Decision Tree vs. Adaptive

---

When developing a survey, one of the biggest concerns is the burden placed on the respondent to complete it. Respondent burden can be measured either in the amount of time required to complete it or the number of questions displayed to the respondent.

People who feel burdened by a survey will likely not respond. In the past, organizations completing longer surveys had to budget for incentives to complete it, or had to accept the lower response rates. When reviewing the competition's technology for survey administration, you might find that their answer to this problem is to create surveys using decision-tree logic to reduce the number of redundant or irrelevant questions people have to answer.

Decision-tree logic is called this because when mapped out it looks much like tree braches.

For example:
1. What is your marital status?
A. Single   B. Married   C. Divorced   D. Widowed

2. How fulfilling is your marriage?
A. Not at all  B. Sparsely  C. Moderately  D. Very fulfilling

3. How long have you been married?
A. Less than 1 yr.  B. 1 to 5 yrs.  C. 6 to 10 yrs.  D. 11 to 15 yrs.  E. Over 15 yrs.

4. In what state do you live? ________________

Surveys developed by all other companies would use decision-tree logic to remove irrelevant questions. For example, if the response from Question 1 is "Single" or "Widowed" to Question 1, the survey could skip the following two questions for all respondents and move directly to Question 4.

In this example, decision-tree logic makes sense. However, consider a survey used to understand the activities of your job:

1. How important is knowledge of statistics on your job?
A. Not at all  B. Minimally  C. Moderately  D. Very  E. Extremely

2. How frequently do you work outdoors?
A. Never  B. Rarely C. Sometimes  D. Often  E. Always

3. How important is written communication to the performance of your job?
A. Less than 1 yr.  B. 1 to 5 yrs.  C. 6 to 10 yrs.  D. 11 to15 yrs.  E. Over 15 yrs.

4. How often in your job do you have to speak in front of groups?
A. Never  B.Once a year  C. Once a month  D. Once a week  E. Daily

How would you go about using decision-tree logic to shorten this survey? Without knowing any other information, it is impossible to shorten this survey this way because the questions that are irrelevant for one person might be very relevant for another. For example, you might be tempted to say that if someone answers "Extremely Important" to Question 1 (knowledge of statistics) that you will have them skip Question 2 because most people who use statistics work indoors.

If the person responding was a Civil Engineer, Land Surveyor, or Environment Scientist, you would be making an incorrect assumption. Without knowing prior information about your respondents and the relationship among the items, it would be nearly impossible to create survey rules that apply to everyone.

With the Adaptiqs platform, customers are capable of reducing the length of almost any survey. Instead of applying decision rules for everyone, our technology makes intelligent choices based on prior knowledge or data collected from previous administrations. The result is a smart survey process that limits the respondent burden and reduces your administration costs.