ASSIMILATION AND CONTRAST EFFECTS IN PART-WHOLE QUESTION SEQUENCES: A CONVERSATIONAL LOGIC ANALYSIS

A theoretical model of the emergence of assimilation and contrasteffects in part-whole question sequences is presented. Whenone specific question precedes a general question and the twoare not assigned to the same conversational context, respondentsuse the information primed by the specific question to formthe general judgment. This results in part-whole assimilationeffects. If both questions are perceived as belonging to gether,however, conversational norms of nonredundancy prohibit therepeated use of information that has already been provided inresponse to the specific question when making the general judgment.Accordingly, respondents interpret the general question to referto aspects other than the ones covered by the specific question.Contrast effects may emerge in that case under specified conditions.If several specific questions precede the general question,however, the general one is always interpreted as a requestfor a summary judgment. This results in assimilation effects,even under conditions that would foster contrast effects ifonly one specific question is asked. The model is supportedby experimental data and provides a coherent account of apparentlycontradictory findings previously reported in the survey literature.     

Question Order Effects on Subjective Measures of Quality of Life1

The effects of question order on respondents' ratings of general and specific aspects of community life were assessed using data from separate mail surveys in Montana and Pennsylvania, The samples differed in locale, composition, and size and the relevant questions varied in number, format, and specific focus. Nevertheless, for both data sets the general question was more likely to be answered and more likely to receive positive responses when it was asked after—rather than before—the specific questions. There was some indication that carryover from the specific items to the general question responses were somewhat greater for those questions asked most recently and less for those asked earlier. Implications of these findings are discussed.     

Presentation matters: how mode effects in item non-response depend on the presentation of response options

Comparability of data across modes is an important issue in survey research. In this paper we discuss item non-response to attitudinal questions in telephone and web surveys. We present results from a survey experiment conducted in Italy and in Spain that compares different presentations of response options in an online setting with a benchmark telephone survey. In line with earlier studies we find that (A), the share of non substantial answers in the online survey depends on how the response option is presented. Comparing different presentations in an online survey to the standard approach of telephone surveys using propensity score matching, we find that (B), the share and pattern of non substantial answers, is most similar across the two modes in the online survey when the it is captured in an instruction on each screen for the online survey. Our findings are of particular relevance for the design of multiple mode or mix-mode surveys of attitudinal questions using online and telephone modes.     

Context Effects on Survey Responses to Questions About Abortion

The proportion of favorable responses to a general questionabout abortion was significantly smaller in an NORC survey thanwere responses to the same item in an SRC survey. We hypothesizedthat questionnaire context was the main source of the difference–inone survey the general item followed a more specific questionabout abortion–and carried out two split-ballot questionorder experiments that provide strong support for the hypothesis.The context effect seems to arise from the generality of theabortion question; the specific question itself showed no changedue to context. Furthermore, there is some evidence that ambivalencetoward the issue of abortion is a factor in the sensitivityof the general question to context. The context effect in theseexperiments is also notable for producing contrast rather thanconsistency between responses to adjacent questions on the sameissue. Implications of the findings for the study of socialchange are emphasized.     

Effects of Question Order on Survey Responses

Two question orders were used in a Kentucky statewide randomsurvey. In nonoverlapping sections, four questions which askedthe respondents to express their general interests in politicsand religion and their general evaluations of the gravity ofthe economic and energy situations either preceded or followedseries of specific questions on the same issues. The respondentsexpressed significantly greater interest in politics and religionwhen these general questions followed the specific questions,but evaluations of the economic and energy crises were not significantlyaffected by question placement. Implications for the arrangementof questions within surveys are discussed.     

A PARTIAL INDEPENDENCE ITEM RESPONSE MODEL FOR SURVEYS WITH FILTER QUESTIONS

In many surveys, responses to earlier questions determine whether later questions are asked. The probability of an affirmative response to a given item is therefore nonzero only if the participant responded affirmatively to some set of logically prior items, known as "filter items." In such surveys, the usual conditional independence assumption of standard item response models fails. A weaker "partial independence" assumption may hold, however, if an individual's responses to different items are independent conditional on the item parameters, the individual's latent trait, and the participant's affirmative responses to each of a set of filter items. In this paper, we propose an item response model for such "partially independent" item response data. We model such item response patterns as a function of a person-specific latent trait and a set of item parameters. Our model can be seen as a generalized hybrid of a discrete-time hazard model and a Rasch model. The proposed procedure yields estimates of (1) person-specific, interval-scale measures of a latent trait (or traits), along with person-specific standard errors of measurement; (2) conditional and marginal item severities for each item in a protocol; (3) person-specific conditional and marginal probabilities of an affirmative response to each item in a protocol; and (4) item information and total survey information. In addition, we show here how to investigate and test alternative conceptions of the dimensionality of the latent trait(s) being measured. Finally, we compare our procedure with a simpler alternative approach to summarizing data of this type.     

A Comparison of Survey Techniques on Sensitive Sexual Behavior in Italy

This article compares two national surveys carried out through the most commonly used procedures in Italy: CATI (computer-assisted telephone interviews) and SAQ-FI (self-answered questionnaires following interviews). Both surveys ask two identical questions concerning sensitive sexual behavior: early age at first intercourse and same-sex attraction. The SAQ-FI survey had both unit non-response and item non-response rates much lower than the CATI survey. Moreover, in the CATI survey, the groups with highest item non-response rates were also the groups with the lowest proportions of early intercourse and homosexual attraction. In addition, a differential analysis of the respondents produced diverse results for the two surveys. This is especially true of results by gender for same-sex attraction: Such behavior is more common among men (3.1%) than women (2.9%), according to the CATI survey, whereas the opposite is true of the SAQ-FI survey (6.1% of men vs. 7.7% women). In Italy at the beginning of the 21st century, CATI surveys reveal a lower level of early intercourse and same-sex attraction than SAQ-FI surveys. This article argues that the CATI survey underestimates the true level of these sensitive sexual behaviors in the Italian population.     

MEASURING HAPPINESS IN SURVEYS: A TEST OF THE SUBTRACTION HYPOTHESIS

Responses to an item on general happiness can change when thatitem follows one on marital happiness. Asking about maritalhappiness first sometimes reduces reported levels of overallhappiness. This reduction may result from a change in the interpretationof the general happiness item. According to this hypothesis,when the general item follows the item on marital happiness,respondents take the item to ask about aspects of their livesother than their marriages-in effect, the respondents subtracttheir (mostly happy) marriages in answering the general item.The study reported here tests this subtraction hypothesis byasking versions of the general happiness item that correspondto the different interpretations. A version of the general itemthat asked about general happiness "aside from your marriage"yielded responses that were quite similar to those given tothe standard item when it followed the item on marital happiness.Another version that asked about general happiness "includingyour marriage" elicited responses quite similar to those elicitedby the standard item when it preceded the marital happinessitem. However, reanalysis of the studies that originally demonstratedthe impact of the order of the two happiness items casts doubton the subtraction hypothesis and related models as explanationsof the earlier findings.     

The Measurement of a Middle Position in Attitude Surveys

Five split-ballot experiments, plus replications, were carriedout in several national surveys to compare the effects of offeringor omitting a middle alternative in forced-choice attitude questions.Explicitly offering a middle position significantly increasesthe size of that category, but tends not to otherwise affectunivariate distributions. The relation of intensity to the middleposition is somewhat greater on Offered forms than on Omittedforms (less intense respondents being more affected by questionform than those who feel more strongly), but in general formdoes not alter the relationship between an item and a numberof other respondent characteristics. Finally, in one instancethere is evidence that form can change the conclusion aboutwhether two attitude items are related, but the results areof uncertain reliability.     

Can We Measure Immigrants' Legal Status? Lessons from Two U.S. Surveys

This research note examines response and allocation rates for legal status questions asked in publicly available U.S. surveys to address worries that the legal status of immigrants cannot be reliably measured. Contrary to such notions, we find that immigrants' non‐response rates to questions about legal status are typically not higher than non‐response rates to other immigration‐related questions, such as country of birth and year of immigration. Further exploration of two particular surveys – the Los Angeles Family and Neighborhood Survey (LAFANS) and the Survey of Income and Program Participation (SIPP) – reveals that these data sources produce profiles of the unauthorized immigrant population that compare favorably to independently estimated profiles. We also find in the case of the SIPP that the introduction of legal status questions does not appear to have an appreciable “chilling effect” on the subsequent survey participation of unauthorized immigrant respondents. Based on the results, we conclude that future data collection efforts should include questions about legal status to (1) improve models of immigrant incorporation; and (2) better position assimilation research to inform policy discussions.     

When near means related: evidence from three web survey experiments on inter-item correlations in grid questions

In this article, we present the results of three question design experiments on inter-item correlations, which tested a grid design against a single-item design. The first and second experiments examined the inter-item correlations of a set with five and seven items, respectively, and the third experiment examined the impact of the question design on inter-item correlations of follow-up questions. We found that the inter-item correlations in grid designs were higher for both the set of five and seven items than for questions in a single-item design. Additionally, the inter-item correlations of the set of follow-up questions were higher if the preceding set of questions was presented in a grid. According to the findings of the present research, respondents seem to apply the interpretative near means related heuristic when answering grid questions, and also when answering follow-up questions preceeded by grid questions.     

CONTRAST EFFECTS AND SUBTRACTION IN PART-WHOLE QUESTIONS

Responses to a question about the state economy changed whenthe question followed a similar item about the economy of therespondent's local community. The overall distribution of theanswers changed, as well as the correlation between the twoitems. Prior work has suggested that respondents may excludespecific aspects of a general issue from consideration whena question about the general issue follows a question aboutthe specific aspect of it. Responses to an open-ended item indicatethat respondents may have excluded a salient characteristicof the community economy when subsequently assessing the stateeconomy, even though this characteristic figured prominentlywhen respondents evaluated the state economy first.     

Prevalence of Sexual Trauma History Using Behaviorally Specific Methods of Assessment in First Year College Students

There are several challenges associated with evaluating the prevalence of sexual trauma, including child sexual abuse and adult sexual assault. The aim of this study was to assess sexual trauma prevalence rates among first year college students (N = 954) using behaviorally specific questions and a more representative recruitment sample that did not rely on self-selection. Participants completed a list of sexual trauma questions, including general questions containing labels such as “rape” or “abuse” as well as behaviorally specific questions that describe specific behaviors that qualify as sexual trauma without labels. Results indicated that 6.7% of the sample reported at least one incident of child sexual abuse, with similar rates for men and women. Women were more likely to report a history of adult sexual assault, which was reported by 12.4% of the total sample. Participants were also more likely to endorse a history of sexual trauma when answering behaviorally specific rather than general “label” questions. Women survivors in particular were more likely than men to identify their experiences as abuse/assault (66.7% versus 21.1% for child sexual abuse), which may help explain prevalence differences between men and women in prior research. Men may be less likely than women to label their experiences as abuse and may be underidentified in sexual trauma research without the use of behaviorally specific questions. Overall, the results of this study suggest that the prevalence of sexual trauma is better assessed using behaviorally specific questions and that this is an important topic of study among both men and women.     

Comparing the Quality of Data Obtained by Minimally Balanced and Fully Balanced Attitude Questions

When measuring attitudes with questions that offer dichotomous,mutually exclusive response options, researchers can ask "fullybalanced" questions (which fully state both competing pointsof view) or "minimally balanced" questions (which fully stateone viewpoint and only briefly acknowledge the second viewpoint).The two studies reported here investigated whether the greaterefficiency of the latter approach brought with it reductionsin the quality of the data obtained. Two experiments embeddedin national sample surveys showed that minimally balanced andfully balanced attitude questions yielded similar distributionsof responses and that responses to the two question forms wereequivalent in terms of concurrent validity. These studies suggestthat greater efficiency can be achieved via minimal balancingat no cost in terms of data quality.     

Estimating sexual behavior parameters from routine sexual behavior data

In mathematical models for predicting the spread of sexually transmitted diseases (STDs), the rate of acquisition of new sex partners and concurrency, the number of simultaneous sexual partnerships, are important parameters. Yet, information on these parameters is rarely obtained in routine sexual behavior surveys; instead, questions about the total number of sex partners during specific periods are often asked. We present two methods that estimate the rate of partnership acquisition and concurrency from reported numbers of sex partners in two overlapping periods. The first method assumes homogeneous behavior, while the second method also takes account of heterogeneity in sexual behavior. Both methods assume random (Poisson) partnership acquisition and the possibility of overlap in relationships. For both methods it might be relevant to distinguish individuals who are in a stable relationship from those who are not.     

How unclear terms affect survey data

Although writing clear questions is accepted as a general goal in surveys, procedures to ensure that each key term is consistently understood are not routine. Researchers who do not adequately test respondent understanding of questions must assume that ambiguity will not have a large or systematic effect on their results. Seven questions that were drawn from questions used in national health surveys were subjected to special pretest procedures and found to contain one or more poorly defined terms. When the questions were revised to clarify the definition of key terms, significantly different estimates resulted. The implication is that unclear terms are likely to produce biased estimates. The results indicate that evaluation of survey questions to identify key terms that are not consistently understood and defining unclear terms are ways to reduce systematic error in survey measurement.     

CAN CHANGES IN CONTEXT REDUCE VOTE OVERPORTING IN SURVEYS?

This paper reports the results of two split-sample experimentsdesinged to examine whether question context can reduce voteoverreporting in surveys. The first experiment tests the effectsof preceding the turnout item with a question about the locationof the polling place. If respondents who overreport do notknow where their polling place is, they should be less apt toreport incorrectly about turnout after being asked the locationof the poll. The second experiment tests the effects of precedingthe vote item with a question about lifetime electoral behaviour.If inaccurancy is dueto respondent's wanting to present themselvesin a favourable light, the oppurtunity to report past or usualgood citizenship should reduce the pressure to claim participationin the last election.     

Getting the story straight: evaluating the test-retest reliability of a university health history questionnaire.

This study was designed to establish the reliability of a health history questionnaire used as a screening tool for incoming university students. The authors used a test-retest design, with a test interval of 6 months, on a sample of medical and nursing students. The analysis focused on overall reliability of the questionnaire and reproducibility of specific items, based on question format. Questionnaire items of specific interest were those with dichotomous yes/no response options versus open-ended format questions, those using the words frequently or recently, or those that asked multiple questions. Demographic characteristics of the subjects were considered in the evaluation of reliability. Overall reliability of the questionnaire (93.6%) was above the anticipated level of 90%, and subject sex or program of study did not show any significant differences in reproducibility of responses. Although wording of questions did not affect item reliability, dichotomous format questions demonstrated a higher degree of reliability (96.4%) than the overall reliability of the questionnaire. Recommendations for enhancing the reliability of the questionnaire are based on item analysis and information gathered from interviews with subjects.     

Racialized Incorporation: The Effects of Race and Generational Status on Self‐Employment and Industry‐Sector Prestige in the United States

This study examines how race and generational status shape self‐employment propensities and industry‐sector prestige among the self‐employed in the U.S. It draws on theories of assimilation, racialization, and a combined framework, racialized incorporation, to guide the analysis and interpret the results. It uses data from the U.S. March Current Population Survey (2000–2010) offering the first nationally representative examination of second‐generation self‐employment in the U.S. This study investigates three questions. First, do the odds of being self‐employed decline in the second and third generations? Second, do generational patterns in self‐employment propensities vary by race? And finally, do race and generational status affect the odds of being self‐employed in low‐, medium‐, and high‐prestige industry sectors? Results offer some support for the assimilation perspective: Immigrants are generally more likely than third‐generation groups to be self‐employed with the exception of Asians, where second‐generation Asians have the greatest odds of being self‐employed. However, results also reveal that generational patterns in self‐employment propensities vary by race and industry‐sector prestige. Accordingly, first‐ and second‐generation whites have the greatest odds of being self‐employed (across all levels of industry‐sector prestige), and third‐generation whites are more likely than all generations of blacks and Hispanics to be engaged in high‐prestige self‐employment. These findings suggest that immigrants, their offspring, and native‐born groups undergo a racialized incorporation in which self‐employment is organized along hierarchical and racial lines associated with uneven levels of prestige.