Branching

Branching is a questionnaire design technique used in survey research that utilizes skip patterns to ensure that respondents are asked only those questions that apply to them. This technique allows the questionnaire to be tailored to each individual respondent so that respondents with different characteristics, experiences, knowledge, and opinions are routed to applicable questions (e.g. questions about a treatment for diabetes are only asked to respondents who have been diagnosed with diabetes).

Branching also is used to ask respondents to choose among a large number of response options without requiring them to keep all the response options in working memory (e.g. respondents can be asked whether they identify with the Republican or Democratic party and then asked how strongly they identify with the relevant party in follow—up questions).

Branching can be conditional, compound conditional, or unconditional. In conditional branching, a single condition is met where routing occurs based on the answer to a single question (i.e. if the answer to question #1 is “No,” then skip to question #3). In compound conditional branching, more than one condition must be met. The branching in this case is dependent on multiple answers, and routing occurs based on a combination of answers (i.e. if the answer to question #1 is “Yes” or the answer to question #2 is “Yes,” skip to question #5). Unconditional branching is a direct statement with no conditions, often used to bring the respondent back to a specific point in the main survey after following a branching sequence. The approaches to branching differ depending on survey administration.

As a general rule, computer—assisted data collection (i.e. Internet surveys or computer—assisted self, telephone, or personal interviews) allows for more complex branching than paper-and-pencil data collection. Branching can be accomplished in computer-assisted survey instruments using programmed Boolean logic statements (i.e. if (question #) (state condition, such as =, <, >) (value), then (skip to question #)). Branching in paper-and-pencil survey instruments cannot make use of these technological complexities. Rather, it requires the appropriate placement of visual cues to guide respondents or interviewers through the branching instructions. Some common visual layouts include using arrows, placing the branching instructions within approximately nine characters of text (within foveal view), using enlarged, bold, and/or italicized font, and changing the background color. Two additional techniques that can be employed to guide the respondent or interviewers through paper-and-pencil branching instructions are the prevention technique and the detection technique. In the prevention technique, respondents are educated before reaching the branching instruction by including statements to remind them to look for instructions. In the detection technique, respondents are able to detect any branching errors they may have made through the use of feedback, such as inserting an additional branching instruction before the question that is supposed to be skipped, allowing them to correct the error and follow the instruction as intended.

There are two types of errors associated with branching. Errors of omission occur when respondents skip questions that were intended for their completion and result in item nonresponse for those items that were inadvertently skipped. Conversely, errors of commission occur when respondents provide answers to questions that were not intended for their completion. Accurate computer-assisted survey programming and proper paper-and-pencil survey visual layout of branching instructions can significantly reduce or even eliminate these errors.

...