02 Section A, Key question 3 - Mixed methods and triangulation (part 5)

Data and methodological pluralism

The form of evidence and the method used to collect it are related, but they are not the same thing.

Quantitative data express amounts or categories numerically: hourly earnings, the number of promotions, or the proportion of employees who report discrimination. Qualitative data convey meanings, experiences and social processes through material such as interview accounts, field notes or documents. A worker's account of feeling excluded from informal networking can reveal what exclusion means to them, even though it does not give a population percentage.

Do not classify the data solely by the method's name. An interview using fixed response options can produce numerical data; an open question in a questionnaire can produce a written account. Observation can generate counts of who speaks in meetings as well as detailed notes about how colleagues respond. Coding an account into numerical categories changes its form and can help comparison, but may lose the context that gave a statement its meaning.

For example, a researcher could interview workers and analyse workplace policy documents. Both might produce mainly qualitative data, yet this is methodological pluralism: the methods access different aspects of working life. Reading a policy tells us what the organisation formally promises; interviewing a worker tells us how they understand its operation. Neither is automatically an accurate record of everything that happens.

Mixed methods commonly combine quantitative and qualitative approaches within an overall investigation. Writers sometimes use the label more broadly for combinations of methods, including methods producing the same form of data. The useful sociological distinction is what is combined and why. A numerical survey and a set of interviews become a coherent design when the researcher explains how findings from one will inform, complement or be compared with findings from the other. Merely placing two unrelated results next to each other does not demonstrate integration. This broader vocabulary is discussed in Julia Brannen's NCRM methods paper.

Triangulation involves comparing evidence from different methods or sources to check, refine or deepen an understanding of the same social phenomenon. It is one purpose of combining methods. Pluralism describes the use of multiple methods; triangulation describes a relationship between the evidence they produce. A mixed design can instead aim mainly to develop questions or explain a pattern, without seeking identical answers from both strands.

Designing purposeful integration

The research question determines both what each method contributes and when it should be used.

A sequential design uses one phase to inform a later phase. In an explanatory sequence, quantitative research comes first and qualitative research follows to investigate a pattern. Consider this original teaching proposal: a survey of adult employees measures hourly pay, job grade and caring responsibilities. If a pay difference appears between groups, interviews with workers in contrasting situations could investigate access to overtime, promotion or flexible work. The survey maps measured differences; the interviews explore how relevant arrangements operate. The interviews do not need to reproduce the survey's percentages to be useful.

The phases must actually connect. Researchers could select interviewees with different survey profiles and develop interview questions about an unexpected pattern. Choosing contrasting cases is purposive sampling: selecting participants because their experiences are relevant to the question. It helps investigate variation, but does not make that interview sample statistically representative.

The order can also be reversed. In an exploratory sequence, interviews might first reveal that employees distinguish formal mentoring, which offers guidance, from informal sponsorship, where a senior colleague actively advocates for someone's career. Researchers can then develop separate questionnaire questions about both. Beginning with workers' meanings could improve validity, the extent to which the research captures what it intends to study: a survey asking only about formal mentoring might miss an important process. Researchers should still pilot the resulting questions rather than assume the first interviewees speak for everyone.

Methods can also run during the same period and be brought together afterwards. A researcher might analyse promotion records while interviewing employees about opportunities to apply. These provide complementary evidence about recorded outcomes and experiences. Complementarity means each strand adds a different relevant aspect; it does not mean the strands must make exactly the same claim. A deliberate plan to connect or compare data matters more than simply describing the design as simultaneous. The distinction between sequential and convergent designs is set out in Fetters, Curry and Creswell's methodological article.

These are original design examples, not reports of completed research. In each case, integration requires a specific analytical link. For example, a researcher can compare interview themes across the contrasting survey groups, instead of presenting the survey and interviews as separate projects. However, an interview account suggesting a mechanism is not, by itself, proof that the mechanism caused every measured difference.

The class ceiling as integrated evidence

A real investigation of class inequality illustrates why broad patterns and detailed accounts can answer different parts of a research question.

Sam Friedman and Daniel Laurison's The Class Ceiling (2019) combined analysis of the UK's Labour Force Survey with 175 in-depth interviews across television, accountancy, architecture and acting. The survey analysis investigated class origins and outcomes in higher professional, managerial and cultural occupations; the interviews explored barriers within particular occupational settings. Their 2019 book-launch presentation identifies the four interview case studies and discusses family financial support, sponsorship and workplace behavioural expectations as possible drivers of advantage.

The quantitative strand found class differences in access to, and rewards within, advantaged occupations. Class origin was measured using the occupation of the main income-earning parent when the respondent was 14, rather than simply asking whether the respondent currently felt working class. The LSE account of the 2019 research explains this measurement and the combination of survey analysis and interviews. These are findings from that research programme, not a statement of today's precise pay gap.

Work through the connection. A survey can identify an association between class origin and career rewards, but may not reveal how informal help operates. Interview accounts can suggest how family resources make insecure early careers easier to sustain, or how senior colleagues sponsor people they see as a good fit. This supplies possible mechanisms behind a distribution. It gives more explanatory depth than treating a numerical difference as self-explanatory.