Keywords

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

Introduction

Lavee and Dollahite (1991) and more recently Hawley and Geske (2000) and Taylor and Bagdi (2005) have provided evidence that theory is seldom used by either researchers or therapists. The oft cited quote “there is nothing so practical as a good theory” suggests that if we had “good” theories about the family, we would find them useful and practical. Furthermore, the 1990s were marked by critiques from various postpositivist schools of thought about the impossibility of knowledge and the relativity of all knowledge claims (White & Mason, 1999). Vargus (1999) has described family theorists as wandering in the wilderness without leadership. Although this chapter covers much of the material framing these claims, it does not directly confront any one of these. Rather, the purpose of this chapter is to provide a description and assessment of the state of our theoretical knowledge.

This chapter covers a diverse range of theoretical material from the philosophy of science to new theoretical methodologies such as “optimal matching.” A general “road map” might assist as we traverse this enormous and varied intellectual landscape. The chapter is organized into sections introducing the basic notions of the philosophy of social science and social science theory; a brief review of the contemporary literature; an application and demonstration of the way theory provides insights, an evaluation of the methods used to construct theory, and a conclusion. The chapter necessarily begins with some fairly abstract discussions involving aspects of philosophy of science and basic problems confronting all of social science theory. Subsequent sections are increasingly concrete and substantive until the section on methodology returns to relatively abstract material.

There are several goals that we would like to reach in most areas of study including the study of families. Perhaps the two that are most immediately relevant to social science theory are the acquisition of knowledge about families and the development of how and why explanations of family phenomena. Each of these goals is more complicated than might appear and both are certainly areas for dispute.

Knowledge

There is little agreement about the general nature of knowledge. Although Aristotle had argued for three criteria, over intervening centuries, epistemologists have largely taken these and most other criteria apart as the pendulum swung between idealism and realism, empiricism and rationalism. Wittgenstein in his last work (1969) said “Whether a proposition can turn out false after all depends on what I make count as determinants for that proposition” (p. 2e). Indeed, in regard to general public knowledge, there is certainly room to argue about situational constrained knowledge (see Habermas, 1971), partial knowledge, knowledge as practice (pragmatics, see Haack, 1993, 1998), and the connection between understanding and knowing.

With scientific knowledge, these complications may seem both less pervasive and problematic. Scientific knowledge is simply those facts and information acquired by using the scientific method. Certainly, the scientific method of observation, hypothesis formation, and empirical testing would appear to produce scientific knowledge that is less contentious than vernacular knowledge claims. Indeed, we could apply the same methods to any knowledge claim and produce scientific knowledge. However “truth,” one of the oldest criteria for knowledge, would have to be replaced by the tentative nature of scientific claims since scientific knowledge may always be overturned and is regarded as tentative rather than absolute. Probably one of the staunchest proponents of the special status of scientific knowledge claim as less problematic is Popper (1959). Popper argued that ultimately knowledge claims were those that were not falsifiable. The skeptical and tentative nature of science was nicely captured in his idea of falsifiability. Contemporary philosophy of science, however, is less sanguine about seeing falsifiability as an essential criterion for scientific knowledge claims. The reason for this is pointed out by Okasha (2002) who describes the dilemma that may occur when scientists encounter data that is inconsistent with their theory. Popper would have us reject the theory, but scientists are even more likely to try to maintain their theory and somehow accommodate the anomaly. Of course, when finding after finding refutes a theory it must be abandoned, but there is no clear point at which we know that we should abandon rather than refine and extend a theory. Kuhn (1962/1996), of course, has given many examples in the history of science of exactly this problem where there is reluctance to abandon a theory that is frequently at odds with empirical results.

Although scientific knowledge is not easily defined by any one trait such as falsifiability, it does represent a particular form of knowledge. It is not defined, however, by only one set of rigorous methods. Science is diverse and the methods of physics and chemistry touted by Kuhn (1962/1996) clearly are less shared by biology, archaeology, and neurology. These disciplines are not less well developed scientifically as Kuhn argued but are simply different according to their object of study and the constraints these objects pose for researchers. Certainly, the study of human families is different than studying star nebulae, but also different than studying aquatic invertebrates. Hellemans and Bunch (1988) in their history of science give a much more complete picture of the diversity in science and method than Kuhn could ever have acknowledged because it would mean sacrificing his central arguments (see White, 2004).

The one element that consistently unites these diverse methods and scientific disciplines is the community of scientists. Although this has been discussed as the core of science by philosophers such as Peirce (1877), it has been nicely elaborated by the sociologist Merton (1942/1973). Merton argued that the community of scientists are united by core epistemic values (Allchin, 1998) including skepticism, universalism, open communication, and evidence. Allchin (1998) has added “honesty” to this list. So, scientific methods are those methods that conform to these values as interpreted at any given historical period. As a result, scientific knowledge is knowledge indirectly produced by the expression of these epistemic values. It is, therefore, a different form of knowledge than religious knowledge, literary knowledge, or common sense knowledge. In regard to the study of the family, surely we are after scientific knowledge.

Explanation

If we admit that one of the goals for the study of families is scientific knowledge, it might seem a mere corollary that we would seek scientific explanations. At a simple level, explanations are our attempts to answer “how” and “why” questions. But not all sciences necessarily seek to address these questions. As Hellemans and Bunch (1988) indicate, many sciences are more concerned with observation and description. Indeed, one can make the case that good, detailed description is the sine qua non of many sciences such as botany and biology. In the study of the family, however, we also want to know the answer to such questions as “how” and “why” relationships end and how and why we have children. So in relation to the questions family scholars want to be answered, certainly scientific explanation is a goal.

The most well known and popular definition of scientific explanation is the one first proposed by Hempel and Oppenheim (1948). They divide a scientific explanation into two parts: the phenomenon to be explained (explicanadum) and the propositions that deduce this phenomenon as a special instance of a broader law or set of propositions (explanans). This perspective, also known as the covering law model of explanation, posits that explanation is provided when we show that a particular event or dependent variable outcome can be deduced from broad general principles or laws. For example, dropping this book (explicanadum) is deduced as an outcome from the theory of gravity (explanans). Using rational choice theory as an example, we could argue that Bill and Sue got married (explicanadum) because Bill and Sue are optimizing their rewards and minimizing their costs (explanans) relative to any given context and time period.

This perspective has led many scholars such as Homans (1967) to affirm that explanation is deduction from general principles. Subsequent debate and assessment, however, has yielded a somewhat revised picture of scientific explanation. Most notable has been the effect of the problem of symmetry and causality (Okasha, 2002). The problem of symmetry is simply that the specific deduced explanans may often be replaced by the explicanadum. Okasha (2002) uses the example of explaining the shadow of a flag pole (explicanadum) by the general laws of light and angle of the sun (laws) in conjunction with the specific height of the pole (specific explanan). Note that the symmetry problem is that we can just as easily use the length of the shadow as an explanans in conjunction with the laws and predict the height of the flag pole making it the explicanandum.

This problem is partially resolved by moving to criteria of causality. Indeed, there is a significant argument that scientific explanation should be identical to causal explanation where some action of one unit produces an effect for another unit. Although causality assists with the problem of symmetry, it nonetheless raises other problems that the covering law model did not encounter. Most important among these is the problem that the action or agency cannot be directly observed. Indeed, it can be argued that terms such as “force” and “cause” depend on a metaphysical belief that cannot be physically observed. We cannot see cause but can only see the associated action of entity A followed by the subsequent behavior of entity B. Such nonobservability of “cause” and the positing of nonempirical theoretical entities pose problems for scientific empiricists and rationalists alike due to their avowed antipathy to metaphysical and religious explanations of physical phenomenon.

A problem that has been more tied to the social sciences is the confusion of explanation with understanding (e.g., Daly, 2003). This was only subtly addressed by Hempel in that his major concern was the “logic” of explanation and as a result the “pragmatics” of explanation received much less focus. The pragmatics of explanation would be where we ask about the experience and practice of explanation. Our understanding is a consequence but not a cause of explanation. Indeed, some scientific explanations may only be understood by a few scholars but the lack of understanding of these explanations by the vast majority of us does not keep them from being explanations. In the study of the family, the call by some scholars (e.g., Daly) for everyday and common sense understandings may do more to impede our progress than propel our knowledge. Even Hempel (1966) cautioned us in this regard.

Scientific explanation is not aimed at creating a sense of at-homeness or familiarity with the phenomena of nature…What scientific explanation, especially theoretical explanation, aims at is not this intuitive and highly subjective kind of understanding, but an objective kind of insight that is achieved by a systematic unification, by exhibiting the phenomena as manifestations of common underlying structures and processes that conform to specific, testable, basic principles (Hempel, 1966, p. 83).

One particular form of familiar explanation is the “just so” or ex post facto explanation. This form is not really an explanation so much as an interpretation of events. After the event has occurred or the data has been analyzed, we can make up any number of theoretical stories that would fit the data. None of these would have the credibility of prediction and none can be eliminated as possibilities because of the after the fact nature of such claims. Much of social science is plagued by such stories. These ex post facto stories are most in evidence when the individual’s motivation is used to explain behavior. Hempel and Oppenheim warned that “A potential danger of explanation by motives lies in the fact that the method lends itself to the facile construction of ex post facto accounts without predictive force (1948, p. 143).” Indeed, when we ask “Why did Bob and Sally get divorced?”, there are an infinite number of “just so” stories to provide less than credible answers.

In the final analysis, even though Hempel’s portrayal of scientific explanation has received ample critical discussion, it remains as the standard approach to the logic of explanation. Okasha (2002) remarks that even the arguments about causation have met with difficulties. For example, lakes commonly “turn over” in the fall and again in the spring. But it would be inappropriate to say that this is caused by water. Indeed, one could argue that water is just being water at different temperatures and that explains the turn over. The idea of cause as a force outside of the properties of water at different temperatures may be a stretch and clearly temperature is not the cause because only water expands upon freezing. The explanation involves water and temperature, but neither is an exogenous causal force. As such Hempel’s basic notion of explanation remains relevant to today’s researchers and theorists.

Theory

In many ways the definition of theory is simply that which would supply scientific explanation. However, theory is not responsible for all of the statements in scientific explanation, but is responsible for most. We can enlarge upon Rudner’s (1966) definition of a scientific theory as a set of propositions, at least one of which is a law-like statement, and at least one deduced proposition is empirically testable. This definition supposes that even a modest theory should provide us with an empirical regularity (law-like statement) such as “people seek to maximize rewards and minimize costs” or “the probability of a transition out of any family stage is determined as the quadratic of duration in the stage.” Furthermore, the theory must be conjoined with a specific proposition (SP) to deduce an hypothesized outcome (H). For example:

  • P1: People seek rewards and minimize costs.

  • SP2: People in group A have no rewards in situation X t .

  • SP3: People in group A have cost Q in situation X t .

  • H: People in group A will minimize Q in situation X t .

Now we seldom if ever see arguments like the above in our empirical work. In reality, situations are usually more complicated and the example above fails to take into consideration the necessary extensive discussion of what would count as “minimizing Q” for group A. On the other hand, if we tested this hypothesis, we would find that either “H” is false or that we have no reason to reject “H” at this time. Every time we fail to reject similar hypotheses about rewards and costs, the theory gains credibility to some degree.

The example above might inadvertently lead readers to assume that the ultimate goal of theory is prediction but prediction (and historical retrodiction) is just a condition for explanation. After all, some would say, a set of propositions that can’t predict are hardly going to suffice for adequate explanation. Certainly, this is true but many theories have rather humble beginnings. Although we can always assess whether or not a theory adequately explains a phenomena, we can also accept that theories progress and develop over time. Theories may be developed inductively, deductively, or both inductively and deductively (see C.S. Peirce’s abduction, Buchler, 1955). Although there are expectations from theory construction about the clarity of concepts and the production of testable propositions, there exists no magical set of rules about constructing theory. The genetic fallacy is when we judge ideas by where or how they were developed rather than by their logical and empirical adequacy. This has led many philosophers of science such as (Kaplan, 1964) to propose that when ideas are being put together in the context of discovery, it is not appropriate to judge them by where or how they came into being. On the other hand, once a theory is sufficiently well developed to offer knowledge claims and propositions, we should be able to test these in the crucible of the context of justification. This distinction proves useful when examining theory construction approaches such as grounded theory and qualitative methods because these would be firmly in the context of discovery rather than the context of justification.

In the context of discovery, we find that inductive and deductive (and combinations of these two) represent the approaches used to produce propositions. Since there are no special rules that tell us how to construct theory (genetic fallacy), we are only guided by our desire for conceptual clarity and logic in the production of propositions. Although most philosophers of science admit that there are creative inductive leaps where a researcher observes complex reality and is able to see a general process or set of categories that explain the phenomena, there is no way to capture or teach this leap of insight. Despite the perception that deduction is a mechanical operation, deductive theory construction may be just as insightful and creative as inductive approaches. From any large but finite set of existing propositions, there are many possible deductions. Selecting the most productive propositions in a deductive argument is a skill in itself. In reality many researchers work dynamically between induction and deduction to produce theory. Certainly, data requires summarizing (induction) and then those general propositions might link with existing general deductive theory to produce novel propositions. For example, Stets (1992) seemed to go through both of these processes by observing and collecting data on dating couples and relating these observations to the symbolic interaction formulation of role taking. Certainly, Gilgun (2005) recognizes this interplay between deduction and induction in producing theory in her method of Deductive Qualitative Analysis.

In the context of justification, only deduction from the propositions is used. Here we take a set of propositions, derive one or more as necessary consequences (prospective prediction or historical retrodiction) and then assess the falsity or tentative truth of the prediction and by deduction, the theory. If the deduction is true, then that simply means we cannot reject the theory and we have no reason to revise the theory. If, however, the deduction is false, then at least one proposition in the theory is also false and we need to either change the theory or reject it altogether.

Basic Problems in Social Science Theory

Over the past century, many scholars have criticized science in general and the social sciences in particular. Although the following list is not completely inclusive of all of these criticisms, these represent some of the major problem areas. The section relies heavily on the philosophical discussions in Turner and Risjord (2007).

Winch (1958) raised the issue that studying human’s scientifically is neither possible nor desirable. His major point was that a truly nomothetic science of societies was impossible and that all sociology could achieve would be to report the rules people follow. Following Winch, many other scholars such as Habermas (1971) raised questions about similar issues. The prevailing reply by Hempel, Rudner, and others has been that the difference Winch claimed between identifying the rules of rule-governed behavior and nomothetic statements simply doesn’t exist. Being able to identify the rules actors use in fact provides nomothetic statements. Furthermore, most of the life sciences such as biology and botany also deal with goal-directed organic behavior, motive, and social functions.

The problem of using the concept of causation rather than association goes back to Hume. Hume argued that all we observe is association, so the idea of “causation” simply adds a needless metaphysical assumption. Even with the emphasis in contemporary sociology on “social mechanisms” (Hedström & Swedberg, 1998), the idea of cause still surfaces. The basic issue remains that we only have associations and to claim “cause” is to switch to a metaphysical claim of unseen force. Certainly empiricists have been uncomfortable with this notion and rationalists more comfortable (see Turner & Risjord, 2007).

Methodological (or sometimes ontological) individualism vs. holism has been an issue since Durkheim’s argument about the existence of “social facts.” In the study of the family this issue surfaces as concern over levels of analysis and reductionism. If the individual unit of analysis is viewed as the only social reality, we have a form of theoretical reductionism and ontological holism. If the individual level of analysis is seen as the only level that affords meaningful explanation because individuals have purpose and motive, then we have a form of methodological individualism. Methodological individualism has been popularized by theorists such as Weber, Parsons, and Coleman. Some of these arguments are captured in discussions of “normative explanations vs. individual rational choice” and in discussions “structure vs. agency.” Strict methodological individualism is often associated with individual meanings, choices, and action as the prime subject matter. On the other hand, methodological holism assumes that social structure and social facts are at least equally efficacious in producing and constraining behavior. Concepts such as community and culture are associated with more holistic approaches. We will have the opportunity to return to this discussion later in this chapter.

Theoretical, cultural, and ethical relativism have remained issues for the social sciences. Although discussions of ethical relativism are best discussed by ethicists and cultural relativism best treated by anthropologists, theoretical relativism is relevant to all the social sciences. At its most extreme, theoretical relativism argues for the incommensurability of knowledge claims (see Knorr-Cetina, 1999; Kuhn 1962/1996 Longino, 2002). This extreme version argues that all knowledge claims are embedded in particular epistemic cultures and language. There is no possible determination of which of any two competing claims is correct because of the incommensurability of the claims. On the other hand, pragmatists would argue that if you can demonstrate any real difference in consequences from holding one belief over another, then they are not incommensurable, and if you cannot demonstrate any different consequences, then they are the same and no dispute exists (see Haack, 1998).

The issue of reflexivity (e.g., Beck, Giddens, & Lash, 1994; Bourdieu, 2001/2004; Habermas, 1971) in the social sciences has argued that social science and social scientific knowledge interact in the social world so as to change the very knowledge claims initially supplied by the social sciences. Probably no one doubts the reflexive nature of knowledge because that is captured in the aphorism that we learn from our mistakes. More important is the claim that reflexivity in some way makes nomothetic science impossible. Certainly at the most simplistic levels this might raise problems such that studying Goffman (1959) might make one more aware of being “on stage” and might transform some previous behavior. On the other hand, it is doubtful that processes tied to social mechanisms and social institutions will be so easily modified. Furthermore, reflexivity would tend to argue in favor of more dynamic conceptualizations of social phenomena, a trend in social theory that should be lauded (see Tuma & Hannan, 1984).

Basic Concepts in Theory

At an informal level, we all use theory every day. We use theory to formulate expectations (predictions) regarding the behavior of physical objects such as “dropping the glass or dropping the pen” (gravity). We also use theory to predict social behavior such as “she will follow the shorter path” (least costs). This informal use of theory assists us in understanding our world and accurate expectations help to reduce our stress. For example, once I know how to ride an escalator, that knowledge can be applied to riding escalators in Vancouver or Dallas. Knowledge is simply a general proposition that works across diverse contexts.

A theory is the expression of ideas by means of concepts and relations some of which have empirical content so that those empirical components of the theory can be assessed by evidence. Theories thus have the components of propositions, concepts, and relations.

All theoretical propositions (including causal and social mechanism propositions) may be parsed to a “concept tied by a relation to another concept.” A concept is simply a unitary idea or collection (set). Relations are what do most of the work in theory, but we seldom notice or credit the importance of relations in theory. All relations are from one set to another set (domain and range). Imagine we have two simple concepts such as males and females. We can imagine a number of important relations between males and females such as “likes,” “knows,” or “is married to.” Relations have properties distinguishable by three major properties: reflexive-irreflexive, symmetry-asymmetry, transivity-intransivity (see http://www.abstractmath.org/MM/MMRelationsProps.htm). When we state that two ideas, sets, or findings are identical, we are arguing that there are specific properties defining the relation such as reflexive, transitive, and symmetric. For example, when we say that two things are “equal” (a = b), we are stating a relation that is symmetric, reflexive, and transitive (a = b, b = c, a = c). Likewise, more complex statement such as “cause” and “is a relative of” and “is married to” are also characterized by a particular vector of these properties. This may not seem important in regard to some relations such as “equals;” however, theorists need to “unpack” many of their complex statements and analyzing the relations in these statements is critical for meaningful knowledge claims.

When theorists form a knowledge claim, they need to break it into the propositions they wish to assert. In science, we can best use propositions where both concepts are clearly defined and where the relations asserted to be linking the concepts are consciously analyzed as to the properties. The reason for this is that eventually, we would like to have theories that are logically true and empirically sound. The empirical part of this is best handled by research methodologist; however, it is extremely important that theorists provide statements that are logically connected. When we test a theoretical proposition and find that it is tentatively supported, we have little reason to change, improve, or revise our theory. It is only when our findings suggest that the proposition being tested does not hold or is false that we gain new knowledge and revise our thinking. We gain the most when our propositions are logically connected. Take for example a simple logical set of statements.

  • P1: All married people are happy.

  • P2: Judy is a married person.

    Therefore: Judy is happy.

The logical form of this argument is known as modus ponens and is of the form P→Q|P therefore Q. In this form, if P1 is true and P2 is true, the conclusion follows logically. A second form of valid argument is modus tollens. It is of the following form:

  • P1: All married people are happy.

  • P2: Judy is not happy.

    Therefore, Judy is not a married person.

In this argument known as modus tollens (P → Q|∼Q therefore ∼P), we deny the consequent that Judy is happy and therefore deny that the case is included in the major premise. Now imagine that the major premise is actually composed of an entire set of logically connected theoretical propositions regarding marital happiness. When we find that the consequent is not true, it entails that at least one proposition in our theory is not true. This in turn requires revision and, in drastic cases, abandonment of the theory. In other words, the logical connections of a deductive theory maximize the goals of reaching more accurate knowledge. Even though our research hypotheses are always couched in probabilistic terms, if there is a general uniformity across empirical contexts then this would provide a general statement such as found in the major premise. In social science, we have many theoretical propositions that can supply major premises such as the ratio of rewards to costs as a motive (rational choice), anticipatory socialization for a role decreases role strain (role theory), and that stage of the family and duration in that stage determine the probability of an event transition (life course). It is the job of theorists to produce logically coherent sets of such propositions and for researchers to determine the empirical adequacy of particular general propositions (and by logical implication the set from which it was derived).

Review of Literature

There are three dominant ways in which scholars review the theoretical literature. Each of these provides an alternative treatment of similar material. Certainly one of the oldest ways of approaching theoretical literature is to review the work of a single author. For example, in philosophy it is not unusual to see entire classes devoted to the works of Plato or Aristotle or Kant. In sociology, we might see classes devoted to Weber, Durkheim, or Simmel. In the area of family theory, this is less likely to be the case. A second approach is to focus on one issue. For example, we find philosophy courses on theories of knowledge (epistemology) or on theories of being (ontology). In sociology and family studies, we are much less likely to find issue-oriented courses in theory such as theories of cohabitation or theories of mate selection. Finally, scholars might review theoretical material by the schools of thought the material represents. For example, in philosophy there could be a course in analytic philosophy or phenomenological thought. In sociology we might find courses in symbolic interaction or rational choice approaches, and in family studies, we might find ecological, family development, and family resilience approaches.

It is this last approach, schools of thought, that has become almost traditional in reviewing theories about families. Early efforts such as Hill and Hansen (1960), Christensen (1964), Nye and Berardo (1966/1981), and Broderick (1971) used a conceptual frameworks approach to capture the various schools of thought. The Burr, Hill, Nye, and Reiss (1979) two-volume work used schools of thought to structure Volume II (deductive theories), but used issues such as family communication and marital quality to structure Volume I (inductive theories). Holman and Burr (1980) examined only frameworks. The subsequent Sourcebook of Family Theories and Methods (Boss, Doherty, LaRossa, Schumm, & Steinmetz, 1993) largely followed a schools or frameworks approach. Winton (1995), Klein and White (1996) and the White and Klein (2002, 2008), Ingoldsby, Smith, and Miller (2003), Smith, Ingoldsby, Miller, and Hamon (2007), Chibucos, Leite, and Weis (2005) have all used theoretical frameworks. The Bengtson, Acock, Allen, Dilworth-Anderson, and Klein (2005) Sourcebook of Family Theory and Research used the issue approach exclusively rather than the frameworks approach.

In the area of family theory, there are few if any individual authors with a sufficient body of theoretical work that would justify an authors approach. The issue approach was used in Volume I of Burr et al. (1979) and the 2005 Sourcebook (Bengtson et al., 2005). Both of these use similar categories such as family violence and marriage, but only the Burr et al. (1979) actually produced theoretical propositions. Indeed, one of the challenges for issue approaches is to stay focused on producing theoretical generalizations and not get too mired in empirical detail or new methodologies. On the other hand, most of the many framework approaches have produced some exemplary propositions to demonstrate various applications of theories to empirical questions. Indeed, it seems that frameworks approach has been preferred by most scholars in their presentation of family theories.

There are several advantages to the framework approach. The study of the family is an interdisciplinary undertaking. Nursing, home economics, family studies, geography, urban studies, political science, psychiatry, psychology, and sociology are only some of the academic and professional disciplines studying families. As a result of this diversity, we have a certain degree of independence in the production of theories about families. The frameworks approach assists us in unifying these diverse theories under common intellectual assumptions. We might also unify these by issue, but then the intellectual assumptions might be quite diverse. The fact that multiple theories share common intellectual assumptions further helps us in seeing that a few assumptions and propositions might explain across disciplines and issues (parsimony). With an issue approach, just the opposite would be achieved; we would show how one phenomenon might be explained in a myriad of ways. It would seem that the route of intellectual parsimony afforded by the frameworks approach might better fit with the goals of science.

Another advantage of the framework approach is that it allows us to unite efforts and innovations from diverse areas by using a common intellectual lens. For example, while Bronfenbrenner (1979, 1989, 2004) contributed much to ecological theory of the developing individual, there were many aspects of social institutions and social processes left undeveloped. When we see Bronfebrenner’s work as part of the ecological framework including a sociologist such as Hawley and a biologist such as Emlen (1995), it allows us a more fully developed picture of ecological processes and possible theoretical propositions. Furthermore, when the theoretical framework is used to unite empirical research, it assists us in seeing common processes across a range of family issues. So the framework approach tends to maximize the theoretical components rather than the empirical components focused upon with an issue approach.

The principal weakness of the frameworks approach is that it is abstract and the high level generalizations may seem remote to those with a more problems based or empirical orientation. However, it has been said that nothing is so practical as a good theory and the frameworks approach certainly emphasizes the theory over particular issues to be explained. Readers will get far more intellectual capital from a general proposition about motivation than a particular proposition about the motivation behind family violence.

Theoretical Frameworks

A theoretical framework is distinguished by the fact that a relatively parsimonious set of assumptions and general propositions characterize the more particular theories that are included under the aegis of the framework. Although individual theoretical formulations such as resilience theory might desire to claim status as a framework, the aim here is to maintain frameworks where several diverse areas of family behavior are explained rather than only one area such as family stress and coping. The frameworks that are reviewed below are also selected because they include some of the most used specific theoretical formulations in the area of family studies.

Rational Choice and Exchange Framework

The rational choice and exchange framework is unified by the common assumption that individual motivation for profit explains choice. Individuals are motivated to choose those outcomes that maximize profit (rewards/costs ratio). In situations where there are only costs, individuals seek to minimize costs. Exchanges are valued as rewarding according to their profitability. So exchanges needn’t be always profitable, but relative to other available relationships over the same time frame, the ones chosen should offer the greatest profit (or the least costs where there are only costs).

It could easily be argued that this framework really embodies two distinct frameworks. One body of theoretical literature deals with exchange relationships and the major unit of analysis is the relationship (e.g., Cook & Yamaguchi, 1990; Emerson, 1962, 1976; Kelley et al., 1983; Sabatelli, 1988; Sprecher, 2001; Van de Rijt & Macy, 2006). The other body of literature is from the rational choice perspective and is clearly more focused on the individual as actor (e.g., Becker, 1981; Coleman, 1990; Donnelly & Burgess, 2008; Friedman, Hechter, & Kreager, 2008; Teachman, Paasch, & Carver, 1997). There are however some major rationales for continuing to treat these two relatively distinct frames as one unitary framework. Most importantly, rational choice has provided the social mechanism for the formation, continuity, and demise of social exchange relationships (see Amato & Hohmann-Marriott, 2007; Coleman, 1990; Donnelly & Burgess, 2008). Furthermore, some of the most popular concepts such as social capital and networks cross-cut and are shared by both perspectives (Teachman et al., 1997). Indeed it is doubtful that questions about why we form relationships can be answered without the individual agency approach in rational choice and it is also doubtful that we can understand how these choices are made without the constraints of social networks and relationships being added into the equation. Thus, these two approaches are complementary and symbiotic and to split them would be to remove much of the explanatory power in this framework.

This framework clearly traces some of its intellectual heritage to the early Greek hedonists and more recently to the utilitarians such as Mills and Bentham. There is little doubt that much of the recent impetus for rational choice derives from the microeconomic work of Becker (1981) and the sociological work of Coleman (1990). Indeed, these two men were at the center stage of Faculty Seminar on Rational Choice at the University of Chicago during the decade of the 1980s. In the area of the family, Coleman’s (1988) paper on social capital gave rise to its use as a major conceptual tool in family studies (see Coleman, 1988; Teachman et al., 1997). Even though the concept of various capitals originates with Bourdieu (1979/1984), it was in the hands of Coleman that it became integrated into rational choice theory. Coleman also provided sociologists and family scholars with one of the most clear formulations regarding the formation of social norms from individual choice (see Coleman, 1988, 1990). Finally, Coleman (1990) laid the foundation for our notions of social institution as rooted in individual rational choice. Indeed, today’s scholars cannot even be afforded the pretense of being theorists without having a firm acquaintance with these formulations.

Without a doubt, the majority of research using this framework (choice) has focused on marital relationships and divorce. Levinger (1965, 1966, 1982) and Lewis and Spanier (1979) first developed the idea that even high-quality marriages could end in divorce if these marriages were in contexts that provided for high levels of alternative attractions to the marriage and low levels of barriers to separation and divorce. More recently researchers have sought to integrate the idea of marital commitment (Johnson, 1985; Rusbult, 1983) to the marital stability equation (Amato & Hohmann-Marriott, 2007). Furthermore, Sabatelli and Ripoll (2004) have argued that barriers and alternative attractors have changed over time and are not stable. On the other hand, Bodenmann et al. (2006) in a retrospective study of barriers and alternative on three countries, report that barriers and alternatives appear to be perceived as stable. The applicability and import of this particular application of choice and exchange theory suggests that it should receive more attention from theorists. As we shall see in a subsequent section of this paper, there are some areas of this application that require intense theoretical development.

Another area of theoretical application has been in regard to the actor’s choice to invest in certain relationships. As previously noted, this particular area is critical to the theoretical link between choice and exchange relationships. For example, Friedman et al. (2008) argue that grandparents’ differential investment in grandchildren is explained by which children are perceived as most likely to give support to the aging grandparents. Donnelly and Burgess (2008) use costs and rewards to explain why people stay in sexless relationships. Rhatigan and Axsom (2006) use the rational choice investment model to understand battered women’s commitment to and staying in abusive relationships.

The concept of social capital has been popular with family and community researchers. Since the original Coleman (1988) paper, there have been both theoretical discussions and research applications of this concept. Most scholars are well aware of Putnam’s thesis about the decline of family and social relationships (Putnam, 1995). This perspective has not avoided debate (Boggs, 2001). It has also created theoretical interest in the causes of social capital (Brehm & Rahm, 1997). Bubolz (2001) has provided a theoretical application of the concept to the family. Israel and Beaulieu (2001) examined the influence of family and community social capital on educational achievement, while a significant cluster of scholars have examined social capital and its effect on health (see Kawachi, 1999). One particularly intriguing application is the explanation of fertility in the developed world by the motivation to acquire family and community social capital (Astone, Nathanson, Schoen, & Kim, 1999; Schoen, Astone, Kim, Nathanson, & Fields, 1999; Schoen, Kim, Nathanson, Fields, & Astone, 1997). For theorists there certainly continues to be a need for conceptual and theoretical clarification of social capital (see Lin, 2001; Portes, 1998) and its relation to human and cultural capital (Bourdieu, 1979/1984).

Equity theory is a variant of choice and exchange developed by Walster and Walster (1978). Equity theory proposes that equitable relationships not only follow the “norm of reciprocity,” but are inherently more rewarding as exchanges. Despite the attractiveness of this way of thinking, findings have not supported this contention (see Pina & Bengtson, 1993; Sexton & Perlman, 1989; Sprecher, 2001). Before abandoning this perspective, however, some very central conceptual issues need to be resolved. The difference between some objective rather than perceived equity needs to be elaborated (Braun, Lewin-Epstein, Stier, & Baumgartner, 2008). Furthermore, the conceptualization of equity as fairness may be at odds with measures of equality. Finally, the universality of the norm of reciprocity needs to be addressed. It is interesting that despite the findings and the problems of conceptualization, this theoretical variant remains popular with researchers.

The criticisms of the rational choice and exchange framework can be broadly summarized as focusing on the assumption of the stability of rewards and costs, and the boundary conditions. For example, Sabatelli and Ripoll (2004) raise the issue of the instability of rewards and costs over time. The assumption that rewards for social actors are both relatively stable and general across time is questionable. In the 1950s, the rewards attributed to marriage were more institutional (support, children), whereas today they are more interpersonal (companionate). Even within the life course of any age cohort, rewards might change with age and period. A second and more core criticism is about the boundary condition that exchange and choice only predict “rational” actors (White, 2004). One problem tied to this assumption is that any time the theory fails to predict proponents can simply say that the actors were obviously being irrational because the theory did not predict. Such tautological logic makes this theory relatively difficult to falsify. The classic case of this is with altruism where either the altruistic act is irrational or the term “altruism” simply conceals rational self-interest such as the Boy Scout helping the elderly across the street in order to get a merit badge. Indeed, it can be argued that the study of what is commonly viewed as an emotional and intimate social group, families, should not be led by a theory that is limited to rational action alone.

Life Course/Family Development

Aldous (1990) argued that neither life course nor family development approaches were theories and that the major distinction between these two was that life course research tended to be at the individual level of analysis, while family development was more focused on the family group. In stark contrast to Aldous (1990), White (1991) argued that family development theory was so advanced as to be formalized as a theory. It is fairly obvious from the 1993 Sourcebook chapters (Boss et al., 1993) on these two approaches that Bengtson and Allen (1993) as well as Rodgers and White (1993) believed that these approaches provide theoretical propositions. Indeed, in the Boss et al. (1993) Sourcebook, these are the only two chapters that provide such theoretical propositions. Researchers, however, have for the most part ignored the theoretical components of this theory in favor of the similar descriptive approach provided by each.

The theory in both of these approaches is similar so as to be treated as a unified framework. The theory recognizes that there is a complex interaction between individual ontogenetic development and sociogenic sources of development. The theory first seeks to clarify the sources of sociogenic development. The major sources of sociogenic development are age-graded norms and event sequencing norms. An age-graded norm is where a specific event (marriage, pregnancy, etc.) is consistently viewed by a society or group as more appropriate for some ages but not others. A sequencing norm is usually constructed from age-graded norms so that we are expected to sequence certain events prior to others (marriage before first birth). For any particular age cohort in any particular historical period, there are age-graded and sequencing norms that compose the normative life course. Hogan’s (1978) observation that those that followed the normative life course path met with fewer difficulties in life sparked a host of theoretical thinking about the “off time” effects first noted by Neugarten, Moore, and Lowe (1965). Closely related to this line of thinking is concern with the individual and family adjustment to one transition event (transition to adulthood, marriage, first birth, retirement, etc.). As a result, much of the research in this area has focused on the stress and adaptation to such events and has resulted in the rise of “resiliency theory” as a major theoretical variant.

Much of the life course research is correlational and uses the theoretical concepts rather than testing propositions. For example, Bucx, van Wel, Knijn, and Hagendoorn (2008) examine the intergenerational contact with parents over the life course of young adult children. Baxter, Hewitt, and Haynes (2008) studied the distribution and duration of time spent on housework at two stages of family. Following up on the literature on transition to parenthood (see Cowan & Cowan, 1992/2000), Helms-Erikson (2001) examine the quality of the marriage 10 years after the first birth. Macmillan and Copher (2005) studied ethnic variations in timing of transition to parenthood using latent class analysis on the National Longitudinal Study of Youth (1979). The common ground for all of these studies is that the normative, chronographical stages of family development are used to structure the research. There has been little effort to test or use theoretical propositions such as those about life course deviance, cross-institutional norms, or even investigating the determinants of event transitions. Fortunately, some of the macrovariables can be identified from work by demographers on events such as cohabitation and first marriage. However, the factors within one stage that determine the transition to an adjacent stage are seldom the focus.

Not only was Hill (see Hill & Rodgers, 1964) one of the progenitors of family development theory, but the study of family stress and resiliency can be traced back to an earlier work by Hill (1949). Hill’s original ABCX model was adapted and modified as the “double ABCX model” by McCubbin and Patterson (1983) to more fully account for the pile up of stressor events and time. The linkage between life course events and stress was of course evident in the Holmes and Rahe (1967) stress scale (SRRS) and noted by other theorists (Pearlin, 1980; Pearlin & Schooler, 1978) and was moved over to the resiliency model (e.g., Hawley & de Haan, 2004; Patterson, 2002). The resiliency model as it currently stands is a heuristic conceptual model to assist in identifying some of the variables composing the process of resiliency and focusing on stress adaptation as an outcome. Although this has proved useful for therapists (see Walsh, 2003) and researchers (e.g., Grzywacz & Bass, 2003), it has considerable distance to travel before it offers a coherent set of general theoretical propositions.

The major criticisms of this theory have focused on the normative interpretation of events and event histories. Certainly, life course theory argues that age, timing, and sequencing norms exist and are sufficiently strong to produce behavior. Most but not all critics (Marini, 1984) admit that “first comes love then comes marriage…” is a form of sequencing norm and that the age at which one can get a driver’s license is a formalized age-graded norm; however, the idea that deviance from these norms produces effects is more controversial. In part, this controversy is part of a larger debate about the oversocialized conception of actors (Wrong, 1961) and the degree of agency actors might possess. More particular to this theory is the idea that distinguishing the effects of norms from social organization (age-graded schools) and biological constraints (aging, fecundity) is difficult. On the other hand, those studying particular transitions have consistently associated sequencing and timing normative deviance with particular later life events. However, even White (1991) cautions that this effect for normative life course deviance could be a selection effect whereby “deviant” types of people are drawn into “deviant lifestyles.” It would hence be a mistake to identify the consequent (lifestyle or life course deviance) as the cause rather than the effect.

Symbolic Interaction

Undoubtedly, symbolic interaction theory is firmly rooted in American pragmatism of the early 1900 including Peirce, James, Dewey, and Mead. Most theorists would see Mead as the main progenitor of this approach. Today, it is difficult to assess the symbolic interaction framework’s popularity because so many of the basic theoretical concepts are widely accepted as part of the general sociological vernacular. Concepts like role, role strain, role transitions, self, and identity are commonly used in much of family studies and sociology. In addition, symbolic interaction has supplied the launching framework for many smaller theoretical variants such as role theory (Biddle, 1986) and identity theory (Stryker & Burke, 1994/2000).

Symbolic interaction is a general theory that posits that social behavior can only be understood in relation to the symbols and meanings any behavior has for actors within a context. Although symbols are defined by social agreement, there is also room for negotiating meanings. Indeed, the degree to which symbols are stable and structural (Stryker, 1980) rather than negotiated (Turner, 1980) has provided some heated debate. The structural school tends to emphasize that the difference between signs and symbols is that symbols are abstract and receive their meaning by consensus and convention rather than similarity. So the word “apple” bears no resemblance to an actual apple. It receives its meaning because we agree that this sound will stand for this object. As a result of this perspective, the structural school focuses on how these are symbols learned and transferred intergenerationally. On the other hand, any particular social role such as “husband” or “wife” may not be subject to prescriptive definitions and, hence, these roles allow for negotiated “role making” (see Turner). The rapprochement in identity theory (see Stryker & Burke, 1994/2000) is probably indicative that the arguments between the structuralist and interactionist schools will increasingly be viewed as “half full, half empty” linguistic debates rather than substantive theoretical problems.

Certainly Goffman’s work (e.g., 1959, 1967, 1974) has provided a unique perspective, the dramaturgical perspective, even within the framework of symbolic interaction. Most every undergraduate is familiar with his use of “front stage” and “back stage.” Recently, however, there has been a surge of interest in Goffman’s (1967) ideas about emotional energy and interaction rituals. This interest has in part been fueled by the academic activity concerned with emotion (e.g., Scheff, 1994, 1999; Stets, 2005; Stets & Turner, 2006; Turner & Stets, 2005). The most recent theoretical expression of this surge of interest in Goffman’s ideas about emotional energy and interaction is contained in Interaction Ritual Chains (Collins, 1981, 1987, 2004). Collins argues that all social action is local and situational. Out of the bonding of local cells of emotional energy, we might see aggregate phenomena but the phenomena and our understanding of it are solely available at the local, situational level. Shared emotional energy coalesces groups, but is produced through the enactment of individual level interaction rituals. Even sexual intercourse is seen as an interaction ritual producing emotional bonding. Collins (2004) clearly focuses on the microsocial as producing macrosocial events (see Baehr (2005) review and Collins responses www.cjsonline.ca/reviews/interactionritual.html). To date, few if any family theorists have explored Collin’s perspective and, at present, it does not appear as a major influence on researchers. It would not be surprising, however, to see the 3 decades of work on the sociology of emotion (see Thoits, 1989; Turner & Stets, 2005) cascade into family studies to inform a new generation of theorist and researchers.

This last point is even more poignant in the context of critiques of symbolic interaction. The major foci of criticisms has been the vagueness of concepts and the inability to incorporate emotion. Although some of the concepts in symbolic interaction are vague and abstract such as symbol, other conceptual areas such as role theory are relatively well defined. So this criticism may not always be relevant. More importantly, symbolic interaction has largely failed to incorporate emotion as much other than an ancillary effect. The current work on emotion (Turner & Stets, 2005) should assist in obviating this criticism.

Ecological Theory and Systems Theories

Ecological and systems theories focus on multiple levels of analysis and the interactions between these levels. For example, an individual is embedded in the family, but also the individual and family are both embedded within a community. Certainly ecological theories have their roots in both biological studies but also in the social ecology of the early Chicago School of social ecology that would include work such as that by Burgess (1925). Systems theory arrived in sociology during the 1940s and 1950s with the systems functionalism of Parsons. Throughout the sciences, systems theory was often viewed as a way to unify all the sciences into one approach because of its high level of abstraction.

In most reviews of theoretical frameworks in family studies, ecological and systems theories are viewed as two distinct bodies of thought (e.g., White & Klein, 2008). From the outset, however, the major distinction between these two has been that most ecological theories emphasize the biological basis of social phenomena and most systems theory is more abstract. In both areas, systems and ecology, this difference is increasingly trivial.

Ecological and systems theories focus on multiple levels of analysis and the interactions between these levels. For example, an individual is embedded in the family, but also the individual and family are both embedded within a community. The complex interactions between levels may include physical and social supports and adaptation.

Many family scholars would see Bronfenbrenner (1979) as the major theoretical source for ecological theory in the family area. Certainly, Bronfenbrenner elaborated ecological levels and interactions between these levels. For the most part, Bronfenbrenner supplied a generation of family scholars with sensitivity to these levels and interactions. The second generation of scholars, however, provided more substantive theory and application in the areas of marriage (Houston, 2004), child abuse (Garbarino, 1992), parenting (Bornstein, 1995), and daycare (Belsky, 1990, 2001a; Belsky & Eggebeen, 1991). In the area of parenting, attachment theory with its roots in ethology (see White & Klein, 2008) continues to be the dominant model, but not without some important critiques (see Hays, 1998).

Another area of development in this framework is the emerging bioecological model. At a microlevel, there is the perspective that social interaction is in part determined by the endocrine system (see Belsky, 2001b; Booth, Carver, & Granger, 2000). Furthermore, there is specific data on stress hormones and marital conflict (Kiecolt-Glaser, Bane, Glaser, & Malarkey, 2003) and testosterone (Mazur & Michalek, 1998). This microlevel approach is often related to more macrolevel evolutionary theories of families (Emlen, 1995) and general evolutionary theory.

The most abstract area of theoretical development in this area is systems theory. Systems theory (see White & Klein, 2008) is firmly rooted in the perspective that the whole is greater than its parts and those systems have properties above and beyond their components. Much of systems theory of the family has been aimed at producing useful metaphors for therapeutic purposes more than scientific explanation. One exception to this is the recent attempt to revitalize functional systems theory of the family (Swenson, 2004). However, much of the academic enthusiasm for systems theory of the family has waned (see Bengtson, 2001), leaving family systems approaches mainly in the areas of family practice and therapy. This should not be seen as a statement about the overall usefulness of the systems approach because in other academic areas the systems approach appears to be flourishing.

One such area is organizational theory. Organizational theory, for the most part, has found general system theory (GST) a good fit with the study of highly organized and hierarchical organizations. Kozlowski and Klein (2000) summarize this fit as follows:

Whether one takes a more macro (Parsons, 1950) or micro (Allport, 1954) perspective, the influence of GST on organizational science has been pervasive. Unfortunately, however, that influence has been primarily metaphorical….GST has exhibited heuristic value but has contributed relatively little to the development of testable principles in the organizational sciences (Roberts, Hulin, & Rousseau, 1978) (Kozlowski & Klein, 2000, pp. 6–7).

Kozlowski and Klein’s (2000) assessment seems very close to the general conclusion in family studies except that they pinpoint the problem for this lack of progress from metaphor to theory. They argue that much of the problem is that we have not developed successful ways to measure and test the complex multilevel interactions portrayed in systems theory. They suggest that with the advent of multilevel statistical analysis and multilevel methodologies, systems approaches might finally move from metaphorical to scientific theoretical status. It is too soon to establish if their optimism is justified. Furthermore, the theoretical problems with levels of analysis also plague ecological models. We will return to this point in the section on theoretical methods below.

One major problem with the ecological and systems framework is the failure to develop substantive theoretical statements rather than methodological sensitivities. For example, noting that “everything is attached to everything else,” that there are hierarchical multilevel effects, and that “holistic perspectives” should be used are all examples of methodological caveats rather than substantive theory. For the most part, ecological and systems theory awaits substantive theoretical development.

Conflict and Feminist Theories

Conflict theories argue that individuals and groups compete for resources or rewards and that the ensuing struggles bring about social change. Feminism, as a form of conflict theory, posits that the most basic source of conflict is between men and women. Over time, men have institutionalized certain forms of female oppression as patriarchy. One major goal of feminism is to challenge and remove patriarchy as a form of oppression (see White & Klein, 2008).

Farrington and Chertok (1993) suggested that the older versions of social conflict theory applied to the family did not seem to have much future in the study of the family. Since that time there has been little activity in the more traditional socialist and Marxian conflict perspectives. Whether or not their statement was indicative of their prescience or simply a self-fulfilling prophecy remains a mute point. These authors did predict some of the developments in family conflict theory.

…critical and feminist approaches seem to possess something else more traditional conflict theory does not: the kind of ideological passion that, we argued earlier, helped to bring about the emergence of the family conflict perspective in the late 1960’s and early 1970’s…. (1993, p. 376).

Indeed, most of the vibrant discourse about positivism, patriarchy, power, and sex and gender has been tied to the two joint intellectual movements. The fact that these two, critical theory and feminist theory, are now inseparable is not surprising. The epistemological and social critiques of critical theorists, poststructuralists, and postmodernists are closely aligned with feminist discussions of oppression and gender.

Postmodernists such as Lyotard (1984, 1992) have argued that postmodernism refers to both a social-historical period and an epistemological perspective. The epistemological perspective refers to the idea that, because the truth or falsity of any knowledge claim cannot be unambiguously decided (Derrida, 1976, 1978; Longino, 1990, 2002; Wittgenstein, 1969), all knowledge claims are on an equal footing, with none being more privileged than any other. As a corollary to this epistemological premise, we live in a socio-cultural period marked by a plurality of ways of knowing rather than a period in which one particular form of discourse is more privileged than others and offers its claims as “truth.” In other words, knowledge is relative.

Heywood and Drake (1997) divide feminism into three waves. Certainly, first wave suffragettes and second-wave egalitarian feminists might have some concerns about epistemological “relativism.” However, the current movement in feminist theory that parallels postmodern epistemology is third-wave feminism (Heywood & Drake). Third-wave feminism has been identified with anti-essentialism, critical race feminism, and standpoint epistemologies. For example, De Reuss, Few, and Blume (2005) argue that “Poststructuralist feminists have challenged the unified category of ‘woman’ as the basis for a general theory of oppression of all women because it obfuscates within-group variation (p. 449).” In this view, explanation does not come about through the use of nomothetic general principles that apply across contexts but by understanding the particular intersections of race, class, and gender as constructing oppression.

Third-wave feminists have argued among themselves regarding standpoint vs. poststructuralist epistemologies (De Reuss et al., 2005, p. 450) and in addition have been assailed by more traditional feminists as not providing any useful general knowledge. Even second-wave feminists, however, have identified the “intersections” of race, class, and gender as important in the exploitation of women by women and patriarchy (Ehrenreich & Hochschild, 2003). Ehrenreich (2000) first identified the “intersectionality” of housework where immigrant women of color were exploited as the house cleaners of the upper middle class. She further explored these same intersections with Global Nannies (Ehrenreich & Hochschild, 2003). So regardless of the epistemological arguments, many feminist scholars have found that gender alone is not a sufficient explanatory variable, but is best used in conjunction with other socially ascribed variables such as race and class.

Wills and Risman (2006) report that 25% of the articles in three major family journals have at least some feminist content. Certainly, this represents a major intellectual contribution to the theoretical scholarship in this area (White & Klein, 2008) since all other theoretical frameworks would share the remaining 75%. These two convergent facts seem to attest to the worldwide force that feminism has had and, as well, the fact that feminist theories are making an intellectual contribution to the family literature.

The major criticisms of feminist theory are that it is not a theory but an ideology and that the theory is too narrow to be useful. The first criticism assumes that there is an inherent contradiction between theory and ideology. There should be no problem unless theoretical falsification is blocked by ideological belief. Contrary to this, feminist theories have shown great diversity and innovation through successive waves of ideology. The stronger criticism is that feminist theories tend to be “one trick ponies” where only one major variable (patriarchy) is used to explain all outcomes. Certainly there is some truth to this but there is the extension of third-wave theory to include intersections (or interactions) with other critical variables.

Insights

The purpose of this section is to demonstrate how existing theory supplies insight and understanding about particular family phenomena. This section will simultaneously point out some of the shortcomings of theoretical formulations in this regard. There are many areas that could serve as examples of insightful theory; for example, several theories of the intergenerational transmission of social class, parent–child attachment theory, family resource theory, and family transitions. The largest concentration of research and theoretical work, however, has been in the area of the determinants of marital outcomes: staying married and separating or divorcing.

Marital Relationships and Marital Stability

The attempt to explain why people get divorced has intrigued scholars throughout the twentieth century and into the current century. The original and somewhat common sense explanation of divorce was centered on the assumption that people who got divorced did so because they were not happy or well adjusted to marriage. As a result, much of the literature has had a distinct bias toward the individual level of analysis rather than the dyad or even the macrolevel of society and culture. It is entirely possible that scholars would still be making the determinants of individual relationship satisfaction the prime predictor of divorce if it were not for the anomaly that even happy marriages end in divorce and unhappy marriages may endure (see Lewis & Spanier, 1979). This anomaly forced scholars to pursue more depth in their theoretical formulations and this process is ongoing.

There are two dominant theories used to explain divorce and both are within the larger choice and exchange framework. One theory focuses on the individual’s marital quality (happiness, adjustment, satisfaction) as a major determinant (e.g., Levinger, 1965; Lewis & Spanier, 1979; Previti & Amato, 2003; South & Lloyd, 1995). The second theory focuses on the interdependency in marital exchange sometimes referred to as the “specialization and trading model” (e.g., Becker, 1981; Parsons, 1950; Parsons & Bales, 1955; Scanzoni, 1970, 1972; South, 2001; Teachman, 2003). It would appear that these two theories not only differ in substance, but also in regard to the level of analysis (individual vs. dyad).

Marital Quality and Marital Stability

This theory assumes that low levels of individual satisfaction and happiness with the marital relationship predicts divorce. Of course divorce is a dyadic phenomenon; however, it only takes one disgruntled spouse to file for divorce and when one member of the dyad experiences costs and withdraws rewards, the relationship usually becomes costly for the other. Thus, this theory would argue that it is the individual perceptions of the relationship that should be analyzed. It is further assumed that the individual would desire outcomes that are rewarding and avoid costs. Lewis and Spanier (1979) supplied an extensive set of propositions about the determinants of individual marital quality including premarital factors, lifestyle factors, and interactional factors. Following Levinger, most theorists in this approach have incorporated the alternative attractions outside of the relationship and the barriers to divorce as moderating the relationship between marital quality and divorce. This perspective allows researchers to explain the two anomalous categories: high-quality marriages ending in divorce and low-quality marriages with high marital stability. High-quality marriages end in divorce because they have a high level of alternative attractors and a low level of barriers. Likewise, the low-quality enduring marriages have few alternative attractors and high levels of barriers (White & Booth, 1991). Hence, this approach can explain why people have high or low marital quality (Lewis & Spanier, 1979) as well as what the factors are that moderate (attractors and barriers) the effect of marital quality on marital stability (Levinger, 1965).

As with most theories, there are a host of unresolved theoretical problems. First, a minor problem with this theory is whether it is entirely an individual level theory or if it is multileveled. Although some scholars treat the alternatives and barriers as only a matter of the individual’s perception (Previti & Amato, 2003), other scholars such as Lewis and Spanier (1979) clearly see alternatives and barriers as exogenous to the individual and embedded in the broader context such as the particular sex ratio or norms of a particular religious community. The importance of this is to clarify whether this is totally a social psychological or sociological theory. A second problematic area is pointed out by the debate as to the form (nonlinearity) of the moderating relationship between marital quality and marital stability given various combinations of alternatives and barriers (Thomas & Kleber, 1981) as pointed out by Bartolic, Bulcroft, and White (1997). Most importantly, these authors raise the issue that low alternatives attractors and high barriers might have a distinctly different pattern of moderation than high alternatives and low barriers. Yet another problem is the need for conceptual clarity regarding attractors and barriers and the stability of these across time. For example, being single might well be seen as an alternative when young but much less so in midlife. The definition of attractors and barriers should ensure that attractors are indeed “pulls” throughout the life course. Despite these problems, this theory represents one of the best explanations of marital stability.

Marital Exchange and Divorce

The “specialization and trading” theory originated with Parsons (1950; Parsons & Bales, 1955) and his assumption that the expressive and instrumental roles in the family were increasingly specialized as families became more isolated and nuclear. Scanzoni (1972) added more of an exchange component focusing on the changing pattern of resource power with wives employment. Over time, scholars argued that the social and economic interdependency that was the foundation of marriage is disrupted by female employment (Becker, 1981). South (2001) reports that the empirical findings are substantial on both sides in regard to the negative effect for wives employment on marital stability. This theory, however, remains the most popular macrolevel theory of marital instability.

There are several major problems confronting this theory. First, there is in general no identification of a social mechanism (Hedström & Swedberg, 1998). What mechanisms are working within the association between wives’ work and marital stability? More precisely, most scholars would like to know how wives’ employment functions in the relationship to increase the probability of divorce. Lowered amounts of couple time, alternative attractions at work, and resource differentials are some of the possibilities. A second problem area is the specification of the exchange or interdependence. Certainly Scanzoni (1972) suggested that the expressive dimension of the exchange might suffer since employed wives would have more instrumental power. However, the exact form of the exchange and how it is linked to divorce remains unclear. Finally, South’s research (2001) points out that the effect of women’s employment is within the context of social norms favoring or disfavoring this behavior. He suggests that the direction of the effects changes across periods of time and is more complex and dependent on gender role attitudes and institutional supports and opportunities.

Perhaps the most striking aspect of South’s comment is the change in levels of analysis from employment to the individual situation and choices. This seems to point to an opportunity for theorists to unite these two major theories into one more coherent theory. Certainly, phrases such as “adequate institutional supports” and “alternatives to their current husbands” would seem to suggest the language of barriers and alternatives. In addition, the conflicting findings about the effects of wives’ employment may be because some employment provides alternative attractors such as male coworkers and economic freedom while other forms of employment provide only same sex coworkers and marginal economic returns. South (2001) also appears to see that barriers and alternatives might change over historical period and cohort. This observation fits well with the argument made by Sabatelli and Ripoll (2004) that barriers and alternative attractors may not be stable across time. Even though these two theories offer separate insights into marital stability, it seems that theorists clearly have the opportunity to address the shortcomings of each theory while constructing a single multilevel theory of marital stability.

Methodology

Traditionally, theory construction activities have been divided into inductive and deductive theory construction. This seems to be a false and misleading distinction. Recall that theory construction occurs in the context of discovery. After the theory is developed, it would then be tested in the context of justification. All theory construction activities are aimed at producing general propositions that constitute a knowledge claim. These general propositions can be deduced from a larger set of propositions or they can be produced from inductive observation. Peirce (1878) suggested that scientists actually use abduction which is a combination of deduction and induction. During the 1960s and 1970s, a host of books and papers appeared on the rules of theory construction (White, 2004). These seemed to achieve little other than to stifle theory development. It is critical to science and theory that the approach, inductive or deductive or anything else, not be used to judge the truth or falsity of the proposition. That would be to commit the genetic fallacy we discussed in the first section of this chapter. Furthermore, the important point is that there is no magical formula for the production of theoretical propositions. These propositions can come from observation, folk wisdom, intensive qualitative research, or quantitative research. Insight and creativity are what is needed.

Even though there is no formula for producing theory, there are at least four elements that we would like any theoretical statement to have. First, all theoretical statements should endeavor to have concepts that are clear and well defined (conceptual clarity). Without this element, we would not know how to devise measures, how to establish validity, and our interpretations would be either vague or ambiguous. A second element is that our statements should be expressly formulated as propositions. Surely, researchers have read entire theoretical tomes and come away not being able to identify a specific knowledge claim. Theorists must succinctly state their claims as propositions. A third element is that of logical coherence. Certainly we want our theories to be logical so we can understand. How do we gain understanding from illogical and nonsensical theory? Even more importantly, the logical structure of propositions assists researchers and other theorists to deduce novel propositions or contradictory propositions or testable propositions (see White & Klein, 2008). The fourth and last element that theory should have is to identify the social mechanism that is producing the effect (see Hedström & Swedberg, 1998). It is useful for theorists to describe the mechanism by which action is produced. Not only does this help us address spuriousness in our theoretical statements, but it also helps us focus on the proper level of analysis. For example, this call for social mechanism would help us identify when an aggregate effect such as employment on marital stability actually reduces to the perception of actors in regard to barriers and attractors. On the other hand, if the attitudes and beliefs about divorce shared by actors in one social system but not other systems are related to marital stability, there may be a claim for “divorce culture” as a social mechanism (Yodanis, 2005).

These four elements represent guides for theory development. In actual fact, even our most well-developed theories can be faulted on one or more of these criteria. It is important to recognize that theories are always developing and often what we call a “theory” in the social sciences is actually a model. There are several levels of theoretic models: conceptual models, formal models, and measurement models. A well-developed theory contains but is not limited to all three of these. A conceptual model simply identifies the concepts in the theory and some of the relations. In conceptual models, the emphasis is on the specification of concepts affecting other concepts. Specification error in such models is due to missing important concepts or including unimportant concepts or ambiguous definitions. Formal models focus on the logic or relations between concepts. In such models, it is the relations that are being specified. For example, if money becomes less valued by actors at a certain point of accumulation, then the relation would be a linear quadratic equation. A formal theory focuses on the specification of such relations and misspecification is tied to specifying the wrong relation. Most researchers are familiar with measurement models which entail the valid operationalizations of concepts into variables and identification of theoretic relations as statistical models. Specification error occurs when operations are not isomorphic with theoretical concepts or relations. Clearly a complete theory would contain all three types of models.

Methodological/Theoretical Developments

There are several new areas of theoretical development that have been at least partially instigated by methodological advances. We discuss three of these. First, theoretical propositions have largely focused on relations between two conceptual entities (bivariate). Over the last 4 decades, researchers have tended to research increasingly complex sets of relations and few theorists have discussed how to make our theoretical formulations reflect these changes. Second, in the last 3 decades researchers have increasingly used multilevel modeling to capture processes at different levels of analysis. It is central that theorists accept the challenge of providing multilevel theories and identify mechanisms that transfer effects from one level to another. Third, the observation of event sequences has been used in family development and life course research. One impediment to such research has been the dependence of simple descriptive devices for sequences rather than the identification of theoretical measures.

Multiconceptual Propositions

A major impact of multivariate analysis is the need for theorists to both generalize complex relations captured in such empirical work and to propose multiconceptual propositions for testing. Certainly, early theory constructionists proposed to communicate such complex propositions as either formal models (mathematical) or as conceptual models. The mathematical models, though precise, lack the conceptual meaning while emphasizing the relations. Conceptual models, on the other hand, can be very useful in adding precision to the complex language statements. Language statements always contain some degree of ambiguity and so it is always useful to use conceptual models to further clarify the content of the theory. Shoemaker, Tankard, and Lasorsa (2004) argue that our theories are now “multivariate.” This claim might strike some theorists as humorous since social science theories have always been complex and multiconceptual (e.g., Marx, Weber, Durkheim, etc.). So this complexity is not new. What is new are the increasingly sophisticated methods available to researchers to test complex theoretical propositions.

There are several ways that theorists might make complex ideas containing three or more concepts available for testing (Shoemaker et al., 2004). Aspects of conceptual models and formal models may be usefully summarized as graphic displays. For example, in our previous discussion of marital quality and marital stability, it was noted that there are several distinct arguments about the ways that alternatives and barriers function as moderators. We can assert that the combination of low alternatives-high barriers affect those with low levels of marital quality more than those with high levels. This, however, can be more easily envisioned in a graphic presentation such as provided by Bartolic et al. (1997). Such graphics as Fig. 2.1 immediately show the curvilinear relation and suggest its form.

Fig. 2.1
figure 1

Graph of one possible moderating relationship from Bartolic et al. (1997, Fig. 1.5, p. 32)

Full size image

As theorists develop propositions where there are more than three or four active conceptual components, the use of graphic and mathematical equations detailing these relations will become necessary. As statistical techniques continue to improve so that we can test complex theoretical relations, it is incumbent on theorists that they clearly and unambiguously specify concepts and the more complex and nuanced relations in their theoretic models. The most optimal way to achieve such clarity is that theorists would specify the mathematical model and researchers would then face problems of translating these into statistical models. Certainly early family theorists endeavored to present graphic models (see Burr et al., 1979), but this has been less in evidence in more recent theoretical products.

Multilevel Propositions

Family theorists and researchers have long been interested in multilevel analysis (Bulcroft & White, 1997; White & Teachman, 2005). Theoretical and methodological discussions in the last 2 decades regarding levels of analysis (e.g., Coleman, 1990; Hedström, 2005; Sawyer, 2005) have focused more attention on these issues. Another component that has affected our awareness in regard to levels of analysis is the increasing use of hierarchical linear modeling or more accurately linear mixed models (LMM) among family sociologists (e.g., Sayer & Klute, 2005; Teachman & Crowder, 2002). It seems, however, that awareness of levels of analysis and the methodological and statistical advances we have experienced in the last few decades have “out paced” theoretical developments and our understanding of what is involved in constructing multilevel theories with cross-level interactions. Even in the most obvious case of marriage and families, we seldom find scholars developing multilevel theories or we find multilevel theories that are profoundly vague. This absence of multilevel theory leads scholars to rely on common vernacular understandings and personifications of aggregates rather than well-developed theory (Klein & Kozlowski, 2000; White, 2009).

Although social science theories have often suggested different levels of analysis (e.g., Bronfenbrenner, 1979), these theories most often failed to identify the mechanisms that move effects from one level to another. Furthermore, it has been difficult to meet assumptions of independence when using aggregated scores or scores collected at a different level of analysis. Some of these obstacles are less inhibiting using LMM and particular forms of these such as hierarchical linear models (HLM). These models allow us to simultaneously estimate the effects for individual level variables and higher order variables such as those within the family or community (see Teachman & Crowder, 2002).

The availability of these statistical techniques to researchers puts increased pressure on theorists to provide the mechanisms by which effects traverse levels. For example, if school board funding is reduced and parental investment increases, we have to explain how the reduction in funding is tied to the individual parent’s behavior to become more involved with their child’s school. White (2009) discusses the complexities that confront us in terms of theorizing about “trickle down” linkages (see Yodanis, 2005) as well as emergent phenomena. Although this is a long-awaited development for researchers and theorists, it does provide new challenges and potential hazards for theory (White & Teachman, 2005).

Optimal Matching

The interplay of deductive and inductive theory is no more apparent than with the methodological approach called “optimal matching” (Abbott, 1995). Optimal matching was first used to describe DNA sequences; however, Abbott (1995) adopted the technique for studying life course events. Briefly, optimal matching examines a finite set of elements and episodes over time and reduces them to one summative, reference sequence. The number of iterations to move a sequence of elements and episodes to the same pattern as the reference allows the creation of a measure of distance between each respondent’s sequence and the reference sequence. In previous research, the mode for each event was often used as a reference point but that introduced error when trying to measure sequences. Optimal Matching Analysis or OMA utilizes the distance measures created to establish clusters (Martin, 2009). This approach has been used in the social sciences with some partial application to family and the study of life course sequences (e.g., Han & Moen, 1999; Pollock, Antcliff, & Ralphs, 2002; Shanahan, 2000). Martin (2009) argue that optimal matching can be used to operationalize some of the processes identified in the theory of family life course developmental. He proposes that sequences can be measured as to their distance from reference sequences and off time and out of sequence states identified. This technique can be used to test theoretical propositions or it can be used inductively to capture empirical patterns of sequences. His proposal has the potential of further refining and developing what continues to be one of the most popular approaches to the study of the family.

Conclusion

Is Vargus (1999) correct to state that family theorists are wandering in the wilderness? At one point in time, family theory could be regarded as one of the most theoretically advanced areas in the social sciences. Certainly, the publication of the two-volume work by Burr et al. (1979) demonstrated an incredibly sophisticated range of both inductive and deductive propositional family theory. Since that time, theorists have focused less on producing sound propositional theory and more on debates about the nature of knowledge and relativity of knowledge claims. Regardless of the debates among theoreticians, researchers have continued to produce excellent empirical studies, though seldom have any of these studies really tested theoretical propositions. Indeed, there seems a disconnection between empirical researchers and theorists. Although there should be an historical analysis of this situation (see White, 2004), it is even more important to close this rift.

In order to make theoretical work relevant to researchers, theorists must become more sophisticated in the way they state theory. Clearly defined concepts and propositions that pay attention to the complexity of relations are especially important. In addition, theorists might focus on particular areas such as marital quality and stability, parent–child socialization, intergenerational transmission of social class, or family violence so that there is an obvious overlap between researchers in these areas and the theorists in these areas. This does not imply that theorists should give up their ties to more overarching theoretical assumptions. It is this linkage to a larger perspective that is one of the strengths theory offers. Theorists also offer logical skills and there is some need for them to return to the basics of logic and logical formulation to make sure that what we say is formally established. At the same time, researchers need to realize that there is more to producing knowledge than publishing reports and getting cited. In the long run, the research that will be remembered and celebrated is the research that has theoretical import.

Although there are good examples of researchers using theory and refining theory (Amato & Hohmann-Marriott, 2007), there are far fewer examples of researchers deducing and testing theory. There are some examples where relatively difficult theories, such as Giddens’ globalization of intimacy, have nonetheless been empirically examined (Gross & Simmons, 2002) and evolutionary theory vs. rational choice (Bokek-Cohen, Peres, & Kanazawa, 2007). Furthermore, there are a host of areas that are awaiting theoretical refinement. For example, Bourdieu’s (1979/1984) notion that the early learning of “habitas” is instrumental to social class transmission needs to be reconciled with Kohn’s (1959) theoretic model of social class transmission that emphasizes current conditions of employment as producing social class values. Another area for development is the area of global family change. In particular, the claim by Therborn (2004) that the only major area supporting Goode’s (1959) convergence theory of family is the decline of patriarchy.

In the final analysis, I believe we now can be more sanguine about the future of family theory than during the previous decade. Many of the destructive debates about relativism are largely ignored as frivolous by contemporary philosophy of science (see Okasha, 2002; Turner & Risjord, 2007). Opportunities for furthering theory that is relevant to researchers are abundant. Available texts and academic training in family theory have improved and a generation of young scholars now have a specific publication outlet for theoretical work (Journal of Family Theory and Review). The confluence of opportunity and training suggests a promising decade ahead for family theory.