'Qualitative' and 'quantitative' methods and approaches across subject fields: implications for research values, assumptions, and practices

Characteristically, presentations in research methods textbooks distinguish postivist and interpretivist approaches or paradigms (e.g. Guba and Lincoln 1994; Howe 1988; Denzin and Lincoln 2011) or ‘two cultures’ (Goertz and Mahoney 2012) with associated debates or ‘wars’ (e.g. Creswell 1995; Morse 1991). Quantitative data are shown as ‘numbers’ gathered through experiments (Moore 2006) or mathematical models (Denzin and Lincoln 1998), whereas qualitative data are usually words or texts (Punch 2005; Goertz and Mahoney 2012), characteristically gathered through interviews or life stories (Denzin and Lincoln 2011). Regarding analysis, some sources claim that establishing objective causal relationships is key in quantitative analysis (e.g. Goertz and Mahoney 2012) whereas qualitative analysis uses more discursive and interpretative procedures.

Thus, much literature presents research in terms of two generally distinct methods—quantitative and qualitative—which many students are taught in research methods courses. The binary divide may seem to be legitimated in the titles of many academic journals. This division prevails as designated strands of separated research methods in courses which apparently handle both (cf. Onwuegbuzie and Leech 2005). Consequently, students may follow this seemingly stereotyped binary view or feel uncomfortable to deviate from it. Arguably, PhD candidates need to demonstrate understanding of such concepts and procedures in a viva—or risk failure (cf. Trafford and Leshem 2002). The Cambridge Dictionary defines ‘quality’ as “how good or bad something is”; while ‘quantity' is “the amount or number of something, especially that can be measured” (Cambridge 2022). But definitions of ‘Qualitative' can be elusive, since “a precise definition of qualitative research, and specifically… its distinctive feature of being “qualitative”, the literature is meager” (Aspers and Corte 2019, p.139). Some observe a “paradox… that researchers act as if they know what it is, but they cannot formulate a definition” and that “there is no consensus about specific qualitative methods nor… data” (Aspers and Corte 2019, p40). In general, ‘qualitative research’ is an iterative process to discover more about a phenomenon (ibid.). Elsewhere, 'qualitative’ is defined negatively: "It is research that does not use numbers” (Seale 1999b, p.119). But this oversimplifies and hides possible disciplinary variation. For example, when investigating criminal action, numeric information (quantity) always follows an interpretation (De Gregorio 2014), and consequently this is a quantity of a quality (cf. Uher 2022).

Indeed, many authorities note the presence of elements of one in the other. For example, in analysis specifically, that what are considered to be quantitative elements such as statistics are used in qualitative analysis (Miles and Huberman 1994). More generically, that “a qualitative dimension is present in quantitative work as well” (Aspers and Corte 2019, p.139). In ‘mixed methods’ research (cf. Tashakkori et al. 1998; Johnson et al. 2007; Teddlie and Tashakkori 2011) many researchers ‘mix’ the two approaches (Seale 1999a; Mason 2006; Dawson 2019), either using multiple methods concurrently, or doing so sequentially. Mixed method research logically depends on prior understandings of quantitative and qualitative concepts but this is not always obvious (e.g. De Gregorio 2014); for instance Heyvaert et al. (2013) define mixed methods as combining quantitative and qualitative items, but these key terms are left undefined. Some commentators characterize such mixing as a skin, not a sweater to be changed every day (Marsh and Furlong 2002, cited in Grix 2004). In some disciplines, these terms are often blurred, interchanged or conjoined. In sociology, for instance, “any quality can be quantified. Any quantity is a quality of a social context, quantity versus quality is therefore not a separation” (Hanson 2008, p.102) and characterizing quantitative as ‘objective’ and qualitative as ‘subjective’ is held to be false when seeking triangulation (Hanson 2008). Additionally, approaches to measuring and generating quantitative numerical information can differ in social sciences compared to physics (Uher 2022). Indeed, quantity may consist of ‘a multitude’ of divisible aspects and a ‘magnitude’ for indivisible aspects (Uher 2022). Notably, “the terms ‘measurement’ and ‘quantification’ have different meanings and are therefore prone to jingle-jangle fallacies” (Uher 2022) where individuals use the same words to denote different understandings (cf. Bakhtin 1986). Comparatively, the words ‘unit’ and ‘scale’ are multitudinous in different sciences, and the key principles of numerical traceability and data generation traceability arguably need to be applied more to social sciences and psychology (Uher 2022). The interdependence of the terms means any quantity is grounded in a quality of something, even if the inverse does not always apply (Uher 2022).

The present paper compares representations found in research methods textbooks with the reported practices of established researchers given in semi-structured interviews. The differences revealed between what the literature review of methods texts showed and what the interview study showed both underlines and extends this complexity, with implications for how such methodologies are approached and taught. The interview study data (analysed below) show that many participant researchers in disciplines commonly located within an ostensibly ‘positivist’ scientific tradition (e.g. chemistry) are, in fact, using qualitative methods as scientific procedures (contra Tashakkori et al 1998; Guba and Lincoln 1994; Howe 1988; Lincoln and Guba 1985; Teddlie and Tashakkori 2011; Creswell 1995; Morse 1991). These interview study data also show that many participant researchers use what they describe as qualitative approaches to provide initial measurements (geotechnics; chemistry) of phenomena before later using quantitative procedures to measure the quantity of a quality (cf. Uher 2022). Some participant researchers also say they use quantitative procedures to reveal data for which they subsequently use qualitative approaches to interpret and understand (biology; dendrology) through their creative imaginations or experience (contra e.g. Hammersley, 2013). Participant researchers in ostensibly ‘positivist’ areas describe themselves as doubting ‘facts’ measured by machines programmed by humans (thus showing they feel researchers are not outside the world looking in (contra. e.g. Punch 2005)) or doubting the certainty of quantitative data over time (contra e.g. Punch 2005). Critically, the interview study data show that these participant researchers often engage in debate over what a ‘number’ is and the extent to which ‘numbers’ can be considered ‘quantitative’. For example the data show how a mathematician considers that many individuals do not know what they mean by the word ‘quantitative’, and an engineer interprets any numbers involving human judgements as ‘qualitative’. Further, both a chemist and a geotechnician routinely define and use ‘qualitative’ methods and analysis to arrive at numerical values (contra. Davies and Hughes 2014; Denzin and Lincoln 2011).

Such data refute many textbook and key source representations of quantitative and qualitative as being binary and separately ringfenced entities as shown in the literature review study below (contra e.g. Punch 2005; Goertz and Mahoney 2012). Nevertheless, they resonate with much recent and current literature in the field (e.g. Uher 2022; De Gregorio 2014). They also arguably extend the complexities of the terms and approaches. In some disciplines, these participant researchers only do a particular type of research and never need anything other than clear ‘quantitative’ definitions (Mathematics), and some only ever conduct research involving text and never numbers (Literature). Moreover, some participant researchers consider certain aspects lie outside the ‘qualitative’ or ‘quantitative’ (the theoretical in German Literature), or do research which they maintain does not contain ‘knowledge’ (Fine-Art Sculpture), while others outline how they feel they do foundational conceptual research which they believe comes at a stage before any quantity or quality can be assessed (Philosophy). Indeed, of the 31 participant researchers we spoke to, nine of them considered the terms ‘quantitative’ and ‘qualitative’ to be of little relevance for their subject.

This paper reports and discusses findings from a constructivist grounded approach interview study that interviewed experienced participant researchers (N = 31) in various disciplines (see Table 1 below) about their understandings of ‘qualitative’ and ‘quantitative’ in their subject areas. Findings from this interview study were compared with findings from a research methods literature review study that revealed many disparities with received and often binary presentations of the concepts in much key literature that informs student research methods courses. In this section we outline the review criteria, the method of analysis, and our findings. The findings are grouped according to how the sources reviewed consider ‘quantitative’ and ‘qualitative’ approaches the aspects of positivism and constructivism; the nature of research questions; research methods; analysis; issues of reliability, validity and generalizability; and the value and worth of the different approaches. Following this. We outline the approach, method, and procedure adopted for the interviews with research participants; sampling and saturation; and analysis; beside details of the participant researchers. Subsequently, Theme 2 focuses on contrasts of the interview data with ‘binary’ textbook and key source representations. Theme 3 focuses on what the interview data show about participant researcher perceptions of the value of ‘quantitative’ and ‘qualitative’ methods and approaches. This section outlines where, how, and sometimes why, participant researchers considered ‘quantitative’ and ‘qualitative’ methods approaches to be (or to not be) useful to them. These interview study findings show a surprising range of understandings, usage, and often perceived irrelevance of the terms. In the Discussion section, these findings form the focus of comparison with the literature as well as a consideration of possible implications for approaching and teaching research methods. In the conclusion we summarise the implications for research methods courses, for researchers in different disciplines and interdisciplinary contexts and discuss limitations and suggest future research. Besides adding to the debate on how ‘quantitative’ and ‘qualitative’ are conceptualized and how they are related, the paper appeals to those delivering research methods courses and to novice researchers to consider the concepts as highly complex and overlapping, to loosen constraints, and elaborate nuances of the commonplace binary representations of the terms.

To identify how concepts are presented in key materials we undertook a literature review study by consulting research methods course reading lists, library search engines, physically available shelves in institutional libraries, and Google Scholar. We wanted to encompass textbooks and some key texts which are recommended to UG, PG Masters and PhD students., for example, ‘textbooks’ like ‘Doing Your Research Project: A Guide for first-time researchers’ (Bell and Waters 2014) and ‘Introduction to Research Methods: A Practical Guide for Anyone Undertaking a Research project (5th Edition)’ (Dawson 2019). Such sources were frequently mentioned on reading lists and are freely available in many institutional libraries. We consulted seminal thinkers who have published widely on research methods, such as Denzin and Lincoln, or Cresswell, but we also considered texts which are likely less known such as ‘A tale of two cultures’ (Goertz and Mahoney 2012) and key articles such as ‘Five misunderstandings about case-study research’ (Flyvbjerg 2006). Students can freely find such sources, and are easily directed to them by supervisors. Although a more comprehensively robust search is possible, we nevertheless followed procedures and standard criteria for literature reviews (Atkinson et al. 2015).

Overall, related to what quantitative and qualitative approaches, methods and analysis are, we found many key, often binary representations in this literature review. We outline these here below.

Firstly, 20 of the sources we reviewed stated that quantitative is considered positivist, and qualitative constructivist (e.g. Tashakkori et al 1998; Guba and Lincoln 1994; Howe 1988; Lincoln and Guba 1985; Teddlie and Tashakkori 2011; Creswell 1995; Morse 1991). Even if not everyone doing quantitative research (e.g. in sociology) consider themselves positivists (Marsh 1979), it is generally held quantitative research is positivist. Here, 12 of the sources noted that quantitative is considered ‘scientific’, situating observers outside the world looking in, e.g. through gathering numerical data (Punch 2005; Davis and Hughes 2014) whereas qualitative “locates the observer in the world” (Denzin and Lincoln 2011, p.3). Quantitative researchers “collect facts and study the relationship of one set of facts to another”, whereas qualitative researchers “doubt whether social ‘facts’ exist and question whether a ‘scientific’ approach can be used when dealing with human beings” (Bell and Waters 2014, p. 9).

Secondly, regarding research questions, “qualitative research… typically has… questions and methods… more general at the start, and… more focused as the study progresses” (Punch 2005, p.28). In contrast, quantitative research uses “numerical data and typically… structured and predetermined research questions, conceptual frameworks and designs” (Punch 2005, p.28). Of the sources we reviewed, 16 made such assertions. This understanding relates to type, and nature, of data, which is in turn anchored to particular worldviews. Punch (2005, p 3–4) writes of how “in teaching about research, I find it useful to approach the qualitative-quantitative distinction primarily through…. the nature of the data. Later, the distinction can be broadened to include …. ways of conceptualising the reality being studied, and methods.” Here, the nature of data influences approach: numbers are for quantitative, and not-numbers (commonly words) for qualitative. Similarly, for Miles et al. (2018) “the nature of qualitative data” is “primarily on data in the form of words, that is, language in the form of extended text” (Miles et al. 2018, no page). These understandings in turn relate to methods used.

Such understandings often relate to analysis, and 16 of the sources we reviewed presented quantitative analysis as being statistical and number related, and qualitative analysis as being text based. With quantitative methods, “the data is subjected to statistical analysis, using techniques… likely to produce quantified, and, if possible, generalizable conclusions” (Bell and Waters 2014, p.281). With qualitative research, however, this “calls for advanced skills in data management and text-driven creativity during the analysis and write-up” (Davies and Hughes 2014). Again, the data’s nature is key, and whilst qualitative analysis may condense data, it does not seek numbers. Indeed, “by data condensation, we do not necessarily mean quantification”, however, “occasionally, it may be helpful to convert the data into magnitudes… but this is not always necessary” (Miles et al. 2018, npg). Qualitative analysis may involve stages such as assigning codes, subsequently sorting and sifting them, isolating patterns, then gradually refining any assertions made and comparing them to other literature (Miles et al. 2018). This could involve condensing, displaying, then drawing conclusions from the data (Miles et al. 2018). In this respect, some sources consider qualitative and quantitative analysis broadly similar in overall goals, yet different because quantitative analyses use “well-defined, familiar methods; are guided by canons; and are usually more sequential than iterative or cyclical” (Miles et al. 2018, npg). In contrast, “qualitative researchers are… more fluid and… humanistic” in meaning making (Miles et al. 2018, npg). Here, both approaches seek causation and may attempt to reveal ‘cause and effect’ but qualitative approaches often seek multiple and interacting influences, and effects and are less rigid (Miles et al. 2018). In quantitative inquiry search for causation relates to “causal mechanisms (i.e. how did X cause Y)” whereas in “the human sciences, this distinction relates to causal effects (i.e. whether X causes Y)” (Teddlie and Tashakkori 2011, p.286). Similarly, that “scientific research in any area… seeks to trace out cause-effect relationships” (Punch 2005, p.78). In contrast, qualitative research seeks interpretative understandings of human behaviour, “not ‘caused’ in any mechanical way, but… continually constructed and reconstructed” (Punch 2005, p.126).

Regarding reliability, validity and generalizability, 19 of the sources we reviewed presented ideas along the lines that quantitative research is understood to seek large numbers, so quantitative researchers, “use techniques… likely to produce quantified and, if possible, generalizable conclusions (Bell and Waters 2014, p.9). This means quantitative “research researches many more people” (Dawson 2019, npg). Given quantitative researchers aim, “to discover answers to questions through the application of scientific procedures” it is anticipated these procedures will “increase the likelihood that the information… will be reliable and unbiased” (Davies and Hughes 2014, p.9). Conversely, qualitative researchers are considered “more concerned to understand individuals’ perceptions of the world” (Bell and Waters 2014, p.281) and consequently aim for in-depth data with smaller numbers, “as it is attitudes, behaviour and experiences that are important” (Dawson 2019, npg). Consequently, generalizability of data is not key, as qualitative research has its “emphasis on a specific case, a focused and bounded phenomenon embedded in its context” (Miles et al. 2018, npg). Yet, such research is considered generalizable in theoretical insight if not actual data (Flyvbjerg 2006).

Regarding ‘value’ and ‘worth’, many see this related with appropriacy to the question being researched. Thus, if questions involve more quantitative approaches, then these are of value, and if more qualitative, then these are of value, and 6 of the sources we reviewed presented these views (e.g. Bell and Waters 2014; Punch 2005; Dawson 2019). This resonates with disciplinary orientations where choices between given approaches are valued more in specific disciplines. History and Anthropology are seen more qualitative, whereas Economics and Epidemiology may be more quantitative (Kumar 1996). Qualitative approaches are valuable to study human behaviour and reveal in-depth pictures of peoples’ lived experience (e.g. Denzin and Lincoln 2011; Miles et al. 2018). Many consider there to be no real inherent superiority for one approach over another, and “asking whether quantitative or qualitative research is superior to the other is not a useful question” (Goertz and Mahoney 2012, p.2).

Nevertheless, some give higher pragmatic value to quantitative research for studying individuals and people; neoliberal governments consistently value quantitative over qualitative research (Barone 2007; Bloch 2004; St Pierre 2004). Concomitantly, data produced by qualitative research is criticised by quantitative proponents “because of their problematic generalizability” (Bloor and Wood 2006, p.179). However, other studies find quantitative researchers see qualitative methods and approaches positively (Pilcher and Cortazzi 2016). Some even question the qualitative/quantitative divide, and suggest “a more subtle and realistic set of distinctions that capture variation in research practice better” (Hammersley 2013, p.99).

The above literature review study of key texts is hardly exhaustive, but shows a general outline of the binary divisions and categorizations that exist in many sources students and newer researchers encounter. Thus, despite the complex and blurred picture as outlined in the introduction above, many key texts students consult and that inform research methods courses often present a binary understanding that quantitative is positivist, focused on determining cause and effect, numerical or magnitude focused, uses experiments, and is grounded in an understanding the world can be observed from the outside in. Conversely, qualitative tends to be constructivist, focused on determining why events occur, is word or textual based (even if these elements are measured by their magnitude in a number or numerical format) and grounded in understanding the researcher is part of the world. The sciences and areas such as economics are said to tend towards the quantitative, and areas such as history and anthropology towards the qualitative.

We also note that in our literature review study we focused on English language textbooks, but we also looked at outline details, descriptions, and contents lists of texts in the languages of German, Spanish and French. We find that these broadly confirm the perception of a division between quantitative and qualitative research, and we detail a number of these in Appendix 1. These examples are all research methods handbooks and student guides intended for under and post-graduates in social sciences and humanities; many are inter-disciplinary but some are more specifically books devoted to psychology, health care, education, politics, and management. Among the textbooks and handbooks examined in other languages, more recent books pay attention to online research and uses of the internet, social media and sometimes to big data and software for data analysis.

In these sources in languages other than English we find massive predominance of two (quantitative/qualitative) or three approaches (mixed). These are invariably introduced and examined with related theories, examples and cases in exactly that order: quantitative; qualitative; mixed. Here there is perhaps the unexamined implication that this is a historical order of research method development and also of acceptability of use (depending on research purposes). Notably, Molina Marin (2020) is oriented to Latin America and makes the point that most European writing about research methods is in English or German, while there are far fewer publications in Spanish and few with Latin American contextual relevance, which may limit epistemological perspectives. This point is evident in French and Spanish publications (much less the case in German) where bibliographic details seem dominated by English language publications (or translations from them). We now turn to outline our interview study.

We approached our interview study from a constructivist standpoint of exploring and investigating different subject specialists’ understandings of quantitative and qualitative. Critically, we were guided by the key constructivist tenet that knowledge is not independent of subjects seeking it (Olssen 1996), nor of subjects using it. Extending from this we considered interviews more appropriate than narratives or focus groups. Given the exploratory nature of our study, we considered interviews most suited as we wanted to have a free dialogue (cf. Bakhtin 1981) regarding how the terms are understood in their subject contexts as opposed to their neutral dictionary definitions (Bakhtin 1986), and not to focus on a specific point with many individuals. Specifically, we used ‘semi’-structured interviews. ‘Semi’ can mean both ‘half in quantity or value’ but also ‘to some extent: partly: incompletely’ (e.g. Merriam Webster 2022). Our interviews, following our constructionist and exploratory approach, aligned with the latter definition (see Appendix 2 for the Interview study schedule). This loose ‘semi’ structure was deliberately designed to (and did) lead to interviews directed by the participants, who themselves often specifically asked what was meant by the questions. This created a highly technical dialogue (Buber, 1947) focused on the subject.

Our sampling combined purposive and snowball sampling (Sharma 2017; Levitt et al. 2018). Initially, participants were purposively identified by subject given the project sought to understand different subject perspectives of ‘qualitative’ and ‘quantitative.’ Later, a combined purposive and snowball sampling technique was used whereby participants interviewed were asked if they knew others teaching particular subjects. Regarding priorities for participant eligibility, this was done according to subject, although generally participants also had extensive experience (see Table 1). For most, English was their first language, where it was not, participants were proficient in English. The language of interview choice was English as it was most familiar to both participants and interviewer (Cortazzi et al. 2011).

Regarding saturation, some argue saturation occurs within 12 interviews (Guest et al. 2006), others within 17 (Francis et al. 2010). Arguably, however, saturation cannot be determined in advance of analysis and is “inescapably situated and subjective” (Braun and Clarke 2021, p.201). This critical role of subjectivity and context guided how we approached saturation, whereby it was “operationalized in a way consistent with the research question(s) and the theoretical position and analytic framework adopted” (Saunders et al. 2018, p.1893). We recognise that more could always be found but are satisfied that 31 participants provided sufficient data for our investigation. Indeed, our original intention was to recruit 20 participants, feeling this would provide sufficient saturation (Francis et al. 2010; Guest et al. 2006) but when we reached 20, and as we had already started analysis (cf. Braun and Clarke 2021) as we ourselves transcribed the interviews (Bird 2005) we wanted to explore understandings of ‘qualitative’ and ‘quantitative’ with other subject fields. As Table 1 shows, ‘English Literature’, ‘Philosophy, and ‘Sculpture’ were only explored after interview 20. These additional subject fields added significantly (see below) to our data.

Our analysis followed Braun and Clarke’s (2006) thematic analysis. Given the study’s exploratory constructionist nature, we combined ‘top down’ deductive type analysis for anticipated themes, and ‘bottom up’ inductive type analysis for any unexpected themes. The latter was similar to a constructivist grounded theory analysis (Charmaz 2010) whereby the transcripts were explored through close repeated reading for themes to emerge from the bottom up. We deliberately did not use any CAQDAS software such as NVivo as we wanted to manually read the scripts in one lengthy word document. We recognise that such software could allow us to do this but we were familiar with the approach we used and have found it effective for a number of years. We thus continued to use it here as well. We counted instances of themes through cross-checking after reading transcripts and discussing them, thereby heightening reliability and validity (Golafshani 2003). All interviews were undertaken with informed consent and participants were assured all representation was anonymous (Christians 2011). The study was approved by relevant ethics committees. Table 1 above shows the subject area, years of experience, and first language of the participant researchers. We also bracket after each subject area whether we consider it to be ‘Science’ or ‘Arts’ or whether we consider them as ‘Arts/Science’ or ‘Science/Arts’. This is of course subjective and in many ways not possible to do, but we were guided in how we categorised these subjects by doing so according to how we feel the methodology sources form the literature review study would categorize them.

We present our interview study data in the three broad areas that emerged through analysis. Our approach to thematic analysis was to deductively code the interview transcripts manually under the three broad areas of: where data aligns with textbook and key source ‘binary’ representations; where the data contrasts with such representations; and where the data relates to interviewee perceptions of the value of ‘qualitative’ and ‘quantitative’. The latter relates to whether participant researchers expressed views that suggested they considered each approach to be useful, valuable, or not. We also read through the transcripts inductively with a view to being open to emerging and unanticipated themes. For each data citation, we note the subject field to show the range of subject areas. We later discuss these data in terms of their implications for research values, assumptions and practices and for their use when teaching about different methods. We provide illustrative citations and numbers of participant researchers who commented in relation to the key points below, but first provide an overview in Table 2.