User research: To ensure the user survey design quality control points

As one of quantitative analysis methods, questionnaires are commonly used in our user research to investigate the user's experience of our products or seek advice, so as to improve the amount of product delivery support.
The questionnaire can be used in both basic user research and product user research, and can be used for both horizontal and longitudinal studies. According to different research purposes, it is necessary to determine the different methods of questionnaire quality audit. In order to make the data obtained by the questionnaire can objectively reflect the real problems, we summarize the following points to ensure the quality.
Description
Basic user research: The users according to sex, age, occupation, income, education, District, etc. to classify, from which can be in a specific industry, the basic characteristics of the population and views. But this classification does not help to find data that supports product improvement.
Product user research: Usually from the user's behavioral differences began to analyze, the difference behavior is often behind the driver (motivation) at work, and motivation depends on the target, so the end will usually fall to the user goals. By "goal" to divide the user group, and then to understand the same target user demographic data, which is the opposite of the basic user research, and precisely because of this, the two are complementary to each other.
From product user research and basic user studies, Angela, published in user research

I. Determination of the purpose of research

Determining the purpose of research is the most important aspect of any study, before preparing the questionnaire, we must be clear about the objectives of the questionnaire (that is, what the demand side wants to investigate, the demand point), the target population and channel of the survey.

1. Objectives:

* Control the demand point: because too many questions of the questionnaire will affect the quality of the user answer, according to the results of previous studies and practical experience, we think that the questionnaire set the attitude problem should not exceed 20, the total title of the best not more than 30.
* Data collection methods: Desktop Research (customer service feedback, log analysis, sales, marketing), brainstorming, telephone interviews and other methods to identify the core objectives of the questionnaire, and accumulate data for the preparation of questionnaires.

2. Objects:

Before compiling the questionnaire, we must understand the target group of our questionnaire, in view of the actual research, we need to clarify the classification and definition of the user group, and it is best to distinguish it by objective indicators. For example:
* New users, lost users, continuous users
* Users of different funding channels
* Users of varying degrees of activity

3. Channels:

When the questionnaire is put in place, it needs to consider the representative, recycle period and quantity of samples.
The general launch cycle is one week. You can use directed delivery (e.g., mail delivery) and non-directional delivery (e.g., online systems). The rate of directed delivery should not be less than 5% (the number of people who completed the questionnaire/the total number of participants surveyed).

second, the formation of a preliminary questionnaire

1. Content structure

* 1/2 lie detection questions, used to determine whether the user is serious answer.
* Reverse Question
* Try not to judge a single problem

2. Project form

* Choice Questions & Openness
* Radio & multiple selection
* Isometric scale

3. Some issues to be avoided in the preparation

* The conceptual understanding between the questionnaire designer and the investigator is inconsistent.
* More than two concepts appear in the same topic.
* Use of special terms, lines, and sayings.
* The answer setting misses the comprehensive option.
* There are suggestive questions with some inclination.
* Use words that are not sure.
* The result is false because the object of investigation has not been experienced.
* Use words that can be interpreted in a variety of ways and with vague meanings.
* Statements using negative sentences, respondents ignored and misunderstood the question.
* The problem lacks a restricted premise.
* The selected item in the topic does not contain all the degrees.

4. Guiding language

* The establishment of initial psychological harmony, the intention and motivation to respond, need to indicate the following points:
The identity of the investigator
The purpose and value of the study
Need to investigate the importance of the collaboration of the object
* Questionnaire Answer Request

third, Internal audit

Before the questionnaire is officially online, it needs to be audited and tested.

1. Expert review

* For the content of the questionnaire, make the relationship between the project and the demand point to achieve the best.
Assessment content
* The issues set are closely related to the subject under study
* The topic should be clear, no ambiguity
* Try not to use tendentious issues
* The questions raised should be questions that the respondents were able to answer
* Try not to mention problems with social pressure
* The issues raised should be as specific as possible
* For some difficult to understand problems should try to make some simple explanation
* It's better to ask respondents for their personal experience than the general term
* Use a variety of ways to mitigate the social pressure that the problem may bring to the respondents
Scoring method
*4~5 Experts independent evaluation, more than half of the project can be used
* For the evaluation of the change is not adopted

2. Internal testing

* For the questionnaire language, form. 4~5 , who were involved in the design, could use some of the questions:
How do you feel about the whole questionnaire?
How do you feel when you say "..."?
If you ask this question, how would you answer it?
If you ask "..." Do you have any difficulty in answering it?

3. External test

* Formal forecasts, people , participants are not involved in the design, preliminary analysis of the results of the questionnaire, early detection of problems. For different questionnaires, focus on different aspects.
Structured questionnaires
* Reliability Analysis
For example: The Cologne Baja Alpha project was modified below 0.5 .
* Validity Analysis
The questionnaire reflects the content to be measured and has multiple validity. such as: structural validity, the validity of the scale and so on.
Non-structured questionnaires
For open answer questions, for analysis, whether important options are omitted.

Four, the official launch

For our products, the recovery of the smallest effective unit sample is not less than, analysis is meaningful.