
Beginning-cohort retention
Of the original 100 employees, 93 remain at month end.
Beginning-cohort retention = 93 ÷ 100 × 100 = 93%.
It is not 110 ÷ 100 = 110%, because new hires do not belong in the retained beginning cohort. It is also not simply 100% minus the 7.62% turnover rate: the numerator populations and denominators differ.
New-hire retention
For a different cohort of 20 hires, assume all 20 have reached a full 90-day observation window and 19 remain employed at that point.
90-day retention = 19 ÷ 20 × 100 = 95%.
Do not mix hires who have completed 90 days with people hired yesterday and call the result 90-day retention. Define treatment of transfers, leave, rehires and acquisitions before calculating.
The BLS JOLTS definitions distinguish quits, layoffs/discharges and other separations. An internal company's turnover convention is not automatically comparable with a published BLS series. Match classification, population, time period, denominator and any seasonal adjustment before benchmarking.
Causal boundary: A team with low favorability and high turnover warrants investigation. It does not prove that the survey issue caused departures. Role mix, labor-market conditions, tenure, a reorganization and other changes could matter. Use appropriately protected exit information, stay interviews and operational data to test explanations.
Step 6: choose a small, prioritized action plan
For Northline, start with hypotheses rather than a generic morale campaign. The following plan is an illustration, not a promise of a specific outcome.
Priority 0: handle urgent concerns separately
If a comment contains a credible allegation of danger, harassment or retaliation, route it promptly through the proper process. Do not wait 30 days for a focus group or use the survey to identify and punish critics. Avoid publishing identifying comments. The principles in employee and labor relations govern case handling.
Priority 1: fix access and clarify the night-shift problem
Owner: HR analyst and operations lead.
Within two weeks: Check how night employees received the survey, offer accessible participation and voluntary listening opportunities, and explain the reporting safeguards. Review shift handoff and schedule-change records. Ask what “unclear priorities” means in actual tasks.
Evidence sought: Whether instructions conflict, priorities change without notice, a supervisor is unavailable or the survey was misunderstood. These are different problems and require different remedies.
Success measure: Comparable opportunities to participate and a documented set of specific, corroborated work barriers. A higher survey score alone is not the goal of this step.
Priority 2: pilot one operational change
Owner: Night operations manager, with HR support.
For the next four weeks: If listening and records confirm handoff confusion, pilot one written priority list per shift and a short handoff with an identified escalation contact. Check that the process is usable and does not merely add paperwork.
Process measures: Percentage of sampled shifts with the agreed handoff, number of conflicting-priority incidents, and response time when clarification is needed. Establish the baseline before setting improvement targets.
Guardrails: Monitor workload, safety and service quality. A quicker handoff is not a success if essential information is omitted or employees must work unpaid to complete it.
Priority 3: close the loop and decide what to keep
Owner: Operations lead and HR.
At the end of the pilot: Tell employees what was heard, what changed, what was not changed and why. Repeat the same clarity item with comparable access and timing. Report eligible counts, valid responses and team composition alongside the result.
Keep, adapt or stop the intervention using both operational evidence and employee feedback. If participation remains weak or other changes occurred simultaneously, describe the uncertainty. Do not award managers bonuses solely for a higher favorable score or encourage them to pressure employees into positive responses.
A useful management summary
An evidence-respecting summary of the fictional survey would say:
“Sixty-eight of 100 employees responded. Clarity was favorable for 41 of 68 respondents, or 60.3%. Night operations had both the lowest response rate, 40%, and the lowest observed clarity favorability, 25% among 12 respondents. We will first check access and specific handoff barriers, then test one process change. These results do not establish all employees' views or prove the cause of turnover.”
That is more useful than announcing that “engagement is low” and launching an unrelated perk.
Practice with answers
1. Fifty people answer a survey, but only 45 answer the resources item. Thirty-six rate that item favorably. What is item favorability?
36 ÷ 45 = 80%. The 50 survey respondents are not the item's valid-response denominator. Report the five missing item responses separately.
2. Resources favorability rises by two percentage points while the night shift's participation halves. Is that evidence the intervention worked?
Not by itself. The worked example produces that increase with unchanged team percentages. Check composition and implementation evidence before attributing improvement.
3. A company begins with 200 employees. Ten of those employees leave, 15 people are hired and two of the new hires leave before period end. What is beginning-cohort retention?
190 ÷ 200 = 95%. Ending headcount is 203, but it includes 13 remaining new hires and is not the retention numerator. Total separations are 12; they belong in a separately defined turnover calculation.
4. A manager asks HR to identify the people who gave low survey ratings so they can be coached. What should HR do?
Do not turn confidential feedback into an unsupported individual performance judgment. Follow the disclosed privacy commitments, examine patterns and coach the manager on process improvements. Address any specific misconduct or safety allegation through an appropriate separate process.
Continue studying
- Use the PHR study plan to revisit denominator errors after a delay.
- Apply cost assumptions to retention proposals in total rewards.
- Practice case triage in employee and labor relations.
- Test these distinctions with PHR practice questions.
Primary sources and review date
Checked October 18, 2026. The survey, workforce ledger, calculations, management summary and action plan are original fictional teaching examples, not findings from the sources below.
- HRCI PHR exam content outline, current document marked ©2026; outline effective 2024.
- U.S. Census Bureau: response-rate definitions, methodological context for nonresponse. No Census-specific survey threshold is presented as an employer requirement.
- BLS JOLTS FAQs, definitions of separations and their categories. Internal example rates use the explicitly stated company convention, not a claim to reproduce BLS estimation.