HTM3205 Analyzing and Interpreting Research - Test 2 Comprehensive Vocabulary
Course Overview and Five-Day Examination Review Strategy
Curricular Details and Context
- Subject Code & Title: HTM3205 Analyzing and Interpreting Research (referenced in internal departmental course documents as HTM 3025).
- Academic Term: 2026–27 Semester 1.
- Assessment Details: Test 2; administered on Monday, November 23, 2026; weighting accounts for of the total course grade; conducted on-site in class.
- Curricular Scope: Cumulative examination covering all lecture and tutorial content from Week 1 through Week 12.
- Foundational Source Materials: Compiled across 46 instructional files, encompassing lecture slide presentations, tutorial supplementary notes, workbook exercise keys, and the Week 6 Mock Examination.
- Language of Instruction and Assessment Strategy: Theoretical concepts are explained in Chinese and English cross-references, with formal academic terminology preserved strictly in English alongside standard English answer templates to facilitate written examinations in English.
Five-Day Intensive Sprint Schedule
- Allocated Effort: Approximately per day spanning Wednesday, November 18 through Sunday, November 22. Analytical data procedures in SPSS are allocated an independent day due to high assessment weighting and common calculation errors.
- November 18 (Wednesday) — Modules 1 to 4 (Foundations of Research, Process, Literature Review, and Data Sources):
- Target: Close all notes and articulate aloud in English the distinctions between pure and applied research, deductive and inductive logic, the three research designs, and validity versus reliability.
- November 19 (Thursday) — Modules 5 to 8 (Sampling, Data Collection & Questionnaires, Qualitative Methods, Ethics & Reporting):
- Target: Write out from memory the four probability and four non-probability sampling techniques with illustrative hospitality examples; memorize the four fundamental ethical principles.
- November 20 (Friday) — Module 9 (SPSS Comprehensive Statistical Analysis):
- Target: For each statistical test, verbally formulate without notes the sequence of inspection: which exact metric to inspect first, the decision rule applied, and the phrasing of the final analytical conclusion.
- November 21 (Saturday) — Modules 10 to 11 (Comparative Analysis and Bilingual Glossary):
- Target: Cover the right-hand column of the confusable concepts matrix to practice active recall; randomly test knowledge on 20 terms from the terminology glossary.
- November 22 (Sunday) — Module 12 (Critical Pitfalls and Full Practice Examination):
- Target: Complete the practice examination within a strict time limit, cross-check answers against the provided keys, and review only the incorrectly answered concepts.
The Nature and Types of Research
Underlying Purpose of Research
- Formal Definition: Research is conducted to solve a problem. Its explicit purpose is to reduce uncertainty, facilitate better-informed decisions, and resolve operational or theoretical difficulties.
- Core Axiomatic Assumption: If no problem exists, there is no justification for conducting research.
- Four Primary Justifications for Studying Research Methods:
- Enables researchers and practitioners to comprehend, critically evaluate, and interpret published research reports and academic articles.
- Equips future researchers with methodological procedures to investigate, analyze, and resolve academic problems systematically.
- Empowers managers and executives to evaluate whether a commissioned market research report is methodologically rigorous or flawed.
- Eliminates common myths and misperceptions surrounding the research process.
Pure (Basic) Research vs. Applied Research
- Pure (Basic) Research:
- Verbatim Definition: Attempts to expand the limits of knowledge; not directly involved in the solution to a pragmatic problem.
- Operational Characteristic: Features no immediate real-world application ().
- Example: Investigating whether biological gender differences influence overall preferences when selecting international travel destinations.
- Applied Research:
- Verbatim Definition: Conducted when a decision must be made about a specific real-life problem.
- Operational Characteristic: Aims to resolve an immediate operational issue ().
- Example: Assessing consumer attitudes toward a newly introduced airline frequent flyer loyalty program to decide on service modifications.
Primary Research vs. Secondary Research
- Primary Research: The systematic collection of original empirical data () utilizing recognized, formal research instruments.
- Secondary Research: The synthesis and interpretation of existing literature and pre-collected data without gathering original empirical evidence; practically all research projects incorporate an element of secondary research during early stages.
- Analytical Distinction between Research Types and Data Types:
- When classifying research, evaluate whether the project collected original empirical data.
- When classifying data (Module 4), evaluate for whose specific purpose the data was originally gathered.
Deductive vs. Inductive Approaches
- Deductive Research Approach:
- Point of Departure: Originates with an established theory or abstract concept; gathers empirical observations to analyze and subsequently accept or reject the hypothesized model.
- Timing of Conceptualization: Theorizing occurs prior to the research study ().
- Three-Step Sequence: Hypothesis / Theory Empirical Observation Accept or Reject Hypothesis.
- Core Logic: Top-down reasoning moving from the general rule to the specific case (theory-first, empirical verification).
- Course Benchmark Example: "All HTM students are human; David is an HTM student; therefore, David is human" represents a deductive syllogism (Major Premise Specific Case Logical Conclusion).
- Inductive Research Approach:
- Point of Departure: Investigates a specific operational dimension of tourism or hospitality; gathers descriptive qualitative or quantitative data and abstracts general patterns to formulate a new conceptual framework.
- Timing of Conceptualization: Theorizing occurs after the research data collection ().
- Three-Step Sequence: Empirical Observation Pattern Identification Formulation of Tentative Conclusions/Theory.
- Core Logic: Bottom-up reasoning moving from specific individual instances to broad generalizable theories (data-first, theory generation).
- Course Benchmark Example: "All managers who have ever existed were human; therefore, all managers are human" represents inductive generalization based on cumulative historical observations.
Quantitative vs. Qualitative Research Paradigms
- Quantitative Research: Empirical data gathered and presented in structured numerical format.
- Qualitative Research: Non-numerical empirical data focusing on subjective meanings, linguistic expressions, behavioral nuances, and lived experiences.
- Sequential Mixed-Method Logic: Researchers frequently deploy a sequential design where qualitative methods explore and explain why underlying relationships exist, followed by quantitative methods to verify the statistical significance and empirical magnitude of those relationships.
Problem Definition and the Iceberg Principle
- Primacy in the Research Hierarchy: Problem definition represents Step 1 in the empirical cycle; it is universally acknowledged as the most critical and challenging phase. It pinpoints the operational difficulty and bounds the inquiry via explicit research questions.
- The Iceberg Principle:
- Conceptual Axiom: The most dangerous, complex dimensions of an organizational problem are submerged beneath the surface—neither immediately visible nor intuitively understood by managers.
- Methodological Imperative: The researcher must systematically isolate and identify the underlying problem itself, rather than merely treating visible surface symptoms ().
- Illustrative Scenario: A television executive encounters plummeting viewership ratings and commissions a study on "how to increase ratings." Declining viewership is a superficial symptom; the root problem may stem from shifting audience demographics, counter-programming by competitors, suboptimal broadcasting timeslots, or obsolete content positioning. Investigating solely "how to raise viewership" misdirects the inquiry.
- Cascading Impact: An erroneous problem definition invalidates every subsequent operational step: the research design choice (exploratory versus causal), target population boundaries, survey item formulation, and statistical tests will be structured around a false premise.
The Research Process and Research Design
The Six Chronological Stages of Research
- Stage 1 — Problem Discovery and Problem Definition: Articulate the formal problem statement and delineate overarching research aims and specific operational objectives.
- Stage 2 — Research Design: Select the overarching framework (Exploratory, Descriptive, or Causal); determine the operational methodology from four primary classes:
- Experiment: Controlled laboratory testing or real-world field experimentation.
- Survey: Structured face-to-face interviews, telephone inquiries, or self-administered questionnaires.
- Observation: Systematic tracking of subjects in naturalistic or simulated environments.
- Secondary Data Study: Systematic aggregation and analysis of existing records.
- Stage 3 — Sampling: Define the target population, establish an operational sampling frame, choose between probability and non-probability sampling techniques, and determine the sample size.
- Stage 4 — Data Gathering: Deploy the selected measurement instrument in the field (distribute questionnaires, execute interviews, record observations, run laboratory protocols, or extract administrative data).
- Stage 5 — Data Processing and Analysis: Complete data editing data coding computer data entry statistical/thematic data analysis, followed by rigorous error verification.
- Stage 6 — Conclusions and Report: Interpret statistical/thematic outputs, extrapolate theoretical and practical implications, provide actionable recommendations, and compile the final research report.
Four-Step Formulation of a Problem Statement
- Step 1: Identify the broad, general research topic.
- Step 2: Formulate specific research questions addressing the core issues.
- Step 3: Convert the interrogative research questions into a formal declarative statement.
- Step 4: Refine the statement to encapsulate the negative nature of the problem ()—explicitly highlighting what empirical knowledge is lacking, what operational deficit exists, or what negative outcomes are currently occurring.
- Practical Case Illustration (Hotel Jaystaycationer Study):
- Topic: Hotel employee perceptions of "jaystaycationer" behaviors.
- Questions: What are these aberrant behaviors, what causes them, how do they impact staff, and how can they be mitigated?
- Declarative Statement: "The study aims to investigate jaystaycationer behaviours from the hotel worker’s perspective and how these behaviours affect the overall working experience in hotels."
- Context & "Stating the Negative": Incorporate contextual dynamics (the COVID-19 pandemic forcing hotels into local staycation markets, romanticized marketing of staycations, and incidents such as the Peninsula Hotel "wild birthday party") while underscoring that prior hospitality research has entirely neglected the frontline employee perspective.
SMART Criteria for Research Objectives
- Specific (): Concise, unambiguous, and focused on defined operational variables.
- Measurable (): Capable of being quantified or systematically observed through standardized metrics.
- Achievable (): Realistic and feasible within practical field conditions.
- Realistic (): Backed by the requisite resources, funding, specialized skill sets, and technological access.
- Time-bound (): Achievable within a defined, realistic research schedule.
- Comparative Critique of Research Objectives:
- Flawed Formulation 1: "To examine the extent to which are different behaviors and motivations related to food tourism among people from different social classes and nationalities." Critique: Conflates behaviors with psychological motivations and leaves target groups overly broad.
- Flawed Formulation 2: "To consider the future importance of the Internet for holiday promotion by major Asian tour operators." Critique: The verb "consider" is broad and unmeasurable; "promotion" fails to specify whether it means consumer advertising, distribution, direct reservation engines, or brand marketing.
- Exemplary Formulation 1: "To investigate the ways in which small- and medium-sized accommodation providers in Hong Kong make use of the Internet in their businesses."
- Exemplary Formulation 2: "To examine the nature of the relationship between customer satisfaction and repeat patronage in fine dining restaurants of Hong Kong."
- Core Attributes of Strong Objectives: Narrow, clearly bounded geographically and demographically, structured around explicit constructs, leaving no ambiguity for misinterpretation.
The Three Primary Research Designs
- Exploratory Research:
- Problem Definition Level: Completely ambiguous; the researcher is unaware of the exact nature of the problem ().
- Core Objective: Serves as initial research to clarify ambiguities and define the nature of a problem; does not provide conclusive evidence; presupposes subsequent follow-up research.
- Typical Questions: "Sales have dropped significantly, but we have no idea why"; "Would consumers be receptive to this brand-new, unprecedented product concept?"
- Applied Example: Assessing how hospitality employees react to a newly designed hotel budgeting procedure.
- Descriptive Research:
- Problem Definition Level: Partially defined; the researcher is aware that a problem exists ().
- Core Objective: Systematically characterizes the traits, demographics, or behaviors of a population or phenomenon; answers the descriptive questions of , , , , and ; evaluates consumer attitudes, purchasing intentions, usage patterns, or competitor strategies.
- Typical Questions: "What demographic segments purchase our products?"; "Which hotel amenities do business travelers value most?"
- Applied Examples: Evaluating frontline hotel staff job satisfaction levels; profiling the demographic characteristics of Hong Kong Disneyland's primary target visitors.
- Causal Research:
- Problem Definition Level: Highly structured; the problem is clearly and unambiguously defined ().
- Core Objective: Identifies cause-and-effect relationships between isolated variables; determines whether manipulation of one factor causes predictable shifts in another.
- Typical Questions: "Will redesigning our packaging increase retail sales?"; "Which of these two promotional advertising campaigns generates higher booking conversion rates?"
- Applied Example: Quantifying the direct impact of digital advertising expenditure on room night sales.
- Uncertainty Continuum: Research design positions along a spectrum of environmental uncertainty: Exploratory (absolute ambiguity) Descriptive (transitional awareness) Causal (absolute clarity and structural control).
- Analytical Assessment Rule: Do not classify research design based on the count of variables; evaluate the degree of clarity surrounding the underlying research problem.
Constructs: Independent Variables and Dependent Variables
- Independent Variable (): The variable that influences, causes, or accounts for variation in another construct; the presumed causal agent manipulated or measured by the researcher.
- Dependent Variable (): The outcome variable that changes in response to alterations in the independent variable; the measured effect.
- Analytical Parsing Rule: In typical causal formulations structured as "", factor represents the and factor represents the .
- Applied Exercise: "The performance of employees is influenced by their motivation to work." In this framework, motivation to work is the , and the performance of employees is the .
Variables and Three Analytical Levels
- Variable: Any measurable characteristic, trait, or attribute that varies across individuals or shifts over time (e.g., chronological age, study hours, classroom preferences, satisfaction scores).
- Univariate Analysis: Analyzes a single variable in isolation; primarily descriptive rather than explanatory; used to identify baseline distributions and screen for data entry errors. Common techniques include frequency counts, mean, median, mode, range, and standard deviation.
- Bivariate Analysis: Investigates the statistical relationship or difference across two variables. Focuses on testing group mean differences (independent-samples t-test, paired-samples t-test) or assessing inter-variable association (Pearson correlation, cross-tabulation with chi-square).
- Multivariate Analysis: Simultaneously investigates complex interrelationships across three or more variables (e.g., multiple linear regression, exploratory and confirmatory factor analysis).
Four Sequential Steps of Data Processing
- Editing: The systematic inspection and adjustment of raw data to correct omissions, illegibility, and logical inconsistencies, preparing data for coding and storage.
- Coding: The technical process of identifying and assigning a numerical score or other character symbol () to edited data values (e.g., coding Yes as , No as ; categorizing motivations on a scale).
- Data Entry: Transferring coded values into a computer software package (e.g., SPSS).
- Data Analysis: Executing statistical procedures, coupled with post-entry error verification and validation sweeps.
Literature Review
Academic Definition and Strategic Purposes
- Verbatim Definition: A literature review involves identifying and engaging with previously published research relevant to the topic of interest.
- Core Conceptual Function: Establishing what academic knowledge already exists regarding a specific topic, and highlighting gaps that remain unexplored. Its primary scholarly objective is systematic gap identification ("fault-finding").
- Three Core Academic Functions:
- Prevents researchers from duplicating previously completed studies ().
- Illustrates the developmental trajectory of prior scholarship (), uncovering future research trajectories while refining the present inquiry's conceptual framework.
- Demonstrates the researcher's mastery over a defined body of knowledge and establishes academic credibility ().
- Foundational Structural Role: The literature review serves as the structural bedrock upon which the entire research inquiry is built.
The Five Typologies of Literature Reviews
- Self-study Reviews: Conducted by researchers to educate themselves and build baseline personal literacy within a specific research domain.
- Context Reviews: Positions a study within the broader academic landscape, showing how the present investigation contributes to existing scholarship.
- Historical-theoretical Reviews: Traces the chronological evolution of theoretical paradigms and analytical models used to explain a phenomenon over time.
- Methodological Reviews: Evaluates and critiques the empirical methodologies, measurement scales, and research designs used in prior studies. Benchmark Example: Comparing the (Perceptions minus Expectations) gap model of against the performance-only () measurement model of in quantifying perceived service quality.
- Integrative Reviews: Synthesizes disparate, unconnected research streams into an integrated whole to assess the state of an emerging field. Benchmark Example: Pearce's (1982) formulation of the Travel Career Ladder model adapting Maslow's hierarchy of needs to tourism motivation.
Seven Evaluative and Compositional Principles
- Principle 1 — Structural Mapping: Begin with an introductory paragraph outlining the thematic organization of the review, using descriptive subheadings to guide readers (e.g., opening with: "This chapter … is organised into four sections: (1)… (2)…").
- Principle 2 — Synthesis Essay Format: Compose the review as an integrated, flowing synthesis essay rather than an annotated bibliography of isolated summaries. Group related studies within shared thematic paragraphs.
- Principle 3 — Focus on Empirical Findings: Emphasize the core empirical findings of previous researchers rather than merely reciting their methodologies and variable lists (e.g., synthesizing findings from Swan & Combs, Maddox, Hausknecht, and Cadotte & Turgeon collaboratively).
- Principle 4 — Trace Thematic Trends: Map broad conceptual trajectories and methodological movements within the literature (e.g., categorizing cross-cultural hospitality research into three distinct schools of thought).
- Principle 5 — Identify Literature Gaps: Point out gaps in prior scholarship where phenomena remain unexamined or inconclusive (e.g., explicitly stating "we still know relatively little about…").
- Principle 6 — Delineate Study Distinctiveness: State how the current investigation diverges from earlier work regarding conceptualization, operational measurement, sample composition, or analytical techniques (e.g., contrasting US-based service models with emerging Chinese consumer behaviors).
- Principle 7 — Quantify Problem Magnitude: Provide empirical metrics, market volumes, or population percentages to illustrate the real-world scale of the problem (e.g., citing international tourist arrival statistics and foreign exchange earnings in China).
Information Retrieval and Digital Citations
- Source Hierarchy: Prioritize peer-reviewed academic journal articles, foundational textbooks, and authoritative systematic review papers. Balance classic landmark studies with contemporary publications.
- Consultative Channels: Consult thesis advisors, academic mentors, subject librarians, and colleagues.
- Digital Citation Protocol: Ensure retrieval from authoritative versions; list volume and inclusive page numbers; append Digital Object Identifiers () to the citation string without a trailing period; provide the operational homepage URL when a is unavailable.
American Psychological Association (APA) Reference Protocols
- Reference List Compilation: Include only sources cited within the text; order entries alphabetically by the first author's surname; position the list at the conclusion of the manuscript.
- Single-Author Book Format: Surname, Initial(s). (Year). Book title in italics. Location: Publisher. Example: Jennings, G. (2001). Tourism Research. Sydney, Australia: John Wiley & Sons.
- Multi-Author Book Format: Link co-authors using an ampersand (&). Example: Bordens, K., & Abbott, B. (2002). Research design and methods. Boston, MA: McGraw-Hill.
- Edited Book Chapter Format: Chapter Author(s). (Year). Chapter title. In Editor Initial(s). Surname (Ed.), Book title in italics (pp. inclusive page numbers). Location: Publisher. Example: Pine, R. J. (1996). Globalization in the hotel industry. In R. Kotas (Ed.), The international hospitality business (pp. 96-103). London: Cassell/HCIMA.
- Academic Journal Article Format: Author(s). (Year). Article title. Journal Title in Italics, Volume in Italics(Issue in parentheses), inclusive page numbers. Example: Heung, V. C. S., & Leung, T. K. P. (1998). Co-operative approach to marketing… Journal of Travel & Tourism Marketing, 7(2), 105-118.
- Published Conference Proceedings: Editor(s) (Ed.). (Year). Proceedings Title in Italics. Conference Name, Date: Location. Example: Heung, V. C. S., Ap, J., & Wong, K. (Eds.). (1999). Conference Proceedings: Tourism 2000… Asia Pacific Tourism Association, Fifth Annual Conference (August 23-25): Hong Kong SAR, China.
- Individual Conference Paper: Author(s). (Year). Paper title, inclusive page numbers. In Editor Initial(s). Surname (Ed.), Proceedings Title in Italics. Example: Wong, S. (1996). Cross-cultural management in the Hong Kong hotel industry, pp. 249-263. In G. Prosser (Ed.), Conference Proceedings: Tourism and Hospitality Research…
- Government Institutional Report: Institutional Author. (Year). Document Title in Italics. Location: Publisher. Example: Census & Statistics Department. (1998). Hong Kong economic trends. Hong Kong: Hong Kong Government Information Services.
- Technical / Industry Report: Author or Agency. (Year). Report Title in Italics. Location: Publisher. Example: Hong Kong Tourism Board. (2002). A statistical review of tourism 2002. Hong Kong: Hong Kong Tourist Association.
- Doctoral Dissertations / Master's Theses: Author. (Year). Dissertation title in italics (Doctoral dissertation or Master's thesis). Awarding University, Location. Example: Ap, J. (1992). Understanding host resident’s perceptions of the impacts of tourism through social exchange theory (Doctoral dissertation). Texas A&M University, USA.
- Newspaper Article: Author. (Year, Month Day). Article title. Newspaper Name in Italics. Example: Smith, P. S. (1999, October 23). The finest dining places in Hong Kong. South China Morning Post.
- Electronic Journal Article with DOI: Formatted identical to a journal article, concluding with the unpunctuated identifier. Example: Brickera, K. S., & Berno, T. (2002). Sustainable tourism curriculum development. Annals of Tourism Research, 29(2), 562-564. doi:10.1016/S0160-7383(01)00045-7
- Electronic Journal Article with URL: Formatted identical to a journal article, concluding with the retrieval declaration. Example: Davenport, F. (2010). Hotel operators proved successful social media strategies. Hospitality Practical Advice, 6, 4-13. Retrieved from http://www.hospitalityPA.org
- Common APA Formatting Errors:
- Book entries: Omitting parentheses around the publication year (writing "Jennings, G. 2001." instead of "Jennings, G. (2001).").
- Journal entries: Erroneously italicizing the individual article title instead of the journal title and volume number.
- Conference papers: Placing pagination details at the end of the entry rather than directly adjacent to the paper title prior to the "In" statement.
- Government documents: Inverting the entry by listing the document title before the issuing department name.
In-Text Citation Conventions
- Indirect Citations (Paraphrasing): Cite author surname and year; specific page numbers are unnecessary. Examples: Tan (2002) found that… or In a recent study (Ap, 2002), it was argued that…
- Dual Authorship: Explicitly cite both author surnames across all in-text appearances. Examples: Tan and Heung (2002) noted… or (Ap & Pine, 2002).
- Three or More Authors: List all co-author surnames in the initial citation; abbreviate to the first author's surname followed by "et al." in all subsequent citations. Example: First citation: Parasuraman, Zeithaml, and Berry (1988); subsequent citations: Parasuraman et al. (1988).
- Same Author with Multiple Publications in the Same Year: Differentiate publications chronologically by affixing lowercase letters to the publication year. Examples: (Pine, 1995a) and (Pine, 1995b).
- Multiple Independent Sources within Parentheses: Order citations alphabetically by the first author's surname, separating distinct studies with semicolons. Example: (Ap, 2002; Chon, 1994; Heung, 1998; Qu & Tsang, 2000).
- Personal Communications (Letters, Electronic Mail, Telephone Conversations): Excluded entirely from the terminal Reference List; cited exclusively within body text, including full date. Example: K. Chon (personal communication, March 18, 2002) indicates…
- Direct Quotations: Mandatory inclusion of exact page numbers. If five or more consecutive words match the source text verbatim, quotation marks are mandatory; passages exceeding 40 words must be set off as freestanding, indented block quotations.
- Methodological Reasons for Academic Citations: Enhances scholarly credibility, avoids plagiarism, and provides readers with an audit trail to verify source documentation.
Data Sources: Primary vs. Secondary
Operational Definitions
- Primary Data: Original empirical evidence gathered directly by the researcher from research subjects or operational study units specifically to address the research problem at hand.
- Secondary Data: Pre-existing empirical records gathered by an external entity for an alternative primary objective, subsequently repurposed by a secondary researcher for an independent study.
Primary Data: Sources, Strengths, and Weaknesses
- Data Gathering Sources: Structured and unstructured observational protocols, standardized self-administered questionnaires, and individual or group qualitative interviews.
- Methodological Advantages: High topical relevance and currency (); targeted population alignment (); absence of external investigator bias and unverified methodological errors ().
- Methodological Disadvantages: Participant reactivity ()—subjects alter behavior due to the awareness of being observed; invasiveness (); high time investment (); extensive financial and resource expenditure ().
Secondary Data: Sources, Strengths, and Weaknesses
- Data Gathering Sources: Public archives, personal diaries/correspondence, institutional records, formal industry publications, and statistical releases from tourism bodies (Global: World Tourism Organization [UNWTO]; Regional: Pacific Asia Travel Association [PATA]; Local: Hong Kong Tourism Board [HKTB]).
- Methodological Advantages: Longitudinal retrospectivity ()—enables historical analysis across prior time periods; rapid and low-friction retrieval; low financial cost; uniquely accessible source for historical or macro-scale phenomena; established institutional data quality; non-reactive ()—historical data is free from participant observation bias.
- Methodological Disadvantages: Incomplete data coverage (); undocumented measurement biases and methodological errors undermining reliability and validity; unverified population representativeness; limited contextual applicability to the new research problem.
The Foundational Concepts of Validity and Reliability
- Validity:
- Verbatim Definition: The extent to which the information collected by the researcher truly reflects the phenomenon being studied.
- Core Meaning: The accuracy of operational measurement—whether the research tool captures the intended underlying construct.
- Example: If an investigator intends to measure customer "brand commitment" but merely asks "Have you heard of this brand?", the tool measures brand awareness rather than commitment, resulting in low validity.
- Reliability:
- Verbatim Definition: The extent to which research findings would be the same if repeated at a later date and/or with a different sample of subjects.
- Core Meaning: The stability, dependability, and repeatability of empirical measurements over time and across alternative respondent cohorts.
- Example: Administering an identical survey instrument to a comparable cohort of tourists next week and obtaining identical response distributions demonstrates high reliability.
- The Scale Analogy: A diagnostic scale that consistently registers weights as below actual mass exhibits perfect reliability (it yields an identical measurement bias each iteration), but lacks validity because it fails to capture true body mass.
- Typical Secondary Data Profile: Secondary datasets frequently present high reliability (collected via rigorous bureaucratic procedures) but compromised validity relative to the secondary researcher's goals, as the items were designed for a different original objective.
Three Strategic Objectives of Secondary Data Inquiries
- Fact Finding: Systematically compiling descriptive information to underpin operational decisions (e.g., establishing demographic profiles, segmenting consumer behavior, tracing longitudinal market trends).
- Model Building: Repurposing secondary metrics to mathematically estimate structural relationships among variables, generating descriptive or predictive equations (e.g., forecasting regional hotel demand, calculating market potential).
- Database Marketing: Maintaining dynamic customer databases to drive targeted relationship marketing (e.g., compiling contact directories, logging historical booking behaviors, evaluating promotional campaign conversions, integrating multi-channel loyalty metrics such as Asia Miles or Hilton Honors).
Primary vs. Secondary Data Comparison Matrix
- Collection Purpose: Primary data is collected directly for the problem at hand; secondary data is collected for other problems.
- Collection Process: Primary data gathering is very involved; secondary data retrieval is rapid and easy.
- Collection Cost: Primary data collection costs are high; secondary data costs are relatively low.
- Collection Time: Primary data requires a long time horizon; secondary data requires a short timeframe.
- Academic Fieldwork Connection: SHTM's longitudinal international tourist tracking database contains over 6,000 interview records. When students extract a subset of this database to investigate specialized research questions, they are engaging in secondary data analysis.
Sampling Design and Methodologies
Baseline Concepts: Sample vs. Census
- Sample: A representative subset extracted from a larger parent population ().
- Census: An exhaustive investigation examining every individual element constituting the target population ().
- Methodological Justifications for Sampling:
- Pragmatic Operational Constraints: Commercial, social, and academic research initiatives operate under finite budget constraints and stringent project deadlines.
- Data Accuracy and Quality Assurance: Well-managed samples can produce higher overall accuracy than a census, as exhaustive censuses are prone to severe non-sampling and administrative errors.
Error Typologies: Random Sampling Error vs. Non-Sampling Error
- Random Sampling Error:
- Etiology: Statistical fluctuation occurring purely by chance due to the probabilistic selection of specific sampling units.
- Definition: The statistical divergence between the sample outcome and the theoretical result obtained from an identical census of the entire target population.
- Example: A properly executed random draw happens by chance to include a slightly younger demographic cohort than the population distribution.
- Sample Size Relationship: Inversely correlated with sample size; expanding the sample reduces random error (e.g., a sample of yields a margin of error of , whereas yields at a confidence level).
- Non-Sampling Error (Systematic Error):
- Etiology: Flaws in research design, administrative oversights, or execution errors rather than chance probabilistic variation.
- Definition: Systematic operational biases resulting in response errors, recording inaccuracies, data coding mistakes, non-response bias, or sampling frame flaws.
- Example: Systematic loss of working respondents due to conducting interviews solely during morning hours; faulty survey questions.
- Sample Size Relationship: Completely independent of sample size; increasing the sample size will not diminish or eliminate non-sampling errors.
- Exam Pitfall: Sampling frame error () is a systematic non-sampling error, not a random sampling error.
Four Probability Sampling Techniques
- Simple Random Sampling:
- Definition: Every individual element within the defined target population has an equal, non-zero, and independent probability of selection.
- Execution Protocol: Implemented via random number tables or algorithmic selection in software (e.g., SPSS via
DataSelect CasesRandom sample of cases). - Field Application: Drawing 10 candidates from a class register of 50 students using a random number table by selecting an arbitrary starting digit and reading two-digit sequences, discarding duplicates and values outside the 01–50 range until 10 unique cases are obtained.
- Systematic Sampling:
- Definition: Selecting individual units from an ordered sampling frame by choosing every name following a randomized starting position between and .
- Operational Mechanics: Sampling interval calculated as . For example, selecting every 5th passenger from a passenger manifest.
- Methodological Limitation: Violates the strict mathematical assumption of equal, independent probability of selection; once an initial random seed is chosen (e.g., case 43), adjacent cases (such as case 44) have zero mathematical probability of inclusion in that draw.
- Stratified Sampling:
- Definition: Partitioning a heterogeneous population into mutually exclusive, homogeneous subgroups (strata) based on a salient characteristic, followed by independent random sampling within each stratum.
- Field Application: Segmenting an organization across nine functional departments (e.g., Front Desk, Housekeeping, F&B) and drawing five employees randomly from each department.
- Cluster Sampling:
- Definition: The primary sampling unit is not the individual respondent, but rather a naturally occurring, geographically bounded, or organizational aggregate (cluster) of diverse elements.
- Strategic Function: Minimizes fieldwork and travel costs over large geographic regions while preserving probability sampling properties.
- Field Applications: Sampling international air passengers via targeted departure gates or specific flights; sampling sports spectators via stadium seating blocks; sampling university seniors via selected institutions; surveying national adult populations via multistage geographic clustering (State County Census Block Household).
Four Nonprobability Sampling Techniques
- Convenience Sampling (Accidental Sampling):
- Definition: Recruiting research units that are most accessible and convenient for the researcher to access.
- Field Application: A television reporter interviewing passing pedestrians on a street corner: "Do you approve of the current President's performance?"
- Judgement Sampling (Purposive Sampling):
- Definition: An experienced researcher selects sample units based on professional judgement regarding which cases best possess the required characteristics.
- Field Application: A plastics manufacturer selecting 20 key enterprise accounts based on the sales director's experience to forecast regional demand for the upcoming quarter.
- Quota Sampling:
- Definition: Selecting respondents non-randomly to meet specific demographic subgroup quotas predetermined by the researcher, ensuring target proportions are filled.
- Field Application: Mall intercept interviewers directed to recruit 100 respondents under 40 years of age and 50 respondents aged 40 or above, selecting participants at their discretion.
- Snowball Sampling:
- Definition: An initial seed group of qualified participants is selected, and these respondents subsequently refer or recruit additional eligible individuals from their networks.
- Field Application: Researching specialized or hidden populations, such as "serious amateur tennis players": screening 200 general individuals to identify a qualified player, and asking that qualified player to recommend two other serious tennis players.
Comparative Analysis: Stratified Sampling vs. Quota Sampling
- Sampling Classification: Stratified sampling is a probability sampling technique; Quota sampling is a nonprobability sampling technique.
- Intra-Subgroup Selection Mechanics: Stratified sampling uses random probability sampling within each stratum; Quota sampling relies on convenience or purposive non-random selection by fieldwork interviewers.
- Subgroup Proportion Determination: In stratified sampling, sample allocations reflect population proportions or formal disproportional designs; in quota sampling, targets are set at the researcher's discretion.
- Exercise Distinction: Dividing product users into three usage tiers and executing independent random sampling within each tier is Stratified Sampling. Instructing field interviewers to find individuals matching an age quota using convenience methods is Quota Sampling.
The Seven Chronological Steps in Sampling Planning
- Step 1: Define the target population ().
- Step 2: Select an operational sampling frame ().
- Step 3: Determine probability or nonprobability design ().
- Step 4: Plan procedural mechanics for selecting units ().
- Step 5: Determine statistical sample size ().
- Step 6: Select the actual sampling units in the field ().
- Step 7: Conduct fieldwork execution ().
Data Collection Methods and Questionnaire Design
Five Primary Data Collection Methodologies
- Survey Research (Questionnaires):
- Overview: The dominant quantitative method in social and hospitality management sciences.
- Administration Modes: In-person interviews (door-to-door, mall intercepts); telephone interviews (Computer-Assisted Telephone Interviewing [CATI]); self-administered instruments (postal mail, institutional drop-off, electronic mail, or web-based survey platforms such as Qualtrics).
- In-Depth Interviews:
- Overview: One-on-one qualitative encounters designed to elicit rich, detailed perspectives from smaller participant cohorts.
- Administration Modes: Unstructured or semi-structured discussions guided by an issue checklist rather than rigid standardized questions; typically audio-recorded and transcribed verbatim for coding.
- Focus Groups:
- Overview: Group interview format bringing together 6 to 8 participants to discuss a focused topic.
- Group Dynamics: Relies on active interaction among group members. The researcher functions as a facilitator rather than a directive interviewer; sessions are typically video-recorded to capture both non-verbal cues and verbal dialogue.
- Observation:
- Overview: Systematic observation and recording of behavioral patterns, objects, and environmental events without questioning or communicating with subjects.
- Methodological Types: Participant observation vs. mechanical observation (e.g., turnstiles, video tracking). Observable dimensions include physical actions, spatial patterns, verbal statements, and expressive behaviors.
- Experimentation:
- Overview: Manipulating one or more independent variables under controlled conditions to observe and quantify changes in a dependent variable, thereby testing causal hypotheses.
- Applied Field Scenario: Assessing the effect of retail "Sale" signage on soft drink purchasing. Phase 1 features regular signage at (Control Group); Phase 2 displays prominent "Sale" signage at the identical price point (Experimental Group) to isolate the psychological effect of promotional framing.
Temporal Dimensions: Cross-Sectional vs. Longitudinal Designs
- Cross-Sectional Studies: Data collected at a single point in time, providing a cross-sectional snapshot of the phenomenon under study.
- Limitation: In rapidly changing environments, cross-sectional findings can quickly become outdated, often requiring replication studies to verify ongoing validity.
- Longitudinal Studies: Data collected at multiple points across time to track trends and evaluate how phenomena evolve.
- Three Longitudinal Typologies:
- Trend Study: Investigates shifts within a broad population over time; successive cross-sectional samples are drawn from the general population at each measurement point, meaning an individual is rarely sampled twice.
- Example: Surveying successive cohorts of graduating SHTM students annually to assess industry perceptions, using a freshly updated sampling frame each year.
- Cohort Study: Tracks a specific, defined subpopulation (a cohort sharing a common life event) longitudinally, drawing independent samples from that defined cohort at successive intervals.
- Example: Tracking undergraduate students who enrolled in the Bachelor of Science program in the year 2026, sampling from this specific demographic cohort again in 2027 and 2028.
- Panel Study: Measures the exact same sample of specific individuals (the panel) repeatedly across each subsequent time wave.
- Example: Selecting a panel of individuals in 2026 and re-interviewing those identical individuals in 2027 and 2028.
Question Formats and Measurement Scales
- Closed-Ended Questions: Standardized response categories provided in advance. Includes categorical checklists (nominal), ranking scales (ordinal), and Likert agreement scales (interval).
- Open-Ended Questions: Participants formulate their own answers without predetermined constraints. Can yield quantitative ratio metrics (e.g., "How many nights did you stay in Hong Kong?") or rich qualitative commentary (e.g., "Describe the three most memorable aspects of your dining experience").
- Likert Scales: Widely used in the social sciences to measure psychological constructs and attitudes. Typically structured on 5-point or 7-point response formats (e.g., Strongly Agree, Agree, Neutral, Disagree, Strongly Disagree). Constructed by generating multiple related statements and scoring responses across standardized points.
The Four Levels of Measurement
- Nominal Scale: Categorical data where numbers serve purely as classification labels without quantitative value or intrinsic rank order. Permissible statistics: Frequency counts, percentages, mode, and chi-square tests. Examples: Gender (, ), Country of Residence, Airport Departure Terminal.
- Ordinal Scale: Data possessing an inherent rank order, but where the intervals between adjacent values cannot be assumed to be equal. Permissible statistics: Median, percentiles, rank-order correlations. Examples: Ranking four competitor restaurants from , hotel star ratings.
- Interval Scale: Ordered data where intervals between adjacent points represent equal mathematical distances, but lacking a true absolute zero point (). Permissible statistics: Arithmetic mean, standard deviation, Pearson correlation coefficients, t-tests. Examples: Likert response scores (), temperature in Celsius.
- Ratio Scale: Possesses all mathematical properties of an interval scale, along with a meaningful, absolute zero point (). Permissible statistics: All parametric tests, geometric means, and ratio comparisons. Examples: Chronological age in years, body mass, length of stay in nights, visitor expenditure.
- Curricular Grading Note: When age is measured using grouped brackets (e.g., "25 or below / 26–35 / 36–45 / 46–55 / 56–65 / 66 or above"), the course answer key categorizes it as Nominal on the premise that these codes function as discrete group labels without continuous mathematical properties.
Three Criteria for Response Sets
- Exhaustive: Response categories must encompass every possible answer; this is typically achieved by including an explicit "Other (please specify)" category.
- Flawed Example: Categorizing age as (1) 20 or below, (2) 21–30, (3) 31–40, (4) 41–50, (5) 51–60. This scheme omits respondents aged 61 or above.
- Mutually Exclusive: Response categories must be distinct, with no overlap, ensuring each respondent fits into only one category.
- Flawed Example: Income categories structured as \20,000\text{--}\ and \30,000\text{--}\, creating ambiguity for a respondent earning exactly .
- Unidimensional: Each question item must measure a single conceptual attribute across a unified continuum.
- Flawed Example: Structuring an evaluative scale as "The service is: (1) very satisfactory, (2) satisfactory, (3) acceptable, (4) unacceptable, (5) very unacceptable." This conflates satisfaction with acceptability, which represent distinct psychological constructs.
Six Guidelines for Questionnaire Construction
- Guideline 1 — Use Simple Language and Avoid Ambiguity: Words like "often", "regularly", and "usually" carry subjective, inconsistent meanings across respondents.
- Revision: Avoid vague phrasing like "Do you think your supervisor is good?"; refine to "Do you think your supervisor is good at assessing subordinates’ job performance?"
- Guideline 2 — Keep Questions Short: Verbose phrasing increases cognitive burden on respondents.
- Revision: Shorten "What is your usual mode of transportation that you utilise to reach Fishing Island?" to "How do you get to Fishing Island?"
- Guideline 3 — Maintain Single Focus and Avoid Double-Barrelled Questions: Items should not combine two separate inquiries into a single response option.
- Revision: "Are you satisfied with the cost and convenience of the service?" must be split into two separate items: one measuring cost satisfaction, the other measuring convenience.
- Guideline 4 — Avoid Leading Questions: Questions must not nudge the respondent toward a particular answer.
- Revision: "Do you think Mr. ABC is the right person for Director of Rooms, as the occupancy rate of our hotel is very low?" must be revised by removing the leading justification regarding low occupancy.
- Guideline 5 — Avoid Negative Phrasing: Negations and double negatives create confusion about whether "Yes" indicates agreement or disagreement.
- Revision: Change "Did you not go shopping today? Yes/No" to "Did you go shopping today? Yes/No."
- Guideline 6 — Use Filter and Skip Sequencing: Use screening questions to direct respondents only to items relevant to their experiences, avoiding non-applicable responses.
- Application: "Is Hong Kong the MAIN destination of this trip? Yes (go to Q7) / No (continue)."
Questionnaire Structural Sequencing
- The Beginning: Open with general, engaging, non-threatening items (e.g., "Is this your first visit to Hong Kong?") to build rapport and put the respondent at ease.
- The Middle: House the core research questions, complex measurement batteries, and cognitively demanding multi-item Likert scales.
- The End: Place sensitive demographic items (age, household income, educational attainment) at the end. By this point, rapport is established, increasing response rates. If a respondent terminates early, core analytical data has already been captured.
The Four Components of Chapter 3 (Methodology)
- Population and Sample: Describe the target population, operational sampling frame, sampling methodology, and sample size calculations.
- Survey Design: Outline research aims, justify the selection of the survey method, specify the administration mode (e.g., mall intercept, online panel), and state the temporal design (cross-sectional vs. longitudinal).
- Instrumentation: Detail the measurement instruments (whether original, adapted, or adopted); document reliability and validity metrics from prior literature; outline pilot-testing procedures (course guideline: a minimum of 5 individuals is acceptable for pre-testing); detail measurement scales; include the survey instrument and cover letter in appendices; outline survey administration and follow-up protocols.
- Data Analysis Plan: Detail the analytical techniques and statistical tests planned to address each research objective, providing theoretical justifications for their selection.
- Field Methodology Case Study (Qu & Tsang Hotel Study):
- Instrument: 35 service quality attributes.
- Target Population: International tourists visiting Beijing, Shanghai, and Guangzhou.
- Sampling Approach: Convenience sampling across 9 participating hotels (due to access constraints), paired with systematic sampling of hotel guests (approaching every 5th guest).
- Response Metrics: Distributed 270 questionnaires, retrieved 213 ( response rate), discarded 31 incomplete surveys, yielding 182 usable questionnaires ( valid response rate).
- Pilot Testing: Administered to respondents, yielding Cronbach's alpha coefficients between .
- Analytical Framework: Paired-samples t-tests to evaluate the expectation-perception gap, factor analysis for data reduction, and logistic regression for predictive modeling.
Applied Research Scenarios
- Scenario 1 — Call Centre Capacity Management: Survey of 12 financial service call centres; paired unstructured executive interviews with operational administrative data; applied linear and exponential regression models to evaluate forecasting accuracy across four operational dependent variables.
- Scenario 2 — Total Quality Management (TQM) Implementation Barriers: Literature review and qualitative depth interviews generated an attribute list of implementation barriers; administered structured surveys; applied Principal Component Analysis () to uncover underlying barrier dimensions.
- Scenario 3 — Shopping Preferences of Taiwanese Visitors: Structured survey organized into three sections (shopping preferences, purchase drivers, demographics); conducted three iterative pilot tests; used quota sampling (, , matching Hong Kong Tourist Association statistical profiles); administered surveys across Kai Tak Airport departure lounges, Causeway Bay, and Tsim Sha Tsui.
- Scenario 4 — Expectation-Disconfirmation Theory: Conducted 8 exploratory focus groups to identify traveler expectations regarding Spain; deployed a longitudinal pre-trip / post-trip telephone interview design to assess expectation-performance disconfirmation.
Qualitative Methods and Content Analysis
Foundations of Qualitative Inquiries
- Verbatim Definition: The collection, analysis, and interpretation of data obtained through observing what people do and say; focuses on meanings, subjective interpretations, concepts, definitions, characteristics, metaphors, and symbols; subjective in nature, employing data collection strategies distinct from quantitative research.
- Methodological Strengths: Examines underlying feelings and emotional motivations (); addresses complexity, contextual nuance, and depth (); generates insights into real-life settings ().
Four Primary Data Collection Methods: Qualitative vs. Quantitative
- Survey / Questionnaire: Quantitative.
- Observation: Qualitative. Involves observing subjects systematically over time to understand behaviors in context ("watching with purpose"). Examples include tracking guest interactions with theme park mascots or observing emotional responses to pets in urban parks.
- Focus Group Discussion: Qualitative. A small group of participants explores a focused topic guided by a facilitator. Examples include 4 focus groups of 6 visitors discussing theme park character encounters, or 3 groups of 3 business travelers exploring "bleisure" travel motivations.
- In-Depth Interview: Qualitative. One-on-one structured or semi-structured dialogue gathering detailed individual insights ("a structured conversation with purpose").
Content Analysis: Deductive vs. Inductive
- Verbatim Definition: A detailed and systematic examination of the contents of a particular body of material for the purpose of identifying themes; resembles a detective examining evidence.
- Deductive Content Analysis: The researcher begins with an established theory or conceptual framework and classifies raw textual data into pre-existing categories ("sorting into pre-built drawers"). Course Example: Quadri & Fiore (2012) analyzing wine tourism experiences by sorting participant transcripts into an established experience economy framework.
- Inductive Content Analysis: The researcher approaches the data without preconceived theoretical assumptions, allowing categories and themes to emerge directly from the text ("building drawers from scratch"). Course Example: Generating thematic dimensions of aberrant tourist behaviors from hotel staff interview transcripts.
The Five Stages of Inductive Content Analysis
- Stage 1 — Immersion in the Data: Read and re-read the raw interview transcripts repeatedly without preconceptions to understand the phenomenon from the participants' perspective.
- Stage 2 — Open Coding: Break transcripts into meaningful segments and attach descriptive conceptual labels (codes). Open coding is inherently analytical; the researcher interprets the data by asking who, to whom, why, what, and how, grouping semantically related statements.
- Stage 3 — Category Creation: Group related open codes into broader, higher-order categories through constant comparative analysis.
- Stage 4 — Theme Identification: Abstract central themes that capture the underlying essence of the data, linking related categories and codes.
- Stage 5 — Drawing Theoretical Inferences: Formulate conceptual inferences and practical implications grounded in the data and existing literature.
- Quantifying Qualitative Patterns: Following Stage 5, researchers can quantify qualitative patterns by calculating theme frequencies (e.g., using qualitative software such as NVivo 12) to assess the relative prominence of each theme.
Case Study: The Hotel Jaystaycationer Investigation
- Research Objectives: Identify frontline manifestations of jaystaycationer behavior, build a typology of these behaviors, and explore staff coping strategies.
- Methodology: Semi-structured in-depth interviews with frontline hotel personnel ().
- Sampling Strategy: Purposive/judgement sampling was used first to recruit 4 experienced frontline managers with extensive background handling difficult guests, followed by snowball sampling as these managers referred frontline colleagues.
- Assessment Rule: If an exam question asks which sampling method was used in this research, both judgement (purposive) and snowball sampling must be cited.
- Transcript-to-Theme Coding Matrix:
- Raw Transcript Unit 1: Guests refusing to wear masks, hosting unauthorized room parties, and disregarding hotel health regulations.
- Open Coding 1: "Jaystaycationers with low hygienic awareness"; "Not wearing masks"; "Partying in room".
- Category 1: Low health awareness; Non-compliance with hygiene protocols; Violation of hotel policies.
- Overarching Themes 1: Personal Antecedent: Personal Hygiene Standards; Jaystaycationer Behavior: Rule-breaking.
- Raw Transcript Unit 2: Guests exaggerating sanitary concerns, demanding tailored staff interventions, and issuing customer complaints.
- Open Coding 2: "Emphasising high awareness"; "Forcing staff to be attentive".
- Category 2: Self-presentation; Coercing staff into customized service.
- Overarching Themes 2: Jaystaycationer Behavior: Attention-seeking; Staff Response: Explaining, Seeking Supervisor Assistance.
Research Ethics and Report Writing
Foundations of Research Ethics
- Core Definitions:
- Research ethics centers on adhering to societal norms () and professional codes of behavior ().
- Defined as "the rules or standards of conduct" and "conforming to accepted principles of right and wrong that governs conduct (of professions)".
The Four Foundational Ethical Principles
- Principle 1 — Informed Consent:
- Participants must be informed of the nature, procedures, and potential risks of the research before agreeing to participate, either in writing or verbally.
- Participation must be voluntary, with subjects free to withdraw at any time. Deception ()—such as misrepresenting the study's purpose, concealing researcher identity, or making false promises—is prohibited.
- Principle 2 — Protection from Harm (No Harm to Participants):
- Researchers must protect participants from both physical harm and psychological distress.
- In hospitality and tourism research, harm is typically psychological, such as stress or embarrassment caused by disclosing personal income, educational background, or workplace conflicts.
- Principle 3 — Anonymity and Confidentiality:
- Anonymity: The researcher cannot link individual responses to specific participants' identities (e.g., untracked online surveys).
- Confidentiality: The researcher can link responses to participants' identities, but promises to keep this information secure and protected from public disclosure.
- Subjects researching sensitive topics may require full anonymity, whereas others may only require confidentiality.
- Principle 4 — Ethical Analysis and Reporting:
- Data analysis and reporting must be accurate and honest.
- Researchers must report negative or counter-intuitive results, disclose methodological limitations, and refrain from framing post-hoc data explorations as pre-planned hypothesis tests.
Case Study: Ethical Breach in Hospitality Research
- Incident Context: During a Summer 1994 internship in a hotel human resources department, an SHTM student accessed confidential personnel databases without management authorization, copying employee salary records onto floppy disks for her undergraduate thesis.
- Detection: The hotel's computer security system logged the unauthorized file access, timestamp, and terminal ID, leading back to the student's desk where the disks were recovered.
- Student Defense: The student claimed she was unaware the data was confidential and believed it would be useful for her thesis.
- Sanctions: The department placed a formal warning letter in her file and required a written apology. Dissatisfied with this sanction, hotel leadership pressed for further action. The departmental board ruled the infraction a major ethical breach, suspending the student for the entire 1994/95 academic year, permitting her to return only in 1995/96 to finish her final year.
- Ethical Assessment: The student violated participant confidentiality (unauthorized acquisition of proprietary salary data), breached professional ethical conduct in data collection (misusing access privileges), and broke workplace trust. Research utility does not justify bypassing informed consent and organizational authorization.
Primary Objectives of the Research Report
- Clearly communicates the research background, the underlying problem that prompted the inquiry, the study's significance, the methodological procedures used, the empirical findings, and their theoretical and managerial implications.
Structural Blueprint of the Research Report
- Front Matter & Layout Order: Title Page Table of Contents Acknowledgements Executive Abstract Main Report Body Appendices.
- Main Body Chapter Progression:
- Chapter I: Introduction
- Chapter II: Literature Review
- Chapter III: Methodology
- Chapter IV: Findings / Results
- Chapter V: Conclusions / Discussions / Implications
Core Content by Chapter
- Introduction: Contextual background, formal problem statement, statement of research significance, overarching aims, and specific objectives.
- Literature Review: Critical synthesis of relevant historical and contemporary research, underlying theoretical frameworks, empirical literature, and identification of research gaps.
- Methodology: Target population and operational sample characteristics, data collection procedures, measurement instrument design, pilot testing (pre-testing on 5 individuals is acceptable), statistical analysis plan, and optional study limitations.
- Findings / Results: Systematic presentation of statistical and qualitative findings aligned with each research objective, using tables, charts, and diagrams.
- Conclusions / Discussions / Implications: Summary of core findings, interpretation of what the results mean, theoretical contributions, and actionable recommendations for industry practitioners.
SPSS: Data Management and Statistical Analysis
The SPSS Architecture and Variable Definitions
- Primary Windows: The Data Editor (
.sav), where datasets are managed and statistical procedures are initiated, and the Output Viewer (.spv), where analytical results are displayed. SPSS can only display one active dataset per Data Editor session. - Data View Window: Rows correspond to individual cases or respondents; columns represent variables; individual cells store the value of a specific case on a given variable. By convention, Column 1 serves as an identification code (
ID). - Variable View Window: Rows represent variables, while the ten columns configure specific variable attributes:
- Name: Variable identifier; maximum of 64 alphanumeric characters; should use descriptive labels (e.g.,
purpose) rather than generic tags (e.g.,Q1) to prevent default naming (e.g.,var00001). - Type: Data format; primarily Numeric (quantitative numbers with defined decimals) or String (alphanumeric text). String variables cannot be used in inferential statistical tests (t-tests, ANOVA, regression).
- Width: Maximum character length allocated for data storage.
- Decimals: Number of decimal positions displayed for numerical data.
- Label: An expanded description of the variable (e.g., labeling variable
purposeas "The main purpose for visiting Hong Kong"). - Values: Maps numerical codes to descriptive text labels (e.g., assigning Value Labels , , ).
- Missing: Designates specific codes for missing data points (e.g., ) that fall outside the valid range, distinguishing between "Refused to Answer" and "Not Applicable".
- Columns: Visual display width of the column in Data View.
- Align: Text alignment within Data View cells (Left, Right, Center).
- Measure: Measurement level: Scale (continuous Interval or Ratio), Ordinal, or Nominal. This must align with the survey design to prevent using inappropriate statistical tests.
- Primary Windows: The Data Editor (
Univariate Descriptive Analysis
- Core Objective: Analyzes a single variable in isolation to describe baseline distributional patterns, summarize sample characteristics, and screen for data entry errors by checking minimum and maximum values.
- Analytical Metrics: Frequency counts and percentages, measures of central tendency, and measures of dispersion/variability.
- Measures of Central Tendency:
- Arithmetic Mean: The sum of all values divided by the total number of observations. It serves as the mathematical balance point of the distribution, but is sensitive to extreme values (outliers).
- Median: The midpoint of an ordered distribution, dividing the upper of observations from the lower . For an odd sample size, it is the center value; for an even sample size, it is the average of the two central values. It is robust against outliers, making it preferred for skewed metrics such as household income and real estate prices.
- Mode: The most frequently occurring score in a distribution; distributions may be unimodal, bimodal, multimodal, or have no mode.
- Measures of Dispersion / Variability:
- Range: The difference between the highest and lowest observed values; it is crude and sensitive to extreme values.
- Standard Deviation (): Quantifies the average dispersion of individual data points around the arithmetic mean; higher values indicate greater spread.
- Interpreting Percentages in SPSS Frequency Tables:
- Percent: Number of occurrences divided by total sample size, including missing values.
- Valid Percent: Number of occurrences divided by total valid responses, excluding missing values.
- Cumulative Percent: Running cumulative total of Valid Percentages up to and including the current value.
Bivariate Analysis Selection Matrix
- Comparing Two Independent Group Means: Independent-Samples t-Test (requires one continuous dependent variable measured on a Scale level and one two-group categorical independent variable on a Nominal level).
- Comparing Repeated Measures from the Same Sample: Paired-Samples t-Test (requires two continuous Scale measurements taken from the same participants under different conditions or time points).
- Assessing the Correlation Between Two Continuous Variables: Pearson Correlation Coefficient () (requires two continuous variables measured on Interval or Ratio scales).
- Testing the Association Between Two Categorical Variables: Cross-Tabulation paired with the Chi-Square Test of Independence () (requires two categorical variables measured on a Nominal scale).
- Descriptive vs. Inferential Statistics: Descriptive statistics summarize sample characteristics through numerical indices and visual graphs; inferential statistics use sample data to draw inferences about broader populations, test hypotheses, and evaluate findings using probability thresholds.
The Foundations of Significance Testing (-Value)
- Statistical Significance: Assesses whether an observed sample difference or correlation reflects a true population pattern or chance variation, quantified via the probability value (-value).
- Scale Range: The -value ranges from (impossible by chance) to (complete certainty of chance).
- Decision Thresholds: A result is statistically significant when , representing a confidence level. More stringent thresholds include ( confidence) and ( confidence).
- Substantive Meaning of : If an evaluation of employee training and job satisfaction yields , the relationship is statistically significant at the level (). We have confidence that the finding reflects a true population pattern; if the study were repeated 100 times, chance would produce this result in roughly 4 iterations.
- Two Common Misinterpretations:
- The -value does not indicate the strength or magnitude of an effect; it only indicates whether the effect is statistically significant. Effect magnitude is evaluated using correlation coefficients () or mean differences.
- An insignificant finding () does not prove the absence of a relationship; it merely means the current sample provides insufficient evidence to demonstrate that relationship.
Bivariate Correlation Analysis (Pearson )
- Coefficient Range: Pearson's correlation coefficient () ranges from (perfect negative linear relationship) through (no linear relationship) to (perfect positive linear relationship).
- Evaluation Benchmarks for :
- : Very strong association
- : Substantial association
- : Moderate association
- : Low association
- : Negligible association
- Three-Step Interpretation Sequence:
- Step 1 — Statistical Significance: Inspect the significance value (, labeled
Sig.). If , conclude that no statistically significant relationship exists and stop; do not interpret the strength of an insignificant coefficient. - Step 2 — Direction of Association: If , inspect the mathematical sign of the coefficient. A positive value indicates a direct relationship (variables increase together); a negative sign indicates an inverse relationship (as one increases, the other decreases).
- Step 3 — Strength of Association: Evaluate the absolute value of the coefficient () against the evaluation benchmarks.
- Applied Interpretation Scenarios:
- IQ and Academic Performance (TEE): An SPSS correlation table reports , , . Conclusion: There is a statistically significant, positive, and very strong relationship between IQ and TEE (, ).
- Lung Function and Cigarette Consumption: Analysis yields , . Conclusion: There is a statistically significant, inverse, and very strong relationship between daily cigarette consumption and pulmonary capacity.
- Correlation vs. Causation: Correlation does not imply causation. Two variables may vary together without one causing the other, often due to an unmeasured third variable influencing both factors. A rooster's crow does not cause the sun to rise.
- SPSS Navigation: Execute via
AnalyzeCorrelateBivariate, move variables into the analysis box, selectPearson, and specifyTwo-tailed(orOne-tailedif the hypothesis has a directional prediction). Scatterplots are generated viaGraphsScatter/DotSimple Scatter, mapping the independent variable to the X-axis and the dependent variable to the Y-axis.
Independent-Samples t-Test
- Analytical Purpose: Compares the arithmetic means of two independent groups to determine whether their difference is statistically significant.
- Statistical Hypotheses: Null Hypothesis () posits no significant difference between group means; Alternative Hypothesis () posits a significant difference.
- Two-Step Interpretation Sequence:
- Step 1 — Levene's Test for Equality of Variances: Inspect the significance value of Levene's Test (
Sig.). If Levene's , the assumption of equal variances is supported (); read results from the top row. If Levene's , the assumption of equal variances is violated (); read results from the bottom row. - Step 2 — Evaluate Mean Equality: Inspect the two-tailed significance value (
Sig. (2-tailed)) in the row identified in Step 1. If , reject and conclude the two group means differ significantly. If , fail to reject , concluding there is no significant difference between the groups. - Applied Interpretation Scenario (Automotive Engine Efficiency): Comparing engine efficiency between manual and automatic transmissions without an additive yields a Levene's test of , directing interpretation to the top row (
Equal variances assumed). The t-test reports . Conclusion: Fail to reject the null hypothesis; there is no statistically significant difference in engine efficiency between manual and automatic transmissions (, ).
Paired-Samples t-Test
- Analytical Purpose: Compares the means of two related observations gathered from the same participants or matched pairs (e.g., pre-test vs. post-test scores, customer expectations vs. perceived service quality, monthly sales across successive periods).
- Two-Step Interpretation Sequence:
- Step 1 — Direction of Differences: Inspect the
Paired Samples Statisticstable to compare the arithmetic means of the two conditions, noting which condition registered the higher value. - Step 2 — Statistical Significance: Inspect the
Paired Samples Testtable to evaluate the two-tailed significance value (Sig. (2-tailed)). If , the difference between the two conditions is statistically significant. If , the difference is not statistically significant. - Applied Interpretation Scenarios:
- Corporate Sales Performance: Comparing last month's sales to this month's sales yields , indicating no significant change in sales performance across the two months.
- Fuel Additive Performance: Testing 22 vehicles across two conditions (without additive vs. with additive) yields . Because the mean efficiency with the additive is higher than without it, we conclude the fuel additive significantly improved engine performance ().
- Independent vs. Paired t-Test Distinction: Use a Paired-Samples t-Test when analyzing repeated measures or before-and-after designs from the same subjects. Use an Independent-Samples t-Test when comparing two distinct, unrelated groups.
Cross-Tabulation and the Chi-Square Test ()
- Analytical Purpose: A non-parametric test evaluating the statistical independence or association between two categorical (nominal) variables ().
- Cross-Tabulation Architecture: A contingency table crossing the categories of two nominal variables, displaying observed response counts and percentages across combinations.
- Core Theoretical Mechanics: Compares Observed Frequencies (, actual counts in the sample) with Expected Frequencies (, the counts expected if the two variables were completely independent).
- Three-Step Interpretation Sequence:
- Step 1 — Statistical Significance: Inspect the asymptotic significance value (
Asymptotic Significance (2-sided)) in thePearson Chi-Squarerow. If , reject the null hypothesis of independence and conclude the two variables are significantly associated. If , conclude there is no significant association. - Step 2 — Percentage Distributions: To understand the pattern of association, inspect percentage distributions within the contingency table. Use row percentages (
Row %) when comparing across demographic groups; use column percentages (Column %) when comparing across behavioral categories. - Step 3 — Formal Documentation: Summarize results in a contingency table, reporting observed counts, percentages, degrees of freedom (), the chi-square statistic (), and the exact -value.
- Applied Interpretation Scenarios:
- Gender and Learning Format Preferences (Online vs. Textbooks): Analysis yields , , . Because , there is no significant association between gender and preferred learning format; males and females show comparable preferences for online learning.
- Residential Setting (Urban vs. Rural) and Magazine Choice: Analysis yields , , . Because , there is a statistically significant association between residential location and magazine preference; rural readers predominantly prefer Financial Review (), whereas urban readership is roughly evenly split between publications ( vs. ).
- Gender and Beverage Preference (Wine vs. Beer): Analysis yields , , . Because , there is a significant association between gender and beverage choice; of male respondents prefer beer, whereas of female respondents prefer wine.
Data Transformation: Compute and Recode
- Analytical Purpose: Modifies raw survey entries into new variables or operational formats to facilitate statistical analysis.
- The COMPUTE Command: Applies mathematical formulas to existing variables to calculate new metrics.
- Example 1: Constructing an index average across items: .
- Example 2: Calculating daily tourist spend: . Calculating average spend without this step yields the average expenditure per trip rather than the average expenditure per day.
- The RECODE Command (Recode into Different Variables): Reorganizes raw values into new categorization schemes. Common uses include collapsing 5-point Likert scales into 3-point categories, or converting continuous measures into discrete brackets (e.g., categorizing age into and by defining Range and Range ).
- Recode to Cross-Tabulation Case Exercise: Raw nationality data was recoded into three regional categories (, , ) and cross-tabulated against intent to revisit. The test yielded , , demonstrating a significant regional association. Proportions indicating no intention to return were for European travelers, for American travelers, and for Southeast Asian and other travelers.
Quick-Reference Statistical Decision Guide
- Two continuous variables (Correlation): Check the
Correlationstable for Pearson and its -value (Sig.). A significant relationship requires . Next, check the sign () for direction, and the absolute value () for strength. - Two independent group means (Independent t-test): Check
Levene's Testfor Equality of Variances. If Levene's , read the top row; if , read the bottom row. A significant difference in group means requiresSig. (2-tailed). - Repeated measures from the same sample (Paired t-test): Check the
Paired Samples Testtable forSig. (2-tailed). A significant difference requires . Then, compare means in thePaired Samples Statisticstable to determine which condition scored higher. - Two categorical variables (Chi-Square): Check the
Chi-Square Teststable for the Pearson Chi-Square asymptotic -value. A significant association requires . Then, check row or column percentages in the crosstabulation table to describe the nature of the association. - Data entry error checks: Check the
Frequenciestable for anomalous minimum, maximum, or range values (e.g., a recorded age of 200). - Correlation vs. Causation rule: A statistically significant correlation does not demonstrate a causal relationship; correlation only establishes that variables co-vary.
- Two continuous variables (Correlation): Check the
Comparative Matrix of Confusable Concepts
Pure Research vs. Applied Research
- Pure Research: Focuses on expanding the boundaries of human knowledge; driven by academic inquiry; has no immediate real-world application.
- Applied Research: Focuses on addressing specific, immediate problems to guide managerial and operational decisions.
Primary Research vs. Secondary Research
- Primary Research: The direct collection of original empirical data using research instruments designed for the study.
- Secondary Research: The synthesis, review, and re-analysis of existing literature and pre-collected data without gathering new empirical evidence.
Primary Data vs. Secondary Data
- Primary Data: Data collected directly by the researcher from original sources to address the specific research problem at hand.
- Secondary Data: Data originally collected by an external party for an alternative purpose, repurposed by a secondary researcher.
Deductive vs. Inductive Approaches
- Deductive Approach: Top-down reasoning moving from theory to empirical observation to test, accept, or reject hypotheses.
- Inductive Approach: Bottom-up reasoning moving from qualitative or empirical observations to pattern recognition, drawing tentative conclusions and building new theory.
Quantitative vs. Qualitative Research
- Quantitative Research: Collects and analyzes data in structured numerical forms, relying on mathematical and statistical procedures.
- Qualitative Research: Collects and analyzes non-numerical data (text, audio, video) to explore subjective meanings, experiences, and contexts.
Exploratory vs. Descriptive vs. Causal Research Designs
- Exploratory Design: Conducted when the research problem is ambiguous; aims to clarify the nature of the problem, but does not provide conclusive evidence.
- Descriptive Design: Conducted when the researcher is aware of the problem; systematically describes the characteristics of a population, answering questions of who, what, when, where, and how.
- Causal Design: Conducted when the problem is clearly defined; isolates and manipulates variables under controlled conditions to establish cause-and-effect relationships.
Cross-Sectional vs. Longitudinal Designs
- Cross-Sectional Design: Gathers data from sample subjects at a single point in time, providing a static cross-sectional snapshot.
- Longitudinal Design: Gathers data across multiple points in time to evaluate changes and track longitudinal trends.
Trend vs. Cohort vs. Panel Longitudinal Studies
- Trend Study: Draws successive, independent samples from the general population at each time point; individual subjects are rarely surveyed more than once.
- Cohort Study: Draws successive, independent samples from a specific, defined demographic cohort over time.
- Panel Study: Tracks and surveys the exact same sample of specific individuals repeatedly across all measurement waves.
Probability vs. Nonprobability Sampling
- Probability Sampling: Every element in the population has a known, non-zero probability of selection, allowing sampling error calculations and population generalizations.
- Nonprobability Sampling: The probability of selecting any individual element is unknown, relying on researcher convenience, quota rules, or subjective judgement.
Stratified Sampling vs. Quota Sampling
- Stratified Sampling: A probability method that divides the population into homogeneous strata and applies random sampling within each stratum.
- Quota Sampling: A nonprobability method where interviewers recruit respondents non-randomly until predetermined demographic quotas are filled.
Stratified Sampling vs. Cluster Sampling
- Stratified Sampling: Divides the population into homogeneous subgroups and samples randomly from every stratum to maximize precision.
- Cluster Sampling: Divides the population into heterogeneous clusters and randomly selects entire clusters to minimize fieldwork costs.
Systematic Sampling vs. Simple Random Sampling
- Systematic Sampling: Selects every element from an ordered sampling frame following a random start point.
- Simple Random Sampling: Selects cases completely at random, ensuring every element has an equal and independent chance of inclusion.
Random Sampling Error vs. Non-Sampling Error
- Random Sampling Error: Statistical fluctuation occurring purely by chance due to random variation across samples; decreases as sample size increases.
- Non-Sampling Error: Systematic error arising from flawed research design, sampling frame errors, non-response bias, or data entry mistakes; unaffected by sample size increases.
Sampling Frame vs. Target Population
- Sampling Frame: The actual operational list, directory, or database from which sample units are drawn.
- Target Population: The broader theoretical aggregate of all elements to which the researcher intends to generalize the study findings.
Validity vs. Reliability
- Validity: The degree to which a measurement instrument accurately captures the intended construct.
- Reliability: The consistency, dependability, and repeatability of a measurement instrument when administered across different times or samples.
Anonymity vs. Confidentiality
- Anonymity: Research conditions where the researcher cannot link individual responses to specific participants.
- Confidentiality: Research conditions where the researcher knows participants' identities, but protects them from disclosure to third parties and the public.
Independent-Samples t-Test vs. Paired-Samples t-Test
- Independent-Samples t-Test: Evaluates mean differences between two distinct, unrelated groups of respondents.
- Paired-Samples t-Test: Evaluates mean differences between two related measures obtained from the same respondents or matched pairs.
Pearson Correlation vs. Chi-Square Test of Independence
- Pearson Correlation: Evaluates the linear relationship between two continuous variables measured on interval or ratio scales.
- Chi-Square Test: Evaluates the association between two categorical variables measured on nominal scales.
Correlation vs. Causation
- Correlation: Demonstrates that two variables co-vary in a predictable linear direction.
- Causation: Demonstrates that changes in one variable directly produce changes in another, requiring temporal precedence, correlation, and the elimination of confounding variables.
Measurement Scales: Nominal, Ordinal, Interval, and Ratio
- Nominal: Discrete categories used purely as labels without quantitative order.
- Ordinal: Ordered categories with unequal or unknown intervals between points.
- Interval: Ordered categories with equal mathematical intervals between points, but no absolute zero point.
- Ratio: Ordered categories with equal intervals and an absolute meaningful zero point.
Closed-Ended vs. Open-Ended Questions
- Closed-Ended Questions: Provide predetermined response categories, simplifying quantitative coding.
- Open-Ended Questions: Allow respondents to formulate unconstrained answers, providing rich qualitative data or exact quantitative values.
Inductive vs. Deductive Content Analysis
- Inductive Content Analysis: Codes, categories, and themes emerge directly from textual data through constant comparative analysis without preconceived frameworks.
- Deductive Content Analysis: Pre-existing theoretical frameworks guide the analysis, sorting textual data into established categories.
Bilingual Terminology Glossary
Core Conceptual Terms
- Research (研究): A systematic, controlled empirical investigation aimed at resolving problems and reducing decision-making uncertainty.
- Pure / Basic Research (純研究): Research designed to expand the boundaries of human knowledge without immediate commercial or real-world application.
- Applied Research (應用研究): Research undertaken to address specific, immediate problems to support operational decision-making.
- Deductive Approach (演繹法): Top-down reasoning moving from established theory to empirical observation to test, confirm, or reject hypotheses.
- Inductive Approach (歸納法): Bottom-up reasoning moving from qualitative or empirical observations to pattern identification and theory construction.
- Quantitative Research (量化研究): Research that collects and analyzes data in numerical form using mathematical and statistical methods.
- Qualitative Research (質性研究): Research that explores non-numerical data (words, meanings, lived experiences) to examine subjective phenomena.
- Problem Definition (問題定義): The first stage of research, pinpointing the problem and bounding the inquiry via explicit research questions.
- Iceberg Principle (冰山原則): The observation that dangerous dimensions of a problem are submerged beneath the surface; researchers must isolate core problems rather than superficial symptoms.
- Problem Statement (問題陳述): A concise declarative statement highlighting what empirical knowledge is lacking, what operational deficit exists, or what negative outcomes are currently occurring.
- Research Objective (研究目標): Clear, operational statements outlining the specific, feasible outcomes the study intends to produce.
- SMART Criteria (目標設定準則): Guidelines requiring research objectives to be Specific, Measurable, Achievable, Realistic, and Time-bound.
- Exploratory Research (探索性研究): Initial research conducted when a problem is ambiguous; helps clarify the problem's nature, but does not provide conclusive evidence.
- Descriptive Research (描述性研究): Structured research that describes the characteristics of a population or phenomenon, addressing who, what, when, where, and how.
- Causal Research (因果性研究): Research designed to isolate and test cause-and-effect relationships between specific variables under controlled conditions.
- Independent Variable (自變項): The variable manipulated or observed as the causal factor accounting for changes in another construct.
- Dependent Variable (依變項): The outcome variable that changes in response to alterations in the independent variable.
- Variable (變項): Any measurable trait, attribute, or property that varies across individuals or shifts over time.
- Editing (資料編輯): Inspecting and adjusting raw data to correct omissions, illegibility, and inconsistencies prior to coding.
- Coding (編碼): Assigning numerical scores or character symbols to categorized data values for computer processing.
Literature Review and Citation Terms
- Literature Review (文獻回顧): A systematic identification, critical evaluation, and synthesis of published scholarship relevant to a research topic.
- Research Gap (研究空白): Unexplored, under-researched, or contradictory areas in existing scholarship that warrant further investigation.
- APA Format (APA 格式): The referencing and formatting style developed by the American Psychological Association, standard in hospitality and tourism research.
- In-Text Citation (文內引用): Author-date citations embedded in the text to credit original sources and substantiate claims.
- Reference List (參考文獻表): An alphabetical list at the end of a manuscript detailing full publication data for all sources cited in the text.
- Plagiarism (抄襲): Using another author's work, data, ideas, or words without proper citation, presenting them as one's own.
Data Sources and Measurement Metrics
- Primary Data (一手數據): Original empirical data collected directly by the researcher to address the specific research problem at hand.
- Secondary Data (二手數據): Pre-existing records collected by an outside entity for another purpose, repurposed for a new study.
- Validity (效度): The degree to which a measurement instrument accurately captures the intended construct.
- Reliability (信度): The consistency, dependability, and repeatability of a measurement instrument across administrations.
- Reactive (反應性): The tendency of study participants to alter their natural behavior when they are aware of being observed.
- Obtrusive (侵入性): Research methods that interrupt, disrupt, or interfere with subjects in their natural environment.
- Retrospectivity (可追溯性): The capacity of secondary historical records to provide longitudinal insight into past phenomena.
Sampling Methodologies and Terminology
- Population / Universe (人口/母體): The complete aggregate of elements sharing common characteristics to which a researcher intends to generalize findings.
- Sample (樣本): A selected subset drawn from the target population to represent the larger group.
- Census (普查): An exhaustive study examining every individual element in the entire target population.
- Sampling Frame (抽樣框): The operational list, directory, or database from which sample elements are drawn.
- Sampling Error (抽樣誤差): The statistical discrepancy between a sample estimate and the true population value arising from chance variation.
- Non-Sampling Error (非抽樣誤差): Systematic error arising from flawed research design, sampling frame errors, non-response bias, or operational mistakes.
- Simple Random Sampling (簡單隨機抽樣): A probability sampling method where every element has an equal, independent chance of selection.
- Systematic Sampling (系統抽樣): Selecting every element from an ordered sampling frame following a randomized starting position.
- Stratified Sampling (分層抽樣): Dividing a population into homogeneous strata and sampling randomly within each stratum.
- Cluster Sampling (群集抽樣): Selecting geographically bounded or organizational groups of diverse elements to serve as primary sampling units.
- Convenience Sampling (便利抽樣): A nonprobability method recruiting accessible and convenient participants.
- Judgement Sampling (判斷抽樣): A nonprobability method where an experienced researcher selects sample units based on professional judgement.
- Quota Sampling (配額抽樣): A nonprobability method recruiting non-random samples to fill predetermined demographic quotas.
- Snowball Sampling (雪球抽樣): A nonprobability method where an initial seed group of qualified participants refers additional eligible contacts.
Data Collection and Instrument Design Terms
- Survey (問卷調查): A structured data collection technique gathering quantitative data from respondents using questionnaires or interviews.
- In-Depth Interview (深度訪談): A qualitative one-on-one dialogue using semi-structured prompts to explore a topic in depth.
- Focus Group (焦點小組): A qualitative discussion among 6 to 8 participants guided by a facilitator, relying on group interaction.
- Observation (觀察法): Systematically watching and recording behaviors, events, and physical settings without questioning participants.
- Experiment (實驗法): An empirical investigation manipulating independent variables under controlled conditions to evaluate causal effects on dependent variables.
- Cross-Sectional Study (橫斷研究): A study gathering data from sample subjects at a single point in time.
- Longitudinal Study (縱貫研究): A study tracking and gathering data from subjects across multiple points in time to evaluate changes.
- Trend Study (趨勢研究): A longitudinal design drawing independent samples from the general population at each measurement point.
- Cohort Study (世代研究): A longitudinal design tracking a specific subpopulation defined by a shared life event over time.
- Panel Study (固定樣本 Panel): A longitudinal design surveying the exact same sample of individuals across every measurement wave.
- Questionnaire (問卷): A structured collection of questions designed to elicit standardized responses for empirical analysis.
- Closed Question (封閉式問題): Survey items that restrict respondents to selecting from predetermined response options.
- Open-Ended Question (開放式問題): Survey items that allow respondents to formulate their own answers in their own words.
- Likert Scale (李克特量表): A measurement scale assessing attitude agreement across standardized interval points (typically 5 or 7 points).
- Nominal Scale (名目量表): A categorical measurement scale using numbers purely as classification labels without rank or value.
- Ordinal Scale (次序量表): A measurement scale with an inherent rank order, but unequal intervals between points.
- Interval Scale (等距量表): A measurement scale with equal intervals between points, but no absolute zero point.
- Ratio Scale (比例量表): A measurement scale with equal intervals and an absolute meaningful zero point.
- Exhaustive (窮盡): A response set that provides options for all possible answers, often using an "Other" category.
- Mutually Exclusive (互斥): A response set designed so that each respondent fits into only one category without overlap.
- Unidimensional (單一維度): Survey items that measure a single underlying concept or attribute.
- Double-Barrelled Question (雙管問題): A flawed survey item that combines two separate inquiries into a single question.
- Leading Question (引導性問題): A biased survey question phrased to prompt or nudge the respondent toward a particular answer.
- Filter Question (過濾問題): A screening question used to direct respondents to relevant subsequent items based on their experiences.
- Pilot Test (預試): A small pre-test of a survey instrument administered before full fieldwork to detect flaws and refine questions.
Analytical and Statistical Analysis Terms
- Univariate Analysis (單變項分析): Statistical analysis of a single variable in isolation to describe baseline distributions.
- Bivariate Analysis (雙變項分析): Statistical analysis of two variables to examine mean differences or associations.
- Central Tendency Metrics (集中趨勢指標): Arithmetic mean, median, and mode, summarizing the central distribution of a dataset.
- Standard Deviation (標準差): A metric quantifying the dispersion of individual data points around the arithmetic mean.
- p-Value (p 值): The probability that an observed sample difference or correlation occurred purely by chance; values below indicate statistical significance.
- Correlation Coefficient (r) (相關係數): A standardized metric ranging from measuring the direction and strength of a linear relationship between two variables.
- Levene’s Test (Levene 檢定): An inferential test assessing whether two independent groups have equal population variances.
- Independent-Samples t-Test (獨立樣本 t 檢定): A parametric test comparing the arithmetic means of two unrelated groups.
- Paired-Samples t-Test (配對樣本 t 檢定): A parametric test comparing the means of two related measures obtained from the same sample.
- Chi-Square Test (卡方檢定): A non-parametric test evaluating whether two categorical variables are statistically independent.
- Cross-Tabulation (交叉表): A contingency table displaying joint frequency distributions across two categorical variables.
- Observed vs. Expected Frequency (觀察/期望次數): The actual sample count observed in a cell versus the count expected if the variables were independent.
- Data Transformation (資料轉換): Computing new variables or recoding raw data values to prepare a dataset for analysis.
- Compute / Recode (計算/重新編碼): Calculating new composite metrics using formulas, or binning variables into new categorical brackets.
- Content Analysis (內容分析): The systematic qualitative examination of textual, audio, or visual materials to identify themes.
- Open Coding (開放編碼): Breaking qualitative text into segments and attaching descriptive conceptual labels during early analysis.
- Theme (主題): A central organizing concept that captures the core meaning running across categories and codes.
Ethics and Research Reporting Terms
- Informed Consent (知情同意): Ensuring participants understand the study's purpose and procedures, and agree voluntarily to participate.
- Anonymity (匿名): Ensuring that individual survey responses cannot be linked back to participants' identities, even by the researcher.
- Confidentiality (保密): Protecting participant identities from public disclosure, even when known to the researcher.
- Deception (欺騙): Misleading participants regarding the true nature, researcher identity, or procedures of a study.
- Implications (啟示/意義): Explaining the broader theoretical contributions and practical recommendations arising from research findings.
- Limitations (研究限制): Inherent methodological constraints, sampling boundaries, or external factors that qualify research conclusions.
Methodological Pitfalls and Self-Assessment Exercises
Ten Common Methodological and Analytical Errors
- Confusing Correlation with Causation: A statistically significant correlation () establishes linear co-variation, not causation. Causal claims cannot be made from correlation alone without experimental controls and temporal ordering.
- Confusing Independent-Samples and Paired-Samples t-Tests: The key distinction is sample dependency. Studies analyzing matched pairs, repeated measures, or before-and-after conditions from the same subjects require a Paired-Samples t-Test, whereas distinct, unrelated groups require an Independent-Samples t-Test.
- Skipping Levene's Test in Independent-Samples t-Tests: Do not inspect the t-test significance value (
Sig. (2-tailed)) before checking Levene's Test (Sig.). If Levene's , read the top row (Equal variances assumed); if Levene's , read the bottom row (Equal variances not assumed). - Misclassifying Sampling Frame Errors as Random Error: Sampling frame errors are systematic non-sampling errors caused by flawed research design, not chance probabilistic fluctuations.
- Conflating Stratified Sampling with Quota Sampling: While both segment the population into subgroups, Stratified Sampling uses random probability selection within each stratum, whereas Quota Sampling relies on non-random convenience selection.
- Interpreting the p-Value as Effect Strength: The -value indicates statistical significance, not effect size. Effect strength must be evaluated using metrics like the correlation coefficient () or mean differences.
- Treating as Proof of No Relationship: An insignificant finding () does not prove the null hypothesis; it only indicates that the current sample provides insufficient evidence to demonstrate an effect.
- Conflating Validity with Reliability: Validity concerns measurement accuracy (measuring the intended construct); reliability concerns measurement consistency (producing repeatable results over time).
- Structuring a Literature Review as an Annotated Bibliography: A literature review should be written as a thematic synthesis essay linking studies and identifying gaps, rather than a list of standalone summaries.
- Misplacing Italics in APA Citations: In APA reference lists, italics apply to book titles, journal names, and volume numbers, but never to the titles of individual journal articles or book chapters.
- Misidentifying Measurement Levels: Nominal variables cannot be analyzed using means or t-tests, but require frequency counts and chi-square tests. Interval and ratio variables support parametric analysis.
- Assuming All Open-Ended Questions Are Qualitative: Open-ended questions can gather quantitative ratio data (e.g., length of stay in nights, visitor expenditure) as well as qualitative text.
Self-Assessment Section A: True / False Items
- Item A1: Secondary data are collected for some purpose other than the problem at hand.
- Answer: True. Secondary data originates from previous research conducted for other purposes.
- Item A2: The numbers assigned in a nominal scale reflect the relative amount of the characteristic being measured.
- Answer: False. Nominal numbers serve purely as classification labels and convey no quantitative value or rank.
- Item A3: Causal research is a type of conclusive research whose major objective is to obtain evidence regarding cause-and-effect relationships.
- Answer: True. Causal research uses controlled designs to establish cause-and-effect relationships.
- Item A4: A census is always more accurate than a sample because it covers the whole population.
- Answer: False. Large-scale censuses are prone to non-sampling and administrative errors, making well-designed samples potentially more accurate.
- Item A5: In systematic sampling, every element of the population has an equal chance of being selected.
- Answer: False. Once the initial random seed is chosen, subsequent selection intervals fix unit inclusion, giving adjacent units zero chance.
- Item A6: Stratified sampling and quota sampling both divide the population into subgroups, so they are the same technique.
- Answer: False. Stratified sampling uses random probability sampling within strata; quota sampling uses non-random convenience selection.
- Item A7: If the p-value is .03, the relationship is statistically significant at the 95% confidence level.
- Answer: True. A -value of falls below the threshold (), indicating significance at the confidence level.
- Item A8: A correlation coefficient of −.85 indicates a stronger relationship than a coefficient of +.60.
- Answer: True. Effect strength is determined by the absolute value of the coefficient (); the minus sign indicates direction, not strength.
- Item A9: In a paired-samples t-test, the two observations must come from two different groups of subjects.
- Answer: False. Paired-samples t-tests evaluate two related measures from the same subjects or matched pairs.
- Item A10: Chi-square is an appropriate test when both variables are measured on a nominal scale.
- Answer: True. The chi-square test of independence is the standard non-parametric procedure for two nominal variables.
- Item A11: If Levene’s test reports Sig. = .02, we should read the “equal variances assumed” row.
- Answer: False. Because Levene's , equal variances cannot be assumed; read the bottom row (
Equal variances not assumed). - Item A12: Open-ended questions can be used to collect quantitative data.
- Answer: True. Open-ended items can capture unconstrained numeric ratio metrics, such as total spend or age.
- Item A13: A literature review should be written as an annotated list of the studies reviewed.
- Answer: False. Literature reviews should be composed as synthesized, thematic essays, not annotated lists.
- Item A14: Validity refers to the consistency of results when a study is repeated with a different sample.
- Answer: False. That is the definition of reliability. Validity refers to whether an instrument accurately measures the intended construct.
- Item A15: Exploratory research does not provide conclusive evidence and is usually followed by further research.
- Answer: True. Exploratory studies clarify ambiguous problems and typically require subsequent descriptive or causal studies.
- Item A16: Pure research is conducted when a decision must be made about a specific real-life problem.
- Answer: False. That describes applied research. Pure research expands theoretical knowledge without immediate real-world application.
- Item A17: In inductive research, theorizing takes place after the research, and it is a theory-construction approach.
- Answer: True. Inductive inquiries collect data first and build theory through bottom-up analysis.
- Item A18: A sample is a subset of a larger population.
- Answer: True. A sample is a representative subset extracted from a larger target population.
- Item A19: Sampling frame error is a type of random sampling error.
- Answer: False. Sampling frame error is a systematic non-sampling error caused by flawed operational lists.
- Item A20: The mean is less affected by extreme values than the median is.
- Answer: False. The arithmetic mean is sensitive to extreme outliers; the median is robust against them.
- Item A21: A variable defined as String in SPSS cannot be used as the test variable in an independent-samples t-test.
- Answer: True. Inferential parametric t-tests require continuous numeric variables, and cannot run on string variables.
- Item A22: If the chi-square p-value is greater than .05, there is no statistically significant association between the two variables.
- Answer: True. A significance value exceeding indicates insufficient evidence to establish an association.
- Item A23: When pointing out the scale of a problem, specific figures should be given rather than vague phrases such as “many people”.
- Answer: True. Academic writing requires empirical metrics and percentages rather than vague quantifiers.
- Item A24: Anonymity means that the researcher keeps respondents’ identities from the public but can still link responses to them.
- Answer: False. That describes confidentiality. Anonymity means responses cannot be linked back to participants' identities, even by the researcher.
- Item A25: A focus group usually involves 6 to 8 participants, and the interviewer acts as a facilitator rather than an interviewer.
- Answer: True. Focus groups typically bring together 6 to 8 participants, with the researcher functioning as an interactive facilitator.
Self-Assessment Section B: Multiple Choice Items
- Item B1: The process of identifying and assigning a numerical score or other character symbol is called ______.
- Options: (A) Editing (B) Coding (C) Data entry (D) Test tabulation
- Answer: B. Coding is the technical process of assigning numerical scores or symbols to data categories.
- Item B2: What is the major difference between applied and pure research?
- Options: (A) Pure research takes longer to complete (B) Pure research has no immediate application (C) Applied research is less important (D) Pure research is more traditional
- Answer: B. Pure research focuses on theoretical advancement without immediate commercial application.
- Item B3: Which of the following is NOT a probability sampling technique?
- Options: (A) Simple random sampling (B) Systematic sampling (C) Convenience sampling (D) Cluster sampling
- Answer: C. Convenience sampling is a nonprobability technique relying on researcher accessibility.
- Item B4: A researcher selects every fifth name from a customer list. This is ______.
- Options: (A) Simple random sampling (B) Systematic sampling (C) Stratified sampling (D) Snowball sampling
- Answer: B. Selecting every element from an ordered list is systematic sampling.
- Item B5: A plastics manufacturer identifies twenty key accounts, based on the manager’s experience, to forecast next quarter’s sales. This is ______.
- Options: (A) Convenience sampling (B) Quota sampling (C) Judgement sampling (D) Cluster sampling
- Answer: C. Selecting units based on professional expertise represents judgement (purposive) sampling.
- Item B6: A study is conducted when the problem is clearly defined and the objective is to test whether price affects sales. This is ______.
- Options: (A) Exploratory research (B) Descriptive research (C) Causal research (D) Historical research
- Answer: C. Testing whether manipulation of price causes shifts in sales is a causal research design.
- Item B7: Which measure of central tendency is most affected by extreme values?
- Options: (A) Mode (B) Median (C) Mean (D) Range
- Answer: C. The arithmetic mean incorporates all values, making it sensitive to extreme outliers.
- Item B8: A Pearson correlation of r = .45 with p = .02 indicates ______.
- Options: (A) a weak negative relationship that is not significant (B) a moderate positive relationship that is significant (C) a strong positive relationship that is significant (D) no relationship because r is below .50
- Answer: B. A value of represents a moderate positive correlation (), which is statistically significant ().
- Item B9: Which test compares the same group of subjects before and after a training programme?
- Options: (A) Independent-samples t-test (B) Paired-samples t-test (C) Chi-square test (D) One-way ANOVA
- Answer: B. Comparing before-and-after conditions from the same subjects requires a paired-samples t-test.
- Item B10: A cross-tabulation reports Pearson Chi-Square = 12.31, p = .001. The correct conclusion is ______.
- Options: (A) there is no association between the two variables (B) there is a statistically significant association between the two variables (C) the two variables are causally related (D) the result is not meaningful because chi-square is parametric
- Answer: B. Because , the association is statistically significant; chi-square does not demonstrate causality.
- Item B11: Which analysis CANNOT be performed on a variable defined as String in SPSS?
- Options: (A) Frequencies (B) Independent-samples t-test (C) Crosstabs (D) Descriptive statistics
- Answer: B. String variables cannot be analyzed using inferential parametric tests that require continuous numerical data.
- Item B12: A seven-point Likert scale (1 = strongly disagree to 7 = strongly agree) is usually treated as ______.
- Options: (A) Nominal (B) Ordinal (C) Interval (D) Ratio
- Answer: C. Multi-point Likert scales are typically treated as interval data in social science research.
- Item B13: Which set of income categories violates the principle of mutually exclusive response sets?
- Options: (A) Under $10,000 / $10,000–19,999 / $20,000 or above (B) Under $10,000 / $10,000–20,000 / $20,000 or above (C) Under $10,000 / $10,001–20,000 / $20,001 or above (D) $0–9,999 / $10,000–19,999 / $20,000–29,999
- Answer: B. Option B overlaps at exactly , violating the requirement for mutually exclusive categories.
- Item B14: In which section of a research report should the limitations of the study be stated?
- Options: (A) Introduction (B) Literature review (C) Methodology (D) Conclusions
- Answer: C. Methodological limitations, sampling constraints, and instrumentation boundaries are formally detailed within Chapter 3 (Methodology).
- Item B15: Which of the following is an advantage of primary data?
- Options: (A) Low cost and quick access (B) Relevance, recency and an appropriate population studied (C) Non-reactivity of respondents (D) Retrospectivity and high quality
- Answer: B. Primary data provides high topical relevance, currency, and alignment with the target population.
- Item B16: Which of the following is a disadvantage of secondary data?
- Options: (A) It is time-consuming and expensive to collect (B) It may be incomplete and its validity may be questionable (C) It is obtrusive and reactive (D) It cannot be used for trend analysis
- Answer: B. Secondary datasets may lack complete records and often present questionable validity for the secondary researcher's goals.
- Item B17: In SMART objectives, the letter “R” refers to ______.
- Options: (A) Relevant to the literature (B) Realistic — whether the necessary resources are available (C) Repeatable in future studies (D) Reliable in measurement
- Answer: B. The "R" in SMART denotes Realistic, evaluating whether the necessary budget, skills, and tools are available.
- Item B18: “Are you satisfied with the cost and convenience of the service you received?” is a poor question because it is ______.
- Options: (A) leading (B) double-barrelled (C) negatively worded (D) too long
- Answer: B. The question combines two distinct attributes (cost and convenience) into a single item.
- Item B19: A researcher studying “serious tennis players” asks 200 people whether they qualify, then asks those who qualify to suggest two other serious players. This is ______.
- Options: (A) Quota sampling (B) Cluster sampling (C) Snowball sampling (D) Stratified sampling
- Answer: C. Using initial qualified participants to refer additional eligible contacts is snowball sampling.
- Item B20: Which of the following is a measure of variability rather than central tendency?
- Options: (A) Mean (B) Median (C) Mode (D) Standard deviation
- Answer: D. The standard deviation quantifies data dispersion, whereas the mean, median, and mode measure central tendency.
- Item B21: Inferential statistics are used to ______.
- Options: (A) summarise the characteristics of the sample using graphs and averages (B) generalise from a sample to a population and test differences or relationships (C) check for data entry errors in a frequency table (D) describe a single variable only
- Answer: B. Inferential statistics allow researchers to generalize sample findings to broader populations and test hypotheses.
- Item B22: In SPSS Variable View, which column is used to assign labels such as 1 = “vacation trips”?
- Options: (A) Label (B) Values (C) Measure (D) Missing
- Answer: B. The
Valuescolumn maps numerical codes to descriptive labels. - Item B23: A study collects data from the same respondents every quarter for two years. This is a ______.
- Options: (A) trend study (B) cohort study (C) panel study (D) cross-sectional study
- Answer: C. Surveying the exact same cohort of individuals across multiple time points represents a panel study.
- Item B24: An Independent Samples Test reports Levene’s Sig. = .683 and t-test Sig. (2-tailed) = .495. The correct interpretation is ______.
- Options: (A) equal variances are not assumed, and the two groups differ significantly (B) equal variances are assumed; read the top row; there is no significant difference between the groups (C) equal variances are assumed; read the bottom row; there is a significant difference (D) the test is invalid because Levene’s test is not significant
- Answer: B. Because Levene's , equal variances are assumed (top row). Because
Sig. (2-tailed), the groups do not differ significantly. - Item B25: A Correlations table reports r = .767 and Sig. = .000 between IQ and TEE (N = 25). The correct interpretation is ______.
- Options: (A) a weak positive relationship that is not significant (B) a very strong positive relationship that is statistically significant (C) a very strong negative relationship that is statistically significant (D) the relationship cannot be interpreted without a scatterplot
- Answer: B. A coefficient of reflects a very strong positive correlation () that is statistically significant ().