Step-Ladder Question

A step-ladder question refers to a type of question sequence that yields more complete and accurate data than would a single question on the same topic. Step-ladder questions are used by survey researchers in an attempt to reduce item nonresponse (missing data), measurement error, or both, although they add slightly to the cost of gathering the data since more than one question needs to be asked.

For example, asking someone into which of the following income categories his or her 2007 total household income fell—less than $20,000; $20,000-$39,999; $40,000-$59,999; $60,000-$79,999; $80,000-$99,999; $100,000 or more—will lead to a good deal of “Don't Know” or “Refused” answers. Researchers have found that a step-ladder question about income will substantially reduce item nonresponse and thus the need to impute those missing values.

A step-ladder question sequence for the income variable referenced in the previous paragraph, that was programmed to be asked in a computer-assisted interview, would be as follows:

• Ql. Was your total household income from all sources in 2007 more than $19,999?
  • <1> YES (GO TO Q2)
  • <;2> NO (GO TO Q6)
  [Page 848]
  • <;8> REFUSED (GO TO Q6)
  • <;9> UNCERTAIN (GO TO Q6)
• Q2. And was it more than $39,999?
  • <1> YES (GO TO Q3)
  • <;2> NO (GO TO Q6)
  • <;8> REFUSED (GO TO Q6)
  • <;9> UNCERTAIN (GO TO Q6)
• Q3. And was it more than $59,999?
  • <1> YES (GO TO Q4)
  • <;2> NO (GO TO Q6)
  • <;8> REFUSED (GO TO Q6)
  • <;9> UNCERTAIN (GO TO Q6)
• Q4. And was it more than $79,999?
  • <1> YES (GO TO Q5)
  • <;2> NO (GO TO Q6)
  • <;8> REFUSED (GO TO Q6)
  • <;9> UNCERTAIN (GO TO Q6)
• Q5. And was it more than $99,999?
  • <1> YES (GO TO Q6)
  • <;2> NO (GO TO Q6)
  • <;8> REFUSED (GO TO Q6)
  • <;9> UNCERTAIN (GO TO Q6)

In this example, after the income sequence has been administered all respondents are taken to Q6 (i.e. whatever is the next logical question in the questionnaire after the income sequence). Of note, even though the entire step-ladder sequence comprises five questions, any one respondent would be asked more questions in the sequence only up until he or she said “No.” As such, proportionally few respondents would be asked four or five of the questions (i.e. only a minority will have incomes exceeding $80,000). The majority would only be asked one, two, or three of the questions in this step-ladder sequence. This step-ladder sequence will lead to far fewer missing income values than a single income question that presents to the respondent essentially the same income categories all at once. It also will yield data that the researchers can combine to form a single income variable with the desired six categories in the original one-question income example.

Step-ladder questions can be used for other constructs that are measured on some form of numerically ordered scale. They are particularly useful when there are many response choices on the response scale and the cognitive burden on many respondents is too great to present all the choices at once. If all choices were presented at once, primacy, recency, and/or other satisficing effects would likely lead to errors in the data.