Why Employees Lie on Engagement Surveys (and How to Get Honest Answers)

Employees do sometimes lie on engagement surveys. More often, they edit themselves. They choose a safer score, avoid the most sensitive question, remove identifying detail, write a bland comment, or decline to participate.

That behaviour is easy to blame on cynicism. It is more useful to ask what the employee believes will happen after pressing submit. If the answer could be traced back to them, used against their manager, dismissed by leadership, or exposed through a recognisable comment, holding back may be a sensible act of self-protection.

Honest feedback therefore depends on two beliefs:

1. It is safe to say. The response will not create an unacceptable personal risk.
2. It is worth saying. Someone will understand the pattern and respond constructively.

A well-designed survey must support both.

What does “lying” on an employee survey look like?

Not every inaccurate answer is a deliberate falsehood. Survey responses can be distorted by fear, ambiguity, memory, fatigue, social pressure, or a response scale that does not fit the employee's experience.

Response behaviour	What may be happening	What the result loses
Giving a more positive rating	Avoiding conflict, identification, or disloyalty	The severity of a real problem
Selecting the middle option	Choosing the least exposed position	A clear positive or negative signal
Skipping a question	Protecting privacy or avoiding a poor-fit item	Evidence on the most sensitive topics
Writing a vague comment	Removing details that could reveal the author	Context needed to understand the issue
Giving the same answer throughout	Rushing, disengagement, coercion, or fatigue	Meaningful variation between issues
Not participating	Fear, lack of access, low confidence in action, or simple absence	Representation of part of the workforce
Choosing the nearest available option	The response set does not match reality	Measurement accuracy

An employer cannot look at one response and confidently decide which explanation applies. Trying to identify or interrogate the respondent would damage the very conditions needed for future candour.

The right response is to improve the system and examine patterns at a safe group level.

Why do employees hold back on engagement surveys?

1. They believe their answer can identify them

An invitation can say “anonymous” while the surrounding experience suggests otherwise. Employees may receive unique links, see questions about a small team, be asked to describe a distinctive incident, or know that managers can filter results by role and location.

Their concern is not limited to names. A project, shift, job title, writing style, timing, or combination of demographic filters can reveal who wrote a response. If the employer can open raw comments or export participant-linked answers, employees are being asked to trust that access will not be misused.

The 2024 American Economic Journal study Secure Survey Design in Organizations examines settings where fear of retaliation makes truthful reporting difficult. Its theory and laboratory experiments show that the security of the elicitation design changes what information people transmit. A privacy promise is not equivalent to a design that constrains access.

2. They expect a personal or social cost

The feared consequence does not have to be formal retaliation. An employee may worry about being labelled difficult, damaging a relationship with a manager, losing access to desirable work, exposing a colleague, or making life harder for the team.

The research literature calls withholding relevant concerns employee silence. Elizabeth Morrison's 2023 review of employee voice and silence research synthesised 158 papers selected from research published between 2014 and 2021. It found that leadership, perceived safety, status, relationships, workplace context, and beliefs about impact all shape whether employees speak up.

Silence also has different motives. A person may remain quiet because they are afraid, resigned, or trying not to embarrass or harm someone else. Treating every cautious answer as bad faith hides these distinctions.

3. They believe speaking up is pointless

Safety is only half the decision. An employee may trust the survey's privacy and still decide that a frank answer is not worth the effort.

Morrison's review highlights evidence that psychological safety has a stronger relationship with silence, while perceived impact has a stronger relationship with voice. In other words, reducing fear helps remove an inhibitor, but employees are more likely to contribute when they also believe their input can matter.

Repeated surveys without visible action teach a clear lesson: the organisation collects opinions but does not use them. Employees may then skip the next survey, complete it mechanically, or provide the answers least likely to create follow-up work.

This is why a “you said, we did” poster is not enough when nothing substantial changed. Credible follow-through includes decisions that leadership could not accept as well as those it adopted. Employees need to know what was heard, what evidence informed the decision, who owns the next action, and when progress will be reviewed.

4. The socially acceptable answer is obvious

Survey respondents often try to present themselves in an acceptable light. In a workplace, the approved answer can be particularly visible:

• “I support the transformation.”
• “My team always puts customers first.”
• “Senior leaders communicate effectively.”
• “I intend to remain here for several years.”

A question framed as a loyalty test invites a loyalty answer. Agree-disagree statements can also encourage acquiescence, the tendency to agree with the proposition as written.

Pew Research Center's questionnaire-design guidance explains that wording, response options, mode, and question order can all alter answers. It recommends clear, specific questions, complete and non-overlapping options, one concept at a time, and pretesting.

A 2026 systematic review of methods for reducing social-desirability bias examined 121 experiments across 79 peer-reviewed papers. Bias was significantly reduced in 55% of experiments overall, but effectiveness varied substantially by approach and topic. There is no single wording trick that guarantees honesty.

5. The question is impossible to answer accurately

Some apparent dishonesty is a measurement failure.

Consider: “Leadership communicates clearly and makes fair decisions.” An employee may think communication is clear but decisions are unfair. One rating cannot represent both judgements. The same problem appears when a question asks about “management” without saying whether that means the direct manager, executive team, or organisation as a whole.

Other common defects include:

• No “not applicable” option for an experience the employee has not had.
• A time period that is absent or too broad.
• Vague words such as “regularly”, “supported”, or “effective”.
• Scales whose labels change between questions.
• Questions that assume the desired practice already exists.
• A forced yes/no choice for an experience that is conditional.
• Two issues joined in one question.

When the instrument cannot represent reality, the nearest selectable answer may be inaccurate without being deceptive.

6. Completion pressure changes the task

Managers understandably want a strong response rate. Pressure becomes counterproductive when employees are chased individually, asked to complete the survey in front of a supervisor, or treated as non-compliant for declining.

At that point, the task can shift from “tell us what work is like” to “make the reminder stop”. A rushed, straight-line submission increases the response rate while reducing the value of the data.

Operational systems may need to know whether a private invitation has been used so they can send reminders and enforce one response. That status should remain separate from answers. Employees should be told exactly who can see delivery and completion, and managers should not be able to connect those events to ratings or comments.

7. Written comments feel uniquely exposing

Free text can reveal more than a score. A recognisable phrase, rare role, named project, customer incident, roster pattern, or personal circumstance may identify the author even after names are removed.

Employees often respond by stripping useful context from the comment or avoiding it entirely. Publishing raw comments then reinforces the concern, especially when leaders discuss who “must have written” a particular sentence.

A safer reporting model analyses restricted source comments and gives the employer only broader findings supported by several independent answers. Distinctive wording and unsupported observations should not cross the reporting boundary.

How can you get more honest employee survey answers?

Make anonymity enforceable

Decide what the employer is technically prevented from seeing, not merely what administrators promise not to inspect.

A strong employer-anonymous design should:

• Keep participant identity and contact operations separate from submitted answers.
• Prevent employer access to participant-linked responses.
• Apply a fixed minimum before any group result is reported.
• Block filter combinations that reduce a result below that minimum.
• Avoid raw response and individual-comment exports.
• Convert written feedback into paraphrased, supported findings.
• Explain whether delivery and completion status are visible and why.

If anonymity can be disabled silently, say that the survey is confidential instead. Accurate language builds more trust than an absolute claim the system cannot enforce.

Explain the boundary before asking for trust

Employees should not need to understand survey software to assess their risk. The invitation and opening screen should answer:

1. Can the employer see who was invited and who completed the survey?
2. Can the employer see individual ratings or comments?
3. What is the minimum reporting group?
4. What happens to groups below that minimum?
5. Will written comments be quoted or paraphrased?
6. Who processes the source responses?
7. When will results and next steps be shared?

Avoid saying only “your answers are anonymous”. Describe the mechanism in plain language.

Ask questions people can answer

Each question should measure one idea, name a useful time period, and offer every reasonable response. Use “not applicable” or a legitimate skip path when the employee may lack the relevant experience.

Compare these examples:

Weak question	Safer, clearer question
Management communicates effectively	In the past three months, how clear have executive updates been about decisions that affect your work?
I have a manageable workload and enough resources	I can usually complete my expected work within my paid hours
Do you support the new operating model?	What effect has the new operating model had on your ability to complete your work?
My manager supports my development	In the past six months, has your manager discussed a specific development goal with you?

Pretest new questions with people from different roles and work settings. Ask them what they thought each question meant and whether any answer option was missing.

Make participation voluntary in practice

Invite everyone who is eligible, provide paid time and accessible devices, and send neutral reminders. Do not ask managers to identify non-participants or demand proof of completion.

Track participation coverage by safe operational groups so you can improve access without opening answers. If one shift has a much lower response rate, investigate timing, device access, language, and supervisor behaviour rather than trying to uncover what particular employees thought.

Show that difficult feedback survives the process

Honesty grows when employees see that negative findings are not edited into harmless language. Reports should distinguish strengths from pain points, retain uncertainty, and state when evidence is mixed.

Leadership then needs to respond without trying to identify critics. A useful response is:

• “This is what the group result shows.”
• “This is what we need to understand better.”
• “This is what we will do now.”
• “This is what we cannot change and why.”
• “This is when we will report progress.”

The reaction to the first uncomfortable result is part of the design of the next survey.

How can HR detect unreliable results without identifying respondents?

No data-quality signal proves that an individual lied. Use them to assess the instrument and collection process at aggregate level.

Review:

• Question-level skipping: Sensitive or confusing items may have higher non-response.
• Response concentration: An extreme ceiling can reflect genuine strength, but also leading wording or pressure.
• Straight-lining: Identical answers across a long battery may indicate speed or fatigue, especially when reverse concepts are included carefully.
• Completion time distribution: Very fast completion can suggest low attention, but accessibility tools and familiarity also affect time.
• Written-feedback participation: A sharp gap between scaled and written answers can indicate exposure concerns.
• Group representation: A company-wide average may hide that one shift, location, or employment group barely participated.
• Trend breaks: Sudden improvement after a change in wording, survey mode, privacy design, or manager pressure may be a measurement change rather than a workplace change.

Do not create a secret “honesty score” for employees. The purpose is to find where the survey process needs improvement, not to turn protected feedback into performance monitoring.

A pre-launch honesty checklist

Before sending an engagement survey, confirm that:

• The privacy claim matches the system's actual access controls.
• Employer users cannot open participant-linked answers.
• Small groups and unsafe filter combinations are suppressed.
• Written feedback will not be exposed as raw, attributable comments.
• Employees are told what delivery or completion status is visible.
• Each question measures one concept in clear language.
• Time periods and subjects such as “your manager” or “executive leadership” are specific.
• “Not applicable” and skip behaviour work where needed.
• The survey has been pretested with representative roles.
• Reminders are neutral and participation is voluntary in practice.
• Leaders have agreed how and when they will report results.
• Difficult findings can be published without searching for their authors.
• There is a separate confidential route for urgent individual matters that require follow-up.

If several boxes remain unchecked, the organisation is not ready to interpret a high response rate as honest engagement.

The honest answer is a product of the system

Employees are not passive measuring instruments. They assess risk, usefulness, wording, and the organisation's history before deciding what to share.

The most productive question is therefore not “How do we stop employees lying?” It is “What does our survey make an honest answer cost?”

Reduce that cost with an enforceable privacy boundary, protected group reporting, clear questions, voluntary participation, and visible action. Employees can then spend less effort protecting themselves and more effort explaining what the organisation needs to understand.

Candora is designed so employers receive privacy-safe group results and supported findings rather than participant-linked answers or individual comments. Read Are employee surveys really anonymous? for the practical anonymity test, or use What to tell employees before a survey to prepare the invitation and follow-up.