Observing Children in Their Natural Worlds

A Methodological Primer, Third Edition

Anthony D. Pellegrini, Frank Symons, John Hoch

• Publisher: Psychology Press
• Year: 2012
• ISBN:
  9781136236792
• Edition: 3
• Topic: Psicologia
• Subtopic: Psicologia dello sviluppo

This book shows readers how to conduct observational methods, research tools used to describe and explain behaviors as they unfold in everyday settings. The book now uses both an evolutionary and a cultural perspective. The methods presented are drawn from psychology, education, family studies, sociology, and anthropology, but the author's primary focus is on children in school, family, and social settings. Readers learn how to make observations in real contexts to help them create a verbal picture of behaviors they see. The importance of considering reliability and validity factors while testing within each environment is emphasized throughout. The author draws from the literature that provides methods for observing animals in their natural habitats, but emphasizes the use of observational methods to solve human problems. The book is organized in the way a researcher conducts observational studies—conceptualizing of the idea, designing and implementing the study, and writing the report. "Things to think about" sections provide an opportunity for students to solidify their understanding of the material and the Glossary defines the key terms introduced in the book.

Highlights of changes in the new edition include:

• The introduction of the cultural perspective in chapter 4 along with the evolutionary (epigenetic theory) perspective and the integration of cultural examples throughout the book.
• More varied examples from developmental psychology, family studies, and education.
• Extensively revised chapter (3) on ethics reflects the current revelations of scientific fraud and the push for researchers to maximize scientific integrity in their community.
• Updated chapter (12) reflects the latest computer technologies used in observational methods including iPhones and Blackberrys for conducting observation, ABC Data Pro and Behavior Tracker for evaluations, and Excel for constructing observational templates.
• Expanded chapter (13) on writing the research report and more on issues of plagiarism (ch. 3).
• The latest on minimizing observer effects on participants and testing their effectiveness.
• New environmentally friendly design, the Things to Think About sections were retained, but the blank pages for answers were eliminated.

Intended as a supplementary text for advanced undergraduate and/or graduate courses in research methods and/or developmental research or developmental/child psychology taught in psychology, education, human development, and nursing, educators and researchers concerned with assessing children will also appreciate this book's introduction to observational methods.

CHAPTER 1

The “Whys” and “Whens” of Observational Methods

This book is concerned with using observational methods, primarily in everyday settings. By observational methods, I mean methods which are based on direct observations of behavior. These methods contrast with other methods, such as questionnaires and interviews. This book will help you learn the skills to construct a verbal picture of behaviors as they unfold in time. The descriptions of these methods are very useful for a number of important ventures, such as basic scientific research and more applied enterprises, such as working with children and families.

Direct observation methods are basic to the inductive and deductive process integral to the scientific enterprise (Russell, 1931/1959). In the former case, direct observational methods can be used to both describe and categorize the behavior of individuals and groups of individuals as they interact in real time (Smith, 2011). For example, in observing toddlers being dropped off at child care by their parents, observers note that children are not socially interactive with peers when their parents leave. Descriptions are basic to the generations of hypotheses that can be later tested, deductively. By this I mean, in the process of description, researchers may get ideas of what they think will happen in a specific case (i.e. an hypothesis), such as how a toddler will react when they are left by their mother in a strange situation. This hypothesis can be tested by observing toddlers in a number of different settings to determine their reactions. The specific hypothesis would be supported if the toddlers are distressed in strange situations but not in familiar settings.

Observational methods are also important for use in applied settings, such as working with children and families at home and in schools. For example, if you want to determine the impact of an early intervention program on children, such as Project Head Start, direct observations of children's “social competence” (Pellegrini, 1992; Zigler and Trickett, 1978) are a more reliable and valid indicator than standardized testing procedures. In the case of testing, young children are especially unreliable test-takers (i.e. their scores across time are unrelated) because they are very sensitive to contextual distractions.

An indispensable part of any description of behavior is a thorough explication of aspects of the contexts in which the behavior is embedded. My assumption is that behaviors are rendered more understandable when they are considered in their context. Context can be described at different levels of abstraction, ranging from distal forces, such as the pressures associated with evolution by natural selection, to the more immediate and proximal forces associated with dimensions of the immediate environment, such as the number of boys and girls present, ambient temperature, or room size. The various levels of specification, of course, relate to our theoretical orientation.

Distal factors, for example, can be specified in terms of the forces of natural selection, if you choose an evolutionary orientation. From this view, males' and females' behaviors are influenced by their evolutionary histories. For example, that males are more active and aggressive than females, is influenced by historically polygynous mating patterns (i.e. males mate with many females and thus compete with each other for access to females) and physical dimorphism (where males are physically larger than females). This difference, in turn, should relate to social segregation by sex and sex differences in physical activity. Later in this book I will explore the ways in which one theoretical orientation among some evolutionary biologists (epigenetic theorists, such as Gottlieb, 1998, 2003; West-Eberhard, 2003) conceptualize the ways in which aspects of the environment and the individual influence each other.

Another, more cultural, theoretical orientation, such as that proposed by Alex Masoudi, Andrew Whiten, and Kevin Laland (Laland and Brown, 2002; Mesoudi, Whiten, and Laland, 2006; Mesoudi and Whiten, 2008; Whiten and Mesoudi, 2008) and Boyd and Richerson (1985; Richerson and Boyd, 2005), stresses the ways in which culture, defined as information received by individuals through social transmission (Richerson and Boyd, 2005), influences not only the immediate behaviors of individuals, but also may act like genes (or memes [Dawkins, 1976]), and affect future generations through a process akin to natural selection.

Describing the transaction between dimensions of the more immediate environment often refer to the physical (e.g. specific toys available to play with) and social dimensions (e.g. number and composition of the group being observed) of the observational setting can be especially useful not only in scientific research but also for those working with children and families in a variety of settings, because these factors influence behavior. For example, female-preferred toys differentially influence the behavior of males and females to the extent that boys might be less engaged than the girls (Pellegrini and Perlmutter, 1989).

While this is a very general definition of context, useful in understanding the inter-relations between individuals behavior and the situation in which they are embedded (See Hinde, 1976, for an exhaustive description of context), it is also important to recognize that two behaviors may have similar physical features but have very different meaning, depending on context. For example, a playface of a popular child (a dimension of the social context) is a typically interpreted as playful by his peers. The very same face exhibited by an aggressive child, on the other hand, is typically interpreted as aggressive (Pellegrini, 2009). Thus, one dimension of context, the participants, relates to the meaning of the behaviors exhibited.

While the methods discussed in this book are generic, that is, they can be applied to the observations of ground squirrels, rhesus monkeys, shoppers in a super market, or children on playgrounds, I will be concerned, primarily, with children and adults embedded in the contexts of schools, neighborhoods, and families. These methods can be applied to describing children in the many contexts they inhabit. I agree fully with the concerns expressed by Herbert Wright (1960), over 50 years ago, when he noted that we have limited knowledge of children. Our knowledge, now as was the case 50 years ago, is based mostly on descriptions of children in preschool settings or in laboratories. While this concern was echoed by Bronfenbrenner (1979), 20 years after Wright, we still know very little about children outside of school settings. For example, we know virtually nothing about with whom children interact with after school and on weekends. Given today's demographics of increasing instances of dual career families, single parents, and paternal custody, it seems important to generate such descriptions. After all, some of our educational intervention strategies, such as joint book reading between mothers and their children, assume that this is still the primary dyad in children's lives. Observational methods applied to children's natural worlds can be very helpful in answering these sorts of applied and policy questions.

---

Good Descriptions: Or, Maximizing Reliability and Validity

While I suggest that the methods considered here are useful for observing children in their everyday worlds, it must be noted that these same observational methods can be used in experimental laboratories as well. In both cases the aim of good description is the same. “Good” descriptions meet criteria of being both reliable and valid. While reliability and validity will be treated in much greater detail in a separate chapter, it is necessary to provide a general definition at this point. Generally, reliable observations are those in which one observer records in a consistent manner and there is agreement between different observers. Valid observations, on the other hand, are those which actually measure what they purport to measure. While much more will be said about these issues in subsequent chapters, suffice it to say for now, that in both the laboratory and in the field, reliability between and among observers is necessary, but not sufficient, for validity. Simply put, before anything can be said about the truthfulness of a description (i.e. validity), there must be agreement on what is seen (i.e. reliability).

With validity, we are concerned with the usual trinity of concerns: internal validity, external validity, and ecological validity, as well as the overarching issues of construct and content validity. Construct validity refers to the degree to which that which we're observing represents a psychological construct, or an abstract representation of a phenomenon such as aggression or intelligence. While we cannot have direct measures of a construct, because it is by definition abstract, we can have indirect indicators; for example, punch, bite, and kick, are indicators of aggression. Added to this, construct validity must be concerned with any policy or social implications of using the measure as part of a decision making process (Cronbach, 1971, 1980; Messick, 1995). For example, if you are using a measure of aggression to evaluate students' placement into special classes, you must also consider your confidence in the measurement in relation to the impact on students' lives. Content validity addresses the extent to which the observational category measures what it purports to measure. So, does the content, or separate behavioral indicators of a category (e.g. play face, reciprocal role taking, nonstandard behavioral sequences and combinations) all relate to the category “play”?

It is, however, on these first three dimensions of validity (internal, external, and ecological) that laboratory and field methods most often diverge, and often the divergence is abrupt. Laboratory procedures are concerned primarily with internal validity, and less concerned with external validity. By manipulating and controlling variables, experimentalists can make internally valid statements about the effects of independent variables (such as the nature of toys that children are exposed to) and on dependent variables (such as children's play behavior). That is, in internally valid experiments, you can be sure that variation in independent variables alone is affecting the dependent measures. Experimentalists are often less concerned with external validity, or generalizing laboratory results to relevant field settings, and ecological validity, or the comparison of results across a number of different settings.

Field studies (or studies of people in their every day environments), on the other hand, are primarily concerned with describing behavior as it occurs in its natural habitat and recognizing the trade-offs between balancing internal and external validity. Imbalances in both laboratory and field approaches result in incomplete knowledge. Obviously, descriptions of laboratory behavior with minimal ties to its field analogue are of limited use and interest. Similarly, descriptions of behavior in the field, without clear limits on alternative explanations, do not advance our understanding of the ways in which variables affect each other. Alternatives to these two approaches include designing “ecologically valid” experiments, such as comparisons of children's behavior in laboratory and classroom settings (e.g. Bronfenbrenner, 1979) or taking advantage of “natural experiments.”

Examples of natural experiments include examining the effects of androgens (a male hormone) on female fetuses and subsequent sex-typed behavior. More specifically, there are cases where female fetuses are exposed to excessive amounts of male hormone (the syndrome is known as Congenital Adrenal Hyperplasia [CAH]). Research has shown that CAH girls, relative to control girls, prefer to play with male toys and have male playmates (Hines, 1982; Hines and Kaufman, 1994). In this case there is an intervention (in the form of naturally occurring, excessive amounts of male hormones), that could not be able to experimentally manipulated (for obvious ethical reasons). In short, field experiments and natural experiments represent a rapprochement, of sorts, between the lab and the field. At a minimum, field work should precede experimental analogues.

If we as researchers do not work towards some sort of rapprochement between experimental and field approaches, disciplines such as early education and child and educational psychology are sure to atrophy, and probably with good cause. I say with good cause, because experiments should aspire to explaining the ways in which organisms live and develop in their natural world. Similarly, we need some level of explanation and causal inferences so that we can begin to understand these processes more clearly and then use this information to design educational environments. As I will discuss in the next chapter, such a rapprochement has been in the offing for a number of years. Specifically, Bertrand Russell (1931/1959) explicitly outlines the ways in which inductive (e.g. description derived naturalistic observations) and deductive (results of hypotheses tested in experiments) approaches are complementary to the scientific enterprise. While this recommendation has continued to be voiced (e.g. Cronbach, 1957; Dawkins, 2009; Smith, 2011; Tinbergen, 1963), it is generally not heeded. To the detriment of many fields of inquiry, one research method is pitted against the other, with accusations of one being more or less “scientific” than the other. Following Russell, my bias is that description is an important, and necessary first step, of any scientific enterprise.

---

“Mere Description”?

Frequently descriptive research, in which observational methods play an indispensable part, is considered “merely” descriptive, implying that description is either an unimportant or an atheoretical enterprise. Experiments, by contrast, are portrayed as more “scientific” and explanatory. An unfortunate example of this dichotomy was a report on teaching children to read, conducted by the National Reading Panel (2000) and funded by the United States National Institutes of Health. The panel examined only experimental and quasi-experimental research to understand how children learn (and don't) learn to read. Fortunately, this policy was not in place when NIH was funding studies of language development, such as Roger Brown's (1973) ground breaking case studies of Adam, Eve, and Sara!

I, along with others (e.g. Blurton Jones, 1972; Dawkins, 2009; Hinde, 1980; Smith, 2011), contend that thorough descriptions are an important step in hypothesis generation and, as we'll also see in the next chapter, can be used to test hypotheses as well. That is, science is a dynamic process where descriptions are crucial to the inductive phase of the scientific enterprise and experiments to the deductive phase. Further, good descriptions are neither atheoretical nor less “scientific” than approaches offering causal explanations. The ability to adequately describe a phenomenon, especially in complex organisms like humans, is crucial to theory generation (i.e. induction). The complexity of the human organism interacting, even in the simplest ways, necessitates that observers make choices about who, what, and when to observe. For example, take the seemingly simple issue of classifying participants' roles. Kagan's (1994) simple but informative example clearly points to the importance of theory in description: “Zoologists classify cows as mammals, economists classify them as commodities, and some cultures regard these animals as sacred symbols” (p. 11). Further, theory guides us in terms of the levels of specificity of our categories. If I had a biological orientation, I might examine relations between hormones and behavior. If, on the other hand, I had a cultural anthropological perspective, I might examine the match between the cultures of school tasks and those indigenous to children's homes. Given the complexity of the phenomena that can be observed, a clear theoretical orientation is necessary because it guides decisions made about observational methods.

It thus makes sense to foreground, or to make explicit, those decisions you make when you observe. In this way you know the paths that you have traveled in making these decisions. It is naive and unrealistic for observers to think or state that they enter an observational field with no biases in terms of what they will observe and/or how the phenomena will be observed and categorized. Human observers have too many “schema,” or concepts about the ways in which the world works, in their heads to take such a stance. The best we can hope for is to make our biases explicit and to try to minimize them. Similarly, observers cannot go out and observe “everything”, for there is clearly too much to observe. With this in mind, you should then make explicit what it is you are interested in (that is, your “question”). With an explicit question in mind (which may arise from making preliminary observations), you can then begin to consider what it is specifically you will observe.

In terms of the “scientific” nature of naturalistic observationa...