ARTICLE IN PRESS

International Journal of Nursing Studies 43 (2006) 113122 www.elsevier.com/locate/ijnurstu

A multidimensional analysis of the epistemic origins of nursing theories, models, and frameworks
Jason W. Becksteada,, Laura Grace Becksteadb
a

University of South Florida, College of Nursing, 12901 Bruce B. Downs Boulevard, MDC 22, Tampa, FL 33612-4766, USA b University of Tampa, USA Received 9 September 2004; accepted 30 September 2004

Abstract The purpose of this article is to introduce our notion of epistemic space and to demonstrate its utility for understanding the origins and trajectories of nursing theory in the 20th century using multidimensional scaling (MDS). A literature review was conducted on primary and secondary sources written by and about 20 nurse theorists to investigate whether or not they cited 129 different scholars in the elds of anthropology, biology, nursing, philosophy, psychology, and sociology. Seventy-four scholars were identied as having been cited by at least two nurse theorists (319 citations total). Proximity scores, quantifying the similarity among nurse theorists based on proportions of shared citations, were calculated and analyzed using MDS. The emergent model of epistemic space that accommodated these similarities among nurse theorists revealed the systematic inuence of scholars from various elds, notably psychology, biology, and philosophy. We believe that this schema and resulting taxonomy will prove useful for furthering our understanding of the relationships among nursing theories and theories in other elds of science. r 2004 Elsevier Ltd. All rights reserved.
Keywords: Nursing theory; Multidimensional scaling; Epistemic space

Science is the systematic investigation and explanation of the world around us. In order to organize their work and its results, scientists have divided science into major elds. Herbert (1959, p. 11)

1. Introduction 1.1. Epistemic space The progress of science is the work of creative minds. However, every creative mind that contributes to
Corresponding author. Tel.: +1 813 974 7667; fax: +1 813 974 5418. E-mail address: jbeckste@hsc.usf.edu (J.W. Beckstead).

scientic advance works within two limitations. It is limited by ignorance, for each discovery waits upon that other that gave way to it. Discovery and its acceptance are also limited by the habits of thought that pertain to the culture of any region or period; slow change is the rule. Given that advances in science build upon discoveries that have been widely accepted or sufciently established, we can envision scientic knowledge as having a progressive, or positive trajectory, moving from a position of less knowledge to one of more. Different elds of science, or bodies of scientic knowledge, may be said to have originated at different points in history and to be growing at different rates, affected by advances in relevant technology and cultural mores. These developments take place in what we label as epistemic space. Epistemic space is the unlimited expanse

0020-7489/$ - see front matter r 2004 Elsevier Ltd. All rights reserved. doi:10.1016/j.ijnurstu.2004.09.021

ARTICLE IN PRESS
114 J.W. Beckstead, L.G. Beckstead / International Journal of Nursing Studies 43 (2006) 113122

in which all knowledge is located. Individual bits of knowledge and the dened relationships among them constitute the system of inquiry known as science. We assume that the amalgamation of scientic knowledge occurs within such space. According to Kuhn (1977), the term metaparadigm may be used to refer to the set of highly abstract concepts that identify the phenomena of interest to a scientic eld or discipline. The metaparadigm is the most abstract component in Fawcetts (2000) structural hierarchy of nursing knowledge containing the global concepts of interest to the discipline and their relationships. Metaparadigms summarize the intellectual and social missions of a discipline and place a boundary on its subject matter (Kim, 1989). As such, the metaparadigm guiding research in a discipline distinguishes that discipline from others. The metaparadigm of nursing identies for the nurse researcher the study of the wholeness or health of humans, recognizing that humans are in continuous interaction with their environments (Donaldson and Crowley, 1978). The metaparadigm of medicine, in contrast, identies for the physician researcher the study of the causes, diagnosis, and treatment of diseases (Fawcett, 1993). Next we turn to the juxtaposition of these two metaparadigms in epistemic space. Our perspective, or weltanschauung, of nursing and medical knowledge begins with Descartes dualism of mind and body which provides a means for dening the metric of our epistemic space. Believing as he did that the bodys behavior can be in no way guided by purpose, Descartes was naturally eager to show that there exists a sharp distinction between body and mind. He split the world into two types of being, volitional thought and physical or spatial matter. Since his time, the specialties of science have emerged consistent with this distinction. Epistemic space may be represented using a multidimensional coordinate system in which science exists and expands outward from an origin (corresponding to relatively established knowledge) in various directions encompassing new theoretical developments. Consider two perpendicular axes representing the trajectories of scientic knowledge development. The rst, C, corresponds to the sciences of the mind (e.g., psychology, sociology, anthropology, etc.) being spontaneous and teleological. The second, b, represents the sciences of the body (e.g., biology, chemistry, physics, etc.), being wholly mechanical. Using such a metric, all scientic knowledge in nursing and medicine may be positioned in epistemic space by assigning general coordinates that reect the relative inuences of established knowledge from the sciences of psychology, biology, etc., as reected in their metaparadigms. We consider the health sciences, namely nursing and medicine, as comprising the application and adaptation of established scientic knowledge (i.e., knowledge that is relatively close to the

New Theoretical Developments

Science of the Mind ( )

Nursing Medicine Established Knowledge Science of the Body ( ) New Theoretical Developments

Fig. 1. Schematic representation of epistemic space showing the relative placement of metaparadigms of nursing and medicine. Science of the Mind encompasses psychology, anthropology, and sociology. Science of the Body encompasses biology, chemistry, and physics.

origin of our epistemic space) to the promotion, improvement, and maintenance of human health and well-being. Thus, our placement of nursing and medicine in epistemic space (see Fig. 1) is consistent with their metaparadigms; nursing being more concerned with, and more inuenced by, established knowledge derived from work in the social sciences, arguably offering more teleologic explanations, while medicine is more concerned with, and inuenced by, the mechanistic explanations afforded in biology and chemistry. 1.2. Purpose The purpose of this article is to introduce our notion of epistemic space and to demonstrate its utility for understanding the origins and trajectories of nursing theory in the 20th century using multidimensional scaling (MDS). Whether a particular assemblage of nursing knowledge qualies as a philosophy, theory, conceptual model, or an organizing framework, has been debated by various scholars (e.g., Fawcett, 1993, 2000; Meleis, 1997; Reed, 1995; Whall, 1996). The subtleties of such discussions are by no means trivial, however, for our purposes we will use the term nurse theorist to refer to those individuals who have substantively affected the developments of nursing as a science during the 20th century, and the term theory in reference to the body of their published work that has lead to such developments. 1.3. Nursing theory and epistemic space In comparison to other sciences, nursing is in the early stages of scientic development (Bishop, 2002).

ARTICLE IN PRESS
J.W. Beckstead, L.G. Beckstead / International Journal of Nursing Studies 43 (2006) 113122 115

Examination of the published works by various nurse theorists reveals that many of them have built upon the established knowledge amassed in the mindbody sciences, consistent with our notion of nursings coordinates in epistemic space. Different foundational ideas, both conceptual and philosophical, have predisposed nurse theorists to describe and explain the phenomena of concern to the discipline in diverse ways (Kim, 1999). Our concern is with such foundational ideas. We propose that by analyzing patterns in the citation of scholars it is possible to discern, with a reasonable degree of objectivity, similarities in the epistemic origins of nurse theorists foundational ideas. In other words, it is possible to position nurse theorists in epistemic space on the basis of the established knowledge that inuenced their critical thinking and research endeavors. This approach to comparing nurse theorists is taxonomic rather than typologic. These two methods of classication are distinguished succinctly by Hair and Black (1998): a taxonomy is an empirically derived classication, while a typology is a conceptually based one. To conduct this analysis, we employed a quantitative technique known as MDS. This technique is briey explained below.

1.4. Multidimensional scaling MDS comprises a family of geometric models for representation of data in two or more dimensions and a corresponding set of methods for tting such models to actual data. The goals of MDS are (a) to nd a low (e.g., two or three) dimensional space in which points in the space represent the objects being studied and such that the distances between the points in the space match, as well as possible, the original similarities (or differences) between the objects and (b) to represent the space and the relationships among objects as a geometric model or picture. This second goal distinguishes MDS from factor-analytic methods (Shepard, 1972). Factor-analytic methods generally require a solution of greater dimensionality than do MDS methods (Napier, 1972). The MDS approach used in the current study is nonmetric in the sense that the data are nominal (a theorist cited, or did not cite a scholar). Typical (i.e., linear) factor analysis models have been criticized when applied to dichotomous data due to the non-positive denite nature of such covariance matrices, and nonlinear factor analysis models for determining dimensionality are unavailable or limited by sample size (see Meara et al., 2000 for discussion). The basic function of MDS may be illustrated by using travel times between a group of cities in order to reconstruct a map of their geographic locations. For example, Greenacre and Underhill (1982) used ying times between major airports. In this case, a single

attribute, travel time, served to quantify the proximity or similarity between objects (i.e., cities). More interesting applications have involved the use of multiple, and less objective, attributes to quantify the proximities among rare Scotch whiskies (Lapointe and Legendre, 1994) and among Renaissance painters (Davenport and Studdert-Kennedy, 1972). The adequacy of the resulting map may be expressed using a general measure of stress in the system, i.e., the lack of t between the actual proximities and those that result from the analysis. The stress measure is the square root of the normalized residual sums of squares and ranges from 0.00 to 1.00, with smaller values indicating less stress and a better solution (Kruskal and Wish, 1978). MDS was originated in the 1930s when Young and Householder (1938) showed how starting with a matrix of distances between points in Euclidean space, coordinates for the points could be found such that distances are preserved. It was later popularized by Torgerson (1958), who applied it to the problem of psychological scaling. Since then it has been used in nursing research (see Houfek, 1992; Wilson and Retsas, 1997) and numerous other disciplines: biometrics Lawson and Ogg (1989), counseling psychology Fitzgerald and Hubert (1987), ecologyTong (1989), ergonomicsCoury (1987), forestrySmith and Iles (1988), lexicographyTijssen and Van Rann (1989), tourismFenton and Pearce (1988), and brain connectivityGoodhill et al. (1995). Given the inherently Cartesian nature of our notion of epistemic space, it seemed reasonable to us to employ MDS in an attempt to objectively position nurse theorists in multidimensional space.

2. Method We began by compiling a list of 129 scholars in the elds of anthropology, biology, nursing, philosophy, psychology, and sociology from popular text books in each eld. This list operationally dened established knowledge for the purpose of our analysis. Next, a literature review was conducted on primary and secondary sources (see Appendix A) written by and about the 20 nurse theorists listed in Table 1. Although not exhaustive, this list represents the major contributors to nursing scholarship during the 20th century based on our review. One hundred and sixteen of our scholars were cited by at least one of the theorists studied here. A data matrix was constructed with nurse theorists as columns and scholars as rows. If a nurse theorist cited a particular scholar, a 1 was placed in the corresponding cell of the matrix, otherwise the cell was coded 0. (Due to space constraints, the data matrix is not reprinted here; however, it is available

ARTICLE IN PRESS
116 J.W. Beckstead, L.G. Beckstead / International Journal of Nursing Studies 43 (2006) 113122 Table 1 Results of multidimensional scaling of nurse theorists Theorist Symbol Coordinates Dimension 1 Abdellah Hall Henderson Johnson King Leininger Levine Neuman Newman Orem Orlando Paterson and Zderad Parse Peplau Riehl Rogers Roy Travelbee Watson Wiedenbach AD HL HN JN KG LN LV NU NW OM OR PZ PR PP RL RG RY TV WT WB 1.2029 2.4563 0.7511 1.3440 0.0298 1.3451 0.1962 1.3547 0.8655 0.6498 1.8926 1.4397 1.0807 0.5863 0.7939 0.0155 0.1301 0.0632 0.2095 0.7294 Dimension 2 0.8487 0.4174 0.2561 1.6144 0.5143 0.7929 0.0399 1.9066 0.0157 0.3037 0.6526 0.1254 0.5279 0.9732 1.7827 0.4519 0.1194 0.4507 0.1587 2.2737 Dimension 3 0.9220 0.7988 0.7169 0.3485 0.6540 0.4077 0.3417 1.0503 2.3113 0.6928 0.2701 1.4503 1.1461 0.7746 2.0639 0.1511 0.2520 0.0752 0.1711 0.1392

Note: Proximities between theorists were based on the proportion of their shared citations (k 74) using Russell and Raos similarity coefcient for binary data. MDS solution produced a stress value 0.167, and R2 0:826:

from the rst author.) This approach does not distinguish whether a nurse theorist cited a scholar to support, or contradict her theory; such was not our intent. Rather, this approach serves as a proxy indicating whether or not a nurse theorist was inuenced (for good or ill) by the writings of a particular scholar, at least to the extent that she felt compelled to cite his or her work. It could be argued that this approach has some limitations. Scholars might be cited for reasons such as availability rather than because of their intrinsic merit. Theorists may have read and may have been inuenced by scholars they did not cite (and it may also happen that some cited works they had not read!). While recognizing these limitations, this form of citation analysis does have the virtue of providing an objective measure similarity in such a context (Everett and Pecotich, 1993). The data matrix was used to calculate a matrix of proximity scores between all pairs of nurse theorists. Proximity scores reect the degree of similarity or dissimilarity among a set of objects being compared on a set of characteristics (Beckstead, 2002). There are numerous proximity coefcients for various types of data (see Cox and Cox, 2001, for review). For the purpose of quantifying the proximity between nurse theorists based on similarities in their patterns of cited scholars, we used Russell and Raos similarity index.

For any two nurse theorists, this index was dened as the number of shared citations over the total number of scholars qualifying for our analysis (see below). This measure was chosen from among several appropriate for binary data because it does not count conjoint absences as reecting similarity (i.e., two theorists are not considered similar because they omitted the same scholars) and because conceptually it represents the similarity between theorists in terms of the proportion of operationally dened established knowledge that they share. Scholars cited by only one nurse theorist, do not contribute to these proximity scores; their inclusion essentially constitutes noise and degrades the reliability of resulting proximity scores (Hair and Black, 1998). To be included in the analysis a given scholar had to have been cited by at least two nurse theorists. Proximity scores were thus based on 319 citings of 74 scholars: three from anthropology, seven from biology, 15 from nursing, eight from philosophy, 35 from psychology, and six from sociology.

2.1. Data analysis The proximity matrix was analyzed using Alternating Least squares SCALing (ALSCAL) developed by Takane et al. (1977). ALSCAL is an extremely exible

ARTICLE IN PRESS
J.W. Beckstead, L.G. Beckstead / International Journal of Nursing Studies 43 (2006) 113122 117 Table 2 Correlations among coordinates for nurse theorists on the scaling dimensions and the proportion of scholars cited from various disciplines Discipline Anthropology Biology General systems theorya Nursing Philosophy (all) Philosophy (existential)b Psychology (all) Psychology (personality)c Sociology Dimension 1 0.181 0.434 0.441 0.114 0.134 0.058 0.779 0.916 0.232 Dimension 2 0.256 0.733 0.691 0.413 0.098 0.066 0.097 0.226 0.335 Dimension 3 0.117 0.141 0.207 0.098 0.788 0.751 0.097 0.255 0.521

Note: Italicized values indicate the discipline(s) dominating each dimension. a There were seven biologists cited, three of these were cited by nurse scholars for their work in general systems theory. b Eight scholars of philosophy were cited, three of these were noted existentialists. c There were 35 psychologists cited, 12 of these were personality theorists.

procedure that can handle data that are: (i) nominal, ordinal, interval, or ratio; (ii) complete or have missing observations; (iii) symmetric or asymmetric; (iv) conditional or unconditional; (v) replicated or unreplicated (see Cox and Cox, 2001, for further discussion). The procedure is available as part of the SPSS package. Our application specied a nonmetric (ordinal level) analysis employing the Euclidean distance model. This model was applied to a full matrix of symmetric proximity scores.

3. Results MDS was successfully applied to the proximity matrix relating nurse theorists based on similarities in patterns of the scholars they cited. Initially, we examined a twodimensional solution given the nature of epistemic space proposed in Fig. 1. This solution produced a stress index 0.225 and an R2 0:757; however, we could not discern any systematic pattern in the coordinate values. The three-dimensional solution produced a stress index 0.167, an R2 0:826; and afforded a meaningful interpretation of the dimensions in epistemic space as described below.1 The resulting coordinates are presented in Table 1. In an attempt to provide substantive interpretation of the three dimensions resulting from the scaling of nurse theorists, point-biserial correlations between coordinates
1 It is not unusual for a higher-dimensional conguration to be more readily interpretable than a lower-dimensional conguration given the relationship among congurations of different dimensionality is more complex with MDS than for other methods (see Kruskal and Wish, 1978, p. 92, for further discussion).

on each dimension and whether or not a scholar was cited by nursing theorists were calculated. Common citations by theorists located at similar coordinates provide a criterion for determining the meaning of each dimension. This analysis proved cumbersome to interpret given the large number of scholars cited. Instead, we rst grouped scholars according to their discipline and then examined the correlations between the proportions of each nurse theorists citations from the elds of anthropology, biology, nursing, philosophy, psychology, and sociology, and the theorists coordinates on each dimension (see Table 2). This approach is a form of property vector tting (see Jones and Koehly, 1993) and is useful for interpreting MDS solutions and deciding how many dimensions to retain. These correlations suggest that as coordinates on Dimension 1 increase so do the proportions of scholars cited from the eld of psychology (r 0:779). As noted above, there were more scholars cited from this eld than any of the others examined here. The most often cited were Abraham Maslow, Carl Rogers, and Gordon Allport; all noted for their work in the area of personality theory (see Feist, 1985; Hergenhahn, 1980). Hence, we decided to examine more closely the proportion of personality psychologists cited by each nurse theorist in relation to their coordinates on Dimension 1 and found this relationship to be quite strong (r 0:916). Values on Dimension 2 appear to correspond to the proportion of biologists cited (r 0:733). Three of the seven biologists cited were referenced for their work in general systems theory; again we decided to examine more closely a subset of scholars. Coordinates on Dimension 2 were also related to the proportion of these general systems theorists cited (r 0:691). Dimension 3 seems most related to the proportion of scholars cited from philosophy (r 0:788). Closer examination of these

ARTICLE IN PRESS
118 J.W. Beckstead, L.G. Beckstead / International Journal of Nursing Studies 43 (2006) 113122
HL OR

AD PZ PP

1
RL WB

HN
TV WT RY LV RG KG NU -2 -1 JN

0
LN

OM

-1 -2

NW

PR

0 1

-1 0 Biolog y (D ) 2

2 2

Fig. 2. Three-dimensional plot of epistemic space accommodating nurse theorists based on MDS analysis of shared scholars cited from disciplines of anthropology, biology, nursing, philosophy, psychology, and sociology. Symbols corresponding to nurse theorists are identied in Table 1.

dominated by anthropologists and sociologists. The tight conguration of theorists in the central position (near coordinates 0, 0, 0) cited more scholars from a wider range of disciplines than did those located in the periphery of the space. It occurred to us that coordinate values based on patterns in citations might be associated with temporal relationships among the nurse theorists. That is, are the coordinates systematically related to the nurse theorists publication dates? Coordinate values on Dimension 1 were correlated, r 0:523; with the year in which the theorists published their initial works (see Meleis, 1997; Tomey and Alligood, 1998) suggesting that earlier nurse theorists cited a greater proportion of psychologists than did theorists publishing later in the century. Publication dates did not correlate with coordinate values on Dimension 2 or 3. Using Fig. 2 and the coordinates in Table 1, it is possible to make substantive comparisons among particular nurse theorists on the various dimensions. We invite the reader to examine coordinate proles of the nurse theorists with whom they are most familiar in order to assess the utility of our analysis.

Psychology

(D 1)

citations revealed that coordinates on Dimension 3 were consistently related to the proportion of these philosophers considered to be existentialists (r 0:751). The relationships among nurse theorists in multidimensional space are shown in Fig. 2. For ease of interpretation, the dimensions have been labeled according to their descriptions above. The initial utility of our dimensional interpretation of the MDS solution becomes apparent when the coordinates of various nurse theorists are examined. For example, nurse theorists often described as existentialphenomenologists, namely, Newman (NW), Paterson and Zderad (PZ), and Parse (PR) are located at the high end of Dimension 3 (toward the front of the gure), which is consistent with the fact that they cited a greater proportion of existential philosophers (Heidegger, Nietzsche, Sartre) than did other nurse theorists. Nurse theorists identied as systems theorists, Neuman (NU), Johnson (JN), and King (KG) are located more toward the high end of Dimension 2 (at the right-hand side of the gure) consistent with their having cited a greater proportion of scholars in the elds of biology and general systems theory (e.g., Hans Seyle, and Ludwig von Bertalanffy). Nurse theorists Hall (HL) and Orlando (OR) had the highest coordinate values on Dimension 1, consistent with the nature of their theories being inuenced by the work of personality psychologists such as Carl Rogers and Harry Stack Sullivan. At the other end of Dimension 1 we nd nurse theorist Leininger (LN), known for her transcultural perspective. Unlike Hall and Orlando, her prole of citations was

Philo

soph y

(D ) 3

4. Discussion Our original description of epistemic space as two dimensional, stemming from Descartes mindbody dualism, was supported, in part, by the alignment of MDS axes with differing degrees of citation from biology and psychology. However, our initial view may have been too simplistic. The three-dimensional model of epistemic space containing the nurse theorists that emerged from our analysis revealed the systematic inuence of scholars not identied with any branch of science per se, but with writings in existential philosophy. In an effort to integrate this nding into our initial framework, we offer the following. The accumulation of knowledge referred to as science is derived from sense experience and may thus be considered a posteriori in nature. Because sense experience is relative, inconsistent, and not fully reliable, such knowledge is not regarded as certain; rather it is probable knowledge. In contrast, the knowledge offered by philosophy is derived from the function of reason without reference to sense experience and may be considered as a priori knowledge, expressing the denitional relationships among ideas true in themselves due to their own inherent meaning. Such a priori knowledge is, by denition, orthogonal to a posteriori knowledge. To prove useful, our notion of epistemic space must encompass both these categories of knowledge. In hind sight, including this dimension seems obvious if we are to accommodate other a priori knowledge offered by the eld of mathematics.

ARTICLE IN PRESS
J.W. Beckstead, L.G. Beckstead / International Journal of Nursing Studies 43 (2006) 113122 119

The more central the coordinates of a nurse theorist in our multidimensional model, the more scholars, and the wider the range of disciplines she cited. We might describe these theorists as having a broader scope to their inspirational landscape. By contrast, theorists located more toward the periphery may be described as being more polarized or inuenced by a smaller and more homogeneous set of scholars. The themes of adaptation and wholeness, which are found in the works of various nurse theorists, may be traced to the eld of biology, and more specically, to general systems theory which has its own roots in biology. The study of adaptation to stress, or coping, was pioneered in the 1950s by Hans Seyle, the most often cited biologist in our review. General systems theory, which worked its way into many scientic disciplines from 1950 through the 1970s, grew out of the work of biologists Ludwig von Bertanalffy, Anatol Rapoport, and W. Ross Ashby who directed scientists to consider, as the proper object of study, the various arrangements of components that are so interrelated as to form wholes. The notion that the whole is greater than the sum of its parts has inuenced various nurse theorists to expound upon the dynamic context in which patient care takes place and to encourage nurses to move beyond mechanistic thinking. Based on the prevalence of their citations, humanistic personality theorists Abraham Maslow and Carl Rogers appear to have had the most inuence on nurse theorists. Personality theorists differ in their assumptions concerning human nature. According to Feist (1985) these assumptions center on components such as: free choice vs. determinism, pessimism vs. optimism, biological vs. social inuences, and the number of basic motives attributed to the person. Carl Rogers view of humanity is basically positive and optimistic. He believed that people are essentially forward moving and that under proper conditions will grow toward selfactualization. People ordinarily know what is best for themselves and have a high degree of choice in their lives. Rogers person-centered therapy is quite teleologic; people are seen as striving with purpose toward goals they freely set for themselves. Maslows view of people is that they are fundamentally good and that the basic needs that motivate us are not negative or pathological, but precisely those that result in positive growth and health. When peoples needs are not met, they may act in evil or destructive ways. He also believed that people aim toward self-actualization, thus holding a purposive or teleological view. Maslow felt that individuals are shaped by both biology and society and that the two cannot be separated. Both these psychologists shared an optimistic view of people as being capable of self-care and self-determination, given a secure, nurturing environment; themes that pervade many nursing theories.

MDS techniques have some limitations that must be kept in mind when interpreting the ndings reported above. It could be argued that our ndings are an artifact of the sample of nurse theorists chosen, the list of scholars assessed, and/or the nature of the algorithm and proximity measure used. Like any exible modeling framework or data analysis method, MDS is well suited to some types of research questions and less appropriate for others. Nonmetric methods, like the one used here for analyzing binary data are based on the assumption that a monotonic relationship exists between the proximity measures and the interpoint distances derived by the solution algorithm. Different proximity measures summarize relationships between objects in different ways and will produce different MDS results. It is important that researchers using MDS understand these measures and choose the one(s) appropriate to their research goals or questions. Prior to settling on the Russell and Rao index, we compared several proximity measures for revealing known dimensionality in small articial data sets similar in character to the one analyzed here. A reviewer of an earlier draft of this paper raised the issue of sparse data arrays in MDS analysis. Although the problems resulting from sparse data arrays have been studied in relation to logistic regression (see Greenland et al., 2000; Kuss, 2002 for discussion), MDS techniques are not adversely affected in the same way because MDS analyzes the proximity matrix calculated from the raw data array. Our 20 74 raw data array contained 319 1s and 1161 0s. The resulting 20 20 proximity matrix contained 380 proximity scores, 68 of which were 0.00 indicating no shared citations between particular pairs of nurse theorists. So long as the appropriate proximity measure is used to prepare the data for MDS analysis, sparsity in the raw data array does not present the same problem as it does in logistic regression methods. Future generations of nurse theorists, whose work will no doubt build upon the efforts of the nurse theorists examined here, are likely to incorporate and adapt new developments occurring in other elds of science; such is the nature of scientic progress. As the assemblage of knowledge comprising nursing science increases, the dimensionality of epistemic space required to encompass nursing theory may also increase. The similarities among nurse theorists in epistemic space, as examined here, were based on commonalities in the sources they cited for their intellectual inspiration. More frequently, similarities among nurse theorists have been derived through systematic content analysis of their writings (e.g., Fawcett, 1993, 2000; Meleis, 1997; Tomey and Alligood, 1998, 2002). The difference in these two methods may be illustrated with a simple example. Consider two nurse theorists who cite identical sources of intellectual inspiration. Our taxonomic approach would describe them as being quite similar

ARTICLE IN PRESS
120 J.W. Beckstead, L.G. Beckstead / International Journal of Nursing Studies 43 (2006) 113122

to one another, while a typologic (content-derived) approach could describe them as being quite similar or quite different from one another depending on how the two interpreted, adapted, or applied the same sources of inspiration. As Kuhn (1977) explained, although proponents of different theories may be looking at the same phenomena, they may focus on different components, and they may see different relations among the same components. We believe, as do many scholars, that a better understanding of any phenomenon is obtained through the use of multiple methods of measurement and analysis. Our notion of epistemic space is offered as a new schema for examining and understanding the relationships among the various elds of science. We invite the reader who is especially familiar with the works of nurse theorists examined here to explore how our coordinate system relates to their own contentderived judgments of the similarities and distinctions among nurse theorists.

Acknowledgement Dr. Beckstead is an associate professor at the University of South Florida College of Nursing. He works as a statistician/methodologist with nurse researchers and graduate students. Laura Beckstead is a graduate student at the University of Tampa Department of Nursing pursing her MSN. The authors wish to thank Drs. Mary E. Evans and Cecilia Jevitt for their help on an earlier draft of this paper, and Drs. Cecile Lengacher and Linda Moody for access to numerous publications by and about nurse theorists. We are also grateful to three anonymous reviewers for their helpful comments and suggestions.

Appendix A. : sources reviewed Abdellah, F.G., 1957. Methods of identifying covert aspects of nursing problems. Nursing Research 6, 423. Abdellah, F.G., Beland, I.L., Martin, A., Matheney, R.V., 1960. Patient-Centered Approaches to Nursing. MacMillan, New York. Chinn, P.L., Kramer, M.K., 1999. Theory and Nursing: Integrated Knowledge Development, fth ed. Mosby, St. Louis. Fawcett, J., 1993. Analysis and Evaluation of Nursing Theories. F.A. Davis Co., Philadelphia. Fawcett, J., 2000. Analysis and Evaluation of Contemporary Nursing Knowledge: Nursing Models and Theories. F.A. Davis Co., Philadelphia. Fizpatrick, J.J., Whall, A.L., 1996. Conceptual Models of Nursing: Analysis and Application, third ed. Appleton & Lange, Stamford, CN.

Hall, L.E., 1963. Center for nursing. Nursing Outlook 11, 805806. Hall, L.E., 1964. Nursing what is it? Canadian Nurse 60, 150154. Hall, L.E., 1966. Another view of nursing care and quality. In: Straub, M.K. (Eds.), Continuity of Patient Care: the Role of Nursing. Catholic University of America Press, Washington, DC, pp. 4760. Hardy, M.E. (Ed.), 1973. Theoretical Foundations for Nursing. MSS Information Corp., New York. Henderson, V., Nite, G., 1978. Principles and Practice of Nursing, sixth ed. MacMillan, New York. Johnson, D.E., 1959. A philosophy of nursing. Nursing Outlook 7, 198200. Johnson, D.E., 1959. The nature of a science of nursing. Nursing Outlook 7, 291294. Kim, H.S., Kollak, I., 1999. Nursing Theories: Conceptual and Philosophical Foundations. Springer, New York. King, I.M., 1971. Toward a Theory for Nursing: General Concepts of Human Behavior. Wiley, New York. Leininger, M., 1970. Nursing and Anthropology: Two Worlds to Blend. Wiley, New York. Leininger, M., 1976. Transcultural Healthcare Issues and Conditions. F.A. Davis Co., Philadelphia. Leininger, M., 1991. Culture Care Diversity and Universality: a Theory of Nursing. National League for Nursing, New York. Levine, M.E., 1969. Introduction to Clinical Nursing. F.A. Davis Co., Philadelphia. Levine, M.E., 1971. Renewal for Nursing. F.A. Davis Co., Philadelphia. Meleis, A.I., 1991. Theoretical Nursing: Development and Progress, second ed. J.B. Lippincott Co., Philadelphia. Moody, L., 1990. Advancing Nursing Science through Research, vol. 1. Sage, Newbury Park. Neuman, B., 1982. The Neuman Systems Model. Appleton-Century-Crofts, East Norwalk, CT. Neuman, B., Fawcett, J., 2002. The Neuman Systems Model, fourth ed. Prentice-Hall Health, Upper Saddle River, NJ. Newman, M.A., 1972. Time estimation in relation to gait tempo. Perceptual and Motor Skills 34, 359366. Nicoll, L.H., 1992. Perspectives on Nursing Theory, second ed. J.B. Lippincott Co., Philadelphia. The Nursing Theories Conference Group, 1980. Nursing Theories: the Base for Professional Nursing Practice. Prentice-Hall, Author, Englewood Cliffs, NJ. Orem, D.E., 1980. Nursing: Concepts of Practice, second ed. McGraw-Hill, New York. Orem, D.E., 1991. Nursing: Concepts of Practice, fourth ed. Mosby, St. Louis. Orlando, I.J., 1990. The Dynamic NursePatient Relationship. National League for Nursing, New York.

ARTICLE IN PRESS
J.W. Beckstead, L.G. Beckstead / International Journal of Nursing Studies 43 (2006) 113122 121

OToole, A.W., Welt, S.R., 1989. Interpersonal Theory in Nursing Practice: Selected Works of Hildegard E. Peplau. Springer, New York. Parse, R.R., 1981. Man-Living-Health: a Theory of Nursing. Wiley, New York. Parse, R.R., 1987. Nursing Science: Major Paradigms, Theories, and Critiques. W.B. Saunders Co., Philadelphia. Parker, M., 2001. Nursing Theories and Nursing Practice. F.A. Davis Co., Philadelphia. Paterson, J.G., Zderad, L.T., 1988. Humanistic Nursing. National League for Nursing, Pub. No. 412218, New York. Peplau, H., 1952. Interpersonal Relations in Nursing: a Conceptual Frame of Reference for Psychodynamic Nursing. Putnams Sons, New York. Riehl, J.P., Roy, C., 1980. Conceptual Models for Nursing Practice, second ed. Appleton-Century-Crofts, New York. Rogers, M.E., 1964. Reveille in Nursing. F.A. Davis Co., Philadelphia. Rogers, M.E., 1970. An Introduction to the Theoretical Basis of Nursing. F.A. Davis Co., Philadelphia. Roy, C., 1976. Introduction to Nursing: an Adaptation Model. Prentice-Hall, Englewood Cliffs, NJ. Roy, C., Andrews, H.A., 1999. The Roy Adaptation Model, second ed. Appleton & Lange, Stamford, CN. Roy, C., Roberts, S.L., 1981. Theory Construction in Nursing: an Adaptation Model. Prentice-Hall, Englewood Cliffs, NJ. Stevens Barnum, B.J., 1994. Nursing Theory: Analysis, Application, Evaluation, fourth ed. J.B. Lippincott Co., New York. Tomey, A.M., Alligood, M.R., 1998. Nursing Theorists and Their Work, fourth ed. Mosby, St. Louis. Tomey, A.M., Alligood, M.R., 2002. Nursing Theorists and Their Work, fth ed. Mosby, St. Louis. Travelbee, J., 1966. Interpersonal Aspects of Nursing. F.A. Davis Co., Philadelphia. Travelbee, J., 1969. Intervention in Psychiatric Nursing: Process in the One-to-One Relationship. F.A. Davis Co., Philadelphia. Watson, J., 1979. Nursing: the Philosophy and Science of Caring. Little, Brown & Co., Boston. Wiedenbach, E., 1964. Clinical Nursing: a Helping Art. Springer, New York. Wiedenbach, E., 1969. Meeting the Realities in Clinical Teaching. Springer, New York.

References
Beckstead, J.W., 2002. Using hierarchical cluster analysis in nursing research. Western Journal of Nursing Research 24 (3), 307319.

Bishop, S.M., 2002. History and philosophy of science. In: Tomey, A.M., Alligood, M.R. (Eds.), Nursing Theorists and Their Work, fth ed. Mosby, St. Louis, pp. 3241. Coury, B.G., 1987. Multidimensional scaling as a method of assessing internal conceptual models of inspection tasks. Ergonomics 30, 959973. Cox, T.F., Cox, M.A.A., 2001. Multidimensional Scaling, second ed. Chapman & Hall/CRC Press, London. Davenport, M., Studdert-Kennedy, G., 1972. The statistical analysis of aesthetic judgments: an exploration. Journal of the Royal Statistical Society 21, 324333. Donaldson, S.K., Crowley, D.M., 1978. The discipline of nursing. Nursing Outlook 26, 113120. Everett, J.E., Pecotich, A., 1993. Citation analysis mapping of journals in applied and clinical psychology. Journal of Applied Social Psychology 23 (9), 450766. Fawcett, J., 1993. Analysis and Evaluation of Nursing Theories. F.A. Davis Co., Philadelphia. Fawcett, J., 2000. Analysis and Evaluation of Contemporary Nursing Knowledge: Nursing Models and Theories. F.A. Davis Co., Philadelphia. Feist, J., 1985. Theories of Personality. CBS College Publishing, New York. Fenton, M., Pearce, P., 1988. Multidimensional scaling and tourism research. Annals of Tourism Research 15, 236254. Fitzgerald, L.F., Hubert, L.J., 1987. Multidimensional scaling: some possibilities for counseling psychology. Journal of Counseling Psychology 34, 469480. Goodhill, G.J., Simmen, M.W., Willshaw, D.J., 1995. An evaluation of multidimensional scaling for understanding brain connectivity. Philosophical Transactions of the Royal Society, Series BBiological Sciences 348, 265280. Greenacre, M.J., Underhill, L.G., 1982. Scaling a data matrix in low dimensional Euclidean space. In: Hawkings, D.M. (Ed.), Topics in Applied Multivariate Analysis. Cambridge University Press, Cambridge, pp. 183268. Greenland, S., Schwartzbaum, J.A., Finkle, W.D., 2000. Problems due to small samples and sparse data in conditional logistic regression analysis. American Journal of Epidemiology 151 (5), 531539. Hair Jr., J.F., Black, W.C., 1998. Cluster analysis. In: Grimm, L.G., Yarnold, P.R. (Eds.), Reading and Understanding more Multivariate Statistics. American Psychological Association, Washington, DC, pp. 147205. Herbert, D., 1959. Mr. Wizards Experiments for Young Scientists. Doubleday & Company Inc., Garden City, NY. Hergenhahn, B.R., 1980. An Introduction to Theories of Personality. Prentice-Hall, Englewood Cliffs, NJ. Houfek, J.F., 1992. Nurses perceptions of the dimensions of nursing care episodes. Nursing Research 41 (5), 280285. Jones, L.E., Koehly, L.M., 1993. Multidimensional scaling. In: Keren, G., Lewis, C. (Eds.), A Handbook for Data Analysis in the Behavioral Sciences: Methodological Issues. Lawrence Erlbaum Associates, Hillsdale, NJ, pp. 95163. Kim, H.S., 1989. Theoretical thinking in nursing: problems and prospects. Recent Advances in Nursing 24, 106122. Kim, H.S., 1999. Introduction. In: Kim, H.S., Kollak, I. (Eds.), Nursing Theories: Conceptual and Philosophical Foundations. Springer, New York, pp. 17. Kruskal, J.B., Wish, M., 1978. Multidimensional Scaling. Sage, Beverly Hills, CA.

ARTICLE IN PRESS
122 J.W. Beckstead, L.G. Beckstead / International Journal of Nursing Studies 43 (2006) 113122 Smith, N.J., Iles, K., 1988. A graphical depiction of multivariate similarity among sample plots. Canadian Journal of Forestry Research 18, 467472. Takane, Y., Young, F.W., de Leeuw, J., 1977. Nonmetric individual differences multidimensional scaling: an alternating least squares method with optimal scaling features. Psychometrika 42, 767. Tijssen, R.J.W., Van Rann, A.F.J., 1989. Mapping co-word structures: a comparison of multidimensional scaling and leximappe. Scientometrics 15, 283295. Tomey, A.M., Alligood, M.R., 1998. Nursing Theorists and Their Work, fourth ed. Mosby, St. Louis. Tomey, A.M., Alligood, M.R., 2002. Nursing Theorists and Their Work, fth ed. Mosby, St. Louis. Tong, S.T.Y., 1989. On nonmetric multidimensional scaling ordination and interpretation of matorral vegetation in lowland Murcia. Vegetatio 79, 6574. Torgerson, W.S., 1958. Theory and Method of Scaling. Wiley, New York. Whall, A.L., 1996. Current debates and issues critical to the discipline of nursing. In: Fitzpatrick, J.J., Whall, A.L. (Eds.), Conceptual Models of Nursing, third ed. Appleton & Lange, Stamford, CN, pp. 112. Wilson, J.M., Retsas, A.P., 1997. Personal constructs of nursing practice: a comparative analysis of three groups of Australian nurses. International Journal of Nursing Studies 34 (1), 6371. Young, G., Householder, A.S., 1938. Discussion of a set of points in terms of their mutual distances. Psychometrika 3, 1922. Kuhn, T.S., 1977. The Structure of Scientic Revolutions, second ed. University of Chicago Press, Chicago. Kuss, O., 2002. Global goodness-of-t tests in logistic regression with sparse data. Statistics in Medicine 21, 37893801. Lapointe, F.J., Legendre, P., 1994. A classication of pure malt Scotch whiskies. Applied Statistics 43, 237257. Lawson, W.J., Ogg, P.J., 1989. Analysis of phenetic relationships among populations of the avian genus Batis (Platysteirinae) by means of cluster analysis and multidimensional scaling. Biometrics Journal 31, 243254. Meara, K., Robin, F., Sireci, S.G., 2000. Using multidimensional scaling to assess the dimensionality of dichotomous item data. Multivariate Behavioral Research 35 (2), 229259. Meleis, A.I., 1997. Theoretical Nursing: Development and Progress, third ed. J.B. Lippincott Co., Philadelphia. Napier, D., 1972. Nonmetric techniques for summated ratings. In: Shepard, R.N., Romney, A.K., Nerlove, S.B. (Eds.), Multidimensional Scaling: Theory and Applications in the Behavioral Sciences, vol. 1. Seminar Press, New York, pp. 158178. Reed, P., 1995. A treatise on nursing knowledge development for the twenty-rst century: beyond postmodernism. Advances in Nursing Science 17 (3), 7084. Shepard, R.N., 1972. Introduction to volume 1. In: Shepard, R.N., Romney, A.K., Nerlove, S.B. (Eds.), Multidimensional Scaling: Theory and Applications in the Behavioral Sciences, vol. 1. Seminar Press, New York, pp. 120.