Content uploaded by C. Harry Hui
Author content
All content in this area was uploaded by C. Harry Hui on Nov 29, 2015
Content may be subject to copyright.
Effects of Culture and Response Format on Extreme Response Style
Abstract
Do cultural and ethnic groups differ in their extreme response style? To answer this question, Hispanic and non-Hispanic subjects were asked to respond to a questionnaire on 5-point or 10-point scales. As predicted, Hispanics were found to exhibit a stronger tendency for extreme checking (about half the time, on the average) than non-Hispanic, but only when the 5-point scales were used. Use of 10-point scales reduced the extreme responses of the Hispanics to the level of non-Hispanics. Extreme responses of non-Hispanics were not affected by the scales. Implications of the findings for social research are discussed.
... A DIM különösen komoly problémákat okozhat nemzetközi kutatásokban. Különböző kultúrák válaszadói nem csak a fogalmakat értelmezhetik különbözően (Weber, 2011), de olyan válaszadói viselkedések, mint a szélső értékek vagy éppen a középső érték preferálása bizonyos mértékig kulturálisan meghatározott (De Jong és mtsai, 2008;Hui-Triandis, 1989). King és mtsai (2004) klasszikus példájában a kínai válaszadók magasabbra értékelték lehetőségeiket abban, hogy állampolgárként milyen mértékben tudnak beleszólni a politikai folyamatokba, mint a mexikói válaszadók, a szerzők azonban kimutatták, hogy a különbség elsősorban abból adódott, hogy a két minta tagjai egészen eltérően használták a skálát, és "valójában" éppen fordítottan igaz a fenti öszszefüggés (minderről rövidesen részletesen). ...
... A DIM mögött végzettségbeli különböségek is állhatnak (pl. a magasabban kvalifikáltaknak kifinomultabb értelmezése van a skálapontokról), illetve az, hogy az egyének milyen mértékben érdeklődnek a politikai iránt (Weber, 2011). Végül arra is számos bizonyíték van, hogy bizonyos skálapontok (pl. a szélső értékek vagy a középső értékek) választása -a skálapontok szubsztantív tartalmától függetlenül -kulturálisan meghatározottak (De Jong és mtsai, 2008;Hui-Triandis, 1989;Van Vaerenbergh-Thomas, 2013). Egyes elemzések például arra utalnak, hogy a szélső értékek választása gyakoribb a dél-európai országokban, összehasonlítva a nyugat-európai országokkal és különösképpen az ázsiai kultúrákkal (áttekintésért lásd: Van Vaerenbergh-Thomas, 2013). ...
Kutatók világszerte használnak skálákat absztrakt fogalmak mérésére. A politikai szociológiai szakirodalomban a baloldali-jobboldali skála az egyik legelterjedtebb ilyen mérőeszköz. Ezen skálák alkalmazása erősen épít arra a feltételezésre, hogy az egyének ugyanúgy értelmezik a fogalmakat, illetve ugyanazt értik a skála egyes pontjain. Amennyiben a skála jelentése egyénileg és akár valamilyen változó(k) mentén szisztematikusan különbözik (differenciált itemműködés), az torzítást okozhat a mérésekben. Ez a probléma különösen hangsúlyos lehet nemzetközi kutatásokban, de országokon belül is felvethető. Tanulmányunkban egy Magyarországon felvett telefonos kérdőíves vizsgálat adataira támaszkodva vizsgáljuk, hogy mennyire egységes a bal-jobboldali skála jelentése. A válaszadókat arra kértük, hogy helyezzenek el két ideáltipikus személyt (egy baloldalit és egy jobboldalit) a bal-jobboldali skálán. A válaszadók mindössze 69 százaléka pozícionálta a „helyes” sorrendbe a vinyettákat. A személyek „helyes” elhelyezését erősen befolyásolta a végzettség: az érettségivel vagy diplomával rendelkezők jóval nagyobb valószínűséggel értelmezték „helyesen” a vinyettákat. A vinyetták abszolút pozíciója legerősebben végzettségi és politikai csoportok között szóródott: a jobboldaliak és a kormánypártiak minden vinyettát erősen jobbra, míg a baloldaliak és az ellenzéki szimpatizánsok erősen balra „húztak”. Tanulmányunkat egy olyan korrekciós módszer javaslatával zárjuk, amely King és munkatársai (2004) lehorgonyzó vinyetta-megközelítését alkalmazva használja fel a vinyettaválaszokat a saját ideológiai pozíció korrigálására, és amely más skálákra is kiterjeszthető.
... The accuracy of self-report data, particularly dietary behavior, has been questioned (Klesges, Eck, & Ray, 1995). In addition, response problems such as extreme responses, acquiescent responses, and socially desirable responses have been documented in Hispanic samples in proportions higher than those reported for non-Hispanic Whites (e.g., Hui & Triandis, 1989). Hui and Triandis (1989) reported that Hispanics tended to use extreme responses, such as always true, rather than more moderate responses, such as sometimes true. ...
... In addition, response problems such as extreme responses, acquiescent responses, and socially desirable responses have been documented in Hispanic samples in proportions higher than those reported for non-Hispanic Whites (e.g., Hui & Triandis, 1989). Hui and Triandis (1989) reported that Hispanics tended to use extreme responses, such as always true, rather than more moderate responses, such as sometimes true. Acquiescent responding is the tendency of participants to agree with a statement, regardless of its content. ...
The authors explored changes in dietary behavior, nutrition knowledge, and parental support among inner-city, low-income, Hispanic American families. Thirty-eight families were randomly assigned to receive a 12-week, culture-specific dietary intervention or be in a control group. Results showed that parental support was related to changes in diet, nutrition knowledge, and attendance for both mothers and children. Dietary behavior changes (e.g., reduction in dietary fat) were seen only in the treatment group. Distribution of health-related pamphlets to the control group may have promoted cognitive changes (e.g., increased nutrition knowledge) seen in this low-literacy sample. Further research is needed to document behavioral changes after ethnic-specific interventions and the maintenance of those changes over time.
... Further, from the response bias literature on mid-point, extreme, and acquiescent response styles, it is known that the extent to which scales and items are affected by response styles is related to rating scale format (e.g. Weijters et al., 2010;Deng & Bolt, 2016;Moors et al., 2014;Kieruj & Moors, 2013;Hui & Triandis, 1989;Henninger & Meiser, 2020). Little is known, in contrast, on whether manipulations of scale format can be leveraged to curb the occurrence of C/IER (see Robie et al., 2022, for a recent exception studying the effect of response option order.) ...
Common indicator-based approaches to identifying careless and insufficient effort responding (C/IER) in survey data scan response vectors or timing data for aberrances, such as patterns signaling straight lining, multivariate outliers, or signals that respondents rushed through the administered items. Each of these approaches is susceptible to unique types of misidentifications. We developed a C/IER indicator that requires agreement on C/IER identification from multiple behavioral sources, thereby alleviating the effect of each source’s standalone C/IER misidentifications and increasing the robustness of C/IER identification. To this end, we combined a response-pattern-based multiple-hurdle approach with a recently developed screen-time-based mixture decomposition approach. In an application of the proposed multiple-source indicator to PISA 2022 field trial data we (a) showcase how the indicator hedges against (presumed) C/IER overidentification of its constituting components, (b) replicate associations with commonly reported external correlates of C/IER, namely agreement with self-reported effort and C/IER position effects, and (c) employ the indicator to study the effects of changes of scale characteristics on C/IER occurrence. To this end, we leverage a large-scale survey experiment implemented in the PISA 2022 field trial and investigate the effects of using frequency instead of agreement scales as well as approximate instead of abstract frequency scale labels. We conclude that neither scale format manipulation has the potential to curb C/IER occurrence.
... Additionally, this type of scale eliminates the problem of extreme response styles, which is a tendency of individuals to either choose or avoid extreme answer points (Batchelor and Miao, 2016). This problem has been extensively studied in cross-cultural research showing that the tendency may cause significant interpretation problems (Clarke, 2000;Hui and Triandis, 1989;Johnson et al., 2005). The software imposes a control screen (Advanced Button) to ensure that the mouse's position does not bias the registered reaction time (RT) for yes and no answers. ...
... These findings imply that there may be interactions between response tendencies and particular method variables. Acquiescence may increase if the number of response options does not allow individuals to express themselves adequately (Hui & Triandis, 1985, 1989. Particular scales or items might be more ambiguous than others. ...
Response styles constitute a formidable challenge for cross-cultural research. In this article, three different response styles are discussed (acquiescence, extremity scoring, and social desirability). Acquiescence responding (ARS) is then integrated into a larger classical test theoretical framework, which allows for an examination of the various roles that ARS may play in cross-cultural research. A new meta-analytical method is proposed to examine the prevalence and nature of ARS. Preliminary evidence suggests that ARS has only a small, but systematic effect on survey responses. The meaning of ARS is explored through correlations with nation-level indicators. Implications for future research are discussed.
... However, the nature of the cross-cultural differences is not well understood (Fischer, Fontaine, Van de Vijver, & Van Hemert, this volume). Response styles can be dependent on characteristics of the culture, of the participants, of the instrument (e.g., the question format, wording and context) or their interaction (e.g., Hui & Triandis, 1989;Johnson & Van de Vijver, 2003;Schwarz, 1999). The current study focuses on characteristics of the culture and the instrument. ...
Likert-type rating scales are susceptible to response styles, such as acquiescence and extremity scoring. Although it is widely acknowledged that response styles can seriously invalidate findings of cross-cultural research, their theoretical underpinnings are hardly explored. The current study analyzed domain-dependency and country differences in acquiescence and extremity scoring in a large dataset of the International Social Survey Program. The hypothesis that response styles are more likely in domains with a high personal relevance compared to domains with a low personal relevance was tentatively confirmed. Correlations with various cultural, psychological, and economic variables were investigated. We found that acquiescence was negatively related to affluence, individualism, and well-being, while extremity was only negatively related to well-being. Positive associations were found between uncertainty avoidance and both acquiescence and extremity.
... In the present research, we took care in constructing response scales which could help avoiding ambiguity by providing midpoints in the response scale. Further, as previous findings suggest that choosing a response scale format of multiple points reduces extreme responses in Hispanic samples (Hui & Triandis, 1989), we stayed with 10-point scales. ...
Past research has shown that disgust is a heterogeneous category and lacks unity in its defining features. In the two studies reported in this paper, we examined the internal structure of disgust in English, and its translation equivalents of asco in Spanish, Ekel in German, garaf in Arabic, and yanwu in Chinese. In Study 1, 517 participants listed the most accessible constitutive features (definition, elicitors, and physical responses) of the concept of disgust in their culture. In Study 2, 653 participants were asked to judge the extent to which each of the 63 features extracted from Study 1 was typical of the concept of disgust in their respective culture. Results revealed differences in content, as well as internal structures across the five cultural groups: the disgust concepts differed in the degree of typicality of their constitutive features, the relevance of single features, the extent to which they shared features and the structural properties of the features. Taken together, our results question the assumed conceptual equivalence of the disgust concept across five cultures and raise questions about the suitability of deploying direct translations of disgust terms in cross-cultural research.
... Further, within Europe, Spanish and Italian participants showed more extreme responses than British, German, or French participants (van Herk, Poortinga & Verhallen, 2004). In the present research, care was taken in choosing Likert scales with 10 points, as suggested by Hui and Triandis (1989). We nevertheless adhered to the procedure of van Herk, Poortinga, and Verhallen (2004) and computed both negative and positive extreme responses by counting the responses given in terms of 0 (not typical ) and 10 (very typical ) and dividing the number by the number of items. ...
Using a prototype approach to emotion concepts, we mapped the internal structure and content of the everyday concept of envy (as used in the United States) and its translation equivalents of envidia in Spanish and Neid in German. In Study 1 (total N = 415), the features of the concept of envy , e nvidia , and Neid were generated via an open‐ended questionnaire. In Study 2 (total N = 404), participants rated the degree of typicality of the constitutive features on a forced‐choice questionnaire. The prototype analysis of envy, supplemented with network analyses, revealed that the largest connected set of features of envy , envidia , and Neid shared a group of central features, including features related to success or to people with a better appearance. Still, envy , envidia , and Neid did differ with respect to their constituent peripheral features as well as the density of their networks, their structure, and the betweenness centrality of the nodes. These results suggest that a prototype approach combined with network analysis is a convenient approach for studying the internal structure of everyday emotion concepts and the degree of overlap with respect to the translation equivalents in different countries.
Recent studies have shown that the validated dimensions in the best-known models of national culture converge into a two-dimensional default model, yielding a cultural map of the world reminiscent of the geographic one. The revised Minkov-Hofstede model is very similar to that default, whereas Inglehart-Welzel’s model is a rotated and flipped variant of it. However, another popular model - Schwartz’s - differs from the default: it does not have a dimension capturing the cultural contrast between East Asia and Latin America plus Africa. Consequently, it cannot explain national differences in educational achievement and a number of other important national indicators, relevant in international business. This omission in Schwartz’s model is puzzling as its author claims to have analyzed all values with invariant meanings across the world. On the other hand, Schwartz’s model has an idiosyncratic “mastery-harmony ” dimension that is not consistent with any geo-economic pattern and has poor predictive properties, constituting another weakness. We show that these idiosyncrasies of Schwartz’s model stem from Schwartz’s controversial decision to ipsatize his items and use multidimensional scaling: a method which, even without ipsatization, can create spatial opposites of items that are not negatively correlated. A principal component analysis of raw (non-ipsatized) Schwartz value domains does yield a variant of the default model of culture. We argue that although ipsatizing Schwartz value measures is not wrong in an absolute sense, it yields an impoverished and somewhat puzzling image of cultural differences across the globe, whereas raw measures reproduce the Minkov-Hofstede variant of the default model relatively well, although a different selection of values might perform even better.
The notion that social distance exists between an individual and others has been discussed in the social-psychological literature for some time. Social distance is here defined as that distance that is indicated by a person to exist, between himself and another person, by means of endorsement of certain statements. This present paper deals with the determinants. Seventy-eight Greek students and 91 Illinois college students rated 16 stimuli, consisting of imaginary people with characteristics involving different combinations of race, social class, religion, and nationality, on an equal-interval social distance scale. The stimuli were chosen according to a factorial design, which permitted the computation of the percentage of variance accounted for by race, religion, social class, and nationality. The responses of the subjects to several personality and attitude scales, as well as to a questionnaire investigating the child training practices of their parents, were obtained. The data were analyzed in terms of the differences between cultures and between subjects high, medium, and low in social distance towards minority group members in each culture. The data suggest that nationality and religion were the two most important variables for the subjects in the Greek sample, and race and religion for the subjects in the American sample. This suggests that different cultural norms are operating in the two cultures. Thus, the social distance that a given individual experiences towards outgroups is a complex function of his social distance norms and personality.
Notions of equivalence in cross-cultural measurement were related to the abstraction-concreteness and the universality-cultural difference continua. Various methods proposed for attaining satisfactory measurement were reviewed and compared within this framework. Each strategy has its own merits and shortcomings. Moreover, the level of cross-cultural equivalence presupposed, the type of equivalence demonstrated and/or improved, and the equivalence assumptions doubted or explicitly rejected are different for different strategies. It was suggested that the strategies are complementary to each other. More than one strategy should be employed and combined for more meaningful and precise measurement.
The authors offer a rejoinder to Professor Benson's comment on their article, "Rating Scales and Information Recovery-How Many Response Categories to Use?"
The author questions the procedure and the advice given researchers in a previously published analysis of simulated data.
Researchers who use surveys and long psychological instruments assume that response sets are constant over time. Data from
three studies are presented that challenge this assumption. Questionnaires of different response formats, subjects, and topics
were used. Correlations of the response sets across adjacent time periods formed a simplex, suggesting that the response sets
across adjacent time periods were more similar than the response sets across distant time periods. The implications of the
instability of response sets for the interpretation of social science findings obtained with long instruments are explored.
The neglect of the consequences of the Extreme Response Style (ERS) phenomenon in cross-cultural research is documented through a selective literature review. Consideration of ERS differences permitted derivation of three logical consequences that pertain to (a) differences in group means, (b) the internal consistency of measures and the correlation among measures, and (c) results of dimensional analyses. Comparison of U.S. and Korean student samples on a 1 30-item questionnaire revealed stronger ERS in the U.S. sample. Furthermore, the predictions derived from ERS differences were confirmed, thus demonstrating the credibility of the ERS argument. General implications of ERS for cross-cultural research are discussed.
A total of 225 college students rated themselves on how much they knew about 12 foreign countries. The rating scales differed in number of categories and in amount of verbal anchoring of scale points (center defined, ends defined, both center and end defined). Results indicated equal reliabilities for scales having 3, 5, 7, or 9 categories, but a decrease in reliability for 11 categories. Reliability increased as amount of verbal anchoring increased. (PsycINFO Database Record (c) 2006 APA, all rights reserved).
Two separate experiments were reanalyzed to investigate the hypothesis that test familiarity influences style of response. IQ and attitude questionnaire responses of 47 Israeli male Bedouin adolescents were compared to 85 of their more Western-oriented Moroccan peers. Both groups showed evidence of a response style biased toward the extreme right of multiple choice arrays which was, however, more reliable among the Bedouins. English language exposure was presumed to have influenced the differential results. Researchers were cautioned to be aware of response style problems as a legitimate concern.