Application of Questionnaire Survey method

Why to choose the Questionnaire survey method

Product side for the use of research projects proposed by the following:

At first glance, the demand is more and more miscellaneous, it seems impossible to solve all the problems in one way. After repeated and product side of the communication, you can dig out their hearts most want to know content. To sort it out, it becomes the following form:

In addition to showing the above situation of NetEase XX users, the end result also involves a variety of categories of respondents and different categories of users on specific topics of the horizontal comparison-this may be more important.

The questionnaire survey method is especially suitable for investigation: User's use purpose, behavior habit, attitude and opinion, demographic information, which is not suitable to explore new and vague needs of users.

Therefore, this research project is suitable to choose the questionnaire survey method as the main method (simultaneously chooses the interview method, the operational data makes the supplementary reference).

Ii. sources and basis of the questionnaire material

1. Second-hand information (existing theories, reports and industry information)

2. Experience of their own use, competitive analysis

You need to develop yourself into a deep user of the product, and insist on using competitive products. This helps to form their own intuition and perceptual knowledge of the product.

The competitive analysis can help us to have a better understanding of the product, so that we can extract a rational view from perceptual knowledge, while ensuring a comprehensive and meticulous examination (not missing important details).

3. Simple interviews with product designers and ordinary users

"Three people line, there will be my teacher", Self-Study finished, and then let product designers, ordinary users to teach you some knowledge.

At this time, the interview using the form of Easy Chat (semi-structured), the object of the interview is also convenient sampling, the nearest choice-the purpose is not to draw any conclusions, more to help themselves further deepen understanding, exhausted a variety of possibilities for the preparation of questionnaires.

and interviews with product designers, as well as information about product department workflows, organizational structures, staffing, resources available, and so on--seemingly irrelevant--actually make the recommendations of our final investigation report more value-oriented!

Iii. details of the questionnaire

1. Indicative note

Social research textbooks usually require us to write a long paragraph of descriptive text, but how many people on the internet will take a serious look at a large piece of text? Take a look at what the author of Don t makes me do: Subtract from 103 words to 41 words (the second edition of the book, P34)! Non-essential information may not appear in the indicative statement of the questionnaire.

The core information that needs to be mentioned is nothing more than these: A. Purpose: To help improve the NetEase xx, also welcome non-NetEase users to answer B. Time: 3-5 minutes

2. The topic prepares the idea

3. Details to be noted in the topic setting

1 Concise and clear: Sometimes the two are contradictory, but try to be balanced.

"Example" How long have you used XX's history? --Concise

When was the first time you used XX? --Simple, popular

When was the first time you used XX (not necessarily the most commonly used one today)? --Sacrificing "simplicity" a little, but more "clear"!

2 attitude neutrality, no inducement

"Example" are you satisfied with the current commonly used XX?

How satisfied are you with the current XX?

4. Details to be noted for option setting

1 option Mutex

"Example" What do you usually surf the Internet more often?

A. Hot/Headline News B. Entertainment gossip C. Current affairs Politics D. Games/anime ...

Option A is intended to correspond to the behavior of someone who reads headlines every day but doesn't have a particular preference for the content area, but there is an overlap between the A option and the following options, especially where a and C overlap very high. So, delete the C option and change the A option to "current affairs news."

2 contains all possible

The ability to do so reflects the adequacy of the preparatory phase. When you suspect that the option settings may be incomplete, you can find products, design, operators to explore, check.

In addition, you can set the "other ________" option to make sure that the user always has a choice, and that the percentage of the "other ___" option also reflects the overall option setting.

3 The user is easy to understand

"The example" asks the user the mobile operating system, the option "iOS (IPhone), Symbian (Symbian)" is easier to understand than "iOS, Symbian", although the former expression looks nondescript.

4 The measurement type of the variable

In actual work, only the general distinction between "category variable" and "continuous variable" two categories. At the level of statistical measurement, the "continuous variable" is higher than the category variable, contains rich information, and can also use a slightly more advanced statistical method. So, if possible, try to make the option "into a continuous variable."

"Example" want to understand the user usually share the outside of the site content, two options set (the latter is better):

A. Sharing B. Do not share

A. Frequent B. usually C. D. Rarely E. Never

5 attitude neutrality, no inducement

"Example" option "Current affairs politics" changed to "current Affairs news";

6) Random Order rendering

If the option is a factual, category variable, rather than a sequential variable in order of degree, then the option should be presented in random order to the answer, which may exclude the tendency to choose between the end option or the intermediate option.

5. Avoid reaction to the heart

Reaction heart refers to people who answer questionnaires with a relatively fixed tendency to respond. Sometimes it may be conscious: please the person who asked the question and make a good impression on others. Sometimes it may be unconscious: habit answers "yes", "consent", social approbation-avoid unfavorable or negative evaluations in sociocultural norms, avoidance, defense of sensitive privacy issues or social taboo topics.

Under the circumstances of online, face-to-face, and prize-answering, it is easier to have a conscious reaction to the problem of heart. On the internet and anonymously, the conscious response will weaken a lot, and the unconscious response can be mitigated by the following efforts. The following are specific practices:

1 Indicative section: Emphasis on anonymity and confidentiality, emphasis on serious, truthful answers to participate in the lottery.

2 in the questionnaire: Not necessary, users are not required to fill in the status of income, mobile phone number. Even if it is necessary to do so, explain why and retain the right of the user to decide whether or not to answer.

3 The questionnaire: the design of a number of echoes of the topic to prove each other, and the appropriate use of reverse scoring method.

4 Data analysis: related to their own products and other products of satisfaction, goodwill, such as the comparison of attitude evaluation, we must be careful to draw conclusions (there may be reaction to the heart, the user's own product evaluation is very good).

6. Create a response between different questions and options

On the one hand, we can prove each other, improve the reliability of the questionnaire (of course, there is no need to simply increase the number of questions--outweigh the gains). On the other hand, it is convenient to use the Cross analysis/table method in statistical analysis-this will improve the "rich, accurate and valuable" aspects of your survey report.

"Example" to choose the current XX as your most commonly used xx is the main reason for? "," you to the current most commonly used this XX main dissatisfaction place? "The results of these two topics can be used to corroborate each other.

"Example", what kind of person do you care about in XX? What is your interest in XX? "These two topics can be used for cross analysis to identify the most typical and representative combinations of people and content."

7. Ranking of topics

The overall topic ranking principle such as above, sometimes also need to consider some special circumstances, such as "contact product time has a relationship, the operation process has a specific order", at this time the local topic is "Time/process successively" sort.

8. Questionnaire questions

First, look at the difference between questionnaires and scale methods:

Can be so popular to understand: The scale is a lot of experts, spent a lot of energy in the preparation of the link (the tool itself is very high reliability); The questionnaire needs a large number of representative samples to achieve the goal, and the validity of the questionnaire itself is still poor, more needs a large number of samples.

In order to get a large number of samples and hope that the quality of the data is high, the questionnaire itself will be limited-the problem is less. In actual work, the total number of questions in the questionnaire is controlled within 25 questions, the average answer time does not exceed 300 seconds (afterwards statistics).

Iv. Forecast Test

After such meticulous and long-term preparation, there seems to be no more problems, can't wait to put the questionnaire. Wait！ A wise man shall have a loss. At this point, you should find some products/designers, ordinary users to predict the test (convenient sampling, the best to find people who have not participated in the discussion before). According to personal experience, this link can still check out some missing problems.

One of the questions in the "Example" questionnaire: "How do you find someone else on XX?" "There are a number of common ways in which the self perception option has been enumerated, and an interactive designer asked me in a prediction test why there is no" direct search for someone "option? Originally I overlooked this most direct, the original search method.

V. Questionnaire delivery

In addition to the quality of the questionnaire itself, the factors that directly affect the success or failure of the survey are sample quality (which is the consideration of the questionnaire strategy). By the way, the most important measure of scale survey is the tool itself (the quality of the scale, Norm Index) and advanced statistical method; The most important thing in the experimental method is control and manipulation (control of extraneous variables, manipulation of independent variables), and the most important of questionnaire survey is sample (sample size is large enough and representative). Want a large sample size, as much as possible to use the channel, of course, some of the delivery of small skills (the more at the end of the operation process, the entrance copy of the more independent layout, the higher the rate of feedback questionnaires, see Taobao ued, "The seller feedback the impact of quantitative research").

Here we focus on the problem of sample representation. First of all, we have to admit that the online questionnaire is not a complete random sampling, we can not be critical to investigate the various characteristics of the object to match the real life of the overall, but we should strive to cover different categories of user groups to have a greater representation.

"Example" This investigation, we besides want to know NetEase xx user's situation, we also hope to be able to understand Sina, Tencent, other XX user's situation. To this end, we have made the following efforts:

1 In addition to the NetEase portal in the questionnaire, we hope to maximize the proportion of outbound channels, such as: The country's major forums, some search engine pages, video site pages, etc., or even better to be able to put to other portals (of course, in fact, can not spend so much price).

2 when the promotion and marketing department colleagues hope that the questionnaire of the click of the picture contains "NetEase xx" word (can play the role of advertising publicity). In this respect, I firmly oppose, because this will undoubtedly seriously affect the "Sina, Tencent, other XX users" click.

3 in the Forum to launch a questionnaire link, the market colleagues in the form of soft text, the effect is "the current XX products are very hot, we can go through this questionnaire to understand the product of XX", write humorous, exciting to see the people's curiosity. Some colleagues do not endorse this practice and think that the specification which does not conform to the instructions in the questionnaire may have an impact on the user's reply. But I personally think this is a better technique. First of all, in the work is not perfect to follow all the principles, must have "principal contradictions, and contradictions of the main aspects" of dialectical thinking, when necessary to weigh the pros and cons of comprehensive consideration. For example, at this time our most important goal is must be covered to "Sina, Tencent, other XX user group", and the number is not less. Second, look at that post, users are using the "scan" rather than "carefully read" way, at most the use of instantaneous memory, such as jump to the page, and then read the formal questionnaire instructions and fill in the questionnaire, before the impact of instantaneous memory is very small. Third, the questionnaire lets the user express the direct attitude the topic to be few, more is the factual question, these questions itself is difficult to induce the systematic deviation. Finally, when we collate the data, we have a series of postmortem data-checking methods.

4 Set up automatic hidden problem, can distinguish from each put the channel sample quantity (especially differentiate NetEase station inside, station outside).

Vi. Data Collation

Usually the examination of data needs to delete incomplete answers, multiple-choice questionnaires, and logical contradictions; accurate and informative screening methods can help further improve the quality of the data (this step could be done in Excel or SPSS).

In addition, also need to be based on "answer time" to screen out the questionnaire (too short reflects not serious, too long reflects the answer when the external interference more). Using the analysis of box diagram in SPPs, we can clearly see the deviation and extreme value on the index of "answer length", and the questionnaire ID number they reflect, so it can be easily deleted.

VII. statistical analysis

Due to the restriction of the measurement type of the variables, the statistical methods available are lower, descriptive statistics are mostly used, sometimes cross analysis, Card square test (and occasionally can be used for correlation analysis, ANOVA). Looking back, again, it's time to think about the statistical methods that might be used, and how to make the options as contiguous variables as possible.

Although this link is not as direct as the previous link to the success of the investigation (because of the chance of repeated attempts), but the rational and full use of statistical technology, will give the final report the value of a significant increase. This with the professional quality of the Institute has a great relationship, the need for long-term continuous efforts to promote, here is not in detail.

Viii. Summary

Seemingly simple questionnaires, to be able to do well and let it play the greatest value, there are many meticulous, worthy of scrutiny. A research project (whether it is not a questionnaire), early preparation and planning are very important, workers!

Description: In view of the confidentiality of the company's products, the information is hidden.

Source: http://uedc.163.com/8379.html