A note on scientific realism in the context of doing MBA, Housing Studies and Accounting & Finance dissertation projects

A note on scientific realism in the context of doing MBA, Housing Studies and Accounting & Finance dissertation projects (information from perplexity.ai)

My question 1: "Highlight the main ideas of scientific realism, in the context of doing MBA, Housing Studies and Accounting & Finance dissertation projects."

The reply: "Scientific realism treats the world (including social and financial systems) as objectively structured, and holds that good theories in MBA, Housing Studies, and Accounting & Finance research can genuinely latch onto those structures and approximate truth about them.

Core ideas of scientific realism

• Mind‑independent reality
  Scientific realism assumes there is a reality that exists independently of our thoughts, language, or models, whether that is organizational culture, housing markets, or capital structure in firms.
  
  
  In dissertations, this supports studying phenomena (e.g., mortgage default risk, dividend policy, tenant displacement) as having causal features that do not simply reduce to respondents’ perceptions, even though those perceptions are important data.

• Truth as a regulative ideal
  Scientific realism views the central aim of science as discovering approximately true descriptions of both observable and unobservable aspects of the world.
  
  
  For dissertations, this justifies treating constructs like “risk appetite”, “informal norms in property management”, or “earnings management incentives” as real targets for explanation, not merely convenient labels, and frames your empirical work as trying to get closer to how these really operate in your specific context.

• Commitment to unobservables and mechanisms
  Scientific realism holds that many important entities and mechanisms (e.g., causal powers, latent structures) are unobservable yet real, and that successful theories refer to them.
  
  
  In MBA and Housing Studies projects, this aligns with looking for underlying mechanisms (e.g., incentive structures, institutional rules, market frictions) that generate observed patterns like staff turnover, rental affordability gaps, or abnormal returns, not just describing correlations.

• Convergence and improvement of theories
  Scientific realism includes the idea that, in at least some domains, well‑supported theories tend over time to approximate an “ideal” theory that would describe reality accurately.
  
  
  For dissertations, this encourages building on established finance, housing, or management theories rather than treating every study as starting from zero, and positions your project as contributing a small but reasoned refinement to an evolving body of knowledge (e.g., testing whether capital structure theory travels to HK SMEs, or how gentrification mechanisms play out in a specific estate).

• Realism with fallible knowledge
  Scientific realism combines ontological realism with epistemic fallibilism: the world is structured and knowable, but our knowledge is always partial and revisable.
  
  
  In dissertations this fits mixed‑method, iterative, or retroductive designs: you can propose best‑current explanations of housing policy outcomes or earnings patterns while explicitly acknowledging limitations (sample, measurement error, model misspecification) and suggesting future refinement.

Implications for MBA research

• Organizational phenomena as real
  Scientific realism supports treating constructs like leadership climate, employee engagement, or AI‑driven work intensification as real patterns with causal powers, not just subjective stories.
  
  
  This legitimizes explanatory questions such as “What mechanisms link AI adoption to employee despair and performance in HK grocery firms?” and supports methods that seek mechanism evidence (e.g., path models plus qualitative process tracing).

• Mechanism‑oriented causal models
  You are encouraged to go beyond “what works” to “what works, how, for whom, and under what conditions”, emphasising context–mechanism–outcome reasoning in organizational change, HR practices, or corporate governance.
  
  
  Even when using quantitative models, discussion sections should explicitly interpret coefficients as indicative of underlying mechanisms (e.g., monitoring, signaling, power relations) rather than as mere statistical artifacts.
  ​

Implications for Housing Studies

• Structured housing realities
  Scientific realism treats housing markets, planning regimes, and neighborhood change as structured systems with real constraints and causal powers, even if individual actors perceive them differently.
  
  
  This supports research that seeks underlying structures (policy rules, financialisation, tenure systems) behind observed outcomes such as Gen Z purchase confidence, estate gentrification, or rehousing patterns.
  ​

• Layered explanations and context
  Realist work in housing often stresses the importance of context–mechanism–outcome configurations, examining how the same policy mechanism (e.g., means‑testing, home‑ownership subsidies) generates different outcomes in different districts or cohorts.
  
  
  Your dissertation can therefore justify combining statistical analysis of housing data with qualitative interviews or case studies to reveal how local norms, expectations, and power asymmetries activate or block mechanisms.
  ​

Implications for Accounting & Finance

• Reality of financial constructs
  Scientific realism treats constructs like “true economic income”, “fair value”, “systematic risk”, or “information asymmetry” as referring to real underlying features of firms and markets, even where measurement is indirect.
  
  
  In dissertations this supports explaining patterns in earnings management, dividend policy, or credit risk as arising from real incentive and information mechanisms rather than only reporting surface regularities.

• Theory‑laden but truth‑oriented models
  Finance and accounting models are seen as theory‑laden representations that may distort in some ways but can still be approximately true and progressively improved (e.g., asset pricing models, accrual quality measures).
  
  
  This underpins practices like model comparison, robustness checks, and sensitivity analysis, where you argue that some models capture the underlying mechanisms in your sample better than others, rather than treating models as merely convenient summaries.

How to write this into your dissertation

• In the philosophy or methodology section, you can state a scientific realist stance: there is a mind‑independent social and economic reality, your constructs refer to it, and your methods aim at approximately true, mechanism‑oriented explanations.

• You can position scientific realism as compatible with, or feeding into, critical realism and realist evaluation in management and housing, especially where you emphasise causal mechanisms and context‑sensitive explanation.

• You can explicitly link theoretical models (e.g., agency theory, signalling, gentrification theory) to the claim that they aim to describe real but partly unobservable structures, and then use your empirical work to test and refine their fit to your HK data."

My question 2: "Please provide a few examples of how scientific realism inform research design and choices of research methods design quality criteria in part-time 4-month MBA, Housing Studies and Accounting & Finance dissertation projects and comments on the limitations of scientific realism in such research methods design and evaluation practices."

The reply: "Scientific realism informs dissertation research design by emphasizing designs that seek explanatory depth into mind-independent mechanisms, such as those underlying housing policies or financial behaviors, while prioritizing criteria like novel predictive success and causal robustness over mere descriptive fit. In part-time, 4-month projects, this supports efficient mixed-methods approaches that balance quantitative testing with qualitative mechanism probing. Limitations arise from its struggle with pervasive underdetermination and the interpretive complexities of social phenomena, which can strain time-constrained evaluations.

MBA Dissertation Examples

Scientific realism guides MBA research on topics like AI adoption in Hong Kong grocery firms by designing studies that test causal mechanisms (e.g., incentive structures driving employee despair) through surveys and process-tracing interviews, evaluating quality via explanatory scope and novel predictions about engagement outcomes.

• Design choice: Hybrid quantitative (regression models) + qualitative (case studies) to approximate truth about unobservable powers like "organizational norms", feasible in 4 months with secondary data and targeted sampling.

• Quality criteria: Causal robustness (sensitivity tests), consilience (multiple data streams converging on mechanisms), and non-ad hoc retention of core theoretical posits.

Housing Studies Examples

In Housing Studies, scientific realism shapes designs probing real structures like policy frictions in HK estates, using residual valuation models calibrated against empirical data to explain affordability gaps, assessed by predictive accuracy in Gen Z confidence forecasts.

• Design choice: Comparative case studies (e.g., Sham Shui Po vs. Fanling) with econometric analysis of liquidity data, retroductively inferring mechanisms like tenure rules—practical for part-time timelines via public datasets.

• Quality criteria: Mechanism consilience (independent detections via surveys/models), approximate truth (model convergence on observed patterns), and explanatory unification across stakeholder experiences.

Accounting & Finance Examples

For Accounting & Finance dissertations on dividend stocks or earnings management, scientific realism favors designs with positive accounting research methods, like panel regressions testing agency mechanisms, judged by novel retrodictions of abnormal returns.

• Design choice: Archival data analysis (e.g., SGX/HKEX reports) combined with event studies, targeting unobservables like "information asymmetry" via indirect evidence—efficient for 4 months with Excel/ Stata.

• Quality criteria: Referential success (terms tracking real entities), predictive novelty (out-of-sample tests), and truth-conduciveness (iterative refinement mirroring theory evolution).

Key Limitations

Scientific realism's reliance on inference to best explanations falters under underdetermination: multiple theories (e.g., competing housing mechanism models) fit data equally, leaving choices pragmatic rather than truth-tracking in short timelines.

​

It demands resource-intensive causal probing (e.g., robustness checks), often unfeasible in 4-month part-time work, risking superficial approximations over deep mechanism insight.

Social sciences amplify issues like theory-laden observation and value influences, where "neutral" criteria (simplicity, scope) embed biases, undermining objective evaluation in interpretive fields like MBA ethics or housing policy.

Selective Strategies to Mitigate

Adopt explanationist selectivity: focus realism on success-generating posits (e.g., core agency assumptions in finance), ignoring idle parts, to streamline design without full theory commitment.

​

Entity realism aids via causal manipulation evidence (e.g., policy interventions' effects), evaluable through accessible metrics like CMO configurations in realist evaluation.

Structural realism prioritizes relational invariants (e.g., market frictions over entity natures), suiting quantitative-dominant dissertations while dodging nature-underdetermination."

A collection of blog notes on using chatgpt for research purpose.