Modular: The questionnaire content determined by the research questions is divided into different modules based on relevance.

Correlation: The variables determined by the research question are expanded sequentially based on correlation.

Example: Regional inequality in admission opportunities for undergraduates at Peking University, differentiation • Dependent variable: state of inequality • Independent variables: Factors affecting inequality, such as regional economic development level; family socioeconomic status, etc. • Control variables: demographic characteristics; family characteristics; school characteristics

Hard to get answers

Difficult question

If the answer to an interview question will affect the arrangement of a series of interview questions, you should ask first and the jump should not be too complicated.

It is not appropriate to put together questions that require the interviewee to think intensively.

• How to make respondents accept your interview and be happy to accept your subsequent interviews

• Use questions that respondents can answer without thinking and without causing alarm.

• Allow interviewees to relieve their worries and fears

• Build trust between interviewer and respondent

Interview questions that require the interviewee’s memory

Interview questions that require the intellectual cooperation of the interviewee

Interview questions that require the psychological cooperation of the interviewee

Public facts, status quo

Privacy categories, such as friend circles, special behaviors, habits, diseases, etc.

Private information, such as identity information, income information, etc.

Attitude-sensitive information, such as political leanings, religious beliefs, etc.

For example, regarding diseases, ask about health first and then the disease.

The basic structure of an interview question

Existence (fact) type interview questions

behavioral interview questions

Attitude (opinion) interview questions

Ability (test) interview questions

Option arrangement, matrix questions

direct question

matrix problem

It is better to ask positive questions than negative questions (no, no ×)

Ask simple questions, not complicated questions

Neutral questions, not inducing questions

Closed questions, open questions are not appropriate

Ask questions in short sentences, not long sentences (approximately 20 words, except for complex questionnaires)

Closed-option responses facilitate data processing, but often limit response diversity.

Open options allow for the diversity of responses, but the responses are not easily digitized.

Single-choice options facilitate measurement certainty, but often limit the economy of interview questions.

Multiple options are economical, but application of responses often becomes a challenge

Try to use the advantages of the option form as much as possible, which is the principle of choosing form

Ask direct questions about the main measured variables and provide response options, such as:

Under the condition that the options are determined, try to use closed option forms to facilitate data processing.

Ask direct questions about the variable to be measured, giving response options, and leaving an option for "other"

If you use unexpected options, be clear in advance how the "other" options will be handled

Main variables to measure, ask questions directly, and faithfully record the responses

Ask a direct question about the variable to be measured and select an option as a response

Ask a direct question about the variable to be measured and select multiple options as a response.

If it is a multiple-choice question, be sure to explain it in the appropriate position of the question stem.

A set of questions that uses multiple questions to measure one variable, each measuring a dimension

Use multiple questions to measure a variable, and the measurement of the questions is nested in progressive question groups

For some non-universal scenarios, virtual scenarios can be designed. Scholars such as Gary King have proposed the “Anchoring Vignettes” method for measurement.

Social distance scale is a typical virtual scenario method

E1.When do you eat?

—— (please record the time)

E2. Do you agree with the ideas of feminism?

1.Yes 2.No 3.Don’t know

E3. How do you evaluate our school’s logistics services and teaching management?

1. Very good 2. Fairly good 3. Average 4. Bad 5. Very bad

E4. What awards did you receive during your studies at school?

1. National Scholarship 2. National Inspirational Scholarship 3. May Fourth Scholarship 4. School Scholarship

Missing: Such as principal scholarship, financial aid scholarship, etc.

It is advisable to directly ask for the year, month, and day of birth; the degree of accuracy depends on the research variables.

If the interviewees are mainly elderly people, you should prepare a zodiac sign-year of birth comparison table

It is better to record directly and not to ask questions.

Combine among options such as "single, cohabiting, married, divorced, widowed"

It is advisable to design interview questions based on the research variables, or ask about the years of education, or the highest academic certificate obtained so far.

If necessary, also ask whether you are receiving academic education on the job.

If there is no need to know the source of income, use the approximation method to ask

If you need to know the source of revenue, use the approximation method to ask about total revenue first and then ask about category revenue

Expenditure questions are the same

It is not appropriate to directly ask what you are doing. It is better to use a three-stage inquiry and do coding.