Book summary “Business research methods”: summary

Introduction

You face complex management choices every day. When intuition fails, you need reliable empirical evidence to guide your decisions. Poor information leads to costly failures. Business research provides the systematic tools you need to solve this problem. Donald R. Cooper shows you how to apply the scientific method to real-world commercial challenges.

Good business research requires a systematic hierarchy from management dilemmas down to rigorous measurement and analysis.

This book guides you through the entire research process. You will learn how to design studies, collect primary and secondary data, ensure ethical standards, and analyze results.

The book answers critical questions about investigation methods:

• How do you translate a vague management problem into a precise research question?
• When should you use qualitative methods instead of quantitative experiments?
• How do you design valid measurement scales and select representative samples?

1. The Nature of Business Research

Business research follows the scientific method to provide reliable empirical evidence for decision-making.

You must evaluate business phenomena through distinct study types. Reporting studies offer a simple summation of data with little inference. Descriptive studies answer who, what, where, when, and how by creating profiles of variables. Explanatory studies investigate reasons and causal relationships using theories. Predictive studies forecast when events will reoccur using theory-driven models.

In the Air Swiss case, researchers evaluated potential international partner companies using public sources in a reporting study. In the ColorSplash case, the owner investigated inventory warehousing processes to link them with sales losses in an explanatory study.

Scientific Reasoning and Philosophies

Good research must adhere to nine specific criteria, ranging from a clearly defined purpose to fully justified conclusions and high ethical standards.

• Purpose clearly defined
• Research process detailed
• Research design thoroughly planned
• High ethical standards applied
• Limitations frankly revealed
• Adequate analysis for decision-maker's needs
• Findings presented unambiguously
• Conclusions justified
• Researcher's experience reflected

Tom Peters and Robert Waterman initially interviewed 62 companies for In Search of Excellence using a non-probability judgment sample rather than a probability study, and Peters later confessed to faking data. After post-review, the final list was reduced to 43 companies.

Scientific reasoning relies on deduction and induction. John Dewey described this combination as the double movement of reflective thought.

Theory and Variable Relationships

Concepts serve as generally accepted labels for observable properties. Constructs are abstract ideas specifically invented for research when phenomena are not directly observable. Researchers have defined personality in 400 ways in the literature.

Variables are empirical symbols to which numerical values are assigned. Independent variables represent presumed causes, while dependent variables represent presumed effects. Moderating variables have a contingent effect, intervening variables act as conceptual mechanisms, and control variables prevent extraneous bias.

Propositions become hypotheses when formulated for empirical testing. Amazon charged 80 to 100 US dollars for The X-files second series, leading researcher Liu Qingling to study price discrimination. Amazon eventually refunded 6,896 customers an average of US$3.

2. Research Process and Proposals

Research projects begin with a clear question hierarchy and require formal planning and resource budgeting.

The management research question hierarchy starts at the broadest level with a management dilemma representing an organizational symptom or problem. You then translate this dilemma into a management question, narrow it to a specific research question, expand it into investigative questions, and finally reach measurement questions.

SalePro faced unexplained sales variations by territory, prompting a hierarchy of questions from the general dilemma down to specific territory sales evaluations. Unstructured exploration helps you refine the research question through literature searches and expert interviews. You can clarify vague management problems by searching published data, reviewing industry reports, and interviewing well-informed people or domain experts representing extreme opinions.

TechByte investigated future technology potential by reviewing specific books and periodicals, followed by interviews with scientists and engineers.

The Question Hierarchy

To build a rigorous foundation, you follow these sequential steps:

• State the basic management dilemma
• Translate the dilemma into a broad management question
• Formulate a precise research question
• Break the research question into investigative questions
• Design specific measurement questions

Pilot testing detects weaknesses in research design and data-collection instruments prior to full-scale execution. A pilot test draws subjects from the target population to simulate collection procedures, uncovers awkward or unintelligible phrasing, and refines measuring instruments without requiring statistically selected respondents.

Professors designing an interview study for EducTV discovered during pre-testing that two-thirds of their questions were unintelligible to the target group who had a median eighth-grade education.

Proposals and Pilot Testing

When you prepare a research proposal, you create a binding road map and formal offer that aligns the sponsor and researcher on objectives, methods, and responsibilities. Proposals vary in complexity from brief internal memos to multi-page external bids ranging from 1 to 10 pages in length. Every proposal must include a statement of the research question and a description of the methodology.

BankChoice proposed conducting personal interviews with a minimum of 200 residents to determine local banking knowledge, use, and attitudes.

Research project budgets are typically divided equally among three major phases, and organizations use three main budgeting approaches. Data collection often accounts for no more than one-third of the total budget. Organizations fund research using rule-of-thumb budgeting, departmental functional budgeting, or ad hoc task budgeting.

A sample contract research budget for Jason Henry and associates totaled 21,180 euros including direct salaries, travel, office supplies, telephone, publication costs, and overhead support. Sponsors evaluate research proposals through formal point systems or informal qualitative reviews based on criteria derived from guidelines.

3. Literature Review and Ethics

A scientific literature review establishes problem context while strict ethical compliance safeguards the integrity of empirical investigations.

You must embed current research into existing knowledge by mapping problem structure and linking theories. A high-quality review combines basic descriptive ingredients with analytical seasoning.

To conduct a good review, you should:

• include literature related to the problem statement;
• mention different theoretical ideas exploring the problem statement;
• summarize previous studies investigating the problem statement;
• discuss theoretical ideas against the background of previous results;
• analyze and compare previous studies in light of their research design and methodology;
• demonstrate how the current study fits in with previous studies and show its specific new contributions.

Student Patricia Carvalho used EBSCO to search for contract-related terms, retrieving 63 peer-reviewed articles for Contract* and opportunism, 32 for Contract* and mitigation, and 182 for Contract* and signal*. Advanced database searches employ Boolean operators, truncation symbols, and limiters to efficiently refine your results. You can use OR for synonyms, AND for combined terms, and truncation like * to replace characters.

Research Ethics and Participant Rights

Researchers must protect participants from harm and secure informed consent before proceeding with any study.

Deontology dictates that the ends never justify ethically questionable means, whereas teleology weighs study benefits against potential harm. You must protect respondents against discomfort and obtain informed consent.

When introducing an interview, you should follow standard administration protocols:

• introduce ourselves – interviewer’s name and research organization;
• briefly describe the survey topic;
• describe the geographic area we are interviewing or target sample;
• tell who the sponsor is;
• describe the purpose(s) of the research;
• give a good-faith estimate of the time required to complete the interview;
• promise anonymity and confidentiality when appropriate;
• tell the respondent the participation is voluntary;
• tell the respondent that item non-response is acceptable;
• ask permission to begin.

For example, Lakeside University Centre for Survey Research interviews required an estimated completion time of 40 minutes. You must also guard data confidentiality and avoid plagiarism or falsification. A survey of American Economic Association members in Chicago revealed that about 4 percent confessed to falsifying research data.

4. Research Design and Sampling

Selecting appropriate research designs and probabilistic sampling plans ensures accuracy and controls extraneous variation.

Design Classifications and Causation

Research studies are broadly divided into qualitative and quantitative based on the type of information they utilize. Steinbeck's description of a Mexican Sierra fish contrasts counting 17 plus 15 plus 9 spines on the dorsal fin of a pickled fish with experiencing the fish striking the line and pulsing with color.

Research designs are classified using eight distinct descriptors that guide planning, execution, and resource allocation. Exploratory studies use loose structures and qualitative techniques to investigate vague problem areas and develop hypotheses. Sampada Mulagapati analyzed secondary wealth lists and conducted eleven personal interviews with self-made multi-millionaires in the Netherlands to explore success factors. Quote magazine ranks the 500 wealthiest people annually in the Netherlands, and Sampada analyzed 210 self-made wealthy individuals after excluding heirs. The richest respondent interviewed possessed 796 million euros, while the poorest possessed 147 million euros.

Causal studies attempt to explain how one variable produces changes in another, evaluated through inductive reasoning and logical canons. Establishing strong evidence for causation requires proving covariation, correct time order, and the elimination of alternative causes. An alumni fundraising experiment tested emotional versus logical appeal letters across randomized groups against a control group receiving no appeal. The mailing list targeted a total of 50,000 alumni, and you choose a sample size of 300 subjects allocated equally with 100 members to each of the three groups. Alumni receiving the emotional appeal reached a 50 per cent response rate, compared to 35 per cent for the logical appeal and 5 per cent for the control group.

Sampling Strategies and Precision

The unit of analysis specifies the exact level at which research is performed, which is often different from the individual respondents questioned to obtain information. Sampling is generally preferred over census-taking because it offers lower cost, greater speed, greater accuracy, and handles situations where population elements are destroyed or infinite. The 2001 British census cost about £255 million (€387 million) and still suffered from a non-response rate requiring a follow-up Census Coverage Survey of 320,000 households. Non-sampling sources contributed to 90 per cent of total survey error in one study.

The validity of a sample depends on accuracy, which measures the absence of bias, and precision, which measures the closeness of sample estimates through the standard error. Executing a sampling plan requires answering sequential questions:

• Determine the relevant population.
• Determine the parameters of interest.
• Determine the sampling frame.
• Determine the type of sample.
• Determine what size sample is needed.
• Determine how much it will cost.

Simple random sampling is the basic probability sampling design where every population element has a known and equal chance of being selected. You select elements from a population of 95 elements by applying specific operational rules:

• Assign each element within the sampling frame a unique number from 01 to the total population size.
• Identify a random start from the random number table by dropping a pencil point blindly.
• Determine how the digits in the random number table will be assigned to the sampling frame to choose the specified sample size.
• Select the sample elements from the sampling frame using the established process while omitting duplicates and out-of-range numbers.

Complex probability sampling designs improve statistical and economic efficiency when simple random sampling is impractical. A cluster sample of 690 interviews costing €5 each can be more economically efficient than 424 simple random interviews costing €10 each. You draw systematic samples by following these steps:

• Identify the total number of elements in the population.
• Identify the sampling ratio by dividing total population size by desired sample size.
• Identify the random start.
• Draw a sample by choosing every kth entry.

5. Primary Data Collection Methods

Empirical business research acquires data through primary or secondary sources using either communication or observation approaches. Primary data are collected specifically for a current study, while secondary data exist beforehand. The communication approach involves questioning respondents personally or indirectly, whereas the observation approach records perceived phenomena or events.

Primary data collection utilizes communication and observation approaches that require careful instrument design and administration.

Communication Surveys and Interviews

The communication approach offers high versatility for gathering abstract information, but its success depends heavily on the participant's ability and willingness to cooperate. Non-response error occurs when participants' responses differ systematically from non-participants, and it can be mitigated using call-backs, weighting, or substitution. A general study found that 31 per cent of all first calls were completed, compared to 20 per cent of all first calls in major metropolitan areas. Furthermore, 36 per cent of central city residents had still not been contacted after three call-backs.

You can handle non-response error using these procedures:

• Establish and implement call-back procedures.
• Create a non-response sample and weight results from this sample.
• Substitute another individual for the missing participant.

Intercept interviews conducted in centralized locations reduce travel costs but require careful survey design to accommodate limited respondent time and mobility. Questionnaires must be kept to four pages or fewer to remain clipboard-friendly.

When screening participants for intercept surveys, you should follow these rules:

• When screening for multiple studies at the same time, make your questionnaire distinctive by using coloured paper or paper with a distinctive coloured or patterned edge.
• Make surveys clipboard-friendly by never printing questions on both sides of the paper, keeping fonts legible in dim light, and confining questionnaires to four pages or fewer.
• Write the respondent approach section to include answers to what the study is about, what is in it for the participant, and how long it will take.
• Limit the number of screening questions to the point and critical data based on facts rather than assumptions.
• Keep screening questions safe from respondents' eyes by using normal type style in parentheses or separate devices.
• Do not overuse cumbersome skip patterns.
• Do not force the interviewer to remember responses to previous pages.
• Tally where respondents terminate the screening process or survey by including a horizontal string of question numbers at the bottom of each page.

Interviewers must use neutral probing techniques to stimulate fuller answers without introducing bias when initial responses are inadequate. One of the book's authors was interviewed in a travel survey where the interviewer stopped him after five foreign countries and suggested checking don't know for trip costs because nobody knows the exact amount. Participant-initiated error studies typically reveal a 25 to 50 per cent underestimation of cash and liquid assets.

Telephone interviews provide moderate costs, broad geographic coverage, and faster completion times, enhanced significantly by computer-assisted systems. Telephone surveys yield 45 to 64 per cent lower sampling and data-collection costs compared to personal interviews. In Europe in 2010, there were 726 million mobile subscriptions versus 263 million fixed lines, reaching 173 mobile subscriptions per 100 inhabitants in Finland in 2012.

Self-administered surveys and mail surveys achieve high response rates through the total design method. Across 48 surveys, the total design method achieved response rates ranging from 50 to 94 per cent, with a median response rate of 74 per cent.

Web-based surveys deliver rapid turnaround, low operating costs, and rich multimedia integration, but face challenges regarding sample representation and measurement errors. In 2011, 74 per cent of the European population had home internet access, while internet usage across European countries ranged from 50 to 90 per cent in 2012.

Observation and Structured Fields

Structured observation systematically records behavioral and non-behavioral phenomena as they occur, eliminating respondent reporting biases while capturing natural environmental contexts. Tim Denison observed 2,000 shoppers and found that men shop with targeted focus for 72 minutes with rising blood pressure that drops after purchase, whereas women act as collectors with sustained high blood pressure.

Unobtrusive measures utilize physical traces such as erosion and accretion to study behavior indirectly without participant reactivity or contamination. William Rathje founded the Garbage Project in Tucson, excavating landfills across 150 coded categories. He discovered that Tucson mothers who unanimously claimed to make baby food from scratch actually left completely different evidence in their trash, showing 15 per cent beer consumption reported in interviews versus 77 per cent estimated from domestic rubbish cans.

Qualitative interviews use a flexible, unstructured or semi-structured format to explore topics and understand the respondent's viewpoint, unlike rigid structured surveys designed for explanation and measurement. Unstructured interviews utilize distinct question types designed to elicit rich narratives while keeping the researcher's influence minimal. Complex public infrastructure projects can utilize innovative tendering procedures and case study research to achieve cooperation among diverse actors. The Dutch Ministry of Transport decided in 1999 to replace the A2 motorway with a 2.3 km tunnel in Maastricht estimated to cost €500 million, leading to completion in 2016.

Focus groups leverage group dynamics among a panel of respondents to detect diverse views and generate insights that individual interviews might miss. A focus group meeting lasts 1 to 2 hours with an ideal size of 6 to 10 respondents on a panel. Rebecca conducted focus groups with students using her famous home-made cake and coffee to understand student perspectives on group work and free-riding.

Observational studies span a continuum from completely unstructured field investigations to completely structured hypothesis-testing frameworks. Donald Roy conducted a Class 1 study titled Banana Time by taking a punchpress job in a factory to observe work group rituals, while Mary Waller conducted a Class 3 study observing pilot interactions in flight simulators. Gaining organizational access and rigorously training observers are essential prerequisites for conducting valid observational research. Barbara Ehrenreich went undercover as a waitress, maid, and retail worker across the USA for 18 months, and Günter Wallraff posed as a tabloid reporter and a Turkish immigrant for 3 months.

Collecting observational data requires defining precise sampling dimensions covering who, what, when, where, and how. An observer can use time sampling by observing a quality-control inspection for 10 minutes out of each hour in broken down durations, such as 2-minute observation intervals repeated five times.

6. Secondary Sources and Qualitative Approaches

Secondary data consists of information already gathered by someone else, usually for alternative purposes, offering a way to save time and money. You can combine existing internal records or public data with new collections to create rich datasets without undertaking expensive primary data collection.

Kwasi Bohahene combined internal weekly sales data for political books with government election results per constituency to investigate sales volumes against political participation.

Data-mining is the process of uncovering valid, novel, useful, and understandable patterns from large databases stored in data warehouses. It acts as a sophisticated tool for inductive discovery, combining exploration and discovery with confirmatory statistical analysis to support decision-making.

Dutch supermarket chain Edah analyzed retail sales via data-mining and identified products frequently purchased together, such as candy bars and soft drinks.

Secondary Data and Data-Mining

Librarians and researchers evaluate any information source based on five critical factors to ensure its credibility and utility. You must examine these distinct attributes:

• Purpose
• Scope
• Authority
• Audience
• Format

The data-mining process follows a structured five-step methodology to ensure systematic discovery and validation of relationships. You execute these exact stages:

• Sample: decide between census and sample data
• Explore: identify relationships within the data
• Modify: modify or transform data
• Model: develop a model that explains the data relationships
• Assess: test the model's accuracy

Qualitative and Case Studies

Content analysis reduces copious textual or media data into manageable amounts through manual or automated coding, categorization, and meaning condensation. You apply systematic coding procedures to documentary resources.

Apple Inc. started selling the iPhone through its USA stores on 29 June 2007, which received extensive media coverage investigated via content analysis.

Grounded theory is an iterative qualitative research approach that builds new theory directly from raw data through systematic coding and categorization. You proceed through three distinct coding stages without forcing pre-existing theories onto the raw data:

• Open coding to disaggregate raw data into conceptual units and labels
• Axial coding to link categories and develop theoretical explanations
• Selective coding to focus on principal categories and develop a new theory

A case study is an empirical inquiry investigating a contemporary phenomenon within its real-life context using multiple sources of evidence. You maintain the embeddedness of a phenomenon in its real-life context rather than isolating variables.

7. Experimentation and Measurement

Experiments establish rigorous causal links, while valid measurement requires reliable scales and careful instrument phrasing.

Unlike ex-post facto research where you observe variables as they are, experiments allow systematic alteration of variables. Establishing causality requires three types of evidence: correlation between the independent variable and dependent variable, correct time order, and the absence of influence from extraneous variables.

In a study of bystanders and thieves, student subjects witnessed a confederate steal money from a receptionist desk. Bystanders were more likely to report the theft if they observed it alone rather than in the company of another person.

Experimental Designs

You must progress through a structured sequence of activities before executing your final test. To conduct an experiment, you follow these steps:

• selecting relevant variables
• specifying the level(s) of treatment
• controlling the experimental environment
• choosing the experimental design
• selecting and assigning subjects
• pilot testing, revising and testing

Internal validity ensures that your conclusions truly imply cause by guarding against temporal and behavioral threats. External validity concerns generalizability across people, settings, and times. True experimental designs use an experimental group and a control group made equivalent through random assignment or matching to measure the effects of experimental stimuli.

In testing price differentials for Top Cannery canned tomato soup, 18 company stores and 3 price spreads of 7 cents, 12 cents, and 17 cents were used across treatment groups.

Measurement Scales and Validity

Research measurement assigns numbers or symbols to empirical events based on mapping rules. You measure indicants of object properties rather than the objects themselves. Measurement scales are classified into four data types based on mapping rule characteristics: nominal, ordinal, interval, and ratio.

Reliability measures the consistency, accuracy, and precision of a measuring procedure. You construct measurement scales using five primary approaches:

• Collect a large number of statements relevant and favorable or unfavourable to the attitude
• Ask representative people to state their agreement using a Five-point scale
• Add responses to secure a total score for each person
• Array total scores and select extreme portions representing highest and lowest scores
• Conduct comparative analysis of response patterns to determine items that consistently correlate with low and high favourability

8. Data Preparation, Analysis, and Reporting

Raw data must be prepared, edited, and analyzed using descriptive, bivariate, multivariate, and qualitative techniques before final reporting.

Editing raw data is a critical first step to guarantee accuracy, consistency, completeness, and uniformity before analysis. You must detect errors, omissions, and double answers through field and central editing.

• Familiarize yourself with the instructions given to interviewers and coders.
• Do not destroy, erase or make illegible any original entry by the interviewer; original entries should be crossed out with a single line so that they remain legible.
• Make all entries on an instrument in some distinctive colour and in a standardized form.
• Mark with your initials all answers changed or supplied.
• Write your initials and the date of editing on each instrument completed.

Coding converts qualitative or raw responses into numerical or symbolic categories for efficient quantitative analysis. You must ensure categories meet specific structural rules.

• Ensure categories are appropriate to the research problem and purpose.
• Ensure categories are exhaustive.
• Ensure categories are mutually exclusive.
• Ensure categories are derived from one classification principle.

Statistical Testing and Multivariate Modeling

Hypothesis testing follows a structured sequence to evaluate null hypotheses against sample data. You choose tests based on your population distribution and measurement level.

• State the null hypothesis and alternative hypothesis.
• Choose the statistical test based on population distribution and measurement level.
• Select the desired level of significance before collecting data.
• Compute the calculated difference value using the appropriate formula.
• Obtain the critical test value from the distribution table.
• Interpret the test by comparing calculated and critical values to reject or fail to reject the null hypothesis.

You select multivariate techniques based on whether your variables involve dependencies or interdependencies. Multiple regression predicts a metric dependent variable from multiple independent variables, while LISREL models covariance structures using measurement and structural models.

• Choose between forward selection, backward elimination, or stepwise selection methods.
• Check for multicollinearity using the variance inflation factor where values of 10.0 or more suggest problems.
• Set aside a holdout sample of one-fourth to one-third of the data to test how well the equation predicts beyond its database.

Qualitative Analysis and Reporting

Qualitative data analysis requires transcribing oral accounts into written text and coding fragments to build theory. You transcribe interviews while making deliberate methodological choices.

• Decide whether to transcribe the entire interview or perform selective transcription based on clear criteria.
• Choose between verbatim transcription, including repetitions, or a more formal writing style.
• Determine whether to include non-verbal elements such as pauses and emotional expressions.

You manage coder subjectivity and qualitative richness using specific operational steps.

• Keep a logbook documenting codes and the rules used to assign them.
• Use double coding where study material is coded independently by additional researchers.

10 Key Ideas

1. Business research follows the scientific method to provide reliable empirical evidence for managerial decision-making.
2. Research projects begin with a clear question hierarchy that translates a broad management dilemma into specific measurement questions.
3. Researchers must protect participants from harm and secure informed consent before proceeding with any study.
4. Selecting appropriate research designs and probabilistic sampling plans ensures accuracy and controls extraneous variation.
5. Primary data collection utilizes communication and observation approaches that require careful instrument design and administration.
6. Secondary data consists of information already gathered by someone else, offering a way to save time and money.
7. Experiments establish rigorous causal links, while valid measurement requires reliable scales and careful instrument phrasing.
8. Raw data must be prepared, edited, and analyzed using descriptive, bivariate, multivariate, and qualitative techniques before final reporting.
9. Scientific reasoning relies on deduction and induction, which John Dewey described as the double movement of reflective thought.
10. The validity of a sample depends on accuracy, which measures the absence of bias, and precision, which measures the closeness of sample estimates through the standard error.