Dimensionalizing Data: Using the in Vivo,Collaborative,Proliferative,Metaphorical,and Graphic Modes of Dimensionalization as Techniques for Qualitative Analysis

Abstract

Qualitative and mixed methods researchers employ a variety of analytic techniques in their research, borrowing from a range of disciplines and methodologic traditions. Leonard Schatzman described dimensionalization as an analytic technique for use in dimensional analysis. Dimensionalization enables a researcher to expand and enhance their dimensionality, or their ability to note a code or concept’s dimensions – its attributes, context, and meaning. Despite its potential for broad use, dimensionalization has not yet been widely taken up, within or outside of its original methodological context of dimensional analysis. We describe and demonstrate five modes for operationalizing dimensionalization throughout the research process: in vivo, collaborative, proliferative, metaphorical, and graphic. For each mode, we offer concrete examples of dimensionalization in action to support researchers in implementing dimensionalization in their own projects. We discuss the contribution of the five modes of dimensionalization for use in research projects across disciplines and methodologic traditions and discuss the practical implications of utilizing each mode. Dimensionalization has the potential to be broadly applicable within and beyond interpretive qualitative methods, to innovate in naturalistic approaches to inquiry, and maximize the promise of team science.

Keywords

dimensionalization dimensional analysis data analysis interpretive methods team science

Introduction

Dimensionalization is an analytic technique originally situated within Schatzman’s method of dimensional analysis that enables a researcher to enhance their dimensionality. Dimensionality is the capacity to interpret a code or concept’s dimensions – its attributes, context, and meaning. Leonard Schatzman described dimensionalization in his 1991 exposition on the method of dimensional analysis, an Interactionist approach to research underpinned by Symbolic Interactionism (Blumer, 1986). Since its methodological debut, dimensional analysis has been taken up by researchers from a variety of disciplines to examine a range of phenomena. Additionally, subsequent methods papers have helped to clarify and expand upon the operationalization of dimensional analysis, including the use of dimensionalization (Bowers & Schatzman, 2016; Kools et al., 1996).

Dimensional analysis is an alternative approach to developing theory grounded in data, useful for investigating socially complex phenomena (Kools et al., 1996). According to the method, dimensionalization is a means of “abstracting the multiplicity of aspects seen as part of the problematic complexity” (Schatzman, 1991, p. 310), or as Kools et al. describes it:

The operation of dimensionalzing, an early analytic process, entails the designation or naming of data bits and the expansion of those data into their various attributes including dimensions and their properties… Each dimension is a component of the phenomenon under study as identified by the researcher. (Kools et al., 1996, p. 316)

Specifically in the context of dimensional analysis, dimensionalization is used to clarify dimensions that become part of the explanatory matrix, an analytic tool composed of dimensional categories including the perspective, context, condition, process, and consequence (Schatzman, 1991). It is thus the critical process for answering the question, “What all is involved here?” about a given phenomenon, the central question of a dimensional analysis. We believe that dimensionalization has the potential to be useful in the analytic process beyond its original methodologic approach.

Constant comparative technique offers an illustrative analogous case example of broadening the application of a methodologically grounded analytic technique. Constant comparison technique was originally introduced for use in Grounded Theory (Glaser, 1967) and has been taken up broadly and enthusiastically by qualitative researchers across methodological traditions (Fram, 2013). At present, constant comparison is known and used as a technique for analytic thinking that enables researchers to synthesize and interpret a range of qualitative data in iterative fashion throughout the study. Dimensionalization is similarly a tool for analysis. Researchers currently use it to gain a deeper and more complete understanding of their data and analysis. Dimensionalization helps to reveal undiscovered aspects of a phenomenon through application to a code, concept, matrix, or theory. However, in our review of the literature, its use remains circumscribed within the dimensional analysis tradition.

We argue that dimensionalization is a powerful but underused tool for analysis of qualitative data that has potential use beyond dimensional analysis. As experienced dimensional analysts, we believe that offering a menu of approaches to operationalize dimensionalization as an analytic technique will enable researchers to consider integrating it in several ways within their research design and analytic processes. In this article, we describe dimensionalizing in detail and demonstrate five modes for operationalizing dimensionalization throughout the research process both within and beyond dimensional analyses. These five modes are accompanied by a range of concrete examples of dimensionalization at work in our own research projects.

Modes of Dimensionalization

Modes of dimensionalization can be used at any point in designing qualitative or mixed methods research, as well as throughout the processes of generating and analyzing qualitative data. Based on our individual and collective experiences dimensionalizing, we see five distinct modes of dimensionalization: in vivo, collaborative, proliferative, metaphorical, and graphic. For each mode, we describe the purpose, advantages, and specific technique. Selected examples from five different projects using a variety of methodological approaches illustrate the application of each mode.

In Vivo Mode

The in vivo mode of dimensionalization entails undertaking an initial coding process to compliment fracturing data into the smallest meaning units. In practice, this looks like generating an extensive list of in vivo codes in the form of words and short phrases from textual data. Each word or phrase should represent a distinct meaning unit, determined by the researcher without regard to the participant’s storied use of it. The purpose of the in vivo mode of dimensionalization is thus to break data free from their intended context and meaning, thereby enhancing a researcher’s ability to note previously hidden dimensions of the data.

For example, C.W. used the in vivo mode of dimensionalization in a study exploring frontline care clinicians’ considerations regarding the development of machine learning-based clinical decision support tools (Whitney & Preis, 2023), using an interpretive descriptive approach (Thorne, 2016). To demonstrate the operationalization of the in vivo mode, consider the underlined data excerpt from a transcript of a focus group discussion with clinician participants:

Obviously in, in this you know specialty, there are a lot of lawsuits, alleging medical malpractice and the like. So if you have an AI or machine learning tool that indicates the likelihood of an outcome is a specific number and, for some reason, you do not follow the recommendation, or you do not explore whatever it is to its maximum conclusion, and something goes wrong – if it's all over the chart – it becomes more difficult for you to explain your clinical judgment… Sometimes providers may be a little reluctant to feel that something is pointing in a direction other than what their clinical judgment would have allowed them to do.

In the analytic phase of the study, the data were fractured into the smallest words and short phrases that had potential meaning, with the rationale that each generated candidate hypotheses, which might inspire analytic insights or questions (Table 1).

Table 1.

Illustration of the in vivo Mode.

Meaning unit	Candidate hypotheses	Analytic insight/question
Sometimes	There may be circumstances under which this will or will not occur
Providers	This may differ by the type of healthcare professional; power dynamics in the directionality of provider versus receiver of care may be at play
may be	There may be circumstances under which this will or will not occur
A little reluctant	Reluctancy has gradations and is not binary (reluctant/not reluctant)
Reluctant	Reluctant differs from other experiences such as hesitant or resistant	“Reluctant” could be dimensionalized using the proliferative mode
Feel	There may be an emotional or sensing dimension about this experience
Pointing	Pointing implies a definitive direction; pointing is less active than other verbs such as mandating or requiring	“Pointing” could be dimensionalized using proliferative mode
Direction	Direction is less specific than a place or exact location
Other than	There is a non-specific alternative
Their	There is ownership or identity affiliated with this
Clinical	There is a difference between clinical and merely logical or intellectual
Clinical judgment	There is a specific type of decision-making within the clinical realm
Judgment	There is an adjudicator whose ruling carries weight	“Judgment” could be dimensionalized using the proliferative mode
Would have	There may be circumstances under which this will or will not occur; there is an imagined alternative
Allowed	There is a dimension of gatekeeping	AI and clinical judgment are both viewed by clinicians as having a certain dimension of power, guiding, and justifying their clinical actions. How and why do clinicians internalize this power dynamic?
do	There is an action that is possible

Each meaning unit is “large” enough to generate a candidate hypothesis, and each is “small” enough so that it does not equal merely the sum of its parts. For example, while initially considered to be a meaning unit, “pointing in a direction” only generates a candidate hypothesis for the sum of its parts (“pointing” and “direction”) rather than generating a distinct candidate hypothesis. This contrasts with “clinical judgment,” which proved to be greater than the sum of its parts “clinical” and “judgment,” with all three meaning units generating their own candidate hypotheses. Four analytic insights were generated from the in vivo dimensionalizing of the underlined section of data, providing new directions for abductive inquiry. Initiating the coding process with the in vivo mode of dimensionalization enables the researcher to engage with a more comprehensive set of dimensions from which they can generate interpretive findings.

Collaborative Mode

The collaborative mode of dimensionalization involves leveraging multiple perspectives to expose previously undiscovered dimensions of the data. In practice, the process of engaging a variety of individuals in any analytic activity integrates and transcends multiple individual perspectives to enhance researchers’ capacity to note the attributes, context, and meaning of the data. In contrast to a lens which privileges consistency and intercoder reliability, the collaborative mode of dimensionalization prioritizes diversity of thought through dialogic engagement with multiple individuals who have a range of expertise and familiarity with the data and phenomenon of interest. The purpose of this mode is to uncover dimensions of data not previously recognized by a single researcher or small team, amplifying researcher dimensionality by integrating many potentially differing perspectives.

For example, we used the collaborative mode of dimensionalization in the Virtual Supportive Cancer Care Research (ViSuCaRe) Study, a dimensional analysis of clinicians’ and patients’ experiences of virtual supportive cancer care during the early COVID-19 pandemic (Evered et al., 2022, 2023). To operationalize this mode, we formulated a 11-person data engagement team including academic nurse researchers, undergraduate and doctoral nursing students, and a clinician and administrator. We partnered with this group throughout the entirety of the study in dialogic engagement sessions, which served to expand the realm of analytic possibilities within the context of what could be interpreted from the data (Table 2).

Table 2.

Team Composition for the ViSuCaRe Study.

Role	n	Potential perspectives informing analysis
Clinician/administrator	1	Clinical and administrative realities of cancer care delivery
Academic nurse researchers	2	Epistemological, ontological, and methodological underpinnings of the approach to research; content expertise
Undergraduate nursing students	4	Penchant for learning; more limited preconceptions or assumptions about the phenomenon of inquiry
Doctoral nursing students	4	Proclivity for inquiry; theoretical flexibility

The academic nurse researchers, leveraging their methodological expertise, led the production of an explanatory matrix, in accordance with the method of dimensional analysis. In developing the final explanation of the data, however, other members of the data engagement team expanded and refined the matrix, based on their clinical expertise and theoretical flexibility. Specifically, the clinician sub-dimensionalized “the experience of cancer care” into “the pathological presence of cancer,” “cancer treatment,” and “treatment effects” to better capture the clinical realities of engaging in cancer care. Additionally, a doctoral nursing student identified a dimension of the data not initially elucidated through the application of the explanatory matrix, expanding the theoretical explanation of the phenomenon of inquiry. Engaging a large and diverse data engagement team using the collaborative mode of dimensionalization enables researchers to consider more fully “What all is involved?” in the phenomenon under study through the data.

Proliferative Mode

Using the proliferative mode of dimensionalization, a researcher generates a list of dimensions for a given code or concept. In practice, this involves listing the synonyms, antonyms, attributes, homonyms, associations, uses, and characteristics of the code or concept of focus, without regard to its intended context or meaning. One way to operationalize this mode is to set a timer for several minutes, and with a partner or data engagement team, engage in an uninhibited and unfiltered brainstorm aloud, while recording each contributed dimension. While the embodied experience of the proliferative mode of dimensionalization resembles that of a free association of words, the mode is in fact a targeted and intentional generation of dimensions of the selected code or concept. After generating this list of dimensions, individuals reflect together on any new analytic insights, questions, or directions of inquiry sparked by the process, experience, and produced dimensions. The purpose of the proliferative mode is to “explode” the code or concept by expanding the set of dimensions considered by the researcher. This mode enables researchers to gain a more holistic understanding of a code or concept through data expansion.

For example, J.E. used proliferative mode in a mixed methods narrative elicitation (Grob et al., 2019) exploring people’s emotional experiences while hospitalized. To demonstrate its operationalization in developing data collection instruments, we consider a case example: In developing the closed-ended survey and narrative elicitation questions, the research team debated how to ask respondents about the phenomenon of interest. The two options under consideration for the question, “Thinking about the things that mattered most to you about your _______ during your hospital stay, how would you say things went?” were “emotional experiences” and “feelings.” The research team aimed to determine whether and how the two concepts differed meaningfully to capture the intended dimensions of the phenomenon in the finalized question. Table 3, constructed from a 3-min dimensionalization session through dialogic engagement, illustrates how the proliferative mode helps to achieve this aim.

Table 3.

Illustration of the Proliferative Mode.

	Emotional experiences	Feelings
Proliferative mode	Emotional breakdownEmotional supportLife experienceYears of experienceMemoryTraumatic experienceTraumaShe’s emotionalHysteriaLabileTherapyEmotional regulationEmotional dysregulationLogicalRationalReasonableUnreasonableWiseProvokedPremenstrualStigmatizedExperience of a lifetimeEmotional highEmotional lowEmotional rollercoasterBeing emotionalIn a stateUnaffectedFlatNever experienced	EmotionalInside out unfeelingSensingHurt feelingsAmbiguous feelingsMixed feelingsColdWarmNumbBig little feelingsName your feelingsFeel your feelingsFeelFeltShareShared feelingsAmbivalentMutual feelingsReciprocated feelingsRomantic feelingsAngerSadnessJoyGratefulnessWorryComplicatedUncomplicatedDraw your feelingsOwn your feelingsSomaticShut down out of bodyEmbodiedReaction
Analytic insights	Emotional experiences are bounded and discreet – a period marked by an emotion rather than defined by the feeling of an emotion (e.g., “life experience,” “in a state”)Experiences are often circumscribed by others’ perceptions, thus describing external rather than internal states or periods (e.g., “she’s emotional,” “being emotional”)The associated vocabulary with emotional experiences can be technical or clinical in nature (e.g., “emotional dysregulation,” “emotional support”)Emotional experiences can be extreme, replete with dichotomies and highs and lows (e.g., “provoked/unprovoked,” “emotional rollercoaster”)	Feelings are often personal, belonging to the feeler (e.g., “feel your feelings,” “embodied”)Feelings are defined by what is felt (e.g., “sadness,” “worry”)Some feelings are relational, felt in interaction with others (e.g., “mutual feelings,” “romantic feelings”)Multiple feelings can be experienced at once in conflicting and synergistic ways (e.g., “complicated,” “mixed feelings”)

In this case, use of the proliferative mode of dimensionalization revealed a few key analytic insights about “emotional experiences” and “feelings.” Whereas the dimensions of “feelings” suggest an innate, flexible, and embodied experience, the dimensions of “emotional experiences” are more cognitive and bounded by time. Given that each construct offers distinct dimensions, the research team elected to use both in their data collection instrument. Dimensionalization of potentially different constructs helps to distinguish between them and determine which dimensions are overlapping or distinct to one construct. This enables a research team to select the most fitting construct or concept for the data collection instrument. The proliferative mode of dimensionalization is widely applicable across aspects of design, data collection, and data analysis for a range of research methodologies beyond Interactionism, including mixed methods research.

Metaphorical Mode

Using the metaphorical mode of dimensionalization involves the intentional use of metaphor in data analysis to bring to light dimensions of a phenomenon by assessing the fit – the suitability and limitations – of the metaphor when applied to the data. In practice, this involves selecting a metaphor, or comparative analogue, which shares at least some dimensions with preliminary interpretation of the data. Once selected, the researcher notes whether and how the metaphor fits the dimensions inductively identified from the data. Importantly, the metaphorical mode is useful for noting where a metaphor is a suitable fit as well as where its limitations highlight dimensions of the data not previously identified.

For example, we used the metaphorical mode of dimensionalization in a dimensional analysis of cancer care partners’ experiences navigating ethical and moral challenges (Whitney et al., 2024). In constructing the explanatory matrix, we applied the metaphor of “healthcare team as car mechanic.” This example emerged from an in vivo metaphor one participant used to describe his experience caring for his spouse. We applied this metaphor because there were shared dimensions between the scenario of interacting with a car mechanic and our preliminary interpretations of the data pertaining to care partners’ interactions with the healthcare team. Specifically, they both revolve around something or someone to be assessed and fixed (car; patient), they both involved three parallel subjects (car–patient, mechanic–healthcare professional, and car owner–care partner), and they both depend on understanding a new language with its own specialized jargon (mechanical terminology; medical terminology). However, the metaphor incompletely described healthcare team interactions because it did not include the critical inductively identified dimension of shared decision-making. This insight prompted us to ask the analytic question of the data, “What are the conditions under which care partners’ interactions with healthcare teams occur without shared decision-making, if at all?” When we returned to the data, we identified two new dimensions that explained the instances where participants sought interactions with healthcare teams (inclusive of shared decision-making) but were unsuccessful. The metaphorical mode allows researchers to elicit additional dimensions of the data, strengthening the interpretive potential of analysis.

Graphic Mode

The graphic mode of dimensionalization leverages visual representations in analysis to construct and reconstruct the dimensions of the data. In practice, this involves using drawings, diagrams, or other visuals to provoke new analytic insights or questions about the dimensions of the data. Once rendered visually, data interpretations may be expanded, clarified, revised, or refined based on what is learned about previously and newly identified dimensions. The purpose of the graphic mode of dimensionalization is to break the researcher free from the limitations of verbal and textual data.

For example, C.W. used the graphic mode in a dimensional analysis of perinatal health clinicians’ moral considerations in the care of substance-exposed dyads (Whitney, 2023a, 2023b). A graphic memo that visually represented a clinician making decisions for clinical care based on social maxims (Figure 1) motivated the analytic question, “How do clinicians make decisions in the absence of social maxims?” This analytic question was inspired by the artistic question, “In the absence of social maxims, does the clinician look off into the distance, close their eyes, or look toward something else?” Further targeted analysis of the data revealed that clinicians’ perspectives were blocked by their own moral inclinations. This was constructed in a new graphic memo, in which the clinician is blocking their view with their own hand (Figure 2). Thus, the graphic mode of dimensionalization revealed further dimensions for the researcher to consider in analysis of the data by bringing to light graphic dimensions. In this way, the graphic mode of dimensionalization allows researchers to illuminate the dimensions of the data more fully.

Figure 1.

First illustration of graphic mode.

Figure 2.

Second illustration of graphic mode.

Discussion

This article contributes the description and demonstration of the in vivo, collaborative, proliferative, metaphorical, and graphic modes of dimensionalization. Taken together, each mode offers researchers an opportunity and means to enhance their dimensionality, or their ability to note a code or concept’s dimensions (Schatzman, 1991). Importantly, dimensionalization is an approach that is accessible and useful for any member of a research team, regardless of level of experience or content expertise.

Though we have described the five modes of dimensionalization as distinct, they are in practice best leveraged in combination across the life cycle of a research project. For example, codes generated in the in vivo mode may become candidates for dimensionalizing in the metaphorical or graphic modes. The proliferative mode is often enhanced when used in conjunction with the collaborative mode to expand dimensionality through diversity of thought. In addition, these modes of dimensionalization can be flexibly applied to many phases of the research process. For example, the proliferative mode can be used to help researchers shape their inclusion and exclusion criteria by clarifying the concepts of focus for a given study. Whether applying the technique using an Interactionist method, phenomenology, or thematic analysis, any approach to dimensionalizing may be most useful when beginning formal analysis of the data or any time a researcher finds themself analytically “stuck”: confused, stagnant, or having prematurely committed to a particular interpretation.

As with all analytic techniques, there are limitations and boundaries to implementing the modes of dimensionalization. The in vivo mode may be challenging to take up, given its approach differs from more conventional approaches to coding that retain the original context and meaning of an excerpt of data. Considering financial, resource, and practical constraints as well as academic and authorship norms, large data engagement teams may be difficult to assemble and sustain, constraining the applicability of the collaborative mode of dimensionalization. Similarly, the proliferative mode can be difficult to effectively implement independently, and groups can be challenged by hegemonic research norms including membership requirements for a research team. Researchers must be mindful when using the metaphorical mode not to prematurely adopt a deductive framework that has an insufficient fit to the data. Finally, some researchers may not find the graphic mode of dimensionalization to be particularly useful, depending on their inclination toward visual representation. We also recognize that the examples described in this article are limited in that they are not contextualized within the full interpretation of the data from which they were selected.

In offering a robust description of and the modes for dimensionalization, we hope researchers across disciplines and methodological traditions will consider using dimensionalization in ways that enhance their inquiry. Borrowing this technique from dimensional analysis offers utility for many. Research using dimensionalization should explicitly reference the technique in dissemination to encourage the interrogation and clarification of the modes described here and those that may evolve in the future.

As described by Schatzman, dimensionalization is a technique available to researchers to enhance their dimensionality (1991). In this paper, we expand on Schatzman’s description by introducing five distinct modes of operationalizing dimensionalization. These modes enable researchers to note the multiplicity and complexity of data in ways that support their ability to make meaning in their analysis. Dimensionalization has the potential to be broadly applicable within and beyond dimensional analysis, to innovate naturalistic approaches to inquiry, and maximize the promise of team science.

Footnotes

Acknowledgments

We would like to thank Dr Sarah H. Kagan and Dr Isha Dhingra for their contributions refining this manuscript. We are also grateful to our data engagement team members and the research participants in our studies.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Clare Whitney

References

Blumer

(1986). Symbolic interactionism: Perspective and method . Univ of California Press.

Bowers

Schatzman

(2016). Dimensional analysis. In Developing grounded theory (pp. 86–126). Routledge.

Evered

Andersen

Foxwell

Iroegbu

Whitney

(2022). The impact and implications of virtual supportive cancer care during the COVID-19 pandemic: Integrating patient and clinician perspectives. Supportive Care in Cancer, 30(12), 9945–9952. https://doi.org/10.1007/s00520-022-07393-9

Evered

Andersen

Foxwell

Iroegbu

Whitney

(2023). Clinician–patient relationships in virtual care: A dimensional analysis of the symbolic world of cancer care. Qualitative Health Research, 33(4), 284–296. https://doi.org/10.1177/10497323231153793

Fram

S. M.

(2013). The constant comparative analysis method outside of grounded theory (p. 18). Qualitative Report.

Glaser

(1967). Strategies for qualitative research. The discovery of grounded theory, 2–6.

Grob

Schlesinger

Barre

L. R.

Bardach

Lagu

Shaller

Parker

A. M.

Martino

S. C.

Finucane

M. L.

Cerully

J. L.

Palimaru

(2019). What words convey: The potential for patient narratives to inform quality improvement. The Milbank Quarterly, 97(1), 176–227. https://doi.org/10.1111/1468-0009.12374

Kools

McCarthy

Durham

Robrecht

(1996). Dimensional analysis: Broadening the conception of grounded theory. Qualitative Health Research, 6(3), 312–330. https://doi.org/10.1177/1049732396006

Schatzman

(1991). Dimensional analysis: Notes on an alternative approach to the grounding of theory in qualitative research. Social Organization and Social Process: Essays in honor of Anselm Strauss, 303(1), e314.

10.

Thorne

(2016). Interpretive description: Qualitative research for applied practice . Routledge.

11.

Whitney

(2023a). Moralization & optimization in perinatal health: A discovery mode-to- emergent fit dimensional analysis. In Eastern nursing research society 35^th annual scientific sessions.

12.

Whitney

(2023b). Accessible analysis and diversified dissemination: Using graphic medicine in qualitative nursing research. In CANS advanced methods conference 2023. Virtual.

13.

Whitney

Evered

Larson

Pyo

Lee

(2024). You haven’t studied for this”: How individuals navigate moral and ethical challenges in their role as cancer care partners. In American society for bioethics & humanities 2024 conference.

14.

Whitney

Preis

(2023). Anticipatory moral distress: Unexamined barriers to acceptability of clinical decision support tools in perinatal health. In Society for the social studies of science 2023 meeting (4S 2023).