International Journal of Consumer Studies

EDITORIAL PERSPECTIVE

Research Methods in a Nutshell: What, Why, When, Where,

Who, and How?
Justin Paul1,2,3 | Rahul Pratap Singh Kaurav4

1University of Puerto Rico, San Juan, Puerto Rico, USA | 2Korea University Business School, Seoul, South Korea | 3Vizja University, Warsaw,
Poland | 4FORE School of Management, New Delhi, India

Correspondence: Rahul Pratap Singh Kaurav (rsinghkaurav@[Link])

Received: 18 September 2025 | Revised: 6 October 2025 | Accepted: 9 October 2025

Funding: The authors received no specific funding for this work.

Keywords: 5WH framework | artificial intelligence (AI) in research | fsQCA | GQCA | mixed methods | multi-­study and multi-­stage designs |
qualitative methods | quantitative methods | research methods | structural equation modelling (SEM, PLS-­SEM, NCA)

ABSTRACT
Over the years, it has been observed that discussions on research methods in academic writing often oscillate between two ex-
tremes: either abstract philosophical debates or purely technical instructions. Many early career researchers—and sometimes
even experienced scholars—struggle with the simple question: Which method should be used? The answer, however, is rarely
straightforward. It depends not only on the convenience or the disciplinary habit of a researcher but also on the nature of the
research question, the timing of the inquiry, the context in which it is situated, the units of analysis involved, and the resources
available for its execution. Textbooks have contributed richly to our understanding, yet they frequently overwhelm rather than
provide clarity. In this context, we provide a comparative overview of important research methods using the 5WH framework
(What, Why, When, Where, Who, and How), based on years of experience as an Editor. We provide a synthesis of regression-­
based methods, experimental methods, multivariate methods, qualitative methods, mixed methods, configurational and com-
parative methods (QCA family), big data and computational methods, AI-­driven methods and techniques, and finally, literature
review methods.

1   |   Introduction Applied to research methods, each ‘W’ illuminates a vital

dimension: What directs the nature of the inquiry and the
Textbooks cover research methods comprehensively but often do construct under examination; Why relates to the purpose, theo-
not provide a comparative logic that assists researchers in mak- retical justification, or practical relevance; When and Where to
ing informed choices. A recurring concern amongst researchers the temporal and contextual setting; Who identifies the respon-
is the misalignment between the research problem posed and dents, participants, or units of analysis; and How explains the
the method adopted (Paul 2025). This mismatch can weaken the tools, frameworks, or procedures employed. The 5WH frame-
promising contributions that a study intends to offer. To address work helps in simplifying comparative reasoning (Paul and
this, we present an editorial perspective that is straightforward Barari 2022).
in structure and also helps in choosing the best possible method.
The 5WH framework—What, Why, When, Where, Who, and A doctoral student or supervisor designing a study, a reviewer or
How (Paul 2024)—offers such a heuristic by providing a com- associate editors evaluating a manuscript, or a practitioner ap-
parative map for method selection and justification. plying a method can all benefit from this article: Have thresholds

© 2025 John Wiley & Sons Ltd.

International Journal of Consumer Studies, 2025; 49:e70134 1 of 23

[Link]
been met adequately?. This editorial perspective, therefore, pro- variables (such as advertising spend, customer satisfaction, or dig-
poses the 5WH framework as a compact but rigorous guide to ital engagement) on dependent outcomes (such as sales, purchase
method selection. It provides a compass, a means of aligning the intention, or loyalty) (refer to Figure 1). They answer what type of
method with the question, context, and purpose. In doing so, relationship exists between predictors and outcomes and whether
it seeks to bring clarity where there is often confusion, and to that relationship is significant and meaningful (Baber et al. 2020;
foster better alignment between methodological choices and the Kaurav et al. 2015). In accounting, finance, and economics, (ex-
research problems they are intended to address. cept behavioural side) researchers often employ distinct regression
approaches compared with those commonly used in consumer
Using the 5WH framework (Paul, Khatri, and Kaur Duggal 2023), and marketing research (Demydyuk and Carlbäck 2023). They
we outline the diversity of methodological approaches available frequently rely on advanced time-­series models such as vector
to researchers. Research has never been confined to a single autoregression (VAR), GARCH, and ARCH, complemented by
pathway; rather, it has evolved into a broad spectrum of designs robustness diagnostics like multicollinearity, autocorrelation, and
and techniques. For clarity and ease of navigation, these ap- heteroscedasticity checks (for thresholds, refer Table 1). Analyses
proaches can be classified into different families. Quantitative proceed only when these assumptions are met (Kumar et al. 2025).
methods provide tools for measurement, prediction, and hypoth-
esis testing (Paul 2024). Qualitative methods illuminate mean- • Linear regression: assesses how continuous predictors (e.g.,
ing, context, and lived experience (Paul 2025). Mixed methods perceived quality, price) influence a continuous outcome
deliberately integrate the two, enabling both breadth and depth. (e.g., repurchase intention).
Big data and computational approaches extend inquiry into dig-
• Logistic regression: models binary outcomes (e.g., purchase
ital traces, large-­scale analytics, and high-­dimensional model-
vs. no purchase, churn vs. retention).
ling. Finally, AI-­driven methods and techniques have emerged
as a transformative frontier, enabling generative, predictive, and • Hierarchical/multilevel regression: accounts for nested
adaptive forms of research that redefine how data can be gath- structures in marketing data (e.g., consumers within re-
ered and analysed (Sharma et al. 2025). gions, stores within chains, or customers across time points
in panel studies).
Presenting methods within such a classification not only offers
structure but also allows readers to appreciate their complemen-
tarities and distinctions. The intent is not to privilege one family 2.2   |   Why
over another, but to provide a consolidated view that highlights
the choices available, and to prepare the ground for later discus- The rationale for regression in business research, particularly
sion of how researchers can evaluate the merits of these methods in consumer research and behavioural studies, is its ability
and choose the right approach. to quantify effects and inform managerial decisions (Mishra
et al. 2022). For example:

2   |   Regression-­Based Methods • Does customer satisfaction predict loyalty scores? (linear

regression)
2.1   |   What
• What factors increase the probability of adopting a new re-
tail technology? (logistic regression)
Regression methods are employed in behavioural, time series
data, and cross-­sectional data. Researchers using survey data in • How do store-­level promotions interact with regional demo-
consumer research usually examine the impact of independent graphics in shaping sales outcomes? (multilevel regression)

FIGURE 1    |    Basic regression portraying the relationship between a predictor variable (Y) and an outcome variable (X), linear, logistic, and mul-
tilevel regression. (Source: Authors).

2 of 23 International Journal of Consumer Studies, 2025

TABLE 1    |    Thresholds of adequacy for regression-­based methods in marketing research.

Method Key thresholds Use cases for marketing research

Linear Sample size: N ≥ 50 + 8 k (k = predictors) Often applied in survey-­based
regression models (e.g., satisfaction → loyalty).
Adjusted R2 (context-­specific, but > 0.20 often
Low R2 may still be meaningful
acceptable in consumer research)
in consumer behaviour research
Multicollinearity: VIF < 10 due to complex influences.
Residuals: normality & homoscedasticity checks
Logistic ≥ 10 events per predictor variable (EPV) Used in adoption/churn studies
regression (e.g., adoption of mobile payments).
Classification accuracy > 70% desirable
Interpretation often focuses on
Pseudo R2 (Nagelkerke R2 ≥ 0.20 as moderate) odds ratios rather than R2 alone.
Hosmer–Lemeshow test: non-­significant (p > 0.05)
Multilevel/ ≥ 20 groups recommended (with 20+ observations per group) Applied when consumers are nested
hierarchical within stores, or sales reps within
ICC (intraclass correlation coefficient) > 0.05
regression teams. Essential for retailing and
justifies multilevel model
(HLM) service management research.
Group-­level variance significant (p < 0.05)
Sufficient level-­2 sample sizes (e.g., ≥ 30 stores in retail studies)
Source: Authors.

2.3   |   When • In organisational studies (HR/OB), regression helps assess

the effects of workplace policies on satisfaction, loyalty, and
Regression is most suitable when relationships amongst vari- retention.
ables are hypothesised, and data are collected in measurable
form (survey responses, experimental manipulations, or second-
ary data). Researchers typically use regression in: 2.5   |   Who

• Survey-­based studies (e.g., linking attitudes to behavioural The unit of analysis depends on the study:
intentions).
• Individual-­
level: consumers, online users or survey
• Experimental studies (e.g., testing how different ad appeals
respondents.
influence purchase intention).
• Group-­level: retail outlets, sales teams, or product
• Panel or customer-­level datasets (e.g., modelling store loy-
categories.
alty across locations).
• Hierarchical: customers nested within segments, or trans-
• Time series data with daily, weekly, or monthly frequency.
actions within customers.

2.4   |   Where 2.6   |   How

Applications of regression are diverse across the social sciences Regression is normally employed using estimation procedures
and business disciplines: such as ordinary least squares (OLS) for linear regression or
maximum likelihood for logistic regression. For example, Baber
• In psychology and sociology, regression is applied to exam- et al. (2020) evaluated the impact of individual market orien-
ine relationships between attitudes, social influences, and tation (IMO) on selling and customer orientation for insurance
behaviours. agents. Adequacy is assessed through both statistical thresholds
and substantive interpretation:
• In education research, it is used to analyse predictors of
learning outcomes and student performance.
Linear regression formula
• In consumer and behavioural research, linear regression is
applied to test attitude–intention–behaviour models. Y = 𝛽0 + 𝛽1X 1 + 𝛽2X 2 + … + 𝛽kXk + 𝜖

• In digital marketing, logistic regression is used to predict

Example: Purchase intention = β 0 + β1 (Brand Trust) + β2 (Price
click-­throughs or conversions from online ads.
Perception) + ε.
• In retailing research, multilevel regression is applied when
sales data are clustered by store or geography. Logistic regression formula

International Journal of Consumer Studies, 2025 3 of 23

 ( ) both explanatory and predictive power, making it one of the
p
logit(p) = ln = 𝛽0 + 𝛽1X 1 + … + 𝛽kXk most enduring tools in the methodological repertoire. When
1−p
misapplied, regression can produce fragile insights. However,
once aligned with theory and thresholds, it is bound to be a
Example: Probability of adoption of mobile payment = β 0 + β1
powerful tool for both explanation and prediction.
(Perceived Usefulness) + β2 (Social Influence).

Hierarchical regression
3   |   Experimental Research Methods
Yij = 𝛽0 + 𝛽1Xij + uj + 𝜖ij
3.1   |   What
where observations (i) are nested in groups ( j), e.g., customers
Experimental research is used to establish causal relationships
within stores.
by manipulating independent variables and observing their
effects on dependent outcomes under controlled conditions. It
Example: Customer satisfaction at store level = β 0 + β1
answers what happens when a specific factor is changed (Roy
(Individual Perceived Value) + u(Store) + ε.
et al. 2025) (see Figure 2).
Regression models are implemented using widely available sta-
In multivariate research, constructs are abstract concepts (e.g.,
tistical software. Traditional packages such as SPSS and Stata
trust), measured via variables (survey questions), items (individ-
remain popular for their user-­f riendly interfaces and robust re-
ual indicators), and aggregated into factors (latent dimensions).
gression modules. Open-­source platforms such as JAMOVI, R
This distinction clarifies the progression from conceptualisation
(packages: lm, glm, lme4), and Python (libraries: statsmodels,
to measurement and analysis and helps researchers see how
scikit-­learn) provide greater flexibility and reproducibility. For
observed responses are systematically linked to underlying the-
advanced applications such as hierarchical models or survival
oretical ideas. For instance, the construct of Brand Trust may
analysis, Mplus and HLM software are often used. Commercial
be measured through a variable such as ‘perceived reliability,’
solutions like EViews are also common in time-­series contexts
which is operationalised as survey items like ‘This brand de-
(Kaurav et al. 2021; Zheng 2023) (refer Table 1).
livers on its promises.’ These items, when analysed together,
load onto a factor representing the latent dimension of brand
reliability.
2.7   |   Common Misuses of Regression
Following this measurement logic, researchers often test their
• Treating correlation as causation without experimental or
hypotheses through experimental designs. The following are
longitudinal design.
some of the popular types of experiments used in marketing and
• Ignoring key assumptions (linearity, homoscedasticity, consumer research:
independence).
• Lab experiments: conducted in controlled environments to
• Overfitting with too many predictors in small samples.
test cause–and–effect (e.g., exposure to different ad appeals).
• Using R 2 inflation as proof of model strength without
• Field experiments: conducted in real-­world settings, often
diagnostics.
with customers or stores (e.g., A/B testing of online promo-
tions, in-­store signage).
2.8   |   Guidelines for Authors • Quasi-­experiments: lack random assignment but still in-
volve intervention (e.g., pre–post studies in a retail chain).
Regression is not simply a statistical tool; it is a decision
compass. It allows scholars to test models of consumer be-
haviour, evaluate campaign effectiveness, and predict mar- 3.2   |   Why
ket outcomes. However, thresholds must be respected; an
impressive R 2 without diagnostic checks is meaningless, and The purpose of experiments is to move beyond correlation and
significant coefficients without conceptual clarity are mis- demonstrate causation. In marketing research, experiments
leading. Properly aligned, regression provides scholars with help answer:

FIGURE 2    |    Mediation and moderation framework showing how variable Me/Mo transmits the effect of predictor Y onto outcome X. (Source:
Authors).

4 of 23 International Journal of Consumer Studies, 2025

TABLE 2    |    Thresholds of adequacy for experimental research designs.

Type of experiment Key thresholds Use cases for marketing research

Lab experiments Random assignment required; sample size: Useful for testing ad framing, sensory
power ≥ 0.80 (often ~30–50 per condition) perceptions; may lack realism.
Field experiments Natural setting, treatment vs. control groups; Strong ecological validity; used in
sufficient clusters (≥ 20 stores/regions) retailing and service management.
Quasi-­experiments Pre–post measures, matched groups; statistical Applied when randomisation is not
controls (ANCOVA, propensity scores) feasible, e.g., new product launches.
Note: Kumar et al. (2025).
Source: Authors.

• Does emotional appeal in advertising increase purchase in- • Randomisation ensures comparability between groups.
tention compared to rational appeal?
• Manipulation checks confirm that participants perceived
• Do price discounts increase footfall more than loyalty the intervention as intended.
rewards?
• Outcomes are analysed with t-­tests, ANOVA, or regression
• Will algorithm-­driven product recommendations outper- models.
form human-­curated ones?
Formula (simple experimental model):

3.3   |   When Y = 𝜇 + 𝜏i + 𝜖

Experiments are most suitable when researchers need strong ev- where Y is the outcome, μ is the grand mean, τi is the treatment
idence of causal relationships. They are widely used in: effect, ϵ is the error.

• Consumer behaviour studies (e.g., testing message framing Example: Testing whether personalised product recommenda-
effects). tions (T1) increase conversion rates compared to generic recom-
mendations (T0).
• Digital marketing (e.g., A/B testing of website layouts, email
subject lines).
Experiments are often designed and deployed using Qualtrics,
• Retailing (e.g., testing shelf placement or dynamic pricing SurveyMonkey, and Gorilla Experiment Builder (for behavioural
interventions). tasks). Cognitive psychology software such as PsychoPy and
E-­Prime is used for controlled laboratory experiments, whilst
platforms such as MTurk, Prolific, and CloudResearch facili-
3.4   |   Where tate online subject recruitment. Statistical analysis of experi-
mental data can be conducted in SPSS (ANOVA modules, GLM
• Lab settings: allow control over extraneous variables, but family), R (afex, lme4), or Jamovi, ensuring that randomisation
risk artificiality. and power thresholds are verified. PROCESS macro by Prof.
Andrew F. Hayes is a popular tool for mediation, moderation,
• Field settings: offer ecological validity but reduced control.
and continuous process analysis (Hayes 2018; Hayes 2022;
• Online platforms: increasingly used for behavioural experi- Hayes and Rockwood 2020; Igartua and Hayes 2021; Aruldoss
ments (Amazon MTurk, Prolific, Qualtrics panels). et al. 2023) (for thresholds, see Table 2).

3.5   |   Who 3.7   |   Common Misuses of Experiments

Units of analysis vary by context: • Over-­reliance on student samples, limiting external validity.
• Manipulation without adequate checks.
• Individual consumers (lab-­based product trials).
• Ignoring ethical issues in field experiments (e.g., deceptive
• Customer groups (loyalty programme experiments).
pricing).
• Stores/regions (field trials of pricing, assortment, or
• Treating small-­sample lab findings as generalisable to large
promotions).
markets.

3.6   |   How 3.8   |   Guidelines for Authors

Experiments are executed through controlled manipulation of Experiments are the gold standard for causal inference in mar-
variables: keting, but their strength depends on alignment between design

International Journal of Consumer Studies, 2025 5 of 23

and research question. Misapplied experiments risk producing el- • SEM: Examining how ‘customer engagement’ influences
egant but irrelevant findings; well-­designed studies, by contrast, ‘brand advocacy’ through ‘trust.’
provide clarity that few other methods can match. For more in-
• CFA: ‘Minimum digital literacy’ as a prerequisite for adopt-
formation on experimental design-­based research, refer to Stoner
ing ‘mobile payment technologies.’
et al. (2023).

4   |   Multivariate Methods 4.3   |   When

4.1   |   What These methods are used when the research question involves la-
tent constructs or multiple interrelated variables.
Multivariate methods examine relationships amongst multiple
variables simultaneously, often to uncover or validate latent • EFA: early stages of scale development or exploratory work.
constructs that cannot be measured directly. EFA (Exploratory • CFA: once a measurement structure is hypothesised.
Factor Analysis) identifies latent dimensions underlying ob-
served variables (e.g., brand trust scale) (Chowdhary et al. 2020) • SEM: when theory predicts paths amongst latent
(see Figure 3 for categories). constructs.
• NCA: When theory or observation suggests that outcomes
• CFA (Confirmatory Factor Analysis): Tests hypothesised require minimum conditions rather than being gradually
measurement models of constructs (e.g., satisfaction → loy- built by predictors. It is particularly valuable in-­Adoption
alty framework) (Dash and Paul 2021; Kaurav, Kainthola, studies, Retailing and CRM (minimum frequency of en-
et al. 2024). gagement before loyalty).
• SEM (Structural Equation Modelling): Integrates measure-
ment and structural models to test complex causal relation-
ships (e.g., satisfaction → trust → loyalty). There are two 4.4   |   Where
possible techniques of SEM—1. CB-­SEM (Covariance-­based
SEM) (Dash et al. 2021; Mehra et al. 2021) and 2. PLS-­SEM Multivariate methods are applied widely in marketing research:
(Partial Least Squares-­based SEM) (Dul 2024; Ramtiyal
et al. 2025; Saxena et al. 2025; Srivastava et al. 2024). We • Consumer behaviour: validating scales (e.g., perceived
suggest readers to refer Dash and Paul et al. (2021), ti- value, involvement), identifying thresholds of trust, involve-
tled CB-­SEM versus PLS-­SEM, for research in the social ment, or engagement.
sciences. • Services marketing: testing SERVQUAL dimensions.
• NCA (Necessary Condition Analysis): Identifies conditions • Retailing/Customer Relationship Management (CRM):
that are necessary but not sufficient for an outcome to occur linking satisfaction, loyalty, and retention.
(e.g., a minimum level of brand trust required for loyalty).
NCA is often used alongside SEM, adding a complementary • Cross-­national research: testing measurement invariance
logic by detecting bottleneck thresholds that traditional across countries.
sufficiency-­based models cannot capture (Han, Baah, • Digital marketing: Determining the minimum online pres-
et al. 2025; Tiwari et al. 2024). ence required for influencer credibility.

4.2   |   Why 4.5   |   Who

The main reasons why multivariate methods are used are be- The unit of analysis is usually the individual respondent (survey-­
cause they help: based), but data may also come from store-­level metrics or pan-
els. Sample size adequacy is critical, which includes:
• Ensure constructs are reliable and valid (For example, atti-
tudes, perceptions, intentions) • EFA: minimum N ≥ 5–10 respondents per item.
• Permit testing of mediators and moderators. • CFA/SEM: N ≥ 200 is often recommended for stable
• Allow comparison across groups (e.g., cross-­cultural con- solutions.
sumer studies). • NCA: Individual-­level: consumers, employees, or respon-
dents in survey data, N ≥ 200 recommended for stability
Examples:

• EFA: Uncovering underlying factors in ‘customer experi- 4.6   |   How

ence’ surveys.
• CFA: Testing whether ‘service quality’ is represented by re- • EFA: Principal Component or Principal Axis Factoring
sponsiveness, empathy, and assurance. with rotation (Varimax/Oblimin).

6 of 23 International Journal of Consumer Studies, 2025

FIGURE 3    |    Models of multivariate analysis: Structural equation modelling (SEM), confirmatory factor analysis (CFA), necessary condition anal-
ysis (NCA), and LV Score, latent variable score. (Source: Authors).

• CFA: Maximum likelihood estimation of measurement • NCA: The whole process involves—Plotting indepen-
models. dent variable (condition) against dependent variable (out-
come); Identifying the ceiling line (upper-­left empty space);
• SEM: Combines CFA with structural paths, estimated
Calculating effect size d to assess necessity strength;
via covariance-­ based SEM (theory-­
driven) or PLS-­
SEM
Producing a bottleneck table showing required minimum
(prediction-­driven).
levels.

International Journal of Consumer Studies, 2025 7 of 23

Formulas: 4.8   |   Guidelines for Authors

EFA/CFA (measurement model): Multivariate methods are the most dependable and frequently
used tools in behavioural research, where latent constructs
X = 𝛬𝜉 + 𝛿 dominate. Their strength lies in rigor, but they are also prone to
misuse when thresholds are ignored, or models are fitted with-
where X is the observed indicators, Λ is the loadings, ξ is the out theoretical clarity. Properly applied, they transform abstract
latent constructs, δ is the uniqueness or specific error variance concepts into measurable, testable structures that advance both
associated with a particular observed variable. theory and practise.

SEM (structural model): CB-­SEM and PLS-­SEM are not substitutes but complements.
CB-­SEM excels in theory testing with strong models, whilst PLS-­
𝜂 = B𝜂 + 𝛤 𝜉 + 𝜁 SEM is advantageous for prediction and exploration in emerging
areas. The choice should be driven by the research question, data
where η is the endogenous constructs, ξ is the exogenous con- conditions, and purpose—not by software convenience.
structs, B, Γ is the path coefficients.
NCA shifts attention from ‘drivers’ to ‘prerequisites.’ In mar-
Example: Customer Satisfaction (latent) → Loyalty (latent), me- keting, this perspective reveals bottlenecks that managers must
diated by Trust (latent). cross before other strategies can work. whilst it complements re-
gression and SEM, it should not replace them; necessity is one
NCA: piece of the causal puzzle.

CE
d=
T
5   |   Qualitative Methods
where CE is the ceiling zone (space above the ceiling line), T is
5.1   |   What
the total empirical data space.
Qualitative methods are designed to capture meanings, pro-
Exploratory factor analyses are commonly conducted using
cesses, and contexts that are often overlooked by quantitative
SPSS. AMOS, SmartPLS, ADANCO, LISREL, Mplus, and
measures. In social sciences, focusing on behavioural research,
Stata are widely used for confirmatory factor analyses. Open-­
they help answer what lies behind consumer perceptions, organ-
source alternatives include R (psych, lavaan, semTools) and
isational practises, and cultural patterns.
Jamovi (SEMLj plugin). For SEM specifically, AMOS, Mplus,
EQS, and LISREL dominate theory-­d riven applications, whilst
• Case study (single-­or multiple-­case): Examines a bounded
SmartPLS, WarpPLS, and ADANCO are widely used for
system (e.g., a retail chain, a start-­up ecosystem) in depth.
prediction-­oriented PLS-­SEM. NCA is supported in R (NCA
Integrating the tenets of a theoretical framework in the con-
package) and in SmartPLS, enabling easy integration with
text of a case study is important. Researchers are suggested
SEM analyses (Rana et al. 2023). Refer Table 3 for thresholds
to refer to Paul and Sahadev (2018) to learn how to do this.
values, Table 4 for differentiating CB-­SEM and PLS-­SEM and
Table 5 to understand should you extend to NCA, when it is • Ethnography: Immersive exploration of consumer or organ-
suitable. isational culture (e.g., studying digital/online fan commu-
nities via netnography) (Sima et al. 2025).
• Phenomenology: Focuses on lived experiences of individu-
4.7   |   Common Misuses of Multivariate Methods als (e.g., how consumers experience luxury or sustainabil-
ity) (Bhalla et al. 2025).
• Performing EFA on small samples with unstable factors.
• Grounded theory: Builds theory inductively from data
• Using CFA/SEM with poor model fit but still ‘accepting’ (e.g., developing a model of consumer trust in AI systems)
results. (Chawla et al. 2025; Demydyuk et al. 2024; Kainthola
• Ignoring discriminant validity and reporting overlapping et al. 2024).
constructs. • Focus groups: Facilitated group discussions to sur-
• Treating PLS-­SEM (predictive) as equivalent to CB-­SEM face shared opinions and language around products or
(theory testing). campaigns.

• Treating necessary conditions as sufficient (ignoring that

other factors still matter). 5.2   |   Why
• Using NCA on very small samples leads to unstable ceiling
• Qualitative methods address why consumers behave as
lines.
they do by uncovering meanings, narratives, and symbolic
• Misinterpreting weak necessity (d < 0.1) as meaningful. processes.

8 of 23 International Journal of Consumer Studies, 2025

TABLE 3    |    Thresholds of adequacy for multivariate methods, including EFA, CFA, SEM, and NCA.

Method Key thresholds Notes for marketing research

EFA KMO > 0.60 Used in scale development;
exploratory studies of constructs
Bartlett's test significant (p < 0.05)
like customer engagement.
Factor loadings ≥ 0.40
CFA Factor loadings ≥ 0.50 (Preferably ≥ 0.60) Applied in consumer surveys to validate
measurement models (e.g., service quality).
Cronbach's Alpha ≥ 0.70 (Acceptable), ≥ 0.80 (Good)
Composite reliability (CR) ≥ 0.70 (Acceptable)
Average variance extracted (AVE) ≥ 0.50 (Acceptable)
Chi-­Square/degrees of freedom ( χ 2/
df) ≤ 5.0 (Acceptable), ≤ 3.0 (Good)
Goodness-­of-­f it index (GFI) ≥ 0.90 (Acceptable), ≥ 0.95 (Good)
Adjusted goodness-­of-­f it index (AGFI) ≥ 0.80 (Acceptable)
Comparative fit index (CFI) ≥ 0.90 (Acceptable), ≥ 0.95 (Good)
Tucker-­Lewis index (TLI) ≥ 0.90 (Acceptable), ≥ 0.95 (Good)
Root mean square error of approximation
(RMSEA) ≤ 0.08 (Acceptable), ≤ 0.06 (Good)
Standardised root mean square residual
(SRMR) ≤ 0.08 (Acceptable), ≤ 0.05 (Good)
Discriminant validity: HTMT < 0.90
SEM CFA + N ≥ 200 for stable estimation multi-­group Common in marketing to test
analysis requires measurement invariance tests mediators (trust, satisfaction) and
moderators (culture, demographics).
NCA CFA + SEM + effect size (d): ≥ 0.1 (small), Indicates how much space is
≥ 0.3 (medium), ≥ 0.5 (large) empty above the ceiling line.
Ceiling line should be visually and statistically checked Ensures necessary but not
sufficient condition is present.
Bottleneck table required for levels of condition Helps identify managerial thresholds (e.g.,
for different levels of outcome minimum satisfaction score for loyalty).
Source: Compiled by Authors from multiple sources, some of the classical inspirations are Anderson and Gerbing (1988); Fornell and Larcker (1981); Bagozzi and
Yi (1988).

• Case studies reveal complex causal chains and context-­ • Case studies and ethnography: when depth and context are
specific strategies (e.g., why certain retailers succeed in om- essential.
nichannel adoption).
• Phenomenology: when the researcher seeks to understand
• Ethnography explains cultural consumption practises (e.g., deep subjective experience.
why communities engage with brands as identity symbols).
• Grounded theory: when existing theory is insufficient.
• Phenomenology captures subjective meaning (e.g., why ‘au-
• Focus groups: when preliminary insights are needed for
thenticity’ matters to craft beer consumers).
survey or campaign design.
• Grounded theory builds middle-­range theories when exist-
ing frameworks are inadequate.
5.4   |   Where
• Focus groups surface why target segments perceive products
differently, providing insights for, i.e., campaign design.
• Case studies: firms, markets, or industries (e.g., Tesla as a
single-­case on disruptive branding).
5.3   |   When • Ethnography: cultural settings, physical or digital (e.g., in-­
store observation vs. netnography of online communities).
These methods are most suitable when the research is explor-
atory, theory-­building, or context-­dependent. For instance, • Phenomenology: consumer interviews in natural settings.

International Journal of Consumer Studies, 2025 9 of 23

TABLE 4    |    Comparison of CB-­SEM and PLS-­SEM.

Aspect CB-­SEM (covariance-­based SEM) PLS-­SEM (partial least squares SEM)

What it is Theory-­testing approach; estimates model Prediction-­oriented approach; estimates model
fit based on covariance matrix. components using variance explained.
Primary aim Confirmatory—tests whether data fit Exploratory & predictive—focuses on
a hypothesised theory/model. maximising explained variance (R2 , Q2).
When to use Strong theory exists Theory is less developed or is exploration phase
Hypotheses require confirmation Focus is on prediction or exploration
Model fit indices are critical Models are complex or data quality is limited
Sample size Preferred larger samples (N ≥ 200 typical). Smaller samples acceptable; rules
of thumb: ‘10 times rule’ (10 × max
arrows pointing to construct).
Assumptions Multivariate normality, continuous data, larger N. More flexible: handles non-­normal data,
smaller N, formative indicators.
Evaluation criteria Global fit indices: χ 2/df ≤ 3, RMSEA Prediction-­focused: R 2 , Q2 , f 2 effect size, path
≤ 0.08, CFI/TLI ≥ 0.90, SRMR ≤ 0.08. coefficients, SRMR (recently adopted).
Strengths Gold standard for theory testing Suitable for exploratory & predictive aims
Rich model fit diagnostics Handles complex models with many constructs
Supports mediation/moderation with rigour Works with small to medium samples
Limitations Sensitive to sample size & non-­normality Lacks established global fit indices
Less suited for formative constructs Sometimes misused as a
substitute for weak theory
Marketing use cases Validating well-­established constructs Exploring new constructs (e.g., customer
(e.g., SERVQUAL, brand equity). engagement, digital trust) or predictive
models (e.g., churn, adoption).
Note: Dash and Paul (2021).
Source: Authors.

TABLE 5    |    Comparison of SEM and NCA.

Aspect SEM (Structural equation modelling) NCA (necessary condition analysis)

Logic Tests sufficiency: Do predictors Tests necessity: Is a minimum level of a
explain variance in outcomes? predictor required for an outcome?
Purpose Explains how much a factor Identifies bottlenecks that make
contributes to outcomes. outcomes possible or impossible.
Data Needs Larger samples (N ≥ 200 recommended). Moderate samples (N ≥ 50 sufficient
for stable ceiling lines).
Output Path coefficients, R 2 , fit indices. Effect size (d), ceiling line, bottleneck table.
Use Case Testing theory, mediation, moderation, Detecting thresholds: e.g., minimum trust for
cross-­group comparisons. loyalty, minimum awareness for adoption.
Complementarity Shows strength and direction of relationships. Reveals critical prerequisites
that SEM may overlook.
Source: Developed based on the inputs from Han, Baah, et al. (2025).

• Grounded theory: iterative data collection in field contexts 5.5   |   Who

(e.g., start-­up accelerators).
• Case studies: organisations, managers, policy-­makers.
• Focus groups: neutral venues, online panels, or brand
communities. • Ethnography: communities, tribes, consumer collectives.

10 of 23 International Journal of Consumer Studies, 2025

 • Phenomenology: individuals sharing a lived experience • Conducting shallow interviews/focus groups without
(e.g., first-­time luxury buyers). saturation.
• Grounded theory: diverse participants until saturation. • Using case studies without triangulation.
• Focus groups: carefully segmented participants (6–10 per • Over-­generalising from context-­specific insights.
group).

5.8   |   Guidelines for Authors

5.6   |   How
Qualitative methods bring voice, meaning, and depth to mar-
• Case studies: triangulate interviews, archival data, and keting research, complementing the statistical rigour of quan-
observation. titative approaches. Their misuse often stems from treating
them as ‘quick fixes’ rather than systematic designs. Applied
• Ethnography: prolonged immersion, thick description, re-
rigorously, however, they uncover the cultural and experiential
flexive notes.
dimensions of marketing that numbers alone cannot capture.
• Phenomenology: in-­
depth interviews, analysis of lived Authors are advised to refer to Crick (2021) for learning more
narratives. about how to execute a classic qualitative study. Similarly, Paul
and Sahadev (2018) show how to apply theoretical tenets in a
• Grounded theory: coding (open, axial, selective), constant
case study context.
comparison.
• Focus groups: guided discussion, content analysis, thematic
coding. 6   |   Mixed Methods

Qualitative data collection and analysis are typically supported by 6.1   |   What

software such as MAXQDA, NVivo (Aggarwal et al. 2023), [Link],
and Dedoose, which provide coding, categorisation, and theme de- Mixed methods combine quantitative and qualitative approaches
velopment functions. Tools like Transana and f4transkript, [Link] within a single study or research programme. They answer what
support transcription and video analysis. For discourse or conversa- happens when numerical breadth is complemented with inter-
tion analysis, specialist packages such as text-­based Python libraries pretive depth.
NLTK (i.e., Natural Language Toolkit), spaCy, Gensim, TextBlob,
and Scikit-­learn can be employed. Increasingly, these platforms in- • Sequential exploratory design (qual → quant): Qualitative
tegrate with AI-­assisted coding modules, enabling semi-­automated findings guide survey design or scale development.
processing whilst preserving interpretive depth. In general, there
is no simple and single way for achieving thresholds; still, refer to • Sequential explanatory design (quant → qual): Survey or ex-
Table 6 for more details. The principle of saturation is followed for periment findings are followed by qualitative exploration to
determining the sample size (Kainthola et al. 2024). explain results.
• Concurrent triangulation: Both strands were collected in
parallel and compared for convergence.
5.7   |   Common Misuses of Qualitative Methods
• Embedded design: One method nested inside another (e.g.,
• Treating anecdotal evidence as theory. interviews within a large-­scale survey).

TABLE 6    |    Thresholds of adequacy for qualitative research methods.

Method Key thresholds Use cases for marketing research

Case study Multiple data sources (interviews, Used for firm-­level strategy or retail
documents, observation). transformation research.
Triangulation essential.
Ethnography Prolonged engagement (months); field Applied in consumer culture theory (CCT),
notes, observations, interviews. digital ethnography in online brand communities.
Phenomenology 6–25 participants often sufficient; Used to explore symbolic consumption
emphasis on depth not size. (luxury, authenticity).
Grounded theory Iterative coding; theoretical Useful for new domains (AI in
saturation (~20–30 interviews). marketing, sustainable practises).
Focus groups 3–5 groups, 6–10 participants per Common in exploratory product testing,
group; moderator neutrality. ad evaluation, segmentation studies.
Source: Authors.

International Journal of Consumer Studies, 2025 11 of 23

 • Multiphase design: Iterative cycles of quant and qual over • Qualitative strand: smaller purposive samples for depth.
time.
• Integration requires transparency on who contributes to
which part.
6.2   |   Why

The rationale is that marketing phenomena are complex—at- 6.6   |   How

titudes, behaviours, and contexts often require both measure-
ment and meaning. Mixed methods are used: • Mixed methods require explicit integration strategies:
• Merging: results compared during interpretation.
• To validate findings across methods (triangulation).
• Connecting: one strand informs the sampling or design of
• To develop scales (qualitative → quantitative). the next.
• To explain surprising results (quantitative → qualitative). • Embedding: one strand provides supportive evidence
• To address multi-­ level problems (e.g., firm-­
level and within the other. Analysis may use joint displays (tables/
consumer-­level perspectives). figures showing how results align).

Examples: Mixed methods research requires platforms that can integrate

qualitative and quantitative strands. MAXQDA, JMP provide
• Using focus groups to generate items for a new brand love modules for joint displays, data linking, and code–statistic in-
scale, then testing via survey (exploratory). tegration. Researchers also use combinations such as SPSS or
Stata (for quantitative strands) together with NVivo/[Link]
• Surveying digital trust, then interviewing consumers who (for qualitative strands). Open-­source platforms such as R and
show outlier responses (explanatory). Jamovi can be combined with RQDA for mixed-­method exten-
• Running parallel interviews and experiments to triangulate sions, making integration more transparent and reproducible
service failure recovery perceptions. (refer Table 7 for thresholds).

6.3   |   When 6.7   |   Common Misuses of Mixed Methods

Mixed methods are most suitable when: • Running two methods independently without integration.

• Research questions cannot be answered adequately with a • Treating the qualitative strand as superficial or symbolic
single method. (e.g., one or two interviews or not taking insights from in-
terview and move ahead).
• Both generalisability (quant) and contextual richness (qual)
are required. • Lack of transparency on sequencing, weighting, or rationale.

• The study aims to develop a new theory and also test it • Using mixed methods only to ‘add flavour,’ without genuine
empirically. complementarity.

6.4   |   Where 6.8   |   Guidelines for Authors

Applications in marketing research include: Mixed methods reflect the reality that marketing problems
are rarely one-­dimensional. When designed rigorously, they
• Consumer research: understanding emerging phenomena bridge numbers and narratives, prediction and meaning. Their
like influencer credibility. misuse comes from a lack of integration; their value lies in
showing how different forms of evidence converge or diverge
• Retailing: combining store-­level sales data with consumer to enrich understanding.
ethnographies.
• Digital marketing: integrating clickstream analytics with
interviews on user experience. 7   |   Configurational & Comparative Methods (QCA
Family)

6.5   |   Who 7.1   |   What

The unit of analysis spans both strands: Configurational methods examine how combinations of con-
ditions jointly lead to an outcome, recognising that different
• Quantitative strand: large samples of consumers, firms, or pathways (equifinality) can produce the same result. They move
transactions. beyond linear sufficiency (as in regression) by incorporating

12 of 23 International Journal of Consumer Studies, 2025

TABLE 7    |    Thresholds of adequacy for mixed method designs.

Design Key thresholds Use cases for marketing research

Sequential exploratory Qualitative saturation (~20–30 interviews) Used in scale development
before survey design; survey N ≥ 200 (e.g., customer engagement,
brand authenticity).
Sequential explanatory Quantitative analysis first (power ≥ 0.80); follow-­up Applied in digital trust/adoption studies.
qualitative sample sufficient to explain anomalies
Concurrent triangulation Comparable weighting of strands; Ensures credibility by testing
convergence analysis required consistency across methods.
Embedded design Nested strand justified (e.g., 10–15 Useful in CRM or loyalty research.
interviews within survey of 500+)
Multiphase Transparent sequencing and integration; Applied in longitudinal
requires resources for multiple rounds brand-­building studies.
Source: Authors.

TABLE 8    |    Thresholds of adequacy for configurational and comparative methods (fsQCA and GQCA).

Method Key thresholds Notes for marketing research

fsQCA Consistency ≥ 0.75 for sufficiency; coverage > 0.25. Used to analyse multiple causal paths
Fuzzy membership scores between 0 and 1. (e.g., omnichannel strategy success).
GQCA Normalised data (0–1 range). Grey relational Applied when data is incomplete/uncertain but
grade ≥ 0.6 indicates strong similarity. Requires patterns must be compared (e.g., retail formats,
reference and comparative series. consumer adoption under uncertainty).
Source: Authors.

causal asymmetry (what drives success may differ from what 7.2   |   Why
drives failure) (Fan et al. 2024).
These methods are valuable because marketing outcomes
• fsQCA (Fuzzy-­ Set Qualitative Comparative Analysis): are often complex, context-­ dependent, and non-­ linear.
Combines qualitative case orientation with quantitative set-­ Configurational approaches allow scholars to capture:
theoretic logic. Cases are assigned fuzzy membership scores
(0–1) to conditions, allowing the identification of sufficient • Equifinality: Different paths to the same outcome (e.g., high
and necessary combinations (Choi 2026; Huarng and Roig-­ loyalty via either brand trust or switching costs).
Tierno 2016; Das et al. 2022).
• Asymmetry: The absence of conditions for success is not
• GQCA (Grey Qualitative Comparative Analysis): Rooted simply the mirror opposite of those for failure.
in grey system theory, GQCA transforms incomplete or
• Threshold sensitivity: Especially relevant in GQCA, which
‘grey’ data into quantitative similarity measures, identify-
quantifies minimum levels of similarity or performance.
ing the degree of closeness between reference and compar-
ative series to evaluate outcomes objectively (Merminod
Examples:
et al. 2022).
• fsQCA: Identifying configurations of digital capabilities
Placing fsQCA and GQCA together as a distinct family of config-
and managerial support that jointly drive successful omni-
urational and comparative methods recognises their shared focus
channel adoption.
on complex causality and alternative pathways to outcomes. Both
methods differ fundamentally from traditional quantitative mod- • GQCA: Comparing competing retail formats under uncer-
els that emphasise linear sufficiency and from qualitative designs tain market data to identify the closest match to consumer
that stress contextual narratives. fsQCA bridges the two tradi- preference patterns.
tions by combining case-­oriented reasoning with set-­theoretic
quantification, whilst GQCA introduces robust comparative
logic for incomplete or uncertain data through grey system anal- 7.3   |   When
ysis (Han, Wong, et al. 2025). Grouping them avoids forcing them
into the conventional ‘qualitative versus quantitative’ divide but Configurational methods are best applied in the following
also highlights their collective contribution: enabling marketing situations:
scholars to capture equifinality, asymmetry, and uncertainty in
ways that other methods cannot (Huarng 2016). The threshold • The research question involves multiple interacting
values are presented in Table 8. conditions.

International Journal of Consumer Studies, 2025 13 of 23

 • Theoretical frameworks suggest different pathways to where Δ0i(k) is the difference between reference and compara-
outcomes. tive series at point k; ζ is the distinguishing coefficient (0–1, usu-
ally 0.5).
• Data may be limited, uncertain, or incomplete (especially
suitable for GQCA).
fsQCA is supported by the dedicated fsQCA software developed
at the University of Zurich, as well as Tosmana (for Boolean
7.4   |   Where analysis) and R packages (QCA, SetMethods) (see Table 9).
These allow calibration, truth table construction, and minimi-
• Consumer research: fsQCA used to explore combinations sation procedures. GQCA is less institutionalised but can be im-
of cultural values and social influence driving product plemented using R (e.g., greyrel libraries), MATLAB, or Python
adoption. packages for grey system analysis. Increasingly, scholars com-
bine fsQCA outputs with SEM/PLS results to strengthen config-
• Retailing: GQCA applied to benchmark alternative store urational insights.
formats under uncertain consumer demand.
• Digital marketing: fsQCA applied to understand different
bundles of social media activities that drive engagement. 7.7   |   Common Misuses of Configurational
Methods
• B2B marketing: Both fsQCA and GQCA applied to analyse
alliance success under uncertain environments. • Treating fsQCA as purely qualitative or purely quantitative,
ignoring its hybrid nature.
7.5   |   Who • Applying fsQCA with too few cases produces unstable truth
tables.
Units of analysis may include:
• Using GQCA without proper data normalisation.
• fsQCA: Firms, consumer groups, or countries (comparative • Over-­interpreting equifinal pathways without theoretical
cases with partial set membership). Sample size required is justification.
a moderate number of cases (often 10–50+).
• GQCA: Indicators or data series that must be compared 7.8   |   Guidelines for Authors
quantitatively (e.g., multiple store performance indices).
Works well with small datasets. Configurational methods provide marketing researchers with a
distinctive lens to address complexity, uncertainty, and causal
asymmetry. fsQCA bridges qualitative depth with quantitative
7.6   |   How set theory, whilst GQCA offers robust analysis under incomplete
information. Their value lies not in replacing traditional meth-
fsQCA: ods but in complementing them, especially where multiple path-
ways exist, or data quality is uncertain.
• Calibrate data into fuzzy sets (0–1 membership).
• Construct a truth table of conditions and outcomes.
8   |   Big Data & Computational Methods
• Simplify logical expressions using Boolean minimisation.
8.1   |   What
• Report consistency and coverage to assess sufficiency and
necessity.
Big data and computational methods address large, complex,
and often unstructured datasets to answer what patterns exist
Formula (Consistency for sufficiency):
in digital traces and how they can be leveraged for marketing
∑ � � insight.
min Xi , Yi
Consistency (X ≤ Y ) = ∑
Xi • Text mining/NLP: topic modelling of reviews, sentiment
GQCA: analysis of tweets (Demydyuk et al. 2024).
• Social media analytics: examining brand conversations on
Normalise data to a 0–1 scale. platforms like Twitter, Instagram, and Reddit.

• Compute grey relational coefficients between reference and • Clickstream/web analytics: analysing user navigation paths
comparative series. on e-­commerce sites.

• Derive the grey relational grade to rank alternatives. • Image/video analytics: computer vision for logo detection,
shelf analysis, influencer content.
Formula (Grey Relational Grade): • Machine learning: supervised (predict churn), unsuper-
n
( ) 1 ∑ min Δ + ζmax Δ vised (segment consumers), deep learning (visual recogni-
𝛾 x0 , x i =
n k = 1 Δ0i (k) + ζmax Δ tion, generative content).

14 of 23 International Journal of Consumer Studies, 2025

TABLE 9    |    Comparison of fsQCA and GQCA in marketing research.

Aspect fsQCA GQCA

Logic Hybrid method combining qualitative case Quantitative method based on grey system
orientation with quantitative set membership. theory, suited to incomplete/uncertain data.
Data type Case-­level data with fuzzy set calibration Time series or indicator data normalised
(0–1 membership scores). to [0, 1] for comparison.
Purpose Identifies sufficient and/or necessary Measures similarity/dissimilarity between
configurations of conditions reference and alternative series to rank options.
leading to an outcome.
Key outputs Truth table, consistency, coverage Grey relational coefficients and grades
metrics; multiple causal pathways. indicating closeness to reference.
Strengths Captures causal complexity, Handles incomplete/’grey’ data;
equifinality, and asymmetry. provides objective rankings.
Limitations Sensitive to calibration and small samples; Less intuitive for theory-­building;
risk of over-­complex models. primarily comparison-­based.
Marketing use cases Examining multiple paths to digital Benchmarking retail formats, consumer
adoption, loyalty, or innovation success. preference structures under uncertainty.
Source: Authors.

• ANN (Artificial Neural Networks): Multi-­layered com- • Network analysis of online brand communities to detect
putational models inspired by the human brain, used for influencers.
pattern recognition, classification, and prediction. ANNs
are widely applied in marketing for churn prediction, cus-
tomer adoption modelling, sentiment analysis of reviews, 8.3   |   When
and recommender systems (Zhao and Zhan 2025).
These methods are most suitable when:
• Predictive modelling: conversion likelihood, recommenda-
tion systems (Kaurav and Thomas 2025).
• Data are high-­volume, high-­velocity, or high-­variety (the
• Network science: mapping social networks, brand commu- ‘3Vs’ of big data).
nities, or knowledge flows.
• Research questions involve prediction, pattern detection, or
digital trace analysis.
8.2   |   Why • Traditional surveys or experiments are insufficient (e.g., an-
alysing millions of tweets).
The rationale is clear: digital marketing generates massive data
streams, and traditional statistical methods are insufficient to
capture their complexity. Big data methods allow marketing re- 8.4   |   Where
searchers to:
Applications span almost all domains of marketing:
• Understand consumer sentiments at scale.
• Consumer behaviour: analysing reviews, social media posts.
• Detect emerging trends in social networks.
• Digital marketing: campaign targeting and optimisation.
• Predict churn, cross-­sell, or conversion probabilities.
• Retailing/CRM: churn prediction, recommendation
• Uncover hidden structures in consumer–brand interactions.
engines.
Examples: • Branding: image recognition in user-­generated content.

• Sentiment analysis of TripAdvisor reviews to predict hotel

ratings. 8.5   |   Who
• Clickstream modelling to optimise checkout flows in
The unit of analysis is often digital consumers (users, custom-
e-­commerce.
ers, communities) or digital artefacts (posts, clicks, images).
• Deep learning on Instagram posts to identify visual drivers Sampling adequacy is less relevant; instead, the focus shifts to
of engagement. data representativeness and cleanliness. For ANN, preferably

International Journal of Consumer Studies, 2025 15 of 23

TABLE 10    |    Thresholds of adequacy for big data and AI-­driven methods.

Method Key thresholds Use cases for marketing research

Text mining/NLP Minimum corpus size: ≥ 1000 Used in review mining,
documents for topic stability; coherence brand sentiment studies.
score ≥ 0.5 for topic models
Social media analytics Adequate time window (≥ 3–6 months Applied in influencer and
for trends); bot filtering essential brand community studies.
Clickstream analytics Session-­level tracking; N ≥ 10,000 Used in website optimisation
sessions for robust models and path analysis.
Image/Video analytics Accuracy > 80% (precision, recall, F1 Applied in shelf recognition, logo
score); labelled dataset quality critical detection, ad effectiveness.
Machine learning (supervised) Train/test split; cross-­validation; Predict churn, conversion, adoption.
accuracy/ROC-­AUC > 0.70 desirable
Machine learning (unsupervised) Cluster stability indices (e.g., silhouette > 0.5) Used in customer segmentation.
ANN Performance Metrics: Accuracy, Precision, Applied to churn prediction, adoption
Recall, F1, ROC-­AUC > 0.70 acceptable. likelihood. Critical when data are
limited. Important for managerial
Overfitting Checks: Cross-­validation,
insights, not just prediction.
dropout, regularisation.
Interpretability: Increasingly addressed
using SHAP, LIME, or feature importance.
Deep learning Requires large datasets (often > 10,000 images Visual analytics, generative
or text sequences); overfitting checks marketing content.
Predictive modelling Out-­of-­sample validation; lift > 10% vs. baseline Widely used in CRM targeting.
Network science Density, centrality, modularity Applied in influencer detection
thresholds; robustness tests and brand community mapping.
Source: Authors.

large datasets (> 1000 observations); more are needed for deep where p(w∣d) is the probability of word, w in document d, z is the
architectures (refer to Table 10 for threshold values). latent topic, K is the number of topics.

Formula sentiment analysis (Naïve Bayes):

8.6   |   How n
∏ (
P(Sentiment | Words) ∝ P(Sentiment) P wi || Sentiment)
• Data collection: APIs, web scraping, CRM systems. i−1

• Preprocessing: cleaning, tokenisation, feature engineering. Formula machine learning: Supervised learning (logistic regres-
• Analysis: machine learning algorithms, deep learning ar- sion, classification):
chitectures, network metrics. 1
P(y = 1| X ) =
• Validation: cross-­validation, out-­of-­sample testing, perfor- 1 + e−(𝛽0+𝛽1X 1+⋯+𝛽kXk)
mance metrics.
Unsupervised learning (k-­means clustering):
Formula (neural network, simpler version): n
∑
MinC ‖x − μ ‖2
y = f (Wx + b) ‖ i Ci ‖
i−1

where W is the weights, x is the inputs (e.g., image pixels), b is where xi is the data point, μci is the cluster centroid.
the bias, f is the activation function.
Deep learning (Feedforward NN):
Formula topic modelling (latent Dirichlet allocation, LDA):
y = f (Wx + b)
K
∑ ∑
p (w| d) = = (w| z = k). (p(z = k| d)) where W is the weight matrix, b is the bias, f is the non-­linear
k−1 activation.

16 of 23 International Journal of Consumer Studies, 2025

A variety of open-­source and commercial platforms support com- • Hybrid AI-­Human Methods: AI-­assisted coding of qualita-
putational marketing research. Python (libraries: NLTK, spaCy, tive data, AI-­augmented survey design.
gensim, scikit-­learn, transformers) and R (tm, quanteda, tidytext)
dominate text mining and sentiment analysis. NodeXL, Netlytic,
Gephi, and UCINET are widely used for social media and network 9.2   |   Why
analytics. Google Analytics and Adobe Analytics provide industry-­
standard clickstream data analysis. Image and video analytics rely AI is being incorporated because marketing data is increasingly
on TensorFlow, PyTorch, Keras, and OpenCV. Predictive modelling unstructured, multimodal, and dynamic. Traditional methods
and clustering are facilitated by RapidMiner, Weka, Orange, and cannot capture this complexity. AI techniques allow:
Python/R machine learning toolkits. For network science, Gephi,
Pajek, and NetworkX (Python) are standard choices. • Discovery of emergent consumer needs through unstruc-
tured text/image/video.

8.7   |   Common Misuses of Big Data Methods • Automated personalisation at scale.

• Simulated consumer decision-­
making under different
• Equating big sample size with representativeness (e.g., scenarios.
Twitter ≠ all consumers).
• New ways to test theory using synthetic or augmented data.
• Using algorithms as ‘black boxes’ without theoretical
grounding. Examples:
• Ignoring data privacy and ethical considerations.
• Using transformer-­based sentiment models to detect emo-
• Overfitting models without proper validation. tions in reviews.
• Applying reinforcement learning to optimise dynamic pric-
8.8   |   Guidelines for Authors ing in e-­commerce.
• Leveraging generative AI to simulate customer responses in
Big data and computational methods expand what marketing
A/B testing.
researchers can study, from millions of consumer conversations
to visual content and predictive analytics. Yet, their misuse often
comes from prioritizing technical sophistication over theoretical
alignment. When integrated with sound research design, these 9.3   |   When
methods open unprecedented avenues for understanding and
shaping markets. ANNs extend the researcher's toolkit for pre- AI methods are most suitable when:
diction and pattern recognition in marketing. Their power lies in
modelling complexity, but their limitation is interpretability. Used • Data is too large, complex, or unstructured for conventional
thoughtfully, they complement theory-­driven methods by reveal- statistics.
ing non-­linear dynamics and generating actionable insights. • Research questions require prediction, recommendation, or
generative outputs.
9   |   AI-­Driven Methods & Techniques • Real-­time decision-­making or adaptive experimentation is
required.
9.1   |   What

AI methods in research involve algorithms that learn from data 9.4   |   Where

to classify, predict, generate, or optimise. They answer what pat-
terns exist in high-­dimensional consumer data and how intelli- • Consumer research: using AI to decode visual consumption
gent systems can be applied to marketing problems (Paul 2024; signals from Instagram.
AlBannai 2025).
• Retailing: reinforcement learning for personalised in-­store
• Natural Language Generation (NLG): Automated text gen- offers.
eration for surveys, product descriptions. • Advertising: generative AI for ad copy and image variants.
• Generative AI (LLMs, diffusion models): Content creation, • Services: chatbots and conversational AI as both objects of
campaign design, synthetic respondents. study and research tools.
• Reinforcement Learning: Optimising sequential decisions
(e.g., personalised promotions).
9.5   |   Who
• Recommender Systems: Collaborative filtering, hybrid rec-
ommenders for customer choice modelling.
Units of analysis shift from individuals to digital traces, syn-
• AI-­enabled Causal Inference: Uplift modelling, counterfac- thetic consumers, or AI agents. In many cases, ‘who’ is both the
tual prediction. consumer and the algorithm interpreting behaviour.

International Journal of Consumer Studies, 2025 17 of 23

9.6   |   How traditions—offering new forms of evidence whilst demanding
even greater attention to thresholds, validity, and ethics.
AI methods are executed via pipelines:

• Data ingestion (text, images, voice, behavioural traces). 10   |   Literature Review as Research Method
• Model training (supervised, unsupervised, reinforcement,
Literature reviews are an integral part of doctoral and academic
generative).
research (Paul and Criado 2020; Tomczyk et al. 2024). A good lit-
• Validation (out-­
of-­
sample tests, fairness audits, erature review offers insights into the evolution of the research
interpretability). field and helps understand theories, context, characteristics, and
methods, and identifies research gaps (Paul and Barari 2022;
• Deployment (as recommender systems, content generators,
Paul 2024). Researchers develop a future research agenda based
predictive models).
on identified research gaps in a typical review paper (Paul,
Ueno, and Dennis 2023; Paul 2025; Kaurav and Mishra 2024;
The popular Generative AI tools in research are ChatGPT
Paul, Ueno, et al. 2024; Kaurav, Kulkarni, and Shee 2024).
(Paul, Khatri, and Kaur Duggal 2023), Claude, Jasper, [Link],
MidJourney, and Stable Diffusion. Transformer-­based NLP mod-
els are available through Hugging Face Transformers, OpenAI
10.1   |   What
API, Google Cloud NLP, and AWS Comprehend. Reinforcement
learning can be implemented in Python (OpenAI Gym, RLlib),
There are 3 Major types of reviews. 1. Domain-­based review.
TensorFlow-­ Agents, or MATLAB Reinforcement Learning
2. Theory-­based review 3. Method-­ based review. Amongst
Toolbox. AI-­assisted qualitative coding is being embedded into
these, domain-­based reviews are more popular, and they can
[Link] and MAXQDA, where machine learning aids the re-
be further classified into 7 different types as follows (Paul and
searcher whilst preserving interpretive validation (Table 11 of-
Criado 2020; Paul, Ueno, et al. 2024)
fers the details for thresholds).
a. Narrative (plain Vanilla) review

9.7   |   Common Misuses of AI in Research b. Framework-­ based review using an Org Framework

(Paul 2024; Hassan et al. 2022; Södergren 2021; Paul and
• Treating algorithms as ‘black boxes’ without theoretical Benito 2018)
grounding. c. Conceptual review aiming for theory development (Paul
• Ignoring bias, fairness, and ethical thresholds in model and Mas 2020 & Majority of the articles in the AMS Review
training. & the Academy of Management (AoM) Review)

• Overclaiming generalisability from synthetic data. d. Bibliometric review (Pattnaik et al. 2020; Paul and
Bhukya 2021; Kaurav and Gupta 2022)
• Using AI outputs as unquestioned ‘truth’ rather than trian-
gulated evidence. e. Hybrid Review (Chakma et al. 2021; Kaurav et al. 2025)
f. Structured, but without an organizing framework
(Dhaliwal et al. 2025; Rosado-­Serrano et al. 2018)
9.8   |   Guidelines for Authors

AI is not just an additional technique but a transformation in 10.2   |   Why

how research is conducted. It redefines ‘data,’ introduces new
units of analysis, and blurs the line between method and tool. Its The rationale for literature reviews as a method lies in their
promise lies in augmenting, not replacing, established research ability to consolidate fragmented knowledge and build future

TABLE 11    |    Thresholds of adequacy for AI-­driven methods.

Method Key thresholds Use cases for marketing research

Generative AI Output validity must be human-­verified; Applied in content creation, campaign testing.
transparency of prompts and parameters
NLP/transformer models Accuracy/recall > 80%; explainability Used in sentiment/emotion detection.
(e.g., SHAP values) desirable
Reinforcement learning Convergence to stable policies; Applied in dynamic pricing,
reward > baseline strategy personalised promotions.
AI-­assisted qualitative coding Agreement with human coders κ > 0.70 Speeds up analysis in netnography, interviews.
Source: Authors.

18 of 23 International Journal of Consumer Studies, 2025

agendas. In social sciences, especially in marketing and busi- 10.5   |   How
ness management research, rapid expansion of subfields, i.e.,
AI, sustainability, and digital platforms means that reviews are Conducting a literature review involves:
vital for:
• Systematic search strategy through databases like Scopus,
• Establishing state of the art. Web of Science, you may use any one or a combination of
these based on your requirements. This step is crucial as
• Synthesising scattered findings.
anyone can replicate the same study, if you are offering a
• Clarifying constructs and measures. clear search strategy.
• Identifying under-­researched gaps. • Inclusion/exclusion criteria for selecting articles.
• Developing integrative frameworks or theories. • Coding and synthesis, either narratively (MAXQDA,
NVivo, [Link]) or quantitatively
Reviews also serve as entry points for doctoral students and new
• Framework development, where constructs, theories, or
scholars, enabling faster immersion into a field.
findings are organised into a structured model (Required
only in the case of Conceptual reviews)
10.3   |   When and Where • Gap identification leading to a future research agenda.

It is important to note that bibliometric and hybrid reviews do

not help identify research gaps; hence, other methods, such as
developing a review using organizing frameworks, are consid-
10.6   |   Common Misuses of Literature Reviews
ered to be better to serve the purpose of deriving research gaps
• Producing descriptive ‘laundry list’ reviews without
to set a future research agenda, which has to stem from research
synthesis.
gaps (Paul et al. 2021; Sahni and Kaurav 2023).
• Over-­
reliance on bibliometrics without conceptual
Researchers would find worth reading specimen articles for contribution.
Theory-­based review (Gilal et al. 2019, 2023; Shahid et al. 2024)
• Failing to derive actionable research gaps.
and method-­based review (Venkatesh et al. 2013; Diwanji 2023).
Reviews are most appropriate when: • Using selective citation rather than systematic coverage.
• Treating reviews as easier or less rigorous than empirical
• A domain has grown fragmented or saturated with primary
research.
studies.
• New theories or phenomena have emerged, requiring con-
ceptual clarity. 10.7   |   Guidelines for Authors
• Researchers seek to provide directions for future research
(e.g., future research pathways). When undertaking a literature review as a method, authors
should begin by clearly articulating the type of review they
Literature reviews are published in Q1 journals including AoM intend to conduct—whether domain-­ based, theory-­based,
Review, Journal of the Academy of Marketing Science (JAMS), or method-­based. Then, researchers can decide what type of
Journal of Business Research (JBR), and in the specialised out- domain-­based review is to be conducted (Paul 2025). This
lets like AMS Review, International Journal of Management clarity not only orients the reader but also sets expectations
Reviews, Human Resource Management Review, Management about the contribution. A rigorous review requires system-
Review Quarterly (MRQ). atic protocols for identifying, screening, and including stud-
ies. Protocols and frameworks such as SPAR-­ 4 -­
SLR (Paul
et al. 2021), POWER (Rana et al. 2022), TMCC, ADO (Sharma
10.4   |   Who et al. 2024) can be adapted to ensure transparency in the se-
lection process. In our opinion, PRISMA is repeatedly used
The unit of analysis depends on the review type: and therefore, it has become redundant and replete without
newness and novelty.
• Domain reviews: focus on industries, consumer segments,
or marketing phenomena. Beyond collecting articles, the review must be designed to
synthesise rather than summarise. Employing conceptual or
• Theory-­based reviews: focus on a theoretical lens in a field.
organising frameworks helps avoid the common pitfall of pro-
e.g., service-­dominant logic stimulus-­organism-­response,
ducing a descriptive ‘laundry list.’ Authors are encouraged to
value-­attitude-­behaviour.
move from reporting what has been studied to explaining how
• Method reviews: focus on research tools and techniques findings connect, diverge, and evolve, thereby offering genuine
across studies, e.g., SEM, FSQCA in a specific field. insight.

International Journal of Consumer Studies, 2025 19 of 23

Finally, every review should culminate in a clear and actionable In a nutshell, the nine families of research methods reflect both
research agenda. Future research directions must be derived the breadth and the evolution of marketing inquiry. Quantitative
explicitly from identified gaps, showing how they connect to approaches provide precision, measurement, and hypothesis test-
theory, methods, or context. By providing specific pathways ing, whilst qualitative methods illuminate meaning, context, and
rather than generic calls, authors demonstrate both scholarly the lived realities of consumers. Mixed methods integrate these
depth and practical utility. A literature review that follows these traditions, demonstrating how numbers and narratives can com-
principles is not an easy shortcut but a robust methodological plement one another rather than compete. Big data and computa-
contribution, capable of advancing the discipline with clarity, tional techniques extend the scope further, enabling the analysis
credibility, and impact. of high-­dimensional, real-­time, and unstructured data that tra-
ditional tools struggle to accommodate. AI-­driven methods ad-
vance this frontier still further, reshaping not only how data are
11   |   Directions for Future Research analysed but also how insights are generated, predictions refined,
and marketing strategies personalised. Each family is distinct, yet
The recent pandemic, geo-­ economic and socio-­ cultural sit- none is sufficient in isolation. The choice of method is not a matter
uations have resulted in structural changes in many sectors of hierarchy but of alignment, ensuring that the what, why, when,
(Alagarsamy et al. 2022; Nayal et al. 2021). i.e., the pandemic where, who, and how of the research question converge with the
has changed the way people think and the resultant consumer method's underlying logic and adequacy thresholds. Taken to-
behaviour (Chakraborty and Paul 2022; Purohit et al. 2022; gether, these families provide scholars with a versatile toolkit that
Chopdar et al. 2022). As an outcome, following experts (Basu balances tradition with innovation, rigour with relevance, and
et al. 2022; Kajol et al. 2022), we concur with the need for devel- theoretical explanation with predictive power.
oping new and novel research methods, scales, and paradigms
to carry out research studies in the post-­pandemic era to anal- Traditional designs, qualitative inquiry, mixed approaches, com-
yse the new processes, patterns, and problems with a unique putational analytics, and AI-­driven techniques all offer unique
approach. possibilities, but their value lies not in novelty or technical

TABLE 12    |    Mapping research methods to the research journey (5WH lens and integration pathways).

Stage of research Primary purpose Typical methods Possible integration

(When) (What/Why) (How/Who/Where) (multi-­stage/multi-­study)
Exploration (early Generate insights, Qualitative methods: case Can feed into survey design (quantitative),
stage) identify constructs, study, ethnography, grounded provide variables for SEM, or
contextual understanding theory, phenomenology, integrate with netnography or digital
focus groups ethnography before large-­N analysis
Measurement Scale development, EFA, CFA, reliability analysis Followed by SEM for structural
development construct validation testing, combined with qualitative
insights for item refinement
Hypothesis testing Test relationships, Regression (linear, logistic, Can be triangulated with qualitative
& causal inference mediation, moderation, multilevel), experiments case findings; often combined with
causal effects (lab, field, quasi) SEM for theory-­driven testing
Comparative/ Understand multiple fsQCA, GQCA Can complement regression/SEM by
configurational causal pathways, showing alternative paths; often combined
analysis ‘equifinality’ with case studies for richer context
Synthesis & Consolidate existing Systematic reviews, Can be integrated as a prelude to
review knowledge, identify gaps bibliometrics, meta-­analysis empirical research, or post-­project
(Tripathi et al. 2024) validation of generalisability
Big data & Analyse large-­scale, real-­ Text mining, social media Can be embedded within survey/
computational time, unstructured data analytics, clickstream, image/ experimental data; often combined with
analytics video analytics, ML/AI qualitative interpretation for context
AI-­driven research Prediction, Generative AI, reinforcement Can be combined with experiments
personalisation, learning, transformer models (AI-­generated stimuli), qualitative
content generation, coding (AI-­assisted), or predictive
adaptive modelling modelling for CRM/retailing
Integration stage Synthesise across Multi-­method/multi-­ Sequential (qual → quant, or quant
methods for robustness study design → qual), concurrent triangulation,
embedded studies, or multiphase designs

20 of 23 International Journal of Consumer Studies, 2025

sophistication alone. The enduring question needs to be answered and Strategic Measures Undertaken for Survival.” Tourism Culture &
scientifically: Is the method aligned with the research problem? Communication 23, no. 2–3: 177–191.
Alagarsamy, S., S. Mehrolia, and J. Paul. 2022. “Masstige Scale:
To further consolidate the 5WH framework, Table 12 provides An Alternative to Measure Brand Equity.” International Journal of
a roadmap that situates different families of methods along the Consumer Studies 48: e12873. [Link]
research journey. It highlights not only when and why a method AlBannai, N. A. 2025. “From Classroom to Boardroom: Ethical and
is most relevant, but also how it can be integrated with others Cognitive Challenges of ChatGPT Use Among DBA Students in Business
in multi-­stage or multi-­study designs. This integrative view em- Problem-­Solving.” International Journal of Management Education 23,
phasises complementarity rather than hierarchy, encouraging no. 3: 101254. [Link] 025.​101254.
researchers to align their methodological choices with both their Anderson, J. C., and D. W. Gerbing. 1988. “Structural Equation Modeling
questions and broader research programmes. in Practice: A Review and Recommended Two-­ Step Approach.”
Psychological Bulletin 103, no. 3: 411–423.
As Table 12 illustrates, methods rarely operate in isolation. Stronger Aruldoss, A., S. Rana, S. Parayitam, and B. Gurumurthy. 2023.
contributions often emerge when exploration leads into measure- “Demystifying Hedonic Shopping Motivation and Consumer Buying
ment, measurement into testing, and testing into integration with Behavior During the Post-­ Global Pandemic: Evidence From a
computational or AI-­driven tools. This journey underscores a cen- Developing Country.” Journal of Marketing Theory and Practice 32, no.
tral point: no method is complete on its own, and methodological 4: 486–505. [Link] 023.​2221442.
rigour in consumer and marketing research increasingly comes Baber, R., R. P. S. Kaurav, and J. Paul. 2020. “Impact of Individual
from thoughtful sequencing and integration across traditions. Market Orientation on Selling and Customer Orientation.” Journal of
Strategic Marketing 28, no. 4: 288–303.
Bagozzi, R. P., and Y. Yi. 1988. “On the Evaluation of Structural Equation
12   |   Conclusions Models.” Journal of the Academy of Marketing Science 16, no. 1: 74–94.
Basu, R., J. Paul, and K. Singh. 2022. “Visual Merchandising and Store
This article provides a simple yet rigorous compass for answer- Atmospherics: An Integrated Review and Future Research Directions.”
ing this question. What clarifies the object of inquiry, why jus- Journal of Business Research 151: 397–408.
tifies its purpose, when and where situates it in context, who Bhalla, R., R. P. S. Kaurav, and N. Chowdhary. 2025. “Meaning Making
identifies the unit of analysis, and how specifies the design and Through Transformative Travel: A Promising Solution to Collective
execution. Thresholds—whether statistical, procedural, or ethi- Madness in Pathogenic Society.” Tourism Recreation Research: 1–20.
cal—ensure adequacy and guard against fragility. [Link] 025.​2533212.
Chakma, R., J. Paul, and S. Dhir. 2021. “Organizational Ambidexterity:
For researchers, the 5WH framework serves not only as a tech- A Review and Research Agenda.” IEEE Transactions on Engineering
nical guide but also as an epistemological bridge: it resists the Management 71: 121–137.
tendency of methods discourse to drift either into abstract philoso- Chakraborty, D., and J. Paul. 2022. “Healthcare Apps' Purchase
phy or narrow technicality. By grounding choices in comparative Intention: A Consumption Values Perspective.” Technovation 120:
reasoning, it helps reconnect the logic of inquiry with the prob- 102481. [Link] 022.​102481.
lems that research is meant to address.
Chawla, G., A. Gupta, R. P. S. Kaurav, and A. Bathla. 2025. “Setting
Quality Standards for Management Education: Lessons From
The framework thus enables authors to justify choices with clar- Internationalization in Indian Business Schools.” International Journal
ity and equips reviewers and editors with a common lens for of Quality & Reliability Management: 1–27. [Link]
evaluating methodological fit. Yet one truth stands out across IJQRM​-­​11-­​2 024-­​0387.
traditions old and new: no single method is ever complete in it- Choi, K. 2026. “Capturing Causal Configurations of Tourism
self. Robust contributions often emerge from multi-­study designs Destination Competitiveness Conditions for Tourism Performance:
or multi-­stage research programmes, where different methods Asymmetric and Symmetric Application.” Journal of Destination
converge to provide stronger evidence. By embracing this in- Marketing & Management 23: 101042. [Link]
tegrative spirit, researchers can balance rigor with relevance, 2025.​101042.
tradition with innovation, and methodological diversity with Chopdar, P. K., J. Paul, and J. Prodanova. 2022. “Mobile Shoppers'
epistemological coherence. Response to Covid-­19 Phobia, Pessimism and Smartphone Addiction:
Does Social Influence Matter?” Technological Forecasting and Social
Change 174: 121249.

Conflicts of Interest Chowdhary, N., R. P. S. Kaurav, and S. Sharma. 2020. “Segmenting the
Domestic Rural Tourists in India.” Tourism Review International 24, no.
The authors declare no conflicts of interest. 1: 23–36.
Crick, J. M. 2021. “Qualitative Research in Marketing: What Can
Data Availability Statement Academics Do Better?” Journal of Strategic Marketing 29, no. 5:
Data sharing not applicable to this article as no datasets were generated 390–429.
or analysed during the current study. Das, M., A. Roy, J. Paul, and V. Saha. 2022. “High and Low Impulsive
Buying in Social Commerce: A SPAR-­4 -­SLR and fsQCA Approach.”
IEEE Transactions on Engineering Management 71: 2226–2240.
References
Dash, G., K. Kiefer, and J. Paul. 2021. “Marketing-­ To-­Millennials:
Aggarwal, P. J., R. P. S. Kaurav, and D. Gursoy. 2023. “Effects of Marketing 4.0, Customer Satisfaction and Purchase Intention.” Journal
(COVID)-­19 Pandemic on Tourism Businesses in India: Operational of Business Research 122: 608–620.

International Journal of Consumer Studies, 2025 21 of 23

Dash, G., and J. Paul. 2021. “CB-­ SEM vs PLS-­ SEM Methods for Huarng, K.-­H., and N. Roig-­Tierno. 2016. “Qualitative Comparative
Research in Social Sciences and Technology Forecasting.” Technological Analysis, Crisp and Fuzzy Sets in Knowledge and Innovation.” Journal
Forecasting and Social Change 173: 121092. of Business Research 69, no. 11: 4721–4726. [Link]
jbusr​es.​2 016.​0 4.​109.
Demydyuk, G. V., and M. Carlbäck. 2023. “Balancing Short-­Term Gains
and Long-­Term Success in Lodging: The Role of Customer Satisfaction Igartua, J.-­
J., and A. F. Hayes. 2021. “Mediation, Moderation, and
and Price in Hotel Profitability Model.” Tourism Economics 30, no. 4: Conditional Process Analysis: Concepts, Computations, and Some
844–875. [Link] Common Confusions.” Spanish Journal of Psychology 24: e49.
Demydyuk, G. V., R. P. S. Kaurav, M. Carlbäck, and H. Vejlgaard. 2024. Kainthola, S., N. Chowdhary, R. P. S. Kaurav, and P. Tiwari. 2024.
“From Galley to Gourmet: Experience Accounting Perspective on the “Motivations of Urban Millennials for Spiritual Travel in India.”
Evolving Dining Choices of Cruise Passengers.” Journal of Foodservice Tourism Recreation Research 49, no. 2: 410–425.
Business Research: 1–33. [Link] 024.​
Kajol, K., R. Singh, and J. Paul. 2022. “Adoption of Digital Financial
2391161.
Transactions: A Review of Literature and Future Research Agenda.”
Dhaliwal, A., D. P. Singh, and J. Paul. 2025. “The Consumer Behavior Technological Forecasting and Social Change 184: 121991.
of Luxury Goods: A Review and Research Agenda.” Journal of Strategic
Kaurav, R. P. S., and P. Gupta. 2022. “Trends in Multidiscipline
Marketing 33, no. 1: 66–92.
Management Research: Past, Present and Future of FIIB Business
Diwanji, V. S. 2023. “Fuzzy-­Set Qualitative Comparative Analysis in Review.” FIIB Business Review 11, no. 4: 382–404.
Consumer Research: A Systematic Literature Review.” International
Kaurav, R. P. S., D. Gursoy, and N. Chowdhary, eds. 2021. An SPSS
Journal of Consumer Studies 47, no. 6: 2767–2789.
Guide for Tourism, Hospitality and Events Researchers. Routledge.
Dul, J. 2024. “A Different Causal Perspective With Necessary Condition
Kaurav, R. P. S., S. Kainthola, R. Baber, and P. Tiwari. 2024. “Assessing
Analysis.” Journal of Business Research 174: 114618. [Link]
Global Perceptions of India: Policy Implications Drawn From Foreign
1016/j.​jbusr​es.​2 024.​114618.
Tourism Narratives.” Turyzm/Tourism 34, no. 1: 57–66. [Link]
Fan, X., S. Zheng, X. Jiang, N. Deng, and P. Xu. 2024. “Understanding 10.​18778/​​0867-­​5856.​3 4.1.​05.
the Configurational Causes of Traveler's Moral Ambiguity in the
Kaurav, R. P. S., S. D. Kulkarni, and H. Shee. 2024. “Bridging Knowledge
Sharing Economy: Evidence From PLS-­fsQCA.” International Journal
Gaps: Exploring Literature Review in Business Management and
of Tourism Research 26, no. 6: 1234–1248. [Link]
Emerging Best Practices.” Abhigyan 42, no. 3: 179–180.
jtr.​2654.
Kaurav, R. P. S., and V. Mishra. 2024. Review of Technologies and
Fornell, C., and D. F. Larcker. 1981. “Evaluating Structural Equation
Disruptive Business Strategies. Emerald Publishing Limited.
Models With Unobservable Variables and Measurement Error.” Journal
of Marketing Research 18, no. 1: 39–50. Kaurav, R. P. S., J. Paul, and N. Chowdhary. 2015. “Effect of Internal
Marketing on Hotels: Empirical Evidence for Internal Customers.”
Gilal, F. G., J. Paul, R. G. Gilal, and N. G. Gilal. 2023. “Advancing Basic
International Journal of Hospitality & Tourism Administration 16, no.
Psychological Needs Theory in Marketing Research.” International
4: 311–330.
Journal of Market Research 65, no. 6: 745–777.
Kaurav, R. P. S., S. Singhania, and M. Kozak. 2025. “Anatolia–Unveiling
Gilal, F. G., J. Zhang, J. Paul, and N. G. Gilal. 2019. “The Role of Self-­
Its Multidisciplinary Landscape and Future Orientations.” Anatolia 36,
Determination Theory in Marketing Science: An Integrative Review
no. 3: 495–512.
and Agenda for Research.” European Management Journal 37, no. 1:
29–44. Kaurav, R. P. S., and A. Thomas. 2025. Marketing Analytics Using Excel:
A Beginner's Guide. Sage.
Han, H., N. G. Baah, S. Kim, X. Chi, and I. Jung. 2025.
“Environmentally Responsible Behaviors in Hospitality and Tourism Kumar, V., R. P. Leone, D. A. Aaker, G. S. Day, and R. P. S. Kaurav. 2025.
Service Employees: An Application of Complexity Theory.” Journal of Marketing Research (13th ed., Indian Adaptation). Wiley.
Service Theory and Practice 35, no. 2: 123–140. [Link]
Mehra, A., J. Paul, and R. P. S. Kaurav. 2021. “Determinants of Mobile
JSTP- ­​02-­​2 024- ­​0 042.
Apps Adoption Among Young Adults: Theoretical Extension and
Han, H., A. K. F. Wong, S. Kim, X. Cheng, and X. Chi. 2025. “Assessing Analysis.” Journal of Marketing Communications 27, no. 5: 481–509.
Asymmetric Selection of Responsible Airlines Based on CSR,
Merminod, V., M. A. Le Dain, and A. G. Frank. 2022. “Managing
Normative, and Emotional Factors: A Complexity Approach With SEM,
Glitches in Collaborative Product Development With Suppliers.” Supply
fsQCA, and NCA.” Asia Pacific Journal of Tourism Research 30, no. 1:
Chain Management: An International Journal 27, no. 2: 210–225. [Link]
1–18. [Link] 024.​2 437630.
doi.​org/​10.​1108/​SCM-­​01-­​2 020-­​0 042.
Hassan, S. M., Z. Rahman, and J. Paul. 2022. “Consumer Ethics: A
Mishra, M., J. Paul, and M. Czinkota. 2022. “Revisiting Models of
Review and Research Agenda.” Psychology & Marketing 39, no. 1:
Internationalization: Pre-­E xport Phase and Lateral Rigidity of Emerging
111–130.
Market Small and Medium Enterprises.” Thunderbird International
Hayes, A. F. 2018. “Partial, Conditional, and Moderated Mediation: Business Review 64, no. 2: 125–138.
Quantification, Inference, and Interpretation.” Communication
Nayal, P., N. Pandey, and J. Paul. 2021. “Covid-­19 Pandemic and
Monographs 85, no. 1: 4–40.
Consumer-­Employee-­Organization Wellbeing: A Dynamic Capability
Hayes, A. F. 2022. Introduction to Mediation, Moderation, and Theory Approach.” Journal of Consumer Affairs 56, no. 1. [Link]
Conditional Process Analysis: A Regression-­Based Approach. 3rd ed. org/​10.​1111/​joca.​12399​.
Gilford Press.
Pattnaik, D., M. K. Hassan, S. Kumar, and J. Paul. 2020. “Trade Credit
Hayes, A. F., and N. J. Rockwood. 2020. “Conditional Process Analysis: Research Before and After the Global Financial Crisis of 2008—A
Concepts, Computation, and Advances in Modeling of the Contingencies Bibliometric Overview.” Research in International Business and Finance
of Mechanisms.” American Behavioral Scientist 64: 19–54. 54: 101287.
Huarng, K.-­ H. 2016. “Qualitative Analysis With Structural Paul, J. 2024. “Publishing in Premier Journals With High Impact Factor
Associations.” Journal of Business Research 69, no. 11: 4727–4731. and Q1 Journals: Dos and Don'ts.” International Journal of Consumer
[Link] 016.​0 4.​110. Studies 48, no. 3: e13049.

22 of 23 International Journal of Consumer Studies, 2025

Paul, J. 2025. “Crafting Impactful Systematic Literature Reviews: Rana, J. Singh, and S. Kathuria, vol. 2, 79–101. Emerald Publishing
Thumb Rules and Suggestions.” International Journal of Consumer Limited.
Studies 49, no. 5: e70103.
Saxena, S., R. P. S. Kaurav, A. Ramasundaram, S. Kataria, and N. B.
Paul, J., and M. Barari. 2022. “Meta-­A nalysis and Traditional Systematic Halvadia. 2025. “AI Enabled Travel: A MOA-­Nificent Journey.” Asia
Literature Reviews—What, Why, When, Where, and How?” Psychology Pacific Journal of Tourism Research 30: 1–19.
& Marketing 39, no. 6: 1099–1115.
Shahid, Z. A., M. I. Tariq, J. Paul, S. A. N. Naqvi, and L. Hallo. 2024.
Paul, J., and G. R. Benito. 2018. “A Review of Research on Outward “Signaling Theory and Its Relevance in International Marketing:
Foreign Direct Investment From Emerging Countries, Including China: A Systematic Review and Future Research Agenda.” International
What Do We Know, How Do We Know and Where Should We Be Marketing Review 41, no. 2: 514–561.
Heading?” Asia Pacific Business Review 24, no. 1: 90–115.
Sharma, A., R. P. S. Kaurav, and S. Koul. 2024. “Understanding Customer
Paul, J., and R. Bhukya. 2021. “Forty-­F ive Years of International Confusion in the Marketplace–A Systematic Literature Review.” Indian
Journal of Consumer Studies: A Bibliometric Review and Directions Journal of Marketing 54, no. 12: 8–28.
for Future Research.” International Journal of Consumer Studies 45, Sharma, N., R. Payal, and J. Paul. 2025. “Hello! How Can I Help You?
no. 5: 937–963. Role of Technology Appetite on AI Adoption in the Post-­Pandemic Era.”
Paul, J., and A. R. Criado. 2020. “The Art of Writing a Literature Review: R&D Management 55: 1560–1581. [Link]
What Do We Know and What Do We Need to Know?” International Sima, H., P. Thaichon, N. Luo, K. Henderson, and J. Chen. 2025.
Business Review 29, no. 4: 101717. “Reimagining Consumption for a Greener Planet: The Role of Online
Paul, J., P. Khatri, and H. Kaur Duggal. 2023. “Frameworks for Communities in Tackling Imperfect Produce and Food Waste.” Journal
Developing Impactful Systematic Literature Reviews and Theory of Business Research 171: 115701. [Link] 025.​
Building: What, Why, and How?” Journal of Decision Systems 33, no. 4: 115701.
537–550. [Link] 023.​2197700. Södergren, J. 2021. “Brand Authenticity: 25 Years of Research.”
Paul, J., and E. Mas. 2020. “Toward a 7-­P Framework for International International Journal of Consumer Studies 45, no. 4: 645–663.
Marketing.” Journal of Strategic Marketing 28, no. 8: 681–701. Srivastava, K., M. H. Siddiqui, R. P. S. Kaurav, S. Narula, and R. Baber.
Paul, J., A. Merchant, Y. K. Dwivedi, and G. Rose. 2021. “Writing an 2024. “The High of Higher Education: Interactivity Its Influence
Impactful Review Article: What Do We Know and What Do We Need to and Effectiveness on Virtual Communities.” Benchmarking: An
Know?” Journal of Business Research 133: 337–340. International Journal 31, no. 10: 3807–3832.

Paul, J., and S. Sahadev. 2018. “Service Failure and Problems: Internal Stoner, J. L., R. Felix, and A. Stadler Blank. 2023. “Best practices for
Marketing Solutions for Facing the Future.” Journal of Retailing and Implementing Experimental Research Methods.” International Journal
Consumer Services 40: 304–311. of Consumer Studies 47, no. 4: 1579–1595.

Paul, J., A. Ueno, and C. Dennis. 2023. “ChatGPT and Consumers: Tiwari, P., R. P. S. Kaurav, and K. Y. Koay. 2024. “Understanding
Benefits, Pitfalls and Future Research Agenda.” International Journal Travel Apps Usage Intention: Findings From PLS and NCA.” Journal of
of Consumer Studies 47, no. 4: 1213–1225. Marketing Analytics 12, no. 1: 25–41.

Paul, J., A. Ueno, C. Dennis, et al. 2024. “Digital Transformation: Tomczyk, P., P. Brüggemann, and J. Paul. 2024. “Variable Science
A Multidisciplinary Perspective and Future Research Agenda.” Mapping as Literature Review Method.” Journal of Marketing Analytics
International Journal of Consumer Studies 48, no. 2: e13015. 12, no. 4: 829–841.

Purohit, S., R. Arora, and J. Paul. 2022. “The Bright Side of Online Tripathi, C. M., R. P. S. Kaurav, and T. Singh. 2024. “How Does Your
Consumer Behavior: Continuance Intention for Mobile Payments.” Cultural Intelligence Contribute to Your Adjustment? Unveiling the
Journal of Consumer Behaviour 21: 523–542. [Link] Link Between Cultural Intelligence and Cross-­Cultural Adjustment
cb.​2 017. Using Meta-­A nalysis.” Journal of Global Mobility: The Home of
Expatriate Management Research 12, no. 1: 167–197.
Ramtiyal, B., G. Sakka, K. Mehmood, S. Narula, and S. Verma. 2025.
“Harnessing the Power of AI for Unlocking Indian Born-­ Global Venkatesh, V., S. A. Brown, and H. Bala. 2013. “Bridging the Qualitative-­
Entrepreneurs in the Global Market Profile.” International Marketing Quantitative Divide: Guidelines for Conducting Mixed Methods
Review: 1–30. [Link] MR-­​10-­​2 024-­​0 428. Research in Information Systems.” MIS Quarterly 37: 21–54.

Rana, S., S. Bag, I. Ghosal, and B. Prasad. 2023. “How Do Influencers Zhao, Y., and Y. Zhan. 2025. “From Hype to Reality: Unveiling Consumer
Moderate Purchase Motives? An Application of S-­O -­R Model to Assess Resistance to NFTs Through a Hybrid SEM-­A NN Approach.” Electronic
the Effects of Reviewers' Rating on Online Purchase.” Journal of Commerce Research and Applications 66: 101537. [Link]
Promotion Management 29, no. 8: 1168–1197. [Link] 1016/j.​elerap.​2 025.​101537.
10496 ​491.​2 023.​2216292. Zheng, L. 2023. “How Racial Diversity and Gender Diversity in Job
Rana, S., S. Kathuria, and J. Singh. 2022. “Presenting the POWER Positions Affect the Economy?” Contemporary Economics 17, no. 2:
Framework of Conducting Literature Review.” In Exploring the Latest 109–127.
Trends in Management Literature, vol. 1. Emerald Publishing Limited.
Rosado-­Serrano, A., J. Paul, and D. Dikova. 2018. “International
Franchising: A Literature Review and Research Agenda.” Journal of
Business Research 85: 238–257.
Roy, A., N. Gera, and G. Das. 2025. “Safety First! Wait, Ambition First!
How Parental Psychology Shapes Purchase Preferences for Their
Children.” Psychology & Marketing: 1–12. [Link]
70055​.
Sahni, S., and R. P. S. Kaurav. 2023. “What? Why? When? How? Where?
of Technology-­Based Bibliometric Review.” In Advancing Methodologies
of Conducting Literature Review in Management Domain, edited by S.

International Journal of Consumer Studies, 2025 23 of 23