Norms, attitudes and behaviour are different things
People often use “norms” and “values” loosely, but researchers tend to separate several ideas that surveys can capture in different ways.
- Personal attitudes are what an individual says they think is right or acceptable.
- Perceived norms are what a person believes others around them think or do.
- Behaviour is what people actually do, which may differ from both.
- Values are usually broader and more stable orientations, such as the importance placed on tradition, autonomy or security.
A shift in one of these does not guarantee a shift in the others. A person may privately accept something while believing their community would disapprove, and may therefore act as if the old norm still holds. This gap between private views and perceived expectations is one reason change can look slow for a long time and then appear sudden.
How change is usually tracked
The most common tool is a question asked in the same way, to comparable samples, at intervals. Large cross-national projects such as the World Values Survey and similar long-running national surveys are built on this idea. When the wording and sampling stay stable, differences between waves can be read as change rather than noise. This is the logic behind tracking surveys, and it is the main reason a repeated question is usually more informative than a one-off poll.
Even so, a few features of the method shape what can be concluded.
Wording must stay put. Small changes in phrasing, answer options or the order of questions can shift results, a problem explored in work on question wording. If a question is reworded to reflect modern language, the trend line may break even if opinion did not.
The sample must stay comparable. If the way respondents are recruited changes, for example from face-to-face interviews to online panels, some of the apparent change may come from who is answering, not from what people think.
Questions must keep their meaning. A term that was once clear can acquire new connotations. Respondents in different decades may be answering slightly different questions without anyone noticing.
Why stated views may not match behaviour
Surveys mostly record what people say. When a topic carries social approval or disapproval, answers can drift toward whatever seems acceptable, a tendency known as social desirability. As a norm strengthens, this pressure can grow, so measured agreement may rise partly because it is more comfortable to voice. The reverse can also occur if a topic becomes polarised and people avoid giving an answer.
Direct behavioural measures are often harder to obtain, which is why research relies heavily on self-reported behaviour. Self-reports are useful but imperfect, particularly for sensitive or socially charged activities. Where administrative records, observation or experiments are available, they can complement surveys, although each covers only part of the picture and has its own blind spots.
Lagging and leading
Attitudes sometimes lag behaviour. People may change what they do because of economics, law, technology or opportunity, and only later adjust what they say is acceptable. For example, a practice can become common because circumstances make it practical, and attitudes catch up once it feels ordinary.
Attitudes can also lead behaviour. People may endorse a principle long before institutions, costs or daily routines allow them to act on it. Surveys on intentions and ideals often run ahead of conduct for this reason.
Which pattern applies to a given topic is an empirical question, and it often varies by group. The general point is that a change in stated views is evidence about opinion, not automatic evidence about conduct, and the reverse holds too. Research on why attitude change on deeply held questions tends to be gradual helps explain why some values appear stable across many survey waves.
Age, period and cohort
When a survey shows differences between older and younger people, three explanations are commonly confused.
- Age effects: people may hold different views at different life stages, and today’s young adults may resemble yesterday’s young adults when they were that age.
- Period effects: an event or broad climate affects everyone at the same time, regardless of age.
- Cohort effects: people who grew up in a particular era carry distinctive views as they age.
Cross-sectional data, a single snapshot, cannot cleanly separate these. Repeated surveys help, but disentangling all three statistically is difficult because they are mathematically linked: knowing someone’s age and the survey year fixes their birth cohort. This is also why popular generational labels can mislead. They bundle people born across many years into a single group and then attribute broad differences to birth year alone, when age and period may explain part of the pattern.
What a trend line leaves out
A national average can hide movement in opposite directions among different groups, or stability in the middle with change at the edges. Regional, religious, educational and urban-rural differences may matter more than the overall figure. A shift in the share holding a view also says nothing about how strongly it is held or how important it is to those people.
Finally, change in measured norms does not tell you why it happened. Rising education, migration, new media, economic conditions or legal change might each contribute, and surveys alone seldom isolate one cause.
Questions worth asking of any claim about shifting values
- Was the same question asked the same way over time?
- Were the samples comparable across waves?
- Does the claim concern personal attitudes, perceived norms or behaviour?
- Could age or period, rather than generation, explain the pattern?
- Which groups does the average hide?
If a source cannot answer these, a headline about a dramatic change in values is best treated as a prompt for further reading, not a settled finding.