Asking Demographic Questions – ChatGPT Tip #1

Published by Ray Poynter, 26 September, 2023


NewMR Tips

One way to use ChatGPT is to check your questionnaire design by asking ChatGPT to generate a questionnaire. In the example below, I have asked ChatGPT to show me the most important demographic questions to include in a survey. I can then compare it with my list to see if I have forgotten any key questions or options.

My question to ChatGPT is a little unfair as I have not described the nature of the study, but that makes for a more general example for this post

The Prompt was
“When conducting a market research study, what are the most important demographic questions to ask? Please show me a questionnaire which includes these questions.”

Here is the reply

Certainly, demographic questions help to segment your study’s respondents into specific groups for analysis. This can help you understand consumer behavior, preferences, and needs more comprehensively. Below is a sample questionnaire that includes essential demographic questions commonly included in market research studies:

# Market Research Questionnaire

## Demographic Information

**Note**: Your answers to the following questions will be kept confidential and will only be used for research purposes.

  1. **Age**

   – Under 18

   – 18-24

   – 25-34

   – 35-44

   – 45-54

   – 55-64

   – 65 or older

 

  1. **Gender**

   – Male

   – Female

   – Non-Binary/Third Gender

   – Prefer not to say

 

  1. **Education Level**

   – High School or less

   – Some College

   – Bachelor’s Degree

   – Master’s Degree

   – Doctoral Degree

   – Prefer not to say

 

  1. **Employment Status**

   – Employed full-time

   – Employed part-time

   – Unemployed

   – Student

   – Retired

   – Homemaker

   – Prefer not to say

 

  1. **Income Level (Annual)**

   – Less than $25,000

   – $25,000-$49,999

   – $50,000-$74,999

   – $75,000-$99,999

   – $100,000 or more

   – Prefer not to say

 

  1. **Geographic Location**

   – Urban

   – Suburban

   – Rural

 

  1. **Ethnicity**

   – Caucasian/White

   – African American/Black

   – Hispanic/Latino

   – Asian/Pacific Islander

   – Native American

   – Other: __________

   – Prefer not to say

 

  1. **Marital Status**

   – Single

   – Married

   – Separated

   – Divorced

   – Widowed

   – Prefer not to say

 

  1. **Household Size**

   – 1

   – 2

   – 3

   – 4

   – 5 or more

 

  1. **Do you have children?**

    – Yes

    – No

    – Prefer not to say

Remember that these are just some of the most common demographic questions. Depending on your market research objectives, you might want to add or remove certain questions. Always ensure that your questionnaire follows ethical guidelines and privacy laws applicable to your region or industry.

There are several of these that I would take issue with and tweak, but as a check, I think this set is really useful.

 

Tower BridgeUnited Kingdom
However, it is clearly a USA list, we can see that in terms of the education levels, the income levels and the ethnicity.

So I asked a new question.

“What would that questionnaire look like for a study in the UK?”

Here is the response

When conducting a market research study in the UK, some demographic categories may differ slightly due to regional and cultural considerations. Below is a sample questionnaire tailored for a UK audience:

# Market Research Questionnaire

## Demographic Information

**Note**: Your answers to the following questions will be kept confidential and will only be used for research purposes.

### Personal Information

  1. **Age**

   – Under 18

   – 18-24

   – 25-34

   – 35-44

   – 45-54

   – 55-64

   – 65 or older

 

  1. **Gender**

   – Male

   – Female

   – Non-Binary/Other

   – Prefer not to say

 

  1. **Education Level**

   – GCSE or equivalent

   – A-Level or equivalent

   – Bachelor’s Degree

   – Postgraduate Degree (e.g., Master’s, PhD)

   – Other (please specify): __________

   – Prefer not to say

 

### Employment and Financial Information

 

  1. **Employment Status**

   – Employed full-time

   – Employed part-time

   – Unemployed

   – Student

   – Retired

   – Homemaker

   – Prefer not to say

 

  1. **Income Level (Annual)**

   – Under £20,000

   – £20,000-£39,999

   – £40,000-£59,999

   – £60,000-£79,999

   – £80,000 or more

   – Prefer not to say

### Location and Lifestyle

  1. **Geographic Location**

   – Urban

   – Suburban

   – Rural

 

  1. **Region**

   – England

   – Scotland

   – Wales

   – Northern Ireland

 

  1. **Ethnicity**

   – White British

   – White Other

   – Asian or Asian British

   – Black or Black British

   – Mixed/Multiple Ethnic Groups

   – Other (please specify): __________

   – Prefer not to say

 

  1. **Marital Status**

   – Single

   – Married/Civil Partnership

   – Separated

   – Divorced

   – Widowed

   – Prefer not to say

 

  1. **Household Size**

    – 1

    – 2

    – 3

    – 4

    – 5 or more

 

  1. **Do you have children?**

    – Yes

    – No

    – Prefer not to say

As with any market research questionnaire, it’s crucial to make sure you’re in compliance with local and national data protection regulations, such as the UK’s Data Protection Act or GDPR, to ensure the ethical collection and handling of personal data.

Arc de TriompheFrance
Well, that helped localize it for the UK, but what about a bigger challenge? I decided to ask it to produce a similar check questionnaire for France.

I asked

“What would that questionnaire look like if we were asking it in France?”

Here is the suggested questionnaire.

When conducting a market research study in France, it’s important to consider both cultural and legal aspects, such as the specific demographic variables that are relevant and allowed to be collected. For example, French law is strict about collecting information on ethnicity and religion. Here is a sample demographic questionnaire tailored for a French audience:

# Questionnaire de Recherche de Marché

## Informations Démographiques

**Remarque**: Vos réponses aux questions suivantes resteront confidentielles et seront uniquement utilisées à des fins de recherche.

### Informations Personnelles

  1. **Âge**

   – Moins de 18 ans

   – 18-24 ans

   – 25-34 ans

   – 35-44 ans

   – 45-54 ans

   – 55-64 ans

   – 65 ans ou plus

 

  1. **Sexe**

   – Homme

   – Femme

   – Non-binaire/Autre

   – Préfère ne pas dire

 

  1. **Niveau d’Éducation**

   – Brevet des collèges ou équivalent

   – Baccalauréat ou équivalent

   – Licence

   – Master ou plus

   – Autre (veuillez préciser): __________

   – Préfère ne pas dire

### Emploi et Informations Financières

  1. **Statut d’Emploi**

   – Employé à temps plein

   – Employé à temps partiel

   – Chômeur

   – Étudiant

   – Retraité

   – Femme/homme au foyer

   – Préfère ne pas dire

 

  1. **Revenu Annuel**

   – Moins de €20,000

   – €20,000-€39,999

   – €40,000-€59,999

   – €60,000-€79,999

   – €80,000 ou plus

   – Préfère ne pas dire

### Localisation et Mode de Vie

  1. **Localisation Géographique**

   – Urbain

   – Périurbain

   – Rural

 

  1. **Région**

   – Île-de-France

   – Auvergne-Rhône-Alpes

   – Nouvelle-Aquitaine

   – Hauts-de-France

   – Grand Est

   – Occitanie

   – Normandie

   – Bretagne

   – Pays de la Loire

   – Centre-Val de Loire

   – Provence-Alpes-Côte d’Azur

   – Bourgogne-Franche-Comté

   – Corse

 

  1. **Statut Marital**

   – Célibataire

   – Marié/Pacsé

   – Séparé

   – Divorcé

   – Veuf/Veuve

   – Préfère ne pas dire

 

  1. **Taille du Foyer**

    – 1

    – 2

    – 3

    – 4

    – 5 ou plus

 

  1. **Avez-vous des enfants ?**

    – Oui

    – Non

    – Préfère ne pas dire

Please note that this questionnaire is compliant with French data protection laws as of my last update in January 2022, which typically prohibit questions about ethnicity, religion, and other sensitive topics. Always make sure to consult local legal experts to ensure compliance with current laws and regulations.

The first thing that jumps off the page is that ChatGPT has removed the ethnicity question from the suggested questionnaire, noting that the laws about collecting ethnicity are strict in France. For anybody familiar with France, this is not a surprise, but for somebody new to research in France it suggests another thing for them to check.

A French Review of These Questions
I am indebted to Amélie de Brem and Cécile Silva Naud, of Platform One for reviewing the quality of the French questionnaire. Their comments included the following.

Positive points included: the conversion into accurate options for education, marital status, and region.

Negatives included: the translation of “Prefer not to say” is not great. It is very literal and even if people can understand what it means, it is not right. “Ne se prononce pas” is a better option.

Debatable, the gender question does not feel right. An option more like Homme/Femme/Autre/Ne se prononce pas would be better.

 

Temple Gate in JapanJapan
Finally, I asked it to create a checklist, demographic questionnaire for Japan. My question was

“And what would the original questionnaire look like if it were asked in Japan?”

Here is the ChatGPT suggestion:

In Japan, the demographic variables to consider might differ slightly due to cultural and social norms. Additionally, some questions that are standard in Western surveys may be considered inappropriate or overly invasive in Japan. Here’s how the original questionnaire might be adapted for a Japanese audience:

# マーケットリサーチアンケート (Market Research Questionnaire)

## 人口統計情報 (Demographic Information)

**注意**: 次の質問へのあなたの回答は、厳密に秘密にされ、研究目的のみに使用されます。

### 個人情報 (Personal Information) 

  1. **年齢 (Age)**

   – 18歳未満 (Under 18)

   – 18-24歳 (18-24)

   – 25-34歳 (25-34)

   – 35-44歳 (35-44)

   – 45-54歳 (45-54)

   – 55-64歳 (55-64)

   – 65歳以上 (65 or older)

 

  1. **性別 (Gender)**

   – 男性 (Male)

   – 女性 (Female)

   – その他 (Other)

   – 無回答 (Prefer not to say)

 

  1. **最終学歴 (Education Level)**

   – 高校卒・以下 (High School or less)

   – 短大・専門学校卒 (Junior College/Vocational School)

   – 大学卒 (Bachelor’s Degree)

   – 大学院卒以上 (Postgraduate Degree)

   – 無回答 (Prefer not to say)

 

### 就労・財政情報 (Employment and Financial Information)

 

  1. **雇用状態 (Employment Status)**

   – 正社員 (Full-time)

   – パート・アルバイト (Part-time)

   – 無職 (Unemployed)

   – 学生 (Student)

   – 退職 (Retired)

   – 主婦・主夫 (Homemaker)

   – 無回答 (Prefer not to say)

 

  1. **年収 (Annual Income)**

   – 200万円未満 (Under ¥2,000,000)

   – 200-400万円 (¥2,000,000-¥4,000,000)

   – 400-600万円 (¥4,000,000-¥6,000,000)

   – 600-800万円 (¥6,000,000-¥8,000,000)

   – 800万円以上 (Over ¥8,000,000)

   – 無回答 (Prefer not to say)

 

### 地域・生活スタイル (Location and Lifestyle)

 

  1. **地域タイプ (Geographic Location)**

   – 都市 (Urban)

   – 郊外 (Suburban)

   – 田舎 (Rural)

 

  1. **都道府県 (Prefecture)**

   – [List of prefectures]

 

  1. **配偶者の有無 (Marital Status)**

   – 独身 (Single)

   – 既婚 (Married)

   – 別居・離婚 (Separated/Divorced)

   – 未亡人 (Widowed)

   – 無回答 (Prefer not to say)

 

  1. **世帯人数 (Household Size)**

   – 1

   – 2

   – 3

   – 4

   – 5人以上

 

  1. **子供の有無 (Do you have children?)**

    – あり (Yes)

    – なし (No)

    – 無回答 (Prefer not to say)

As always, ensure you’re in compliance with local data protection laws and norms. Ethical considerations and privacy norms can differ by country, so being familiar with local guidelines is crucial.

The English appears to have been added for my convenience; it would not be part of a questionnaire in Japan. Again, we see the ethnicity question not being asked. In Japan, something like 98% of people are of Japanese ethnicity, which makes it a less appropriate question in most circumstances.

I asked ChatGPT which prefectures it was referring to in its suggested questionnaire, and it listed the following 47 prefectures.

In a Japanese questionnaire, you would include all 47 prefectures of Japan. It’s essential to list them in a way that’s easy for respondents to understand and select, usually in alphabetical order or grouped by region. Here’s how it might look:

 都道府県 (Prefecture)**

   – 北海道 (Hokkaido)

   – 青森県 (Aomori)

   – 岩手県 (Iwate)

   – 宮城県 (Miyagi)

   – 秋田県 (Akita)

   – 山形県 (Yamagata)

   – 福島県 (Fukushima)

   – 茨城県 (Ibaraki)

   – 栃木県 (Tochigi)

   – 群馬県 (Gunma)

   – 埼玉県 (Saitama)

   – 千葉県 (Chiba)

   – 東京都 (Tokyo)

   – 神奈川県 (Kanagawa)

   – 新潟県 (Niigata)

   – 富山県 (Toyama)

   – 石川県 (Ishikawa)

   – 福井県 (Fukui)

   – 山梨県 (Yamanashi)

   – 長野県 (Nagano)

   – 岐阜県 (Gifu)

   – 静岡県 (Shizuoka)

   – 愛知県 (Aichi)

   – 三重県 (Mie)

   – 滋賀県 (Shiga)

   – 京都府 (Kyoto)

   – 大阪府 (Osaka)

   – 兵庫県 (Hyogo)

   – 奈良県 (Nara)

   – 和歌山県 (Wakayama)

   – 鳥取県 (Tottori)

   – 島根県 (Shimane)

   – 岡山県 (Okayama)

   – 広島県 (Hiroshima)

   – 山口県 (Yamaguchi)

   – 徳島県 (Tokushima)

   – 香川県 (Kagawa)

   – 愛媛県 (Ehime)

   – 高知県 (Kochi)

   – 福岡県 (Fukuoka)

   – 佐賀県 (Saga)

   – 長崎県 (Nagasaki)

   – 熊本県 (Kumamoto)

   – 大分県 (Oita)

   – 宮崎県 (Miyazaki)

   – 鹿児島県 (Kagoshima)

   – 沖縄県 (Okinawa)

Note that the list above is arranged more or less from north to south rather than alphabetically. This is often how they are listed in official documents and surveys in Japan.

A Japanese Review of These Questions
I am indebted to Noriko Kishida of the Japan Market Research Association for reviewing ChatGPT’s suggested Questions. Below is a list of changes Mrs Kishida suggested.

# マーケットリサーチアンケート (Market Research Questionnaire)

## 人口統計情報 (Demographic Information)

**注意**: 次の質問へのあなたの回答は、厳密に秘密にされ、研究目的のみに使用されます。

次の質問へのあなたの回答内容は、秘密保持の対象として厳密に守られ、研究目的のみに使用されます。

  1. **地域タイプ (Geographic Location)**

   – 都市 (Urban)  

   – 郊外 (Suburban)

   – 田舎 (Rural)  it should be農村

  1. **配偶者の有無 (Marital Status)**

   – 独身 (Single)

   – 既婚 (Married)

   – 別居・離婚 (Separated/Divorced)

   – 未亡人 (Widowed)   未亡人 is only for women, it should be死別

   – 無回答 (Prefer not to say)

 

Asking Better Demographics in International Studies
If you really want to ask the best demographics, especially in a multi-country study then you should read the ESOMAR best practice guidelines. There are guidelines on Working Status, Age, and Gender – with more in the pipeline. Check them out by clicking here. 

So What?
When you are busy it is easy to forget something. If you are writing a questionnaire, you might spend a long time on the critical parts — for example, the questions relating to product preference or attitudes and behaviours. However, you also want to get the basics right. By asking ChatGPT to produce a ‘typical’ set of demographics, you can cross-check it with your questionnaire. This will help you spot a question you might have missed or an answer option you might have missed. It might help you spot areas where your questions might be outdated, for example, if you are still asking about gender as the binary pair of Male/Female.

One thought on “Asking Demographic Questions – ChatGPT Tip #1

  1. Thanks Ray. As you mention elsewhere a usefull beginners list but also limited.
    My usual bugbear … the age grouping lumps everyone over 65 together. Given esomars great demographic guideline on age last year finally ( after 10 years of waiting and nudging ) recognised 5 or 10 year breaks sho I ld apply at least to 100 years old. A 67 year old is certainly as different from an 87 year old as a 47 year old. And with many countries now seeing over a quarter of the population over 60 it is essential marketers, researchers and GPT recognise this.

Comments are closed.