Why people believe something is sensible but still do not do it

Ajzen's (1991) theory of planned behaviour distinguishes attitudes, perceived social norms and perceived behavioural control as factors relevant to intentions. That is a useful starting point. Employees might value safety but assume that nobody in their team reports small incidents. Others may want to report yet believe the digital system cannot be used during a busy shift.

These elements do not amount to a complete explanation for every action. Habits, contextual signals, competing demands, organisational decisions and actual ability to act may also matter. It is more useful to treat the model as a framework for questioning than as a machine for predicting an individual's next decision.

There is a further distinction between endorsing a general value and forming a concrete intention. 'Safety matters to me' describes a broad objective. 'I will record every relevant near miss during my shift' specifies an action. Even the latter statement does not establish whether the employee can reach the form, knows what qualifies, or trusts the handling of their report.

Effective interviews therefore reconstruct recent events rather than rely only on abstract agreement. What did the person see? What happened next? Who was nearby? When was the decision not to report made? Answers about a specific episode can reveal context that a general opinion question cannot. They also reduce the temptation to interpret declared goodwill as reliable proof that action is already possible.

Organisations often measure behaviour far away from the point at which it should occur. A survey completed on Friday asks what happened on Tuesday. Memory, social desirability and retrospective explanations all enter the response. The answers may still be informative, but they should not be confused with an observation made at the relevant moment.

Define the target behaviour precisely. For our fictional packaging factory, it is not 'raising safety awareness'. It is recording a near miss through an accessible channel, within an agreed time frame, with enough information for responsible follow-up. Specify who is expected to act and what constitutes a genuine opportunity to report. Someone who has no opportunity should not automatically be described as unmotivated.

Reconstruct the event as a sequence: a signal is noticed, interpreted, evaluated and either acted upon or postponed. Practical barriers can enter at any stage. Perhaps the reporting criteria are unclear. Perhaps the workstation requires a password the employee does not have. Perhaps the form has too many fields or the team assumes permission must be obtained first.

At that level of detail, a seemingly psychological problem can become a straightforward process problem. That is a valuable finding. A sophisticated explanation of hidden resistance adds little if the most plausible cause is a system that employees cannot use while working.

The matrix shows how identical visible behaviour can arise from very different causes. These are research hypotheses, not measured Neurofactor scores or findings from a real client.

The rows are not six fixed personality categories. One employee may face several barriers; another may experience no psychological resistance yet still lack the required information. The matrix is intended to structure a diagnosis, not to classify people permanently.

Use it as a guide to the next research question. If the reporting route is unavailable, make it available before trying to improve attitudes. Once the action is feasible, the rules are clear, and follow-up is credible, there is a stronger basis for investigating why the action still does not occur.

An association map can help explore the meanings employees attach to the word 'report'. Possible associations include prevention, professional care, shame, control, loyalty, criticism and administrative burden. The objective is not merely to produce a colourful collection of words. The important questions are whose associations were measured, in which situation, against which alternatives and with what measurement quality.

Separate the objective from the behaviour. 'Safety' may evoke strongly positive associations, while 'filing a report about a colleague' may evoke uncomfortable ones. If these phrases are collapsed into a single survey item, the tension disappears. The same applies to 'learning from mistakes' compared with 'making my own mistake visible'.

Methods for measuring associations can include explicit ratings, properly designed association tasks and qualitative conversations. An implicit measure is not a direct readout of a person's hidden truth. Interpretation depends on the stimuli, comparison categories, task design and reliability. Those conditions must be documented before a pattern is translated into practical advice.

The greatest value of an associative lens is the generation of alternative, testable explanations. Perhaps reporting is associated with professional pride; perhaps with public embarrassment. That hypothesis becomes more credible when independently gathered behavioural and contextual data point in the same direction. Without that step, the map remains a possible interpretation rather than a demonstrated causal diagnosis.

Fictional illustration - not a Neurofactor client project, field study or measured outcome. A packaging manufacturer wants to identify near misses earlier so its teams can improve machinery and work processes. Employees agree that safety should take priority. Yet the formal reporting system is rarely used for events that could have caused harm but did not.

Management initially considers another awareness campaign, with clearer signs and monthly reminders. But imagine that exploratory conversations reveal several possible concerns. Some operators want to get the production line moving again after a small disruption and view an official report as unnecessary paperwork. Others expect an argument with colleagues or worry that management will ask who was at fault. Some simply do not know how significant an event must be before it qualifies. Others remember submitting reports without ever hearing what happened next.

These statements are invented examples of potential explanations. They are not findings from a survey, and no percentages or brain measurements are being claimed. A genuine study would need to examine actual reporting opportunities, the definitions used by the organisation, the reporting steps, previous follow-up and employees' experiences. It would also need to assess the reliability of the underlying records. Few registered incidents cannot automatically be interpreted as proof that few incidents happen or that many events are deliberately concealed.

Different findings would point towards different improvements: simpler access to the reporting process, clearer examples of reportable events, more transparent follow-up, or fairer organisational responses. Which change is appropriate cannot be known in advance. The choice should be based on the diagnosed barrier and evaluated under realistic working conditions.

A behavioural diagnostic framework helps keep possible explanations distinct. Within Michie and colleagues' (2011) Behaviour Change Wheel, COM-B groups relevant conditions under capability, opportunity and motivation. These are not separate areas of the brain, nor are they diagnostic scores obtained from a single task. It is a conceptual framework for deciding what to investigate.

First specify the behaviour. 'More reporting' is vague when the number of relevant opportunities is unknown. Ask when an incident can be recorded, what information is required, whether employees can complete the action, and whether the system provides adequate follow-up. Record what observation would support each proposed barrier and what evidence might contradict it.

When studying motivation, consider not only willingness but also the perceived consequences of taking action. When studying opportunity, audit the physical, digital and organisational environment. When studying capability, examine knowledge, recognition and skill. Finally, look at habits: what happens almost automatically when production resumes and another task demands immediate attention?

Do not compare departments without accounting for different exposure to incidents, team sizes, work patterns and reporting systems. A league table built from raw reports can wrongly suggest that one team has a 'better safety culture' simply because it generates more entries. A meaningful diagnosis needs the opportunity and the denominator, not just the recorded numerator.

Start with interviews about specific recent situations. Ask what happened and what people expected, rather than collecting only general opinions. Compare these accounts with process observations and an audit of the reporting channel. Such basic work may identify an obvious barrier more quickly than a complex psychological experiment.

Association tasks can add another perspective when words such as responsibility, reporting, mistakes and criticism appear to have competing meanings. But scores are generally relative to a task and its conditions. Triangulation in research can strengthen an interpretation when multiple independent methods converge. Contradictory evidence should not be hidden; it tells researchers that the initial explanation may be incomplete or wrong.

EEG records electrical brain activity, while eyetracking records patterns of visual attention. Both can be useful for clearly defined research questions. Neither directly proves whether an employee will report an incident, how psychologically safe an organisation is, or what a person's true intention must be. Greater visual attention to a safety message does not demonstrate an improvement in reporting behaviour.

A simple principle follows: measure actions with appropriate behavioural records, meanings with suitable association and qualitative methods, and organisational conditions through observation and process evidence. Document the limitations. An impressive-looking instrument should never displace a more directly relevant observation.

A communication campaign can clarify rules, raise awareness and make a topic easier to discuss. But when people already know what is expected, repeating that expectation is not automatically the best next step. A workshop may ask employees to agree once again with something they have already endorsed, while leaving the actual barrier untouched.

Existing routines can also compete with a new goal. Wood and Neal (2007) describe how habits are connected with familiar contexts and learned response patterns. Lally and colleagues (2010) studied the development of automaticity in everyday actions and found considerable variation in how habits developed. There is no universal rule that a behaviour becomes automatic after a fixed number of days, and a strong intention does not immediately overwrite an established routine.

For the packaging factory, the familiar routine is to resolve a disruption and continue production. If reporting is to become part of the workflow, the team must identify a safe and reliable point at which it can happen. A short report may be possible once the immediate hazard has been handled. However, recording should never delay urgent safety measures or conflict with professional procedures. Safety expertise, not a communications campaign, must determine the correct operational order.

The goal is not to manipulate people into compliance. It is to make the desired action clear, safe and achievable while addressing the organisational consequences associated with it. If the employer does not keep its promises about how reports are handled, employee motivation is not the only issue that needs to change.

The second matrix guides research priorities, not automatic intervention selection. The following article in the series examines how to choose between communication, process and environmental changes; a later article focuses on real-world evaluation.

An option should only be preferred when it plausibly addresses the identified issue without creating greater harm. A default setting can simplify some actions but cannot repair unfair management responses. Rewards for higher reporting counts can create undesirable incentives if the quality and meaning of reports are ignored.

Michie and colleagues' work on the Behaviour Change Wheel and the behaviour change technique taxonomy (2011; 2013) helps describe interventions precisely. That is important for later evaluation: 'we launched a behaviour programme' says little about what changed, when it changed and which parts employees actually encountered.

A person may intend to act but fail to recognise the appropriate start signal. Implementation intentions connect a specific situation to a response: 'If situation X occurs, I will do Y.' Gollwitzer (1999) and Gollwitzer and Sheeran's (2006) meta-analysis describe how this type of planning can support goal achievement under certain conditions. Planning cannot automatically make a workplace safe, remove a technological barrier or correct unfair treatment.

An appropriate example might be: 'If an immediate hazard has been handled according to the safety procedure, I will register the near miss through the accessible reporting channel.' The order matters. Protecting people and following relevant safety procedures must always come before an administrative or research objective.

Planning is especially relevant when someone wants and is able to act but forgets, misses an opportunity or becomes distracted. If employees face social punishment, lack permission, or cannot access the system, an if-then statement does not resolve the root cause. Sniehotta and colleagues (2005) examined planning, self-efficacy and action control in a different behavioural setting. That supports further questions about planning; it does not provide an industrial safety outcome guarantee.

Before recommending a prompt, check whether the cue is recognisable, the action is allowed, the resources are present and the organisational response is trustworthy. Only then is it sensible to test whether a reminder or a specific plan improves execution.

An organisation might celebrate a rise in stated willingness to report. But what happens after a real near miss? Behavioural measurement requires a defensible understanding of the opportunities in which reporting could occur. Near-miss reporting is especially difficult because unreported events are, by definition, often absent from the database. The raw number of reports is therefore not a clean measure of safety or willingness to report.

Where feasible, combine reporting records with responsible safety observations, clearly defined reporting categories, process information and qualitative follow-up. Examine usefulness as well as quantity. Can the reports help identify relevant conditions and prevent recurrence? A rise in reports might be good news because staff are speaking up. It might also accompany more actual incidents. Those explanations cannot be separated using the count alone.

If the organisation wants to claim an intervention caused an improvement, it needs an appropriate comparison. A staged implementation or controlled field study may be possible where it does not compromise safety and where the design can handle differences between teams. Blog G-069 covers this in greater depth. Here the important distinction is that more positive attitudes, stronger associations or increased perceived safety are not the same as demonstrated behaviour change.

Monitor unintended consequences as well. Excessive administration, accidental identification of reporters, a misleading reduction in recorded incidents or reduced trust can damage the very outcome the intervention was meant to improve. Useful evaluation looks beyond a single headline number.

The first mistake is treating all non-action as resistance. An employee who has no access to the reporting system does not necessarily have a persuasion problem. The second is the reverse: assuming all missing action is technical friction while ignoring what the action could mean socially.

The third mistake is interpreting an average score as one shared cause. Departments and situations can differ, and positive and negative associations may coexist within the same group. The fourth is presenting a rise in intention as completed behaviour change. Research into the intention-behaviour gap shows why that leap is not justified.

The fifth mistake is allowing an appealing research method to become the explanation itself. An association score, EEG response or attention to a campaign does not determine whether the real reporting process works. Nor should an inferred psychological characteristic be used to judge or select an individual employee.

A stronger diagnosis asks different questions: was the action possible, what did it mean to the employee, what cues and routines competed with it, and what evidence would distinguish one explanation from another? Without those questions, the intervention remains a guess dressed as behavioural science.

Use this sequence to design an initial investigation without deciding on an intervention before understanding the problem.

1. Specify the action. Identify who should do what, at which recognisable moment, where and under which conditions, then define a meaningful behavioural outcome.
2. Identify the opportunity. Determine how frequently relevant moments occur and whether people have a safe and practical way to act when they arise.
3. Reconstruct real situations. Combine recent experiences, workflow observation and process evidence rather than relying exclusively on statements about importance.
4. Explore competing associations. Compare positive meanings attached to the objective with potentially negative meanings attached to the actual action in each relevant context.
5. Separate motivation from possibility. Use a framework such as COM-B to examine capability, opportunity, habits, norms and expectations as alternative explanations.
6. Formulate a small falsifiable hypothesis. State the suspected barrier, the proposed change, the expected behavioural signal and the unwanted effects that must be monitored.
7. Observe what actually happens. Choose a responsible comparison, review implementation and unintended consequences, and revise the diagnosis if the expected result fails to appear.

This article provides the diagnostic part of a larger process. G-013 goes further into selecting communication, process or environmental interventions. G-069 addresses testing those interventions under real-world conditions. A plausible explanation and a demonstrated improvement are separate achievements, and each deserves its own evidence.

What if the reason someone wants to act also makes the action uncomfortable? An employee may sincerely want to protect colleagues while fearing that an incident report will cause one of them trouble. Both meanings can be real. Persuasion alone cannot necessarily remove that conflict.

Believing something is sensible and doing it are different things. The bridge between them depends on opportunities, abilities, routines, social context, expected consequences and the meaning attached to the action. Association research can help formulate relevant hypotheses, provided it is not mistaken for proof of a hidden cause or a guaranteed intervention effect.

The practical conclusion is straightforward: do not increase the volume of persuasion before establishing where action actually breaks down. Investigate the moment. Make the necessary change. Evaluate what people can and do, not only what they say they intend to do.

For a starting point, explore the behaviour-change association hub and its introduction. Neurofactor can help specify the association research and behavioural question without claiming more than the available evidence can support.