Contents
What exactly are we talking about?
Three closely related mechanisms combine in a customer survey:
- Courtesy: the customer does not want to offend the interviewer, the adviser who has just helped them or the company asking for feedback. They soften their answer.
- Acquiescence: the tendency to agree with a statement, whatever its content. Smith (2004) and Harzing (2006) define it in this way and show that it varies from country to country.
- Social desirability: the tendency to give the answer that seems appropriate rather than the true answer. The review by Tourangeau and Yan (2007, Psychological Bulletin) shows that this type of error is common on sensitive questions.
These mechanisms do not mean that customers lie. They mean that part of the answer reflects the survey situation and not only the experience itself. The metrics themselves are covered in our guide to NPS, CSAT and CES; this page, which complements our customer experience guide, deals with how they are collected.
Distributions that are almost always skewed upwards
Peterson and Wilson (1992, Journal of the Academy of Marketing Science) established a finding that still holds: self-reported satisfaction measures are almost always skewed, with most answers at the top of the scale, and part of this positivity comes from the measurement instruments themselves.
The practical consequence: an average score of 4.3 out of 5 says little on its own. What is informative are the differences (between branches, between channels, between periods) and the share of low answers, which often carries the useful information.
The effect of survey mode
How you ask changes the answer. Three findings converge:
| Study | Comparison | Finding |
|---|---|---|
| Holbrook, Green and Krosnick (2003), Public Opinion Quarterly | Telephone and face to face | By telephone: more acquiescence, more undifferentiated answers and more socially desirable answers |
| Tourangeau and Yan (2007), Psychological Bulletin | Self-administered and interviewer-administered | Self-administration reduces socially desirable answers |
| Dillman et al. (2009), Social Science Research | Telephone, interactive voice response, post, web | On satisfaction questions, aural modes more often produce the most positive answer |
For a contact centre that calls its customers after an interaction, the risk is twofold: the mode is aural, and the call sometimes comes from the very company that provided the service. Our guide to contact-centre quality shows how to avoid an adviser being assessed by a survey they administer themselves.
Response styles across cultures: what we know, and what we do not
Comparative research establishes that response styles vary from country to country. Its detailed findings, however, diverge:
- Johnson et al. (2005, Journal of Cross-Cultural Psychology), across 19 countries, associate high power distance with more extreme answers, but with less acquiescence.
- Harzing (2006, International Journal of Cross Cultural Management), across 26 countries, finds on the contrary more acquiescence where power distance is high, as well as where collectivism and uncertainty avoidance are high. Her sample includes no Arab or North African country.
- Smith (2004) links acquiescence to family collectivism and uncertainty avoidance, and sees it as a meaningful communication style rather than mere error.
For the Arab world, Benstead (2018, PS: Political Science & Politics) shows that certain interviewer characteristics change answers to sensitive questions, while warning against the idea that Arab respondents are less sincere. Jones and Mitchell (2025, Public Opinion Quarterly) observe an interviewer-gender effect in five Arab countries, none of which is Morocco, and on political questions.
In short: we have not identified any published study measuring a response style specific to Moroccan customers, nor the size of any overstatement. Any "correction" figure applied to Morocco would be invented.
Designing a survey that limits bias
- Favour self-administered modes (web link, form, SMS) for measuring satisfaction, and keep phone calls for in-depth interviews.
- Separate the interviewer from the service being assessed: it is not for the adviser or branch concerned to collect the feedback.
- Ask questions anchored in facts: "Was your request resolved at the first contact?", "How many times did you have to call back?" rather than "Are you satisfied?".
- Balance the wording so that systematic agreement does not inflate the score, for example by alternating positive and negative statements.
- Always include a free-text field, and read it: verbatims reveal what the scale smooths over.
- Test each language version (French, Arabic, with Darija accepted in free-text answers) with a few customers before launch. Harzing (2006) observed that a questionnaire in English, a non-native language, produced more midpoint answers: language matters.
- Never mix modes within the same series without distinguishing them in the analysis.
These principles are in line with the methods presented in our guide to customer needs research.
Comparing what customers say with the facts
The best antidote to bias is comparison. If the survey shows high satisfaction while complaints are rising, customers are calling back several times about the same issue or account closures are increasing, it is the survey that needs questioning. De Haan, Verhoef and Wiesel (2015) measured only modest correlations between self-reported metrics and actual loyalty. See behavioural signals and cause-based complaint management.
For banks, these signals already exist in the quarterly complaint statements sent to Bank Al-Maghrib, as Bank Al-Maghrib's figures show. Any combination of survey data and customer data must also comply with Law 09-08: see privacy by design.
What not to conclude
- That Moroccan customers "over-rate" by a given number of points: no data supports this.
- That high power distance necessarily produces acquiescence: the research is divided.
- That a high score is worthless: its trend over time, with the mode held constant, remains informative.
- That changing the scale is enough: mode, interviewer and wording matter more.
Checklist before launching a survey
- Self-administered mode chosen, or aural mode justified and kept constant over time.
- Survey collected by a person or tool independent of the service being assessed.
- At least two factual questions and a free-text field.
- Language versions tested.
- A plan to compare results with complaints and behaviour set out from the start.
- Purpose of processing the answers defined, and customers informed.
Further reading
- 4 min read
- 7 min read
- 7 min read
Frequently asked questions
What is courtesy bias in a survey?
It is a respondent's tendency to give a polite answer rather than their real opinion, particularly when facing an interviewer or the company asking. It is close to social desirability (giving the answer that seems appropriate) and acquiescence (agreeing whatever the question). Its effect is to push satisfaction scores upwards.
Do telephone surveys produce higher scores than online surveys?
Research points that way. Dillman et al. (2009) observed that respondents surveyed by telephone or by interactive voice response chose the most positive answer more often than those who responded by post or online. Comparing scores obtained through different modes without taking this into account therefore leads to wrong conclusions.
Do Moroccan customers rate more generously than European customers?
We have not identified any published study measuring the size of any such effect in Morocco. International research shows that response styles vary across cultures, but its findings diverge, particularly on the role of power distance. Any correction factor should therefore be avoided, in favour of a survey design that reduces the bias.
Should satisfaction surveys be abandoned?
No. They remain useful for tracking a trend and collecting verbatims. They should, however, be designed to limit bias, tracked for their trend rather than their absolute level, and compared with complaints and with customers' actual behaviour.
In which language should customers in Morocco be surveyed?
In the language in which the customer expresses themselves most naturally, which often means French and Arabic versions, with free-text answers accepted in Darija. Each version should be tested with a few customers before launch, because the language of the questionnaire can itself change the distribution of answers.
General information, not legal advice.
Report an error