About The ARDA | Tools | FAQs | Contact Us
Contingent Items

Not all items in a questionnaire apply to all respondents, so often respondents are asked to skip one or more questions, depending upon how they answered earlier ones. For example, questions about the respondent’s spouse will be asked only if the respondent is married. Asking irrelevant questions would waste the respondent’s time, and possibly confuse or annoy the respondent sufficiently to cause him or her to bail out of the survey, or simply to give lower quality responses to the next few items. ARDA usually makes it clear when how question may contingent on another, but using common sense and look at crosstabulations of data from pairs of questions can verify exactly what was contingent on what. For example, in the Baylor Religion Survey, 2005, many questions concern “your current place of worship.” For each of these, around 480 respondents have missing data, and most of them are people who lack a place of worship. While analyzing existing data, it is crucial to understand the implications of the skip pattern in the particular dataset.

Probably the most common contingent items in religion surveys concern the respondent’s religious denomination. For example, here are three items from the 2006 General Social Survey:

130) What is your religious preference? Is it Protestant, Catholic, Jewish, some other religion, or no religion? (RELIG)
131) IF PROTESTANT: What specific denomination is that, if any? (DENOM)
133) IF JEWISH: Do you consider yourself Orthodox, Conservative, Reform or none of these? (JEW)

Question 131 is asked only if the respondent answered “Protestant” to question 130, and question 133 is asked only if the answer to question 130 was “Jewish.” It is quite appropriate to work only with the data from a contingent question, so long as results are clearly reported. For example, using the data from item 131 alone, one could say what percent of Protestants were members of the United Methodist Church. But if one wanted to say what percent of people belonging to any religious denomination were members of the United Methodist Church, it would be necessary to combine data from items 130 and 131.

In designing religion questionnaires, researchers are often faced with a dilemma of deciding which questions are relevant to people of different beliefs. For example, is it worth administering to an atheist a battery of items about how a person conceptualizes God? Even an atheist will have a picture of God in his or her mind, just as people have an image of Santa Claus without believing in his actual existence. Similarly, people who lack religious faith may none-the-less have had personal experiences in their lives that other people would call religious.

Sometimes, a questionnaire will ask a question that some respondents will feel is not relevant to them, but provide an explanation. For example, the American Mosaic Project: A National Survey on Diversity is a telephone interview study that has a battery of social-distance questions, beginning with:

43) People can feel differently about their children marrying people from various backgrounds. Suppose your son or daughter wanted to marry an African American. Would you approve of this choice, disapprove of it or wouldn’t it make any difference at all one way or the other? (IF PERSON SAYS, “I’m African-American”, SAY “We’ll be asking about other groups you don’t belong to as well” and repeat) (MARR1A)

Note that this item begins with an introductory sentence, but then if the respondent happens to be Africa American and does not see the point of the question, the interviewer speaks a contingent explanation then repeats the question.

Contingent questions can become rather complex, as illustrated by an item in the National Study of Youth and Religion, asked of a tiny fraction of parent respondents:

172) P100. [If parent respondent has not been attending religious services AND does not identify with a religion AND is married or living with a partner][(IF P80>6 AND P811) AND P2C<3] Is your spouse/partner religious, or not religious? (PRLSPREN)

1) Religious 0.9%
2) Not religious 2.8%
777) Don’t know 0.1%
888) Refused 0.0%
999) Not asked 96.1%
TOTAL 100.0%

If a questionnaire is administered on paper, it probably should not use elaborate skip patterns, because the respondent will get lost. Even simple contingencies need to be made clear through large text and arrows showing the respondent where to go next. Patterns can be more complex if a trained interviewer is administering the questionnaire. For many years, telephone interviewers have used computers that built the skip pattern into the program, so that the interviewer did not need to worry about it. When the interviewer presses the appropriate key to indicate whether the respondent was Catholic, Protestant or Jewish, the computer would automatically display the appropriate follow-up question, if any. Starting in the late 1990s, many surveys were administered online, and complex skip patterns can easily be built into the program. When very large numbers of online respondents are expected, some surveys have incorporated specialized modules that did not need all the respondents to the full questionnaire, selecting respondents at random rather than on the basis of a respondent’s answer to an earlier question. The key point for secondary analysis is to be clear on how the questions were asked, and be aware when only a subset of respondents were asked certain questions.

QuickSearch The Knowledge-Base

To search the knowledge-base, enter a term below:

Select a Theory below to learn more:

Select a Concept below to learn more:

TCM Contributors

Would you like to be considered for a position on the Theories, Concepts, and Measures contribution team? If so, click on the link below and complete the TCM Online Request Form.

TCM Online Request Form

If you are already a contributor for Theories, Concepts, and Measures and need assistance in managing the content, please click on the link below for instructions.

Site Administration Instructions