Abstract
Survey instruments are often watermarked with the language, thought patterns, experience and expectations of their designers’ cultural world. This creates some evident challenges when using surveys in international research projects (for example, in finding equivalent terms in translation), but also some less evident ones deriving from potentially hidden local variations in social experience, values and cognition. Using examples from the International Self-Report Delinquency Study, this article identifies several potential challenges to measurement that arise when an instrument designed with the youth of industrialized democracies in mind is presented to adolescents in a Latin American country (Venezuela). Some strategies for addressing these challenges are outlined.
Introduction
How well do survey instruments travel? This question nicely captures the challenges identified by many researchers interested in international projects (for example, Berry et al., 2002; Greenfield, 1997). Put simply, will instruments originating in one culture have the same level of validity and reliability when used in other cultures? Or will ‘imported’ instruments interact with the cultural characteristics of local respondents to generate changes to measurement that confound the comparisons between survey sites? For example, does the embarrassment reportedly felt by women in Uganda (see Ssamula, 1995) – and probably in many other places – when asked about incidents of sexual victimization lead to comparative under-reporting in that country? Or would items concerning mental maps of crime be affected by the fact that Colombians apparently do not orient themselves in terms of cardinal directions (Ardila, 2007)? These sorts of question call for assessments of the extent to which survey instruments reflect their originators’ cultural world; whether demonstrably idiosyncratic content affects measurement; and how any resulting biases can be overcome.
Such matters can be resolved only by sustained enquiry. Thus, although the notion of travel invokes an image of the manufacture, transport and application of survey instruments, patterns of international usage are evidently more complicated than this. Instruments that ‘travel’ represent the cross-cultural model of survey diffusion, which, in its pure form, is exemplified by a research group in one culture designing an instrument and then using it in other cultures. The alternative model is multicultural, in which researchers from several cultures come together to design a common instrument that they use at each site. Indeed, these two models could be placed at the opposite ends of a scale and specific research initiatives located along it. But, although the cross-cultural model is more likely than the multicultural model to pose questions about the validity and reliability of the research instrument, a detailed assessment of items in terms of their cultural uniqueness or generality is necessary in all international survey projects.
Similarly, whereas assessments of survey items’ cultural specificity can be made with varying degrees of certainty, the effects of that specificity (if any) on the data generated requires systematic investigation. Social scientists have devoted considerable efforts to studying the interaction between instrument design and resulting responses, well exemplified in criminology by the extensive methodological work on surveys that measure offending or victimization (Junger-Tas and Haen-Marshall, 1999; Lynch and Addington, 2010; Thornberry and Krohn, 2000). However, the exploration of these effects in the international context has been absent from criminology and is only beginning to be addressed in other fields, such as development studies and health (for example, Beaman and Dillon, 2012; Das et al., 2012).
In this paper, we use the application of the International Self-Report Delinquency Study (ISRD) in Venezuela as a case study to explore the particularities of an international survey instrument and assess their implications for the validity and reliability of measurement. Our argument will be that the ISRD instrument is quite strongly imprinted with the culture that spawned it, a culture that reflects the highly industrialized societies of Northern Europe, the USA and Canada. Although Venezuela is an urbanized middle-income and oil-rich country, in many studies of the world’s major cultural types it is often classed with less industrialized societies – for example, as a ‘collectivist’ rather than an ‘individualist’ culture (Hofstede, [1980] 2001) or as a ‘high-context’ rather than a ‘low-context’ culture (Hall, 1976). The particularities of the ISRD survey instrument are identified through an exploration of differences between the experiential and cognitive world conveyed by its items and the experiential and cognitive world likely to be inhabited by many Venezuelan adolescents. Some of these differences are relatively easy to detect and substantiate, originating in varying organizational arrangements for activities and institutions. Others are fairly easy to detect but more difficult to substantiate, for example those arising from the normative ambiguity that surrounds specific behaviours, which puts their validity as measures of delinquency in doubt. Still others are difficult both to detect and to substantiate, requiring a keen eye for social arrangements and cognitive frameworks and extensive research to document them. Drawing on a variety of sources and approaches, we suggest some Venezuelan particularities at each of these levels and speculate on the possible impact of these particularities on the validity and reliability of the data collected with the ISRD instrument. At each level we briefly propose a strategy for dealing with measurement issues.
The International Self-Report Delinquency Study: A brief biography
Like the International Crime Victimization Survey (ICVS) before it (see Van Dijk, 2015), the ISRD project was launched from the Netherlands (Junger-Tas, 2010). The initial methodological experiments in surveys of self-reported delinquency had been conducted in the USA and had led to the conclusion, later summarized by Thornberry and Krohn (2000: 33), that ‘self-report [offending] data appear acceptably valid and reliable for most research purposes’. Indeed, self-report surveying of delinquency had been institutionalized in the US National Youth Study (NYS) since the mid-1970s (Elliott and Huizinga, 1989). This experience provided some of the intellectual and methodological antecedents to the ISRD (for example, Klein, 1989), which was initiated by the Dutch Research and Documentation Centre in 1990 (Aebi, 2009).
Like the NYS, the first ISRD project (ISRD-1, carried out between 1991 and 1992) focused on youths (12–18) and sought out respondents in their household or school. Thirteen countries participated in the study, mainly from Europe but also including the USA. The second ISRD project (ISRD-2) conserved many of the characteristics of the first project while also instituting some changes (Junger-Tas et al., 2010). It narrowed the target age group being studied to students attending 7th to 9th grade (roughly 12–16 years old depending on the educational system of each country); mainly involved city samples in each country; and used exclusively school-based samples. Project leadership was provided by an international group of scholars based in Western Europe and North America, who had principal responsibility for deciding on the sampling strategy and design of the instrument. Other countries were invited to participate and a total of 31 undertook data collection between 2005 and 2007. Of these, 25 were located in Europe and six in the Americas. The latter included Canada and the USA but also two Caribbean island nations (Aruba, Netherlands Antilles) and two South American countries (Surinam, Venezuela).
The third ISRD project (ISRD-3, 2012–14) covered a similar number, and spread, of countries as the second sweep (again, including Venezuela). It retained the same target age group, school-based sampling, and a similar city-based sampling strategy as ISRD-2, along with many of the latter’s survey items. In line with the project coordinators’ ambitions to use the ISRD project to test current criminological theories, questions from previous sweeps on self-control, family, school and neighbourhood bonding were accompanied in ISRD-3 by questions relating to situational action theory (Wikström, 2009), institutional anomie (Messner and Rosenfeld, 2007) and procedural justice (Tyler, 2006).
In terms of the two models of international survey work identified previously (cross-cultural, multicultural), the ISRD is an intermediate case. Its roots in US criminology have undoubtedly exercised a strong influence on the choice of variables to be measured in the survey and on the cognitive framework employed for eliciting information about the prevalence and incidence of the behaviours of interest. The European sponsors and associated participants in the ISRD have engaged in fairly extensive discussions about the instrument’s design, particularly regarding the items for measuring delinquency (Junger-Tas et al., 2010: 2; Junger-Tas et al., 2003: 4). However, the extent to which varied social arrangements, expectations and mindsets across the countries participating in the survey could affect the validity and reliability of the instrument has not been fully considered, even as the project has extended to Eastern Europe and Russia and to the Caribbean and South America. This is undoubtedly owing to the considerable effort required to develop comparable samples (Haen-Marshall, 2010) and the obvious requirement to use a common, standardized, instrument. In meeting this latter requirement, attention has focused on translation rather than item design. In the following sections, we discuss some of the potentially ethnocentric characteristics of the ISRD-3 survey instrument by offering contrasting examples – some easily detected and substantiated, others less so – from Venezuela.
Cultures in contrast
As shown in Table 1, the ISRD-3 instrument consists of 11 modules containing 73 main questions. Many of the latter include itemized repetitions of the main question; for example, Question 7.1 on offending behaviour has 15 separate sub-sets of questions, each referring to a specific kind of offence. As is to be expected, many of the items in the instrument do not raise concerns about cultural equivalence – at least from the vantage point of urban Venezuela. 1 For example, asking adolescents from Caracas to indicate their age in years, whether they have lost a parent through death and how frequently they do homework looks to be no different from asking the same thing of adolescents in most other big cities around the world. Our analysis of the cultural particularities in the ISRD instrument focuses on selected questions or sub-items, proceeding from examples that are fairly obvious, easily substantiated and easy to resolve, to cases that are suggestive, less easily substantiated and somewhat more difficult to resolve.
ISRD-3: Overview of the instrument.
Obvious differences
When referring to objects, activities and institutions, survey items may incorporate the typical experiences of their designers. For example, one of the behaviours itemized in the ISRD-3 delinquency inventory is stealing a bicycle. This presumably reflects the prevalence of bicycles (and their theft) in Western Europe, especially the Netherlands. By contrast, despite a growing minority of sports enthusiasts, Venezuela is a country where cycling is almost entirely absent from urban streets and roads, perhaps partly because of the perceived risks of riding in anarchical traffic and partly because of the tropical climate. Thus, asking Venezuela adolescents about bicycle theft is probably of little relevance when attempting to measure the quantity and variety of their delinquent behaviour. Similarly, another item asks respondents if they have taken part in ‘a group fight in a football stadium, on the street or other public place’. Here, the prominent mention of football stadiums evokes not just the world of European soccer but its potential significance for youth violence. As a sporting country, Venezuela is more Caribbean than Latin American, with a stronger development of baseball than football. Although there has been some violence at soccer matches, the majority of Venezuelans are more likely to have attended a baseball game, if they have been to a sports event at all – for rates of attendance at sporting events may be lower than in Europe or Anglo America.
Finally, a subsequent question asks respondents if they have ever had contact with the ‘police’ concerning the types of offending listed earlier in the survey. Additionally, Module 10 has eight questions concerning perceptions of the ‘police’. The difficulty here concerns who a Venezuelan adolescent might think of when asked about the ‘police’. This is because, as in most Latin American countries, there are a number of different agencies that engage in police work, but not all of them have the word ‘police’ in their title. In Caracas, for example, adolescents may encounter officers attached to the National Police (Policía Nacional Bolivariana), created in 2009 through the amalgamation of state, municipal and traffic police. However, they will also see ‘national guards’ from the Guardia Nacional Bolivariana, a hybrid military and civilian corps assigned to street patrol and public order, along with ‘detectives’, ‘functionaries’ or ‘agents’ from the ‘scientific police’ (Cuerpo de Investigaciones Científicas, Penales y Criminalísticas) who investigate crime and undertake some patrol activities. So, when answering the questions in the ISRD-3 instrument, would an adolescent think of those who are called ‘police’ or those who engage in police work? And might perceptions and experiences vary quite markedly for different agencies? 2
Fortunately, the superficial nature of this type of problem, deriving as it does from different arrangements regarding activities and institutions, suggests relatively simple solutions. Thus, the varied importance of bicycles and soccer across different cultures points to the need to frame questions in terms of more generalized referents. For example, respondents could be asked about the theft of ‘things’, with some generalized examples mentioned (for example, phones, wallets or money) or several more particular examples included (for example, bicycles [Europe] or backpacks [Latin America]). In relation to group fights, football stadiums could be changed to ‘sports events or concerts’ and relegated to the end of the question. 3 And, for questions on the police, the relevant agencies would need to be briefly indicated in each country.
Going deeper: Normative ambiguity
As commentators on the self-reported assessment of delinquency have pointed out, the operationalization of youth offending is not without problem (Elliott and Huizinga, 1989; Junger-Tas and Haen-Marshall, 1999; Thornberry and Krohn, 2000). Within industrialized countries, discussion has focused mainly on the validity of status offences and trivial delinquent behaviour as measures of offending, on the number and range of offences included in surveys, and on the way in which the incidence of offending is counted. International projects raise the additional matter of varied normative and legal perspectives on behaviour: what is a problematic or delinquent conduct in one country may not be so in another. Although the ISRD coordination team has done much to develop a set of items that are common to all participating countries, the inclusion of Venezuela in ISRD-2 and ISRD-3 highlights the need for continued discussion about the cultural equivalence of some of the questions.
For example, one item asks respondents if they have ever ‘painted on a wall, train, subway or bus (graffiti)’. We note, in passing, that trains are another example of a superficial difference: the single suburban train line in operation in Caracas since 2006 contrasts with the dense and longstanding networks of trains in Western Europe and some other parts of the world. But more problematic is the reference to graffiti. Although the term is known in Venezuela and used to evaluate street painting as either art or vandalism (for example, El Universal, 2011), there is also a rich variety of street ‘messaging’, including political protest and propaganda, moral slogans, gang-style tagging and even declarations of love (see Alonso, 1998). Numerous and highly varied groups and individuals fill many walls in public spaces, which often include the boundary walls of private properties. Painting on walls might seem unsightly to those who worry about urban aesthetics (or whose private walls have been affected), but public spaces are more a concourse of competing private interests and activities than sites for satisfying collective values of order or taste. Although there is some debate about their desirability (for example, El Universal, 2009), street painting and messaging vary from the institutionally sponsored/artistic/moralistic to the defacing and downright insulting and are rarely considered (or controlled) as illegal in Venezuela. Thus, ‘painting on walls’ could have multiple connotations for adolescents. Indeed, those who have painted on walls are as likely to have done it as part of an organized school, community or political activity, as they are to have gone out at night to leave a furtive sign.
A second item in the survey asks about illegal downloading of music or films from the Internet. Unauthorized and unlicensed copies of music, film, video games and software are widely available in Venezuela’s informal commercial sector and rarely confiscated by the authorities (Lugo and Sampson, 2008). Despite being commonly referred to as ‘pirated’ (for example, La Patilla, 2012), their existence possibly reinforces the doubts about the strict illegality of these copies and probably reflects the widespread ‘neutralization’ of this offence by reference to the relatively high cost and limited accessibility of legitimate copies. With the expansion of Internet and associated technologies, young people are quickly socialized into a culture of downloading that carries few negative consequences. Indeed some Venezuelan respondents to ISRD-3 expressed surprise that downloading could be considered illegal.
Both of these examples reveal the normative ambiguity surrounding behaviours that, in other contexts, might be more clearly defined and recognized as illegal. They suggest that responses to these two items should be set to one side in the analysis of delinquency among Venezuelan adolescents, and that in future sweeps of the ISRD further thought needs to be given to a set of delinquency items common to all countries.
Hidden differences
For the heading to this section, we borrow the title of one of anthropologist Edward Hall’s popular books for business (Hall and Hall, 1990). As Hall continuously emphasized, many cultural differences are not immediately apparent and are revealed only through careful study. His own approach was largely qualitative and based on extensive interviews with business leaders, leading to characterizations and classifications of key cultural features. Quantitative approaches, for example that based on the World Values Survey (for example, Inglehart and Carballo, 1997), lead to more complicated pictures. Additionally, it is important to recognize the space for individual (for example, Triandis et al., 1988) or group departures from the prevailing cultural ‘type’ found in a country. Indeed, the existence of cultural differences is ultimately a matter for careful research. Hence our exploration is speculative rather than confirmatory. We use Hofstede’s ([1980] 2001) concept of ‘collectivist culture’ as the overarching framework for examining selected characteristics of Venezuelan adolescents’ social and cognitive experiences. 4 In collectivist cultures, the individual is relatively tightly embedded in extended in-groups that provide support and control. By contrast, in ‘individualist cultures’ – found in Northern Europe and Anglo America – people are most closely wedded to their own interests and to those of their immediate family (see, also, Hui and Triandis, 1986). Collectivism and individualism have implications for social institutions such as family, community and friends, and also for cognitive frameworks that govern perceptions of time and, possibly, of the self.
Family
Table 2 provides the outlines of the ISRD-3 questions about family. They focus almost entirely on the nuclear family and – in an empirical exploration of Hirschi’s (1969) theory of delinquency – its functioning as an object of emotional bonding and source of social control. Question 1.6 acknowledges the possibility of alternative arrangements for bringing up children, but the subsequent questions on family refer only to parents. Two characteristics of Venezuelan housing arrangements mean that the focus on parents misses some additional important sources for attracting adolescents’ affection and exerting some control over them. The first is the quite high prevalence of extended families living under the same roof, which may include one or both grandparents, parents’ siblings, or more distant relations. The most recent census (2011) reports that 43.5 percent of Venezuelan households comprised extended families and a further 3.8 percent included non-family members as well (INE, 2012). Internationally, the percentage of children who are probably living with extended families is higher in Latin America (for example, 45 percent in Mexico, 55 percent in Colombia) than it is in Western Europe (for example, 17 percent in France,) and Anglo America (for example, 29 percent in the USA) (Child Trends, 2014). Additionally, the prevalence of extended families is rising in many Latin American countries, possibly because of the increasing economic challenges to establishing an independent residence (Jelin and Díaz-Muñoz, 2003). Within these extended families, responsibilities for bringing up children may be shared.
ISRD-3: Family.
Researchers and other commentators on the Venezuelan family point to its matrifocal character, which places the mother at the centre of a network of consanguineous females who cooperate in many domestic tasks, including child-rearing. In its weaker form, this narrative posits a division between gender roles, with male partners taking charge of external representation, including part or all of the income generation, and sons allowed much more time outside the house than daughters. It also identifies the grandmother as a key source of authority and control over the children (De Lima and Sánchez, 2008; Hurtado, 2003). In its stronger form, the narrative speaks of male partners consigned to little more than a procreative function, under relatively transitory and unstable relationships that never imply more than a marginal role in domestic life (Moreno, 2002). In these circumstances, the female support network is even more strongly developed as, for example, mothers go out to work while grandmothers look after the children (Hurtado, 2003). This latter family model – ‘Caribbean’ (Bovenkerk and Wolf, 2010; Parrado, 2002) – is thought to be particularly common in low-income Venezuelan households.
Income level may also be associated with a second housing arrangement, involving the clustering of related households in specific neighbourhoods. Consider, for example, the three families studied by Otálora-Montenegro and Mora-Salas (2004) living very close to each other in a low-income neighbourhood of Caracas:
Family One: male (29), female (36) – consensual union; three sons (12, 10, 8), a fourth son is the wife’s only (15).
Family Two: female (35, sister of the male in Family One); eight daughters (ages from 2 to 19), two sons (18, 5).
Family Three: male (42, brother of the female in Family One, father of the six oldest children in Family 2), female (37) – married; one daughter (16), two sons (15, 10).
In what Otálora-Montenegro and Mora-Salas call this ‘modified extended family’, the organization of child-rearing was not studied but could have involved one or a combination of intra- and inter-household arrangements. For example, apart from receiving maternal care, some of the younger children in Family Two may have been significantly nurtured or supervised by their older siblings, their father from Family Three, or their uncle and aunt in Family One. Moreover, any such arrangements might have varied quite frequently over time as individual circumstances changed. This example may look exceptional because of the close physical proximity of the three households, but many family sub-groups in Venezuela are co-located in the same city, if not the same neighbourhood, with the ability to establish and maintain networks of support for healthcare, childcare, meals and recreation.
The importance of family in everyone’s lives is corroborated by testimony from the three Caracas households (Otálora-Montenegro and Mora-Salas, 2004: 83–8) and by the results of an international comparative project on young people in 12 countries in which Venezuelan respondents were more likely than any other nationality to name the family as their primary source of help (Felce-DiPaula, 2000). These dense networks of interaction and exchange are facilitated, but not determined, by physical proximity and they underline the difficulty of treating a single household or the nuclear household as the exclusive domain for raising children and young people. As De Lima and Sánchez (2008: 14) observe, ‘families inhabit an enlarged space, where the house of the oldest adult, the house of each descendent, the patios, neighbouring streets and the neighbourhood are organically linked together’. This pattern of organization, which is considered a strength rather than a weakness of the Venezuelan family (Otálora-Montenegro and Mora-Salas, 2004), particularly in the face of economic adversity, is not easily captured by static census or survey categories and requires specialized study. 5
And – returning to Item 2.2 of the ISRD-3 instrument – it also raises doubts about the extent to which eating an evening meal with parents is a valid indicator of family bonding in cities such as Caracas.
Peers
Table 3 shows a sample of the items from the ISRD-3 questionnaire that ask about leisure activities and peers. As reflected in Item 5.3, the key distinction here is between family and friends, with an assumption that the only peers an adolescent will have are friends. The questions then move on to explore the level of bonding with friends and the extent to which the latter engage in delinquent behaviour. In the context of relatively large households and extended family networks, these items fail to capture the contact and shared activities that take place with siblings, cousins and other family members who are close in age to the respondent. Although definitely not friends, because in Venezuelan culture the distinction between family and friends is always recognized, these family members may represent a peer group that is as stable and at least as clearly identified by its affiliates as any other type of youth group. Family-related peers may spend a lot of time doing things together or ‘just hanging out’ and some of them may, too, have stolen from a shop or beaten someone up. Not asking about this group of peers therefore misses a potentially significant element in the lives of most Venezuelan adolescents. A recent survey of Venezuelans aged 15–29 found that 93 percent of respondents reported spending at least some of their free time with family and 79 percent reported going out with friends (Freitez et al., 2014). The extent to which ‘family’ included peers and going out with friends also included going out (time spent) with family are unknowns.
ISRD-3: Leisure and peers.
The previous 12 months
The ISRD-3 instrument probes offending by asking whether the respondent has ever engaged in a set of delinquent activities (lifetime prevalence) and, if so, how often in the previous 12 months (12-month incidence). Of these measures, the 12-month incidence (which can also be converted to a measure of 12-month prevalence) is the most relevant to theory testing. This is because almost all of the other items in the survey refer to the ‘present’ (that is, the moment that respondents give answers). Thus, it is hoped that comparison of the 12-month prevalence of delinquency with these current measures relating to family, school, leisure and attitudes will not lose much validity because of the lack of exact synchronicity between the corresponding reference periods.
The self-report method, of course, relies on the willingness and ability of respondents to recall and report the relevant events. When researchers have focused on recall ability, they have noted that this cannot be considered to be perfect or even to be invariant. The use of retrieval cues (such as a calendar) and shorter reference periods (such as six months or one month) can increase the number of incidents recalled and help to avoid their erroneous placement (‘telescoping’), whether within or outside the reference period (Junger-Tas and Haen-Marshall, 1999; Lynch and Addington, 2010). Although the aspiration of developing a ‘gold-standard’ survey drives methodological enquiry slowly forwards in industrialized research-intensive countries, international self-report surveys raise an additional consideration regarding the extent to which people in less industrialized cultures operate with a similar conception of time.
It is generally accepted that one of the distinguishing features of major cultural groups is the way in which time is conceptualized, used and talked about. For example, in a widely read book, Levine (2006) proposed that people in Anglo America, Western Europe and East Asia think in terms of clock time, originally reflected and certainly reinforced by the invention of mechanical timekeeping. By contrast, people in Latin America and Asia think in terms of event time, which presumably divides temporal experience into irregular units defined by the beginning and end of socially significant bundles of activity such as a task (for example, production, repair, harvesting) or a gathering (for example, party, visit, funeral wake). As Levine (2006: 137) reported it: ‘Many Brazilians we tested were completely out of touch with clock time.’
A similar distinction has been proposed between collectivist cultures that are thought to operate with a ‘polychronic’ conception of time and individualist cultures that work with a monochronic conception (Fulmer et al., 2014). In ‘polychronic cultures’, multiple and simultaneous social interactions and obligations lead to engagement in several activities at once, which presumably reflect or reinforce a nonlinear conception of time. In ‘monochronic cultures’, time is conceived in linear terms and is built through sequences of single activities, undertaken one after another (Hall, 1983; Nonis et al., 2005). Western Europe and Anglo America are considered to be examples of monochronic individualist cultures, whereas South Asia, the Middle East and Latin America are frequently cited as examples of polychronic collectivist cultures. Finally, linguistic research has established that some languages (such as English and Indonesian) express time in linear terms whereas others (such as Spanish and Greek) use volume or quantity (for example, mucho tiempo [much time]) (Casasanto et al., 2004).
Although these lines of enquiry are still at an incipient stage, they are indicative of potentially important cultural differences in conceptions of time that may have significant implications for patterns of recall. As with most self-report surveys, the ISRD-3 uses a linear conception of time, asking respondents to indicate any offending behaviours committed during the previous 12 months. If Venezuelan adolescents live in a polychronic collectivist culture, their ability to place events accurately within the past 12 months may be lower than that of adolescents in the monochronic individualist culture of Western Europe and Anglo America. 6 In the absence of systematic research, any effects of polychronic or event time on the amount and type of telescoping of delinquent behaviour can be only speculative, but cannot be dismissed. We also note that in a polychronic culture the notion of ‘free time’ – one of the focuses of ISRD questions – may have less resonance for respondents than it does in monochronic cultures.
‘Who am I?’ 7
Table 4 shows some of the items from the self-control scale used in ISRD-3, which is a shortened version of Grasmick et al.’s (1993) original tool. Respondents are asked to describe themselves by selecting an option from the Likert range that accompanies each item. The task appears straightforward and, of course, Grasmick et al.’s full or shortened scale has been validated in numerous empirical tests conducted in several countries (Tittle et al., 2003).
ISRD-3: Self-control.
However, cultural psychologists have proposed that conceptions of the self and the sources of information used to construct it differ between collectivist and individualist cultures. For example, Triandis (1989) proposed a distinction between a private self (for example, ‘I act on the spur of the moment’), a public self (for example, ‘People think that I act on the spur of the moment’) and a collective self (for example, ‘My family thinks that I act on the spur of the moment’). Other things being equal, individuals in a collectivist culture may more frequently ‘sample’ their collective self. In a related manner, Markus and Kitayama (1991) distinguished between an independent and interdependent conception of the self. The former, thought to be prevalent in ‘Western’ cultures, is conceived in terms of traits and stable propensities to behave, whereas the latter, thought to be characteristic of Asia, Africa, Latin America and many countries of Southern Europe, is conceived in terms of interpersonal relationships and situational contingencies. Although individuals in all cultures can rate themselves on traits (such as impulsivity, risk seeking), ‘traits are expected to play a more important role in self-concepts, person descriptions, causal attributions, and predictions of behavior in individualistic cultures and individuals as compared to collectivistic cultures and individuals’ (Church, 2000: 670). If this is the case, in collectivist countries such as Venezuela trait-like measures of self-control would presumably show a weaker relationship to delinquent behaviour than behavioural measures of self-control (including behavioural measures captured through survey items). It should be noted, however, that psychologists disagree about the empirical status and significance of broad conceptual categories such as ‘collectivist’ and ‘individualist’ and much work is being conducted to explore cultural variation in conceptions of the self and their relationship to behaviour (for example, Cheung et al., 2011; Church, 2000; Watkins et al., 1998). As such, the implications of this research for survey design are tentative rather than definitive.
Addressing hidden differences
That our discussion of hidden differences is speculative rather than confirmatory reveals the incipient and provisional state of knowledge regarding variations in key dimensions of social experience and cognitive processes around the world. These represent substantial lines of future enquiry for which results will always be conditioned by the level of theoretical (dis)agreement, the number of sites studied, the methods used in data collection and the rate of social and individual change. To suggest that international survey projects should wait for the conclusive results from this research would therefore be to condemn them to a permanent delay. 8 Meanwhile, the topics of these lines of cultural enquiry offer important suggestions regarding matters that merit sustained attention in the design and evaluation of current survey instruments. For example, future surveys of self-reported delinquency in Venezuela could develop new questions regarding household/family arrangements and peer group composition to tap into additional sources of family and peer bonding. The reference period for reporting delinquent behaviour could be shortened and anchored in relation to a salient event (for example, the New Year) in order to encourage accurate recall. And items measuring personal dispositions could focus on behavioural indicators instead of self-assessment of traits. The results of these innovations need to be assessed by systematic comparisons with the existing design in order to identify their impact, if any, on the process of measurement.
Conclusion
Cheung, van der Vijver and Leong (2011: 595) argue that ‘[i]f one starts from an imported instrument, emic [culturally specific] aspects of a construct will remain hidden’. In our own participation in the ISRD-3 project, the need to use a standard method at all international sites led us to focus on equivalence in sampling strategy and instrument translation rather than on the more comprehensive and challenging task of achieving cultural equivalence in measurement. Indeed, sampling of a school-based population with a fairly strong presence of middle-income students may have led us to assume more cultural equivalence than a more detailed consideration suggests. Venezuela has been broadly catalogued as an Andean/Caribbean, collectivist and polychronic culture, and these labels may well capture some important characteristics, particularly among lower-income groups. To the extent that they are confirmed, the inclusion of Venezuela in an international project such as the ISRD raises important questions concerning the measurement equivalence of survey instruments in different cultures. The argument is not that Venezuela is a unique case, simply that it points to the need to explore the interaction between measurement and culture both intra-nationally (for example, in relation to migrant and minority groups) and internationally (for example, in relation to the major cultural regions of the world).
‘Hidden differences’ are as much a hypothesis as a firm generalization. Their investigation demands a different approach to instrument design, involving much exploratory research to test alternative models both within and between countries. Thus, the results of any search for cultural differences that affect measurement may reveal none, few or many that are of consequence. Where significant differences exist, international instruments such as the ISRD will need to be redesigned to take account of them. This would require the involvement of multicultural teams (Greenfield, 1997) in instrument design in order to ‘decenter’ (Hofstede, [1980] 2001) the cultural sources of content. And it would also foster more universal approaches to operationalizing key theoretical concepts – even a putatively general concept such as self-control.
Footnotes
Funding
This research was supported by Grant No. C201200930 from the Fondo Nacional de Ciencia y Tecnología, Venezuela.