Video observations of sensitivity in context: integrating insights from seven cultural communities

ABSTRACT

This integrative discussion of the special issue on video observations of sensitive caregiving in different cultural communities provides a reflection on the seven empirical studies that are at the heart of this special issue. The two main aims of this special issue are highlighted in terms of their overall conclusions: (1) video observations can be useful and reliable tools to assess sensitivity in non-Western cultural contexts; (2) caregiver sensitivity can be observed across very different cultural contexts and can be expressed in various culture-specific ways. Furthermore, the limitations of the set of studies as well as opportunities for future research are discussed.

KEYWORDS:

• Maternal sensitivity
• cross-cultural
• video observation

View correction statement:

Correction

The two main aims of this special issue on sensitivity in cultural context were (1) to provide insight into the feasibility of videotaping parents and children in different cultural contexts for the study of sensitive caregiving in early childhood, identifying both obstacles and potential ways to overcome these; (2) to enhance our understanding of the occurrence, nature and role of caregiver sensitive responsiveness to young children in non-Western cultural. Seven empirical studies using video observations of sensitivity in seven different countries and cultural contexts contributed unique information toward these aims. In my role as supervisor of each of these seven studies, training coauthors to code sensitivity, and coding many hours of video myself, I will reflect not only on the results of these studies as presented in the papers, but also on the general experience of collaborative observation studies in such divergent contexts. In addition, I will discuss some of the issues raised in the insightful commentaries by Ross Thompson (this issue), and Klaus and Karin Grossmann (this issue) throughout this contribution.

Using video to assess sensitivity across cultures

All seven studies confirmed the feasibility of using video observations to assess caregiver sensitivity in diverse cultural contexts. Although there were some exceptions, in most studies there was very little evidence that participants were more uncomfortable with the camera than is typical of Western samples. In fact, I noticed that the more ethnographic approach of following families for several hours without any instructions regarding their activities—as in the studies in rural Peru and rural Kenya—yielded the most naturalistic video observations in which attention for the camera was almost absent and people were clearly just going about their business. This impression was confirmed by the authors who were familiar with these specific communities. The mean levels of sensitivity observe in these two communities with similar observation procedures were quite different, and yielded both within- and between-group variations. The richness of this type of video data, showing the infants in different situations with different caregivers, is immense and probably comes close to some of the intensive observation work that Mary Ainsworth conducted both in Uganda (Ainsworth, 1967) and in Baltimore (Ainsworth, Bell, & Stayton, 1974).

Although even very short observations (even those of just a few minutes) have been shown to also tap into individual variations in sensitivity that are consistent with those found in longer observations (e.g. Joosen, Mesman, Bakermans-Kranenburg, & van IJzendoorn, 2012), longer observations might yield more ecologically valid and reliable data simply because they can capture the variety of interactions across multiple settings that might evoke different levels of sensitivity. One of the Gusii cases in Kenya clearly illustrates this point. If we had only observed this one mother in a free play situation with her child for maybe 15 min, we would have rated her as highly sensitive, but following the pair for several hours made it clear that as soon as customers arrived at her home-based shop, she would abruptly abandon the child and leave it crying in a pen until she was finished with the customer. These recurring sudden and complete breaks in interaction were clearly upsetting to the child whose needs were intermittently met with great care and ignored completely (Mesman et al., this issue). Similarly, if we had chosen to just observe the Gusii mothers bathing their children, most of them would have received very low sensitivity scores. Apparently bathing is conducted more like a chore in this community rather than an opportunity for social interaction, whereas other routine mother-child interactions (such as feeding, but also general “being together”) were much more sensitive and reciprocal in nature.

Of the studies employing shorter and more structured observations, the one conducted in Iran encountered the lowest levels of camera awareness. The studies in Indonesia, Brazil, and South Africa revealed more camera awareness by subgroups of mothers, as evidenced not only by looking at the camera frequently but also expressing insecurity about what to do or trying to make their children “perform” in front of the camera. However, I have also frequently observed these behaviors in short observations in Western samples, such as families in the Netherlands, the USA, or the UK. In typical “free play” sessions of 5–10 min, commenting on the camera and being filmed are not uncommon, and mothers and father often try to elicit certain behaviors (especially smiling in infants) for the benefit of the camera. However, to my knowledge these behaviors have never been systematically coded in these contexts, and direct comparisons of the frequency of such behaviors in Western versus non-Western countries can not be made. It might be worth including assessments of camera-related behavior in all studies using video observations to allow for clarifications of the influence of this methodology on participant behaviors.

Anecdotal observations in the South African study, for example, suggest that young mothers in particular seemed anxious in front of the camera (Dawson et al., this issue). Racial and educational divides between observer and participant were also felt to be potential sources of camera shyness. In the Indonesian sample, camera awareness seemed to also be expressed in the way that mothers dressed themselves and their children. Some dyads were clearly dressed in their nicest clothes, with mothers also wearing far more makeup than seemed typical for the area. However, in other Indonesian families, mothers and children were very casually dressed and videos included, for example, fathers wearing only shorts taking a nap at the edge of the scene, suggesting little concern about appearances. We do not yet know what distinguishes the first from the second type of family, but this is worth investigating because it might be very relevant to the validity of observations in certain subgroups of families.

The study revealing the most clear influence of the camera was the one conducted in the Yemeni slum areas. The contrast between the video equipment and the living circumstances was by far the largest out of all of the studies in this special issue. The homes of the participants were generally no more than bare rooms with cracked concrete walls and floors and broken windows and a mat or mattress to sit on. The families generally had no furniture and very few possessions on display and no evidence of electricity. This was in contrast to the slum area in Indonesia where all families had furniture, toys, and electrical equipment, including big televisions. Even in rural Kenya and Peru living circumstances were generally characterized by more amenities and presence of electrical equipment and furniture than what was observed in Yemen. Thus, both the visitor (Yemeni, but from a very different social background) and the equipment appeared to be very “alien” to the mothers and children. As noted in the paper by Alsarhi et al. (this issue), the entire observation situation was unnaturalistic because mothers and children rarely spend time together inside during the day, as children are almost always outside without much adult supervision. Because of the camera, mothers were fully veiled whereas they would normally not be veiled in their homes without male visitors. Interestingly, however, both the native Yemeni coder and myself had no significant trouble coding sensitivity, and intercoder reliability was obtained with relative ease. Furthermore, sensitivity scores in this sample were significantly ad meaningfully related to other variables (social support and education), suggesting that the observations did tap into some ecologically valid individual differences in parenting. Nevertheless, investing in alternative settings for video observations in communities like these would be a worthwhile endeavor to increase validity and decrease participant discomfort.

So what can we conclude about the influence of camera awareness on sensitivity ratings? In Peru, Kenya, and South Africa camera-related behavior was not rated using the scales, but based on more general impressions. In Iran camera-related behavior was too rare to use those variables in analyses (Asanjarani et al., this issue). But in the three studies in which associations between camera-related behavior and sensitivity could be examined (Indonesia, Yemen, and Brazil), they were unrelated. This is reassuring, because the presence of the camera itself then does not seem to affect the expression of the key parenting variable in this set of studies. In a way this was to be expected based on the notion that sensitivity is thought to be expressed largely at an unconscious and “intuitive” level rather than at a conscious level. Camera awareness might be more likely to correlate with other parenting dimensions such as teaching behavior and discipline. Ideally, assessments of camera awareness would not only be applied to parents (and other caregivers) but also to children. In this set of studies, we did not code camera-related behavior of the children, but there were certainly cases in which the children (including siblings of the target children) appeared more interested in the camera than the adults. In most of these cases their behaviors did not necessarily seem to be very “unnatural,” but some feeling of behavioral inhibition was present in certain cases. Especially in the Yemeni sample, a few children almost froze in front of the camera and did nothing much except for stare at it. Nothing this extreme was observed in any of the other studies, but there were certainly children who for a little while would just stare before going back to their activities. Making the rating of camera-related behavior a standard assessment in video observation research would be helpful in furthering our understanding of which family members in which situations appear to be particularly influenced by the presence of a camera in what ways.

Even though the camera is clearly not an entirely neutral “player” in this type of research, the value of video far outweighed the downsides in this set of studies. First, the fact that intercoder reliability between an expert Western coder (myself) and new coders from the countries in question in these studies was established without significant problems means that (1) the Ainsworth sensitivity construct resonates with people from different cultural backgrounds; (2) the scale can be applied to very different cultural contexts; (3) the reliability of the sensitivity scores is ensured; and (4) the analyses and conclusions are replicable. Furthermore, the fact that the project teams could go over the video materials multiple times was a clear advantage over a single observer making notes. During the coding processes in each of these studies, team discussions about the video materials often led to new ideas about important patterns of behavior (like maternal multitasking in rural Peru). All videos could then be watched again to identify such patterns. Without video it would have been impossible to add different variables to the observation protocol post hoc. In fact, we are still in the process of noticing new aspects of the videotaped interactions that give rise to new ways of looking at the videos that might lead to new insights not just limited to the sensitivity construct. Furthermore, the discussion and re-viewing of video materials was extremely helpful in noticing more subtle aspects of caregiver–child interactions that tend to be particularly salient for coding sensitive responsiveness in non-Western communities (Mesman et al., 2017).

Sensitivity in cultural context

The second aim of this special issue was to enhance our understanding of the occurrence, nature and role of caregiver sensitive responsiveness to young children in non-Western cultural communities. As also noted by Thompson (this issue), the approach of the current set of studies was one in which a specific concept taken from one cultural setting is being tested in other settings to examine whether it can applied to other cultural contexts and whether culture-specific and potentially culture-general aspects of that concept can be identified. This is also known as the etic approach to studying behavior in cultural context and is contrasted with the emic approach that takes each culture as its own starting point (Harris, 1976). The risk of the etic approach is that the use of a predefined concept can lead to tunnel vision and insufficient attention to deviations from the expected behavioral patterns. The advantage is that it guides explorations so that more culture-general patterns can be identified without getting side-tracked by a potential myriad of culture-specific patterns. This special issue employed the etic approach because it explicitly sought to test the applicability of the sensitivity construct to non-Western constructs. The studies did have some emic quality as well, by allowing for culture-specific expressions of sensitive parenting without sticking to predefined behavioral patterns (see also Mesman et al., 2017).

As summarized nicely by Grossmann and Grossmann (this issue), the seven empirical studies confirmed that sensitive responsiveness can be observed in each of the very different cultural contexts represented in this special issue. Nevertheless, scoring patterns varied between studies. Of course differences in observation procedures and sample sizes limit direct comparisons of sensitivity levels between cultural contexts, but a qualitative reflection on the apparent between-context differences is possible. Mean levels of sensitivity were highest in rural Peru with most mothers evaluated as (highly) sensitive. This was interpreted as being due to access to a network of caregivers, flexible caregiving routines, and habits of multitasking (Fourment et al., this issue). These factors in a context of low socioeconomic resources represent a very interesting new avenue of research into contextual risk and resilience in relation to sensitivity. Sensitivity scores were lowest in the Yemeni slums, which also represent by far the most economically deprived context out of the seven communities in this special issue. Next were the rural Gusii and the mothers in Indonesian slums, where sensitivity scores were on average just in the good-enough range. Potentially harsh living circumstances were evident in both communities, but significant within-community variations showed that there are likely to be many other potential factors influencing sensitivity other than the broader socioeconomic context. Indeed, maternal history of childhood maltreatment (Indonesia: Rahma et al., this issue) and access to a network of caregivers (Gusii: Mesman et al., this issue) appeared to be relevant variables explaining variations in sensitive responsiveness. Sensitivity scores for mothers in Brazil, South Africa, and Iran were on average in the adequate range. Results in these studies also revealed significant within-sample variations that could be partly explained by sociodemographic variables in Brazil (Ribeiro-Accioly, this issue), and maternal reflective functioning in South Africa (Dawson et al., this issue).

Regarding the nature of sensitivity in the seven communities, different expressions have been observed. In the samples from urban Brazil, urban Indonesia, and urban Iran, sensitivity was expressed in ways rather similar to those commonly observed in Western samples. Verbal exchanges were common, as were face-to-face playful interactions, and these were often the ways in which sensitivity was expressed. Indeed, even though there were clear culture-specific elements to some of these interactions, I found the general experience of coding sensitivity in these videos rather similar to what I was used to in studies with Western families. This was quite different in the observations of rural Kenya, rural Peru, rural Iran, and rural/urban Yemen. Here, manifestations of sensitivity were far less verbal (quite a few mothers hardly ever spoke directly to their children), and more physical in nature, similar to what I have previously described regarding rural communities in the Philippines, Congo, and Mali (Mesman et al., 2017). Sensitivity was often subtle, and part of routine caregiving interactions or chores that could easily be missed as being important vehicles for sensitive responsiveness.

In urban South Africa, the pattern was mixed, with some mothers showing a more “Western” pattern with high verbosity and face-to-face play, and others showing very little speech and more physical responsiveness. It may be that these within-group differences relate to the context of their own upbringing. Many Alexandra slum residents have migrated from rural areas, whereas others have been urban residents since childhood. A rural background may relate to the more non-verbal pattern of sensitive interactions in this sample. This notion is consistent with findings in the study with samples from urban and rural Iran (Asanjarani et al., this issue). The two samples did not differ regarding sensitivity levels, but were significantly different in the modes of interaction, with rural mothers showing lower levels of verbal expression and warmth during the observations than urban mothers. Thus, the urban-rural distinction may be particularly important to the manifestations of sensitivity rather than the level of sensitive responsiveness.

Regarding the role of sensitivity in parenting in these different contexts, in general the videos did not give rise to specific re-interpretations of attachment theory’s assertion that sensitivity serves to instill basic trust that a caregiver can function as a secure base from which to explore and a safe haven to return to when in need of comfort or support (Ainsworth, Blehar, Waters, & Wall, 1978). Indeed, the videos included many instances of caregivers monitoring and gently guiding their children’s exploration behaviors, as well as caregiver responsiveness to children’s proximity seeking. These patterns represent two central constructs in attachment theories, and Grossmann and Grossmann (this issue) rightly point out that understanding their distinction is crucial. Of course quality of attachment relationships was not assessed so whether these interaction patterns indeed relate to attachment security cannot be gauged based on these studies. In the studies where simultaneous multiple caregiving was commonly observed (Peru, Kenya), such patterns were not necessarily specific to mothers, and the notion of received sensitivity (regardless of who responds), and its role in potentially a more group-level attachment development represent salient issues for further study and theorizing in line with previous work on these issues (Mesman, Minter, & Angnged, 2016).

It is important to note, however, that the studies did not explicitly address the role of sensitivity in parenting interactions. In his commentary, Ross Thompson (this issue) wonders why the papers do not include more culturally informed interpretations of relevant behaviors by local informants that might have elucidated the place and function of sensitivity in the broader caregiving context. He writes:

But what is missing from these reports is the deeper insight that local informants could potentially provide into how sensitivity is culturally constructed by families in the community. Beyond applying Ainsworth’s measure, for example, how would they have commented on the behavior recorded in the videos in light of local beliefs and practices? What could we have learned from them about why caregivers responded as they did in the contexts in which they were observed? (Thompson, this issue, p. x).

I am certainly very interested in the answers to these questions. However, some of the other instruments included in the studies (not reported here for lack of space) made it very clear that asking locals about the goals or meaning of parenting behaviors is not so effective in several of these communities. In Yemen and Indonesia, questions about the goals of parenting were often met with blank stares, and answers not more informative than “because it is my duty.” Inquiries into their own children’s behaviors were equally unsuccessful. Many mothers seemed to not be in the habit of reflecting on any of these issues and were often unable to provide any concrete answers. Indeed, conscious and verbal reflecting on parenting and child behavior can be seen as a rather Western activity particular to certain socioeconomic strata and “modern” times. I would not be surprised if such questions would have led to equally blank stares and uninformative answers if posed to an early twentieth-century rural American mother without formal education. In fact, even when posed to a highly educated twenty-first-century Dutch mother, her answers would not necessarily reflect the actual underlying motives for or meanings of her (in)sensitive behaviors. As mentioned earlier and in the Introduction to this special issue, the rather unconscious nature of many parenting behaviors that constitute sensitive parenting makes it a rather difficult construct to tap into through parent reports.

A more indirect approach to assess mothers’ parenting ideas and goals is the use of vignettes about daily caregiving situations and asking mothers what they think another mother should do in certain situations and why. We tried this in Yemen and Indonesia and the preliminary analyses of some of these vignettes suggest that this method may indeed yield more useful answers than more direct questions. The question of mothers as informants also reminds me of Mary Ainsworth’s variable “mother’s excellence as an informant” as used in her Uganda study (1967). This variable reflected the amount of detail mothers were able to provide about their children’s characteristics and routines. Interestingly, this variable was more closely related to sensitivity than other variables such as warmth. Such information from mothers was not so easy to elicit in some of the communities represented in this special issue. However, Mary Ainsworth spent much more time with the families in her study than any of our teams did with the samples described here. A more embedded (anthropological) approach with closer relations with the families is likely to lead to more casual conversations that do elicit more relevant information from mothers than a more static interview by a very temporary visitor.

The suggestion of asking locals to review the videos and reflect on the parenting behaviors is interesting and I have actually proposed to do so in a previous study. However, further reflection on this option raises some ethical issues and questions regarding methodological validity. Most of these communities are very tight-knit and in several cases quite isolated, which would mean that any local would know the family in question. Their personal relations of the family could easily interfere with their interpretations, either feeling inhibited to disclose sensitive issues, or providing subjective interpretations based on their feelings for the family rather than actual behaviors. Another option could have been to interview (health care) professionals working in the communities. This could have been an interesting addition to the studies and might have uncovered a little bit more “couleur locale” to the interpretations of the video observations. However, for the majority of the communities represented in his special issue, prejudice and discrimination are very real threats to professionals’ objectivity regarding these families’ parenting skills and practices. Families in slums are often seen as inherently unintelligent and crude, and rural families are seen as backward and superstitious, unfortunately also by those who are meant to support and serve these families. Nevertheless, this is probably not the case for all professionals working in such communities, and in future studies their perspectives would be worthwhile to include.

Could the coauthors who grew up in the countries in question have provided more in-depth interpretations of the video observations? Of course to a certain extent they could and also most certainly did in each of the papers. However, in the majority of studies, the coauthors were from the country in question, but not from the region or specific area in question. The likelihood of finding educated scholars who actually grew up in a slum or a remote rural area is understandably rather slim. So even though the authors were knowledgeable about the language and general country-level customs, norms, and values, they themselves were also at times surprised by what they observed in the videos, as they were not intimately familiar with these subcultures. I imagine this would be much the same if I were to visit a family from a traditional fisherman’s community in the town of Urk in the Netherlands. I am intimately familiar with the Dutch language and culture, but in practice mostly within the limited demographic in which I grew up. Nevertheless, I would be better equipped to interpret some of their behaviors and interactions than someone from—say—Switzerland, because the broader societal and linguistic context in which they occur are within my expertise. The coauthors on the Gusii paper actually did grow up in the region represented in the study, and they did provide their own local interpretations of the parenting observations. These however went beyond the main question about sensitivity central to this special issue and its word count limits, and will be addressed elsewhere.

Even though the “local” coauthors were not always intimately familiar with the communities represented in this special issue, they were invaluable to the success of these projects. As outlined in the Introduction to this special issue (Mesman, this issue), their ability to access these communities, understand their language, and to a certain extent interpret the observations were clear assets to the studies. The entire process of these projects was experienced as win-win in the research teams. As a scholar of cross-cultural questions in parenting, I gained access to a wealth of observations from across the globe that have significantly broadened my understanding of salient issues in my research field. My coauthors got the opportunity to be trained in standardized data collection, observation methods, statistical analyses, and academic writing skills that are generally difficult to come by in countries with limited scholarly resources and traditions. As a scholarly community, we could be more adamant that similar models are adopted when data are collected in non-Western countries, to prevent Western appropriation of knowledge development about non-Western childrearing, and rather stimulate true collaboration and knowledge sharing.

In conclusion

The entire process of several years leading up to the publication of this special issue—from meeting with young scholars from different parts of the world to working closely with them to address some of the most salient scholarly questions on early childhood parenting—has been one of the most rewarding and instructive ones in my scientific career. Each project taught me something new about the nature and expression of sensitivity and its place in daily caregiving in different cultures. Each project has led to a myriad of new questions and new avenues of research to explore. Of course, as also noted by Thompson (this issue), most studies in this special issue are in fact pilot studies that are particularly suited for hypothesis generation but not necessarily for big definitive conclusions. They were also necessarily limited in scope to ensure theoretical coherence and fit within the space constraints. Several studies also had (very) small sample sizes, which precludes certain statistical analyses. However, small samples with more extensive observations can be incredibly valuable for generating new ideas that can subsequently be tested in larger samples. Indeed, the richness of the available video materials has lead to the design of several larger follow-up studies that will allow for several expansions in scope and depth.

The more I reflect on the different approaches to studying a concept like sensitivity in different cultural context, the more I value the more anthropological methods of extensive naturalistic observations to get to know the community better and facilitate interpretations of those observations, while retaining some standardized assessment methods (such as the Ainsworth scale) as theoretical anchors. This is indeed what we have tried to do in the studies in Peru and Kenya, but a more embedded approach including extensive conversations with locals would certainly have had added value and might have deepened some of the interpretations of the video materials. Recognition of the value of approaches from historically separate fields of study and their integration is a particularly powerful way of reaching new insights and bridging gaps in the literature. Although such bridges can seem beyond reach sometimes when differences rather than common ground are emphasized, they are certainly within our grasp if we want them to be.

Disclosure statement

No potential conflict of interest was reported by the author.