Skip to main content icon/video/no-internet
icon backReturn to Entries

Interpenetrated Design

An interpenetrated survey design is one that randomly assigns respondent cases to interviewers. This is done to lower the possibility that interviewer-related measurement error is of a nature and size that would bias the survey's findings. This type of design addresses survey errors associated with the survey instrument and the recording of responses by the interviewer. One way to reduce subjective interviewer error is to develop a survey using an interpenetrated design—that is, by ensuring a random assignment of respondents to interviewers. Surveys employing an interpenetrated design, when such is warranted, will tend to reduce the severity of interpretation errors resulting from the conflation of interviewer bias with some other statistically relevant variable that might serve as a basis for assigning respondents. It will also typically reduce the overall standard error of response variance, especially for types of questions that inherently require some judgment or interpretation in recording by the interviewer.

Example of an Interpenetrated Survey Design

Assume a survey of 100 women from known high-risk populations (e.g. low income, substandard education, history of domestic violence), who are being queried about their tobacco use. The survey will be administered face-to-face by five interviewers and will feature a mix of demographic and binary-response questions, as well as several open-ended questions about the respondents' psychosocial triggers for smoking that the interviewer will interpret and assign a clinical risk index score.

In an interpenetrated design, the 100 women will be randomly assigned to each of the five interviewers. This means that any potential skewing of recorded results arising from bias or judgment by any single interviewer will be relatively equally shared by all of the respondents assigned to that interviewer and could therefore be considered “background noise” in terms of finding correlations within and among classes in the data. By contrast, in a noninterpenetrated design, it is possible that a correlating variable or class could be overemphasized or underemphasized by the relative weight of interviewer bias across a nonrandom assignment of respondents to interviewers.

For example, if all pregnant women queried about their tobacco use were assigned to a single female nurse interviewer who believes smoking is a social vice and not a chemical addiction, the nurse-interviewer's own subjective bias might contribute to Type I or Type II error for the class of pregnant women, relative to the survey's working hypothesis, or the bias might introduce systemic error into the response-variation rate for the class of pregnant women assigned to that interviewer. An interpenetrated design, in this example, would decrease the likelihood that one interviewer's behavior will contribute in a statistically significant way to analytic error.

Challenges in Implementing an Interpenetrated Design

The use of an interpenetrated design can mitigate the inflation of statistical error engendered from systematic error in survey design, for surveys with measurement tools or questions that fail to adequately control for interviewer bias, in cases where such bias could affect findings.

It can be difficult to engineer an effective interpenetrated design, however. There may be situations, particularly with large-scale face-to-face surveys, when geography or interviewer expertise with a particular class of respondent reduces the design's capacity to fully randomize the assignment of interviewer to respondent. There may be some benefit to determining whether a mixed strategy might be appropriate, with a partial randomization along respondent demographic, location, or cohort lines that are not believed to be relevant to the hypothesis of the survey or in its final analysis. As with any survey design, the question of which variables should be considered relevant must be approached with great caution.

...

  • Loading...
locked icon

Sign in to access this content

Get a 30 day FREE TRIAL

  • Watch videos from a variety of sources bringing classroom topics to life
  • Read modern, diverse business cases
  • Explore hundreds of books and reference titles

Read next

More like this

Sage Recommends

We found other relevant content for you on other Sage platforms.

close

Have you created a personal profile? Login or create a profile so that you can save clips, playlists and searches

close
close
close Signed In to profile You are signed in as:
close
Logged in Institution Institution