The Family's Conception of Accountability and Competence: A New Approach to the Conceptualization and Assessment of Family Stress†
This research was supported by Grant RO1-MH26711 to Dr. Reiss from the National Institute of Mental Health. Send reprint requests to Dr. Reiss.
Abstract
Clinicians and researchers have a strong interest in understanding how families respond to stress. Often, they begin their analyses by attempts to estimate the seriousness or magnitude of the stressful events impinging on the families they observe. Until now, they have relied on two strategies. First, they attempt to develop objective or external indicators of the magnitude of the stress of the events. The problem here is that the family's own perceptions and experiences are not properly weighted. The second strategy depends heavily or exclusively on the family's perceptions of the events. However, these perceptions are often a product of the family's efforts to cope with the stress since the organization and perception of meaning in events is a fundamental part of family coping. Thus, this approach cannot disentangle the stress inherent in the events from the family's efforts to cope with it. This article explores a third alternative. The social community in which the family lives often provides a coherent frame of meanings for most events. It not only defines the magnitude of the event but it also defines how accountable the family is for producing the event in the first place. A method for assessing these community frameworks is presented. Initial results suggest that there is not only a coherent community framework attributing magnitude and family accountability to a large number of stressful events impinging the family but, also, that these community attributions are embedded in a community concept of family development.
Family stress is a matter of is great importance to family researchers but is of even more importance to families themselves. Just as researchers do, families can observe the occurrence of stress in themselves as well as in other families. And as researchers do, they can observe their own response to stress; they can observe how families they know respond to stress, and they can form their own conclusions about the consequences for themselves and for other families of those responses. Most importantly, families, like researchers, can develop theories to explain their responses and those of others. It seems likely that these intuitive theories of family stress are part of an amalgam of perceptions, values, theories, and philosophies families develop about themselves and the social world in which they live.
If we consider any community of families (a small town; a neighborhood; a small, homogeneous and enclosed primitive culture), we can distinguish two kinds of these intuitive theories. First, there are theories that are relatively unique to each family in the community, theories that are a product of the family's own development and history. For any family, these are visions, values, world views, and subjective convictions that are, more or less, shared by all its members but not necessarily by all or even most other families in the same community. There have been several well-known attempts to distinguish among families on the basis of the form or quality of these theories. For example, Farber and Jenney (10) distinguished between child-and career-centered families; Kantor and Lehr (18) distinguished the meaning systems of families as open, closed, or random; and our research group has empirically defined a coherent world view, characteristic for each family, which we (20) have labeled the family paradigm.
For some of these informal theories, family members may be fully aware of the principles or perspectives that typify their own family unit; at other times, these informal theories must be inferred by the outside observer because the family cannot describe them in words but enacts them through patterned routines, problem-solving patterns, and rituals (21, 28). We have reported evidence that these efforts by families to understand the social world in which they live include their efforts to understand other families (23). Our data suggested that members of a family do indeed discuss and formulate together the characteristics and motives of other families they know. The data also suggested that these discussions reveal, for each family, aspects of its underlying paradigm about how its social world is constructed.
It seems likely, however, that the concepts that families develop about other families do not remain entirely idiosyncratic. There are many opportunities for families to compare their concepts about family life with other families in their community: informal discussion, reading, television, and through a variety of forms of social participation. We might expect, then, that in any community of interacting families these views of family life achieve some consensus through a continuing process of comparison and reconciliation among families. At this level, a consensus among families is akin to what Goffman (12) called primary frameworks, sets of common social assumptions in the community by which its members readily construe the cause, probable intent, and consequence of any conspicuous social act. In a recent review, drawing on ethnographic accounts of family responses to natural disasters, Boss (5) applies this perspective to understanding family stress along lines that parallel our own.
These conceptions of family life, shared by families in the same community, can be referred to as community-level concepts, in contrast to conceptions that are shaped by the individual circumstances of a particular family, which can be called family-level concepts. The level, then, refers to the subject of the concept not its object. Heider (15) was the first to study systematically these community-level conceptions about motives that laypeople attribute to individuals. He called them “naive psychologies,” although a less pejorative term might be intuitive psychologies. Correspondingly, when families struggle to understand the patterns and motives in other families, they are constructing their own intuitive family sociologies.
In this current article, we focus on one aspect of this community-level, intuitive sociology of family life: how families understand the impact of stress on family life. The article deals sequentially with five logically interrelated components of this process:
- 1
We posit some of the forms these community-level views about family stress may take.
- 2
We show how this approach to understanding family stress improves the precision and clarity of assessing, in any community, which events are stressful for families and which are not.
- 3
We present a simple and straightforward quantitative method for assessing these community-level conceptions of family stress.
- 4
We report results of the first use of this method. These results show a consistency across families in how they understand the impact of life events and circumstances on families. Moreover, these results reveal a consensus among families in their conceptualization of two processes that also lie at the heart of many professional theories about families: family boundaries and family development. Indeed, these results suggest that the families in our sample used their concepts of family boundaries and family development to understand the impact of life events on family process.
- 5
We illustrate how these findings may be useful in understanding the impact of personal and community disasters, how they may help to identify high-risk and alienated families within any specified community, and how they may explain difficulties in establishing relationships between therapist and family.
Intuitive Sociological Conceptions of Family Stress
A variety of historical and anthropological analyses suggest that community-level concepts about family stress may vary along two dimensions: the accountability of a family for producing or exacerbating the stressor situation in the first place, and its competence in protecting itself from disruption in the face of the stressor, whether the family can be held accountable for it or not.
Family Accountability
By accountability we mean how an intuitive sociology conceives of family dynamics or family process. We suggest that families in all communities share some notions about causal sequences of interactional events in families: what kind of family interactions and role relationships lead to what kind of consequences for the family as a whole, for its individual members, and for the community. Perhaps the most universal concerns in this regard are child development and health and illness. Community-level concepts may be distinguished by their attributions concerning the family's influence and power over these processes. For example, for child development, historical analyses are most instructive. Aries, in Centuries of Childhood (4), argues that in Europe during the Middle Ages there was no conception of childhood as an extended and vulnerable phase of human development. Quite the contrary, the child was—at an early age—perceived as a little adult whose development unfolded within the social forces of general public life. During that period, to ask a family whether it or any other family could influence this development was to inquire about not only the impossible but also the inconceivable. In striking contrast, as detailed by Demos (8), in Puritan communities the family was charged with full responsibility for the community's children, not just the social and emotional development but their academic and occupational training as well. Sameroff and Feil (26) have shown that intuitive sociologies may be discriminated on much subtler grounds; for example, parents in traditional Mexican communities in the United States view child development as determined by single, simple causes such as good or bad discipline, whereas higher-SES families, fully acculturated to American life, conceive of a variety of interacting and conditional causes for child development, some originating in the family and some without, which may change across the life span.
Two major features of intuitive sociologies emerge from these reports. First, subject families can readily conceive, at least on an implicit level, of an average or typical family in their culture or community, and develop opinions, views, perspectives, or behavior in accord with this community or culture-based notion of a typical family. Second, these lay conceptions of accountability can shape the family's view of the cause of a broad range of stressful life events or circumstances. For example, if a child was caught stealing in the European Middle Ages, it was unlikely that the family would be held accountable. However, in Puritan New England, the family would be held fully accountable not only for these instances of moral turpitude but for an even broader range of educational and occupational failures as well.
Family Competence
By competence we mean distinctions among community-level conceptions in how the community views the family's competence or hardiness in the face of stress. This includes views of how disruptive a given circumstance should be on family life; that is, the community's estimate of the magnitude of the impact of a stressful circumstance on a typical family reflects its conception of the typical or ordinary family's competence to withstand it. Current data suggest that intuitive sociologies address competence in two closely related ways.
First, they shape an understanding of the seriousness of the event or circumstance for the typical family in the community. Intuitive sociologies recognize that, for the average family in its community, some events are more disruptive than others. Scotch's (27) study of high blood pressure in the Zulus illustrates this process. His analysis sought to understand variations in social circumstances that would lead to variations among Zulus in elevated blood pressure. His data revealed that in traditional, rural Zulu communities the number of children a woman had was unrelated to her blood pressure, but the onset of menopause often produced sharply elevated levels. In striking contrast, in urban communities—into which Zulus had migrated in large numbers—the reverse was true: each new child brought the risk of more elevated blood pressure, but menopause was not a risk factor at all. It was clear to Scotch that the birth of a child in traditional Zulu communities was framed by a traditional intuitive sociology as a blessed event assuring the continuity of clan and the community. In urban settings a child was another mouth to feed in a desperately poor and ravaged community; thus, the urban intuitive sociology conceived of the birth of a child as a major disruptive event for the mother and her family.
Intuitive sociologies also conceive of the expectable hardiness of families in the face of serious stressful events. For example, we (19) have tried to reconstruct the family theories of those families who braved the rigors and isolation in settling the Great Plains in the mid-19th century. Primarily from literary sources, for example, Hamlin Garland (11), Laura Ingalls Wilder (30), and Willa Cather (6), it was possible to recognize a shared belief that individual families could be entirely self-sufficient in the face of almost any conceivable threat. Boss, in her recent review (5), paints a very different picture for Turkish families facing natural disaster. In these communities, a pervasive fatalism construes the families as having little power to resist the enormous disruptions of floods and earthquakes.
Analytic Advantages of Mapping Community Conceptions of Family Stress
Our community-based approach to understanding family stress is designed to fill two gaps in current theory and empirical work in this area. First, it offers the prospect of a theoretically coherent, empirically based definition of family stress. According to this new perspective, family stress is defined as an event that by community standards is high in the capacity to disrupt ordinary family routines. This permits us to avoid a partially tautological definition now current in the field, an approach that—following Hill (16)—defines a stressful event according to the meaning of the event as the family itself sees it. Elsewhere (20), we have pointed out that this more typical approach fails to recognize that the family's experience of an event is better regarded as a central part of its coping response rather than as inherent in the event itself. For example, consider a family in rural Zulu culture. Let us suppose that the family perceives the wife's menopause as an expectable transition and takes comfort in the children that have already been born, rather than being terrified about those that, after this transition, will never be born. This family is actively framing its own meaning of an event in a way that contrasts sharply with the surrounding culture. If, in this family, our definition of stress depended on the family's rather than the community's definition of stress, we would be tempted to say simply that the family is not undergoing much stress. By not taking account of the contrast between the family and community definitions, we would would miss the extraordinary competence of the family's coping effort. In other words, the active reframing is itself a central part of the coping effort.
Second, this new approach offers the opportunity for a rational, empirically based distinction between stress arising from processes internal to the family and that arising from forces external to the family. As many critics have pointed out, it is often meaningless to use stressful life events as independent variables in seeking explanations for the development of psychopathology in individual members or the development of relationship disturbances in families. This is because most life-events scales, and the theories underlying them, cannot clearly distinguish between events that themselves may be indices of psychopathology or of relationship disturbances, and events that may cause these disturbances. For example, several research programs attempt to measure stress on families by determining which events, from a standardized list, have happened to the family recently. Several of these lists contain, for example, the item “child's engagement is broken” This event might be a cause of marital distress in the child's parents or a cause of psychopathology in the mothcr. On the other hand, it might reflect a general level of disturbance and distress in the family. Along with mother's psychopathology or marital distress, it might be only an index rather than a cause of family disruption. Here again, community-based definitions offer considerable conceptual and empirical leverage. Researchers would not have to make arbitrary or intuitive distinctions about “outside” or “inside” circumstances but could rely on those distinctions inherent in the cultural or community definitions of reality.
Hypotheses
These concepts of a intuitive sociology of families, developed by families, generate four testable hypotheses:
- 1
In any relatively homogeneous community, there should be consensus or uniformity concerning the impact of events on the family and the accountability of families for producing those events in the first place.
- 2
The importance of families for thuse concepts should be evidenced in two ways: (a) the community-wide consensus should be preserved when we inquire about events stressful to families and about the competence of families, each considered as a unitary group, to respond to those events; (b) the consensus should also be preserved if we query families, as whole groups, about their views on stress rather than querying individuals one at a time.
- 3
This consensus of families about families should hold both for accountability and competence of the average family.
- 4
If families are serving as informants about the perceptions of their culture or community rather than as expressors of their own views, then their perceptions of events should be relatively uninfluenced by characteristics that distinguish them within their community. Of particular interest are the idiosyncratic philosophies, visions, values, and world views that are a unique product of their own history.
Two other hypotheses are conspicuous derivatives of these considerations but are not tested in this study. First, these concepts or intuitive theories of families' responses to stress should vary in predictable fashion from one defined culture or community to another. Second, the community's own social apparatus for assisting families facing stress should be predictable from its intuitive sociology. For example, we might expect more informal support systems for traditional Zulu women facing menopause and more for urban Zulu women bearing children. Alternatively, if support enters the community from the outside (Federal disaster relief to a rural community or Federal health services for Indian reservations), it will be accepted to the extent that it conforms with the community's intuitive sociology about family needs.
Strategies for Measurement
To anticipate, our method represents a relatively straightforward extension of a method first used by Holmes and Rahe (17) in their studies of stress, and later perfected by Dohrenwend and colleagues (9). As they did, we ask our repondents to assign numerical ratings to a broad variety of stressful events and examine the consensus among them. We regard high levels of consensus among many families as evidence that there is a shared or common community conception of stressful circumstances. Masuda and Holmes (19) recognized early that this approach would be sensitive to community and cultural differences, and they published data showing striking differences between Japanese and American respondents; unfortunately, these illuminating cultural differences were never pursued systematically.
By using this procedure, we can extend our observations beyond straightforward tests of the hypotheses we have listed. We can explore the pattern of the ratings our families make. We can ask whether there are underlying themes or contours in these shared images of competence and accountability of families in the face of stress. Even more interesting will be an examination of events using the dimensions of competence and accountability simultaneously. What events, for example, do the families regard as very disruptive but for which the family is not to be held accountable? Will these ratings reveal some underlying assumptions about how families operate?
Our approach to measurement is different from previous ones in four respects. First, we have developed systematically a set of circumstances most likely to be stressful to the family as a group rather than to individuals. Second, we ask families to provide ratings by responding as a group. Third, we not only ask questions about how disruptive events would be-as did Holmes and Rahe and Dohrenwend et al.—but we also ask how accountable the family is for the event in the first place. Finally, to make the task more vivid, we use a variant of Q-sort methodology. In this initial study we use a convenience sample to explore the feasibility and initial yield of our new approach.
METHODS
Sample
In this first study, we recruited 45 White, two-parent step- and nondivorced families with a minimal marital duration of 7 years from two high-school districts in the suburban ring around Washington DC. High-school Parent-Teacher Associations (PTAs) were used in each district as the basis for recruiting families, but families varied greatly in their level of participation in the PTA or other school-based activities. From our informal observation, the cultural perspectives on family life in this ring are relatively uniform. In general, families are at least second-generation Americans, White, middle class, and their employment is often in the Federal government or in the myriad of firms and businesses that serve the government. To the extent that there is more heterogeneity in this mix than these simple indices suggest, the use of this sample will work against the possibility of confirming our first hypothesis.
As Table 1 shows, this sample is squarely in the middle class, with relatively narrow ranges of income and occupational rating. The subsamples from each of the high schools were similar to each other on all demographic measures, and thus we pooled the two for all analyses. We always worked with a family triad: both parents and the high-school student.
| Variable | Mean | SD |
|---|---|---|
| Father, years of education | 17.4 | 2.9 |
| Mother, years of education | 15.2 | 2.2 |
| Father, income in thousands | 37.6 | 17.8 |
| Mother, income in thousands | 10.9 | 11.1 |
| Father, Duncan SES rating | 73.6 | 14.9 |
| Employed mothers, Duncan SES rating | 67.5 | 15.4 |
| Father, age | 47.7 | 6.4 |
| Mother, age | 45.6 | 6.0 |
| Marital duration, years | 20.7 | 5.9 |
| Father, Shipley-Hartford | 66.4 | 13.1 |
| intelligence (maximum score = 80) | ||
| Mother, Shipley-Hartford | 65.3 | 11.2 |
| intelligence | ||
| Child, Shipley-Hartford | 70.0 | 12.7 |
Procedures
Three steps were required by our measurement strategy. First, we needed to produce a set of events that are perceived as happening to families as groups rather than to individuals. Second, these events had to be described in general terms so that a family could judge how disruptive it might be for a typical family without having to know specific details about the event or about the family to whom it was happening. Third, we had to design a procedure for family triads to work together as unitary groups to rate these events on the dimensions of accountability and competence. In effect, we were making it possible for families to serve as cultural informants in a way somewhat analogous to the field procedures of many anthropologists.
Step 1: Eliciting Events
Each family was invited to our research offices for a structured interview in which we asked them to describe events during the last year that in some way had had an impact on their family. After an initial probe, we systematically explored six separate areas of family life: health, jobs, family activities, school, extended-family life, and the family's neighborhood. The family discussion was recorded and transcribed verbatim. Each family produced many events and, at the conclusion of interviewing, 45 families had provided a total of 469 family events. These events, of course, were very specific: “Aunt Sally was in a car accident”; “Fred dropped out of high school”; “Ferris Junior High School is going to close”
Step 2: Translating Events from the Specific to the General
We, as researchers, had to “translate” these particular events into general terms so that they could be rated by families who did not know Aunt Sally or her relationship to her family. If our overall notions about intuitive sociologies are true, then a translation process of this kind is one that must go on in all cultures. In order for an intuitive sociology or for a community frame to work effectively, there must be common symbols or understandings by which particular or idiosyncratic events are coded into common terms. Goffman (12) used the concept of “keying” to refer to this process. This informal translation process would itself be of great interest in ethnographic investigations. Since an ethnographic investigation of this kind was beyond the scope of this initial study, we followed the next best course: to develop an intuitively based set of translation procedures that would follow specific rules. We hoped these rules would be similar to how ordinary families key or code events. It was also important that, in our work, we verified that these translation rules could be reliably applied by our coding staff.
A major objective of the coding was to reduce the original list of 469 highly specific events to a much smaller set that was phrased in more general terms. The coding procedure had three steps. First, the “translator” scanned the verbatim type-scripts of all 45 original interviews to demarcate descriptions of actual events. Second, whenever possible, the event description was divided into a noun phrase and a verb phrase or their nearest equivalents, that is, who did what and what was done. For example, “Grandmother [noun phrase] went to the hospital [verb phrase]” Third, each phrase was translated into more general terms. We limited almost all of the noun phrases to the following seven categories: family member, mother, father, child, parent, relative, friend, neighbor, and pet. We did this with a heavy heart since it involved blurring such critical distinctions as “grandparent” and “aunt.” However, retaining this distinction would have expanded our list of events well beyond 200. We also tried to make systematic distinctions between short- and long-term events, and to distinguish events that arose from direct or immediate experience from events that were learned about indirectly or in advance (for example, as in the distinction between “child's school closes” and “family learns that child's school may close”). Fourth, all events were phrased in the present tense.
Two coders jointly coded 5 (11%) of the typescripts. An agreement was counted if both derived the identically worded event from the typescript. A disagreement was counted when one coder derived an event that was not matched by the other coder (kappa for each jointly coded typescript = 0.7); 170 items were retained for the next phase of the study. Conceptually, most could be grouped into the same six categories as specified in the original interview; however, 57 items fell into a range of other categories including major life-cycle events, moves, experiences with crime, and several items on abortion and miscarriage.
Step 3: Rating Events
We invited the same 45 families back to serve as judges of the entire set of 170 translated items. Each family triad was asked to judge the items as a group. (In some families the ratings were done after full discussion and careful consensus-building; in others a single member dominated decision making. We did not test systematically the effect on our ratings of these different styles of reaching a family decision.) Each item was typed on a separate card. The order of the cards was the same for each family triad but the items were randomized throughout the card deck without making any effort to preserve the logical categories into which they could be grouped.
Family triads rated the events two ways. First, in order to assess competence, was a rating of the magnitude of events. Here, families rated events according to their capacity to disrupt a typical family. This rating reflects the community-level conception of both the salience of the event being rated for the typical family in the community as well the hardiness of the typical family to withstand disruption from that event. Second, we asked our family triads to rate the typical family's accountability for producing the event in the first place. An event for which the family is not held accountable is called an external event. There were, as a result, two dimensions for the family triad to use in rating: magnitude and externality.
Families were first asked to rate the magnitude of each item by placing the cards on one of nine ruled columns. They were asked to consider families they knew and were given this specific instruction: “How much impact would the event have on the average family?” As we anticipated in our theory of intuitive sociologies, no family seemed to experience any difficulty in conceiving of the “average family” and using this conception as a ready basis for judgment. The first, third, fifth, seventh, and ninth columns were labeled “minimal or none,”“mild,”“moderate,”“substantial,” and “severe,” respectively. Unlike conventional Q-sorts, families could place as many cards in each column as they wanted; in effect, the task was equivalent to rating each item on a 9-point Likert scale. Most families assigned one person the role of the “card placer” who acted on behalf of the whole family.
After each of the cards was placed, the same deck was again presented to the family for ratings of externality with these instructions: “What kind of factors lead to the event's occurrence?” A new set of sorting columns was placed in front of the family with the first, third, fifth, seventh, and ninth columns labeled as follows: “Only factors internal to a family,”“Predominantly factors internal to the family,”“A balance between internal and external factors,”“Predominantly factors external to a family,” and “Only factors external to a family.”
The most critical test of our basic hypothesis is the level of consensus achieved among all family triads on these two ratings. If there were no consensus on all or most of the items, then there would be little basis to believe in a cultural or community frame that shapes families' perceptions of events and their impact. However, as we will show, the achievement of consensus does not, by itself, support the notion of a community framework. For example, it may be an artifact of agreement among families on just the most severe or trivial items. Indeed, in order to confirm our hypothesis, it is necessary to show that there is consensus on items across the full range of magnitude and externality, and that there are well-structured patterns in actual ratings of items that achieve high consensus.
Card-Sort Procedure
In addition to procedures for rating events (steps 1–3), each of the families in the sample was assessed for characteristics that might strongly influence their perception of these events; these assessments served as an initial exploration of hypothesis four. The test of this hypothesis is not, of course, a strong one since it can be confirmed only by the absence of associations between these measures and the ratings.
The Card-Sort Procedure measures three aspects of families' beliefs about the social world. Collectively, these shared beliefs have been called the family's paradigm; a family paradigm is an excellent example of family-level conception about the social world. A detailed explanation of the procedure and its theoretical rationale is available (20). The procedure is administered to family members who are isolated in booths where they can talk to each other over a telephone-like apparatus but cannot see each other. The procedure is divided into two phases. In the first, family members work alone to sort a deck of 16 cards into as many as seven different piles. In the second phase the members receive a somewhat different deck and are encouraged to talk to one another. Many families recognize that the cards can be sorted logically according to the patterns of letters on each. For example, they can see that these two sequences are alike: PMSVK and PMSMSMSMSVK. Other families use more superficial sorting systems.
Three measures are derived from this procedure. Configuration reflects the family's success in identifying the logical patterns in the cards. A broad range of research has demonstrated that this behavior reflects the family's shared view that the world is patterned and masterable. Coordination reflects the level of cooperation and communication during the second phase of the procedure, and, according to past research, reflects the family's belief that they are a single unit rather than separate, loosely federated individuals. Closure reflects the family's flexibility in the face of new information. Do they stick with the same pattern as they examine each card in turn or does the system evolve or change as new information is reviewed? High or delayed closure reflects the family's openness to new experience. If ratings made by the families in this study reflect their own philosophies rather than their role as cultural informants, then we might expect to find significant correlations between these variables and the event ratings. For example, high-configuration families might see a number events as less disruptive and high-coordination families might hold families more accountable for their circumstances. In other research, the Card Sort Procedure has reflected a variety of family views about how families and groups function. The absence of significant correlations in this study would support our idea that when we instruct the family to do so, it does serve as a cultural informant.
In addition, as Table 1 indicates, this relatively homogeneous sample still displays a broad range of variation on a number of social status indicators that are ordinarily correlated with social attitudes. Again, an absence of correlation between these variables and the event ratings would support the argument that family ratings are influenced by community rather than family factors.
RESULTS
Analysis
Each item was scored according to the column number in which it was placed. Thus, an item would receive a magnitude rating of 9 if it were placed under the heading “severe,” or an externality rating of 6 if it were placed in the column between “A balance between internal and external factors” and “Predominantly factors external to a family.”
In order to assess the overall agreement among all families on all items, we used an intraclass correlation coefficient, which is essentially a one-way analysis of variance, using item number as the independent variable, ratings as the dependent variables, and families as “judges” or “cases.” A highly significant coefficient reflects high levels of variance between items in comparison to variance within items, which in turn suggests that families were relatively similar to each other in their ratings.
More revealing is an analysis of those items that are rated with particularly high consensus. Here we assess the families' ratings of magnitude and externality. There are two analytic tasks involved. The first was to determine differences among items in the levels of consensus. This enabled us to focus on items that were rated with moderate or high consensus; these items are, presumably, the best reflection of the community-level intuitive sociology that guided the item ratings. In other words, in order to delineate the content of the community's intuitive sociology, we needed to identify which items achieved reasonably high consensus ratings from our informant families. The second analytic task stemmed from hypothesis four. Here we wanted to know whether salient differences among families in the sample influenced their ratings of items.
In order to accomplish the first task, we needed an estimate of the consensus among families on each item. There is no widely agreed-upon method for this assessment on an item-by-item basis; thus, we adapted the chi-square test for this purpose. We reasoned that if, for any item, there were no agreement among families on the rating of either magnitude or externality, the ratings would tend to distribute themselves equally across all nine categories. (A knowledge of the actual sampling distribution of item ratings must await the study of many communities.) Increasing agreement would be reflected in a non-chance concentration of ratings in just a few of the categories. Chi-squares were computed comparing the actual distribution of items against a distribution in which items were equally distributed across all nine categories. Thus, high chi-squares would reflect substantial agreement among families that the items should be placed in just a small subset of the available categories. There is one exception: bi- or multimodal distributions of ratings. For example, a high chi-square could be obtained if 15 families rated an item in the lowest category, another 15 rated it in the middle category, and yet another 15 rated it in the highest. Thus, we discounted all items with other than unimodal distributions of ratings and regarded them as unreliably rated no matter what the magnitude of chi-square.
A second analysis of items related each to measures of family paradigm as well as selected indicators of family status. Here we used simple bivariate correlations.
For both analyses we applied the Bonferroni procedure with a liberal overall Type-I error rate of 0.2. This liberal level reduces the chances of a Type-II error, a more serious error in an exploratory investigation of this kind. It further reflects our conservative decision to use the entire 170-item set in our computation of the Bonferroni inequality with no effort to subgroup the items into subsets (13); the inequality was computed separately for magnitude and externality ratings. Computations yielded a minimum chi-square of 25 and a minimum bivariate r of 0.48 (for both, p= 0.001).
Findings
Levels of Consensus and Item Analyses
The first three hypotheses, all concerning levels of consensus among families, were tested by the overall intraclass correlations for magnitude (the inverse of competence) and externality (the inverse of accountability), as well as a detailed inspection of consensus among families for each of the items. The intraclass correlation coefficient for the magnitude ratings was 0.51 (p < .0001), and for the externality ratings even higher: 0.61 (p < .00001).
Findings for the item analysis for magnitude ratings by the families showed that of the total of 170 items, 61 failed to achieve unimodal chi-squares above 25. As expected, all of these low consensus items had mean ratings in the mid-range, from 3.1 to 6.5, since the ratings were, by definition, scattered across most of the 9 rating categories. Table 2 shows examples of low-consensus items.
| Item Number | Item Content |
|---|---|
| Items low on consensus for both magnitude and externality | |
| 69. | Married child gets pregnant |
| 103. | Family learns that parent will take an extended business trip |
| 156. | Family member develops short-term illness, condition, or disability |
| Some items low on consensus for magnitude | |
| 2. | Demands of parent's job increase |
| 36. | Family learns that neighborhood schools may close |
| 45. | Property near family's home is sold for development |
| 100. | Subway station opens in family's neighborhood |
| 117. | Relative gets separated or divorced from spouse |
| 161. | Mother enters menopause |
| 170. | Relative gets in trouble with the law |
| Some items low on consensus for externality | |
| 32. | Married child has a baby |
| 50. | Mother has a miscarriage |
| 92. | Family member learns about a member's medical problems |
| 125. | Family member is convicted of a crime |
However, there were a large number of mid-range items on which families showed high levels of consensus. For example, of the 122 items that were rated less than 7 or more than 2, 61 achieved unimodal chi-square scores above 25. Even among the 42 highest consensus items—those achieving unimodal chi-squares of above 50—12 were rated lower than 7 or higher than 2. Of these 42, only 3 were rated 2.5 in magnitude or less. Thus, our family judges showed substantial agreement on the items across the full range of magnitude ratings, but with a dip in consensus in the middle ranges and a substantial increase in consensus at the upper end. It is, of course, impossible to say whether the higher consensus achieved at each extreme—particularly the upper end—of the magnitude ratings was due to the pervasiveness of a community frame or because the high magnitude items were so self-evidently disruptive that any group of families would have agreed on them. But it is clear from the large number of mid-range items that achieve high consensus that our high levels of overall agreement among families, as shown by the high intraclass correlations, is not due simply to high levels of agreement on just a few items that are conspicuously stressful or trivial.
The patterns of item agreements are even more pronounced for externality ratings. Here only 28 items of the 170 failed to achieve unimodal chi-squares of 25 or more. These items were all mid-range items with mean externality ratings of 3.2 to 6.5; 82 items achieved unimodal chi-squares of 50 or greater, 24 of which were rated less than 7 or greater than 2. Of these 82 items, only 8 were rated 2.5 or less, again demonstrating that high-consensus items tend to be mostly the very highly rated ones. Table 2 shows examples of items achieving low consensus on externality ratings.
We examined the bivariate correlations among the three dimensions of family paradigm and the families' magnitude and externality ratings on all the items. Of these 1,020 correlations, only 2 exceeded the Bonferroni threshold. We next examined more traditional indicators of social status for both mothers and fathers: education, income, and Duncan occupational rating. We also correlated the mean score for the family threesome on the Shipley-Hartford intelligence scale, using separate scores for the verbal and abstract portions. Of these 2,720 correlations, only 15 were above threshold. Although seemingly very scattered, there was one pattern to note: of these 15 significant correlations, 10 are for mothers and they are all inverse. In general, mothers who have more education and a better income are located in families that see events as generally of lesser magnitude and the family as more accountable (lower externality) for producing them in the first place. There is no consistent influence of father characteristics on these ratings.
Pattern of Item Ratings: Delineating Intuitive Sociology
The central task of the analysis is to draw a map of the community's concept of family competence and accountability that our family informants have provided. We do so in four steps, as illustrated in Figures 1, 2, 3 and 4. The basic strategy is to construct a two-dimensional map, using the mean of the magnitude ratings across families as one dimension and the mean of the externality ratings as the other (as shown in Figure 1). This creates four broad regions: (a) events perceived as having high impact on the family but for which they are not held accountable; (b) high-impact events for which the family is held accountable; (c) low-impact events for which the family is and (d) is not held accountable. The map can be organized by using the highest consensus items first; then its details can be sketched in using items that achieved somewhat lower levels of consensus, for the most part, but still exceeded Bonferroni thresholds on both magnitude and externality ratings.
Very high-consensus items: charting a community map.
Zones of family calamity.
Conceptions of family boundaries.
Conceptions of family development.
Figure 1 locates the 18 highest consensus items, those that achieved chi-squares of above 50 on both magnitude and externality ratings. We assume these are the most secure indicators of the community's conception of families and stress, and hence may be the basis for defining critical sub-zones within the four broad regions of our map. Indeed, there do appear to be five subzones that are almost naturally defined in this graph: four in the high magnitude regions and one in the low.
Farthest to the left (family accountability is very high for these events) are the events “parents have a baby” and “parents adopt a child.” Hansen and Hill (14) would have called these accession stresses. However, to anticipate subsequent figures, this zone may be called Family Transformation. This term fits the special nature of the events in this zone and suggests the central role, as the community ascribes it, the family plays in these events.
A second subzone is just to the right, implying equal magnitude but less clear family accountability. Here we find the events “family member attempts suicide,”“unmarried child gets pregnant,” and “parent stops working.” It is conceivable that the community recognizes all these events as indicative of intense family turmoil for which they are to be held at least partially accountable; hence this may be identified as the subzone of Family Turmoil. All three of these items are phrased as happening to an individual member of the family. There are two possible mechanisms that may account for why our family raters attributed these events to “factors inside the family.” One possibility is that they saw these events as arising out of a matrix of family patterns. In this instance they would be operating as many family therapists do. A second possibility is that the family may rate the event as due to factors inside the family because the act in the event is committed by a “family member” or an “unmarried child” or a “parent.” All of these labels, on the cards they are sorting, refer to people inside the family. Thus, our raters may have been swayed by these labels rather than by a family-systems perspective.
A third zone contains “family's financial situation deteriorates,”“parent loses job,” and “parent gets job demotion.” As do several items, “family member dies” lies at the border, but because of its qualitative distinctions from the next set we place it here as well. The item “increased conflicts at parent's work” may be too low in magnitude to fit into this subzone. Here, our informants are telling us, are events that are bound to produce serious family crisis, but for which the family cannot be held very accountable. To distinguish the lesser responsibility, we can call this the subzone of Family Crisis.
A fourth zone has four high-magnitude items for which the family is not held accountable: “family home burglarized,”“family member assaulted or robbed,”“parent is transferred by employer,” and “family home damaged by storm, fire, etc.” To emphasize the view that these are completely external afflictions, we can call this the subzone of Family Ravagement.
A fifth zone is identified by four high-consensus items that all cluster in a small area in the right, lower quadrant. Because they are perceived as having so little impact on the family and are clearly regarded as arising from the environment and not the family, we can term this the subzone of Environmental Curiosity.
In Figure 2, we have drawn squares to indicate the approximate regions of these subzones so that they embrace the high-consensus items that seem to define the meaning of the zone. We then added all the remaining high-magnitude items that met the Bonferroni criterion for magnitude and externality. Our definition of the zones seemed confirmed for the most part by these additional items. Thus, two items reflecting major internal changes in the family ended up near the Transformation Zone: “parents get separated” and “emotional climate in the family deteriorates.”
Four serious events happening to individuals ended up near or in the Turmoil Zone: “child has an abortion,”“child drops out of high school,”“child gets divorced,” and “unmarried child has a baby.” These are events over which individuals may be said to have some control and for which they, and perhaps their families, are being held to account. The fact that a fourth item referring to the whole family, “family moves,” is grouped here further suggests that our families see these individual misfortunes as equivalent to the misfortune of the entire family group.
One more item ends up in the Family Crisis zone: “family member develops long-term illness, condition, or disability.” Families rate this as only slightly less in magnitude than “family member dies,” and also hold the family slightly more accountable for the event. Its placement here, along with the placement of other high-consensus items in this zone shown in Figure 1, suggests that our family raters see both major reverses in health and employment in similar ways: they are sure to generate a crisis and yet the family cannot be completely exonerated from responsibility for producing or contributing to these events in the first place.
An examination of the remaining Bonferroni items begins to fill out a picture not only of family calamity but also of the structure and development of the family itself. Figure 3, shows all Bonferroni items in which an event ascribed to a member of the family, “child,” or “parent” is also ascribed to a “relative” or “friend.” We made one exception to the Bonferroni restrictions by adding, for completeness, the item “relative develops long-term illness,” which achieved the Bonferroni threshold for externality but just missed achieving a Bonferroni chi-square for magnitude (χ2= 24, p= 0.002). While, as we have noted, the term “relative” is somewhat ambiguous, all subject families were told that this was someone outside the household other than a child. Thus, it must have been clear to all our families that the items discriminated between nuclear family members and all others.
Two patterns in these items can be observed. First, an item happening to a relative is always considerably to the right and below the comparable event happening to parent or child; that is, it is seen as being much less disruptive and much less under the influence of the family. Many of the “relative” items are in or near the zone of Environmental Curiosity. The one exception to this pattern is the item “relative comes to live with family,” which is not too distant from other severe Turmoil items such as “unmarried child becomes pregnant” and “child drops out of high school,” and even surprisingly close to “family member attempts suicide.” This pattern of ratings strongly suggests a community conception of family boundaries as firmly fixed around the nuclear household: the family has little influence over “relatives” and in many cases is barely disrupted by major events in the lives of relatives. When relatives break this boundary by coming to live with the nuclear family, it is clearly a crisis.
Against this background of perceived firmness of family boundaries, the ratings depicted in Figure 4, take on added interest. Here we have displayed all items that relate to major developmental achievements of children in childhood, adolescence, and young adulthood. These items are organized developmentally in Table 3, which shows that we added, again for completeness, three items that failed to achieve Bonferroni thresholds.
| Developmental Achievements | Developmental Setbacks | ||
|---|---|---|---|
| Item Number | Content | Item Number | Content |
| Childhood | |||
| 12. | Child starts preschool* | 53. | Child suspended from school |
| 165. | Child starts elementary school | ||
| Adolescence | |||
| 89. | Child starts junior high school | 107. | Child does poorly in school work |
| 16. | Child starts high school | 158. | Child drops out of high school |
| 121. | Child begins to date | ||
| 25. | Child starts dating steady boyfriend/girlfriend | 137. | Child breaks up with steady boyfriend/girlfriend* |
| 125. | Child graduates high school* | ||
| Young adulthood | |||
| 62. | Child starts college* | 21. | Child drops out of college |
| 153. | Child graduates from college | ||
| 96. | Child moves away from home | ||
| 95. | Child is engaged | 105. | Child's engagement is broken |
| 59. | Child gets married | 163. | Child gets divorced |
- *Ratings of these items did not exceed Bonferroni thresholds.
Figure 4 shows that these items cluster in two groups. In the top cluster are, primarily, normative and non-normative transitions of considerable importance. The non-normative ones—those associated with some stigma and psychological pain—are at the top of the cluster: “child drops out of high school,”“child gets divorced,”“child drops out of college,” and “child is suspended from school.” The bottom cluster is made up entirely of normative transitions into a new status rather than out of a current status; these are seen as much less problematic for families.
Compare these results to those in Figure 3 and note, in particular, the ratings of items that refer to adult children: “child gets married,”“child gets divorced,”“child gets engaged,” and “child's engagement is broken.” In this late-marrying subculture, these “children” are probably perceived as in their 20s and 30s and, for the most part, as living away from home. However, our family triads place these items as part of a cluster that includes items such as: “child does poorly in school work.” This item, and others like it, must be perceived as a stressful event for a much younger child who is still clearly a household resident. Taking these ratings together, our informants seem to be saying that the tight family boundary, noted in the ratings shown in Figure 3, stretches to included the major transitions of young adults who are living outside the home. However, as soon as these adult children have their own children, their own “parents” may be extruded from the inner or nuclear family. That is, when their children have their own children, “parents” become “grandparents.” As “grandparents” they may suffer the same fate as other “relatives” whose major life transitions end up in or near the zone of Environmental Curiosity from the perspective of the family of procreation.
DISCUSSION
Practical and Clinical Import
These findings do suggest an intricate concept of family competence and accountability that achieves high levels of consensus within the community. These concepts seem embedded within even more sophisticated concepts of family calamities, family boundaries, and family development. These intriguing findings from this one community already suggest practical and clinical uses of data of this kind.
First, they allow us to take a much more accurate reading of the relative role of the family and the community in managing events of great stress. When most members of a community agree that a stressful event is both external and of great magnitude, we might expect the community to establish or support agencies or mechanisms for support of families dealing with this event. Consider, for example, the enormous stress engendered by severe, chronic medical illness in a parent. In the face of the enduring demands of the illness, a broad range of ordinarily trivial events have the potential for major disruption, and such events can feel as if they are beyond the afflicted family's control. For example, the entry of child into elementary school may pose a mixture of practical and emotional problems with which burdened families cannot cope. Their ratings for this event would show it as both very disruptive and very external. However, since this is at variance with the community rating, the family is, we would suspect, left alone to cope without the built-in or institutionalized mechanisms for support. One might regard the difference between community readings of events and those of a beleaguered subset of families as central to the isolation and alienation of these beleaguered families.
In contrast are circumstances where there is a high level of synchrony between the family and the community. For example, consider the reading of events that might occur during or after a natural disaster that has equally affected an entire community. In this instance, we might expect families to be brought together with unusual force and intensity—rather than to be forced into a position of alienation and full self-reliance. Families, then, would become part of a community of unusual solidarity, welded together by intense, vivid, and shared conceptions of the calamities before them. Some histories of the impact of World War II on Britain suggests precisely these transformations. The more routine and distinct readings of events that kept British social classes distinct prior to the outbreak of war became unified in the face of the Blitz.
A second, and related use of tools of this kind helps us to spot the minority or alienated family groups whose world of calamity feels to them unique and isolating. Family self-help groups are born from such a matrix. For example, Darling (7) studied the reading of events that were unique to families of children with spina bifida, a brain and spinal cord congenital abnormality that can be severely disabling. The attractiveness of these groups for families was that other families in the group shared their perception of the magnitude and disruptive potential of even simple events. Further, the group helped its member families to perceive clearly that they could not be held accountable for the fact of spina bifida itself or for the disruptiveness of many events that occurred in its wake. Indeed, these groups developed high levels of consensus that the medical and political systems were to be held accountable for many of the stresses they endured. As a consequence, they became highly effective political action groups.
Finally, as implied by the discussion above, tools of this kind are helpful in understanding the therapeutic process. For example, therapists comprise their own communities with their own beliefs about issues such as family boundaries and development. Almond (2) and Reiss, Costell, and Almond (22) have shown how, for example, psychiatric wards function as communities with their own values and perspectives; these perspectives, partially unique to each unit, tend to shape the technical options woven into therapeutic regimens. Much of the initial tangle between a therapist and family may involve clashes in these fundamental concepts. Reiss and colleagues (24) have shown that family members who do not see themselves as enclosed by firm boundaries—when being treated by a therapeutic team that seems to hold exactly such a conception of most families—are at risk of dropping out of therapy.
Toward Improved Delineation of Intuitive Sociologies
The validity and utility of this procedure depends on two research design decisions: the selection of items for rating and the selection of families to do the rating.
Items Selected for Rating
At least four changes in the items used in this study would be useful for future work. First, the exploration of the intuitive concept of family boundary could be much more thorough if finer-grained distinctions among “relatives” were used. For example, in future work, it would seem useful to single out grandparents from other “relatives,” and even distinguish among paternal and maternal grandparents. These two sets of grandparents may take on very different functions in the life of the family, and those differences may be important and distinctive to our cultural informants. For example, families may be closer to the maternal grandparents because of the particularly strong bonds between adult mothers and their own mothers. The special ties between nuclear families and the maternal extended family have been amply documented (1, 3, 25, 29). Thus, families—as raters—may report that events in the lives of maternal grandparents are likely to have a bigger impact on and seem less external to the families in their communities.
Second, new items might make the procedure more sensitive to perceived gender differences in children and adults. Indeed, mothers and girls may be seen as more central and boys and men more at the periphery of family life. For example, we did have an item “unmarried daughter gets pregnant” (which received very nearly the maximum score on magnitude) but did not have a corresponding item “unmarried son gets his girlfriend pregnant.” Interestingly, the failure of the second item to occur reflects the fact that not a single family mentioned that item in our original structured interviews. This in itself might presage a sizeable gender effect, with many calamities being perceived as more disruptive and more internal when they happen to mothers and daughters, in contrast to fathers and sons.
Third, the community's intuitive concepts of development could be explored more systematically, with a greater range of items covering developmental achievements and setbacks across the life span. These would enable us to examine more closely how development is conceived in the intuitive family sociologies we are tracing.
Fourth, we might explore the difference between rare and common events. For example, might rare events—particularly if they are unexpected—be seen as more disruptive and external?
Families Selected as Raters
Obviously, the selection of particular groups of subjects as raters has a big impact on the ratings. For example, strict comparisons among cultures or among communities—clearly a next step for work of this kind—would require systematic, random sampling of larger samples of families. This has already been achieved by Dohrenwend et al. (9) for individual informants. Also, we have only assumed that our ratings take on a special coloration because we have family triads working as a group to do the ratings. We have assumed we are, thereby, tapping into an intuitive sociology constructed by families in order to regulate—on a community level—the relationships among families. But surely these conceptions held by families shade into systems more developed by individual family members. This contrast could be explored by systematic comparisons among different sets of raters: families versus individuals or, among individuals, parents versus adolescents or mothers versus fathers. Such an examination would provide an alternative, and complementary, perspective on the developmental and gender issues that are addressed in part by ratings of selected classes of items. It may also provide a means of studying boundaries within the family, by identifying family subgroups that are defined by patterns of similarity and dissimilarity in judgments of the meaning of salient events in the life of the family.
