Complete Study Notes · All 4 Units · Exam Ready
Fundamentals, Nature, Types, Research Process, Problem, Hypothesis, Design & Literature Review
Research is a systematized effort to gain new knowledge. It is a careful, step-by-step process of investigating a problem, collecting relevant information, analyzing it, and arriving at conclusions that help us make better decisions.
The word Research comes from the French word "recherche" meaning "to go about seeking."
Research Methodology is a way to systematically solve the research problem. It tells us not only WHAT methods are used, but also WHY those methods are chosen.
👉 Method = The TOOL you use (questionnaire, interview)
👉 Methodology = The REASON behind choosing that tool
Real Life Analogy — Doctor Example: A doctor doesn't randomly give medicine. He asks questions (data collection), checks your body (observation), runs tests (analysis), then gives a diagnosis (conclusion). That's exactly what a researcher does — for business problems!
Business Example: Your biscuit company sales dropped 30%. You research: Is it a new competitor? Price increase? New packaging? You collect data, analyze it, find the reason — that's Research!
Research has 6 key characteristics that make it different from ordinary thinking:
Follows a definite, logical order. You cannot skip steps — like baking a cake, you follow the recipe in order.
Based on reasoning. Conclusions must be logically connected to data. If 80% prefer online shopping → invest in e-commerce.
Based on real observation and data — not just opinions or assumptions. You verify, not guess.
Another researcher using the same methodology should get similar results. This builds credibility.
Variables are controlled so you can identify cause and effect clearly. Keep all factors constant except the one you're testing.
Every assumption and finding is questioned and scrutinized. Nothing is accepted at face value.
Explore a new area to understand it better. No specific hypothesis. Example: A company entering a new market does exploratory research to understand it.
Accurately describe characteristics of a group or situation. Answers "What is happening?" Example: Describing buying behavior of millennials in India.
Test whether an assumption is true or false. Example: "Do loyalty points make customers spend 20% more?" — Test it!
Most practical objective. Why are sales dropping? How to improve satisfaction? Which ad works best?
Analyzing past trends to forecast future demand, market changes, economic conditions.
Does training increase productivity? Does price discounting increase sales? Causal relationships help decision-making.
Zomato Example (all objectives combined):
Research follows a definite, logical process. Think of it as a recipe — follow the steps in order!
MOST IMPORTANT step. A well-defined problem is half solved! Must be clear, specific, feasible, and stated as a question.
❌ Vague: "Our company is not doing well."
✅ Good: "Why has sales of Product X declined 25% in North India during Jan–March 2024?"
Read what others have already found about your topic. Understand existing knowledge, identify gaps, avoid duplication, refine your research question. Like reading history before writing the future!
An educated guess about the answer. Stated BEFORE data collection, then tested. H0 (Null): No relationship. H1 (Alternative): There IS a relationship.
The blueprint of your entire research. Specifies how you'll collect data, from whom, what tools to use, how to analyze. Ensures your research is valid, reliable, and efficient.
Select a representative smaller group from the larger population. Population = all elements. Sample = smaller, representative group from population.
Actually gather the information through questionnaires, interviews, observation, experiments, secondary sources. Garbage in = Garbage out!
Editing (check errors), Coding (assign numbers), Classification (grouping), Tabulation (arrange in tables), then apply statistical techniques — percentages, averages, correlations.
Interpret what findings mean and write a report covering: what you studied, how, what you found, conclusions, and recommendations.
Full Example — Flipkart Cart Abandonment:
Problem → Why are customers abandoning shopping carts? | Literature Review → Studies on online shopping behavior | Hypothesis → High shipping costs cause abandonment | Design → Descriptive survey | Sample → 1000 customers | Data → 65% abandon due to high shipping | Conclusion → Offer free shipping above ₹500 | Result → Flipkart Plus launched!
A research problem is a definite expression about an area of concern, difficulty to be eliminated, or a troubling question that exists in theory or practice and points to the need for meaningful investigation.
❌ Bad Research Problem: "How is the Indian economy?" (Too vague, too broad)
✅ Good Research Problem: "What is the impact of GST implementation on the revenue of small retailers in Maharashtra from 2017–2022?" (Specific topic, specific population, timeframe, measurable variable)
A hypothesis is a tentative statement or educated guess about the expected relationship between two or more variables. It is proposed BEFORE research is conducted and then TESTED using data.
Think of it like predicting a cricket match result before the match — research tests whether the prediction was right!
Assumes NO relationship/effect/difference. The "default" assumption — nothing special is happening.
Example: "There is NO significant relationship between advertising expenditure and sales revenue."
States that there IS a relationship, effect, or difference. What the researcher believes to be true.
Example: "There IS a significant positive relationship between advertising expenditure and sales revenue."
States that a relationship EXISTS but does not specify direction. H1: μ ≠ 60 (different, could be higher or lower)
Specifies the DIRECTION of relationship (positive or negative). H1: μ > 60 (specifically higher)
Research Design is the plan, structure, and strategy of investigation conceived so as to obtain answers to research questions and to control variance. It is the ARCHITECT'S BLUEPRINT of research.
Used when problem is not clearly defined. Flexible, unstructured. Methods: Focus groups, interviews, literature review.
Example: "Why do young professionals prefer gig work?"
Describes characteristics of a population. Answers "What is?" More structured. Methods: Surveys, questionnaires.
Example: "What are demographics of Myntra shoppers?"
Establishes CAUSE & EFFECT. Manipulates one variable, observes effect on another. Method: Experiments.
Example: "Does red 'Buy Now' button get more clicks than green?"
A literature review is a comprehensive summary and analysis of existing published research related to your topic. It shows what is already known, what debates exist, and where the gaps are that your research will fill.
Measurement Scales · Scaling Techniques · Questionnaire Design · Primary & Secondary Data · Data Processing
An attitude is a learned predisposition to respond in a consistently favourable or unfavourable manner with respect to a given object, person, or situation. Attitudes influence BEHAVIOUR — if you have a positive attitude towards a brand, you are more likely to buy it!
How does a person FEEL about the object?
"I love Tata products because they are Indian."
What does the person INTEND to do?
"I will always buy a Tata car."
What does the person KNOW or BELIEVE?
"Tata cars are safe and fuel-efficient."
Simply names or labels categories without any order or numerical meaning. Numbers are used ONLY as labels, not as values. You CANNOT do arithmetic on nominal data.
Example: "What is your preferred payment mode?" Cash=1, Debit=2, Credit=3, UPI=4, Netbanking=5. If 210/500 choose UPI → "42% prefer UPI." But you CANNOT say "UPI(4) is better than Cash(1) because 4>1." The numbers are just labels!
RANKS items in order (1st, 2nd, 3rd) but the difference between ranks is NOT equal or known. You know the order, but not by HOW MUCH.
Has equal intervals between values — so you can calculate exact differences. BUT it has NO TRUE ZERO POINT. Zero does not mean "absence" of the attribute.
Has all properties of interval scale PLUS a TRUE ZERO POINT — zero means complete ABSENCE of the attribute. You CAN calculate ratios!
| Scale | Key Property | Example | Statistics Possible |
|---|---|---|---|
| Nominal | Names/Labels only | Gender, Religion, City | Mode, Frequency, % |
| Ordinal | Ranking/Order | Hotel ratings, Satisfaction rank | Median, Percentile |
| Interval | Equal intervals, No true zero | Temperature, Likert scores | Mean, SD, Correlation |
| Ratio | Equal intervals + True zero | Sales, Income, Age | All statistics including ratios |
Scaling means creating a measurement tool that converts attitudes, opinions, or perceptions into numbers so they can be measured and compared.
Presents attitude statements and asks respondents to indicate degree of agreement or disagreement on a 5-point or 7-point scale. The total score = sum of all items (hence "Summated").
Scale: 1 = Strongly Disagree | 2 = Disagree | 3 = Neutral | 4 = Agree | 5 = Strongly Agree
"I am satisfied with my work environment." — Average score of 200 employees = 3.6
Since 3.6 > 3 (neutral) → employees are moderately satisfied!
Measures the meaning of an object by asking respondents to rate it on bipolar (opposite) adjectives on a 7-point scale. Used for brand image studies.
Example: Quality: Good [_][_][_][_][_][_][_] Bad Speed: Fast [_][_][_][_][_][_][_] Slow
Simplified version of semantic differential. Uses a single adjective (not bipolar pairs) and a 10-point scale from -5 to +5. Positive score = descriptor applies; Negative = does not apply.
Example: Rate Jio on AFFORDABLE: +5, +4, +3, +2, +1, -1, -2, -3, -4, -5
Respondents are presented with pairs of objects and asked to choose the preferred one from each pair. Total pairs = n(n-1)/2.
Coca-Cola vs Pepsi → Coca-Cola. Coca-Cola vs Thums Up → Thums Up. Pepsi vs Thums Up → Pepsi.
✅ Very easy — just compare two at a time. ❌ With many objects, pairs become very large.
Respondents are given a fixed number of points (100) and asked to distribute them among attributes according to importance.
"Distribute 100 points for buying a mobile: Price__, Camera__, Battery__, Brand__, Design__ (Total=100)"
Result: Safety=35, Mileage=25, Price=20, Brand=12, Design=8 → SAFETY is top priority! This directly helps product design and marketing.
A questionnaire is any planned set of questions designed to generate data necessary to accomplish the objectives of a research project. Most widely used data collection tool in business research.
Define what info you need. Every question should directly relate to a research objective.
Self-administered, interviewer-administered, online, or by mail. Affects question complexity and length.
Is one question enough? Can the respondent answer? Is it necessary?
Open-ended or closed-ended? What type of scale?
Must be simple, clear, unambiguous, relevant, short, not leading.
❌ BAD: "Don't you think our excellent product deserves 5 stars?" (Leading)
✅ GOOD: "How would you rate our product on a scale of 1 to 5?"
❌ BAD: "Do you use our product daily and find it useful?" (Double-barrelled)
✅ GOOD: Two separate questions!
Start with easy, non-threatening questions. Place complex/sensitive questions (income, age) at END. General → Specific (funnel approach). Group related questions.
Format, font, color, graphics. A visually appealing questionnaire gets higher response rates.
Test with 10–30 people similar to target respondents. Check: Are questions clear? Too long? Confusing? Revise based on feedback.
After revising, finalize. Get reviewed by supervisor. Then deploy for actual data collection.
| Method | What is it? | Advantages | Disadvantages |
|---|---|---|---|
| Observation | Researcher watches and records behaviour without asking questions | Real behaviour captured, no respondent bias | Time-consuming, can't observe attitudes |
| Personal Interview | Face-to-face conversation between interviewer and respondent | High response rate, can observe non-verbal cues | Expensive, interviewer bias |
| Telephone Interview | Interview conducted over phone | Faster, cheaper, wider geography | No visual aids, respondents may disconnect |
| Online Survey | Questionnaire via email/Google Forms/SurveyMonkey | Very cheap, fast, global reach, no interviewer bias | Low response rate, can't verify respondent identity |
| Schedule Method | Filled BY the interviewer (not respondent) after asking questions | High accuracy, usable with illiterates | Expensive, needs trained interviewers |
| Case Study | In-depth investigation of a SINGLE unit — person, org, event | Very rich insights, understands complexity | Cannot generalize, time-consuming |
| Focus Group (FGD) | 6–12 people discuss a topic under a moderator | Rich qualitative data, group dynamics spark ideas | Small sample, dominant participants may influence others |
Data processing is the series of operations performed on raw data to convert it into usable, meaningful information. Raw data is messy, incomplete, unorganized — processing makes it clean, organized, and analysis-ready.
Checking and cleaning collected data for errors, inconsistencies, omissions, and illegible responses.
Editors check for: Completeness | Legibility | Consistency | Accuracy | Uniformity
Assigning numerical codes to responses so they can be entered into a computer and analyzed statistically. Especially important for open-ended questions.
Example: Open-ended question: "Why do you prefer WhatsApp?"
Responses: "Easy to use", "All my friends use it", "Free calls", "Good for groups"
Codes: 1=Ease of use, 2=Popularity/Social network, 3=Free services, 4=Group features
Now "Easy to use" and "Simple interface" both get code 1. Unstructured text → analyzable numbers!
Organizing or grouping data into classes or categories based on common characteristics.
Arranging classified data in the form of a TABLE with rows and columns. Tables are the bridge between raw data and statistical analysis.
Parts of a Good Table: Table Number | Title | Column Headings | Row Headings | Body | Footnotes | Source
Sampling Terminology · Probability & Non-Probability Sampling · Z-Test · t-Test · F-Test · Chi-Square Test
Sampling is a process of learning about the population on the basis of a sample drawn from it. It is the process of selecting a small, representative group from a larger population.
| Term | Meaning | Example |
|---|---|---|
| Population (Universe) | Complete set of ALL elements relevant to the study | All MBA students in Maharashtra |
| Sample | Smaller representative group selected from population | 200 MBA students from 5 colleges |
| Sampling Frame | Complete list from which sample is drawn | List of all registered MBA students |
| Sampling Unit | Basic element selected in sample | Each individual MBA student |
| Sample Size (n) | Number of units selected | n = 150 students out of 1000 |
| Parameter | Numerical value describing the POPULATION (usually unknown) | Average income of ALL MBA students in India (μ) |
| Statistic | Numerical value describing the SAMPLE (used to estimate parameter) | Average income of 200 surveyed students (x̄) |
| Sampling Error | Difference between sample statistic and actual population parameter | Can be reduced by increasing sample size |
| Non-Sampling Error | Errors from mistakes in data collection, not from sampling process | Respondent misunderstands a question |
| Sampling Bias | Systematic error making sample unrepresentative | Certain groups over- or under-represented |
In probability sampling, every element of the population has a known and equal (or calculable) chance of being selected. This allows statistical inference — you can generalize sample results to the whole population with a calculable confidence level.
Every unit has an equal and independent probability of being selected. Like a lottery system!
A teacher wants to select 10 from 60 students. She writes all 60 roll numbers on chits, shakes the box, picks 10 randomly. Each student has exactly 10/60 = 16.7% chance.
✅ Unbiased, simple, equal chance | ❌ Impractical for large populations, requires complete sampling frame
First unit selected randomly, then every kth unit is selected from the sampling frame.
Zomato has 10,000 delivery partners, wants to survey 500. k = 10,000/500 = 20. Random start number = 7. Sample: 7, 27, 47, 67, 87… (every 20th).
✅ Simple, quick, spread evenly | ❌ Periodic pattern in list can cause bias
Population divided into distinct strata (subgroups) based on relevant characteristics, then random sample drawn from EACH stratum. Ensures representation of all subgroups.
Reliance Retail — 40% Jio Mart, 35% Smart Bazaar, 25% Reliance Fresh. For sample of 400: Select 160 from Jio Mart, 140 from Smart Bazaar, 100 from Reliance Fresh (proportionate).
✅ Ensures representation of all subgroups, more precise | ❌ Requires knowledge of population composition
Population divided into clusters (usually geographic), random selection of clusters made, then ALL units within selected clusters are studied.
Government studies digital literacy in rural Maharashtra. Instead of selecting students from all 36 districts, randomly selects 10 districts (clusters) and surveys ALL students in those 10 districts. Saves enormous travel cost!
✅ Very cost-effective for geographically dispersed populations | ❌ Higher sampling error than SRS
Extension of cluster sampling where sampling occurs in multiple stages. Most practical for large-scale national surveys.
NASSCOM surveys IT professionals in India: Stage 1 → Select 10 states | Stage 2 → Select 5 cities per state | Stage 3 → Select 10 companies per city | Stage 4 → Select 20 employees per company.
Selection based on researcher's judgment, convenience, or specific criteria — not random chance. Not every unit has a known probability. Results CANNOT be statistically generalized to the entire population.
Select units that are most easily accessible. Simplest and cheapest method.
Researcher standing outside Nagpur mall stops shoppers and asks about brand preferences. Only those present who agree are sampled.
✅ Very fast, cheap, easy | ❌ High chance of bias, cannot generalize | Best for: Pilot studies, exploratory research
Researcher uses expertise and judgment to deliberately select units most relevant and informative for the study.
Researcher studying AI adoption in Indian companies deliberately selects senior IT managers from TCS, Infosys, Wipro — they have the best knowledge of AI implementation. Only 15 people, all experts.
✅ Information-rich data, efficient | ❌ Researcher bias, not generalizable
Non-probability equivalent of stratified sampling. Population divided into subgroups (quotas); a fixed number from each subgroup must be collected by any convenient means.
HUL survey: Instructions say collect exactly 100 male, 100 female, 75 urban, 75 rural respondents. Field investigators fill these quotas however they can.
✅ Ensures representation of key subgroups, faster/cheaper than stratified | ❌ Within quota, selection is non-random
Starts with a small group who then refer other eligible participants. Sample grows like a snowball. Used for hidden/hard-to-reach populations.
Researcher studying drug addiction: Interview 5 known addicts → each refers 3 more → those 15 refer more. Used because there's NO sampling frame for this population!
Best for: Rare populations, hidden groups (illegal migrants, HIV patients, whistle-blowers)
| Basis | Probability Sampling | Non-Probability Sampling |
|---|---|---|
| Selection | Random, based on chance | Non-random, based on judgment/convenience |
| Equal Chance | Every unit has known probability | Probability of selection unknown |
| Generalizability | Can generalize to population | Cannot generalize statistically |
| Bias | Less biased | More prone to bias |
| Cost | More expensive, time-consuming | Cheaper, faster |
| Use | Quantitative, conclusive research | Qualitative, exploratory research |
| Examples | SRS, Stratified, Cluster, Systematic | Convenience, Purposive, Quota, Snowball |
Hypothesis testing is a statistical procedure for testing whether chance is a plausible explanation of a research finding. It allows making decisions about a population based on sample data.
Like a court case: the accused is assumed INNOCENT until proven guilty. H0 (no effect) is assumed TRUE until data provides strong enough evidence to reject it.
Probability of rejecting H0 when it is actually TRUE (Type I Error rate).
| Decision vs Reality | H0 is TRUE | H0 is FALSE |
|---|---|---|
| Reject H0 | TYPE I ERROR (α) False Positive — wrongly rejected a true H0 | CORRECT DECISION Power of test = 1 - β |
| Fail to Reject H0 | CORRECT DECISION Confidence = 1 - α | TYPE II ERROR (β) False Negative — failed to detect real effect |
Used when: Sample size n ≥ 30 (large sample) AND population standard deviation (σ) is known. Uses the standard normal distribution (mean=0, SD=1).
| Type of Test | α = 0.05 | α = 0.01 | Decision |
|---|---|---|---|
| Two-tailed | ±1.96 | ±2.576 | Reject if |Z| > critical value |
| One-tailed Right | +1.645 | +2.326 | Reject if Z > critical value |
| One-tailed Left | -1.645 | -2.326 | Reject if Z < critical value |
Example — HUL Surf Excel Sales:
Claim: Average daily sales = ₹5000 (μ = 5000). Researcher surveys 36 outlets: x̄ = ₹5200, σ = ₹600.
H0: μ = 5000 | H1: μ ≠ 5000 (two-tailed, α = 0.05)
Z = (5200 - 5000) / (600/√36) = 200/100 = 2.0
Critical value at α=0.05, two-tailed = ±1.96
Since |Z| = 2.0 > 1.96 → REJECT H0
Conclusion: Average sales ARE significantly different from ₹5000.
Used when: Sample size n < 30 (small sample) OR population standard deviation (σ) is UNKNOWN. Uses the t-distribution, which has thicker tails reflecting more uncertainty.
Example — MBA Study Hours:
Principal claims students study 6 hours/day on average. Researcher surveys 16 students: x̄ = 5.2 hours, s = 1.2 hours.
H0: μ = 6 | H1: μ ≠ 6 (α = 0.05)
t = (5.2 - 6) / (1.2/√16) = -0.8/0.3 = -2.67
df = 16 - 1 = 15. Critical t = ±2.131 (from t-table)
Since |t| = 2.67 > 2.131 → REJECT H0
Conclusion: Students study significantly FEWER than 6 hours per day.
Used to compare the means of THREE or more groups simultaneously (ANOVA = Analysis of Variance). Named after statistician Ronald Fisher. Avoids multiple t-tests which increases Type I error.
Example — Tata Motors Regional Sales:
Marketing manager wants to know if monthly sales are different across North, South, West regions.
H0: μNorth = μSouth = μWest | H1: At least one region's mean is different
Using ANOVA: If F-calculated > F-critical → REJECT H0 → At least one region performs differently.
⚠️ F-test only tells you THAT groups differ — NOT WHICH groups differ. Use post-hoc tests (Tukey's HSD) for that.
Chi-Square is the most important non-parametric test. Unlike Z, t, F tests (which deal with means), Chi-Square works with CATEGORICAL data — data classified into categories.
Example — ChatGPT Usage & Study Stream:
Survey of 200 students: Do you use ChatGPT? (Yes/No) × What is your stream? (MBA/Engineering/Commerce)
H0: ChatGPT usage and stream are INDEPENDENT (no association)
H1: They are significantly associated
df = (2-1) × (3-1) = 2. Critical χ² at df=2, α=0.05 = 5.991
If calculated χ² = 8.23 > 5.991 → REJECT H0 → ChatGPT usage IS significantly associated with stream!
| Situation | Choose This Test |
|---|---|
| CATEGORICAL data (nominal/ordinal), testing association | Chi-Square (χ²) |
| QUANTITATIVE data, large sample (n≥30), σ KNOWN | Z-Test |
| QUANTITATIVE data, small sample (n<30) OR σ UNKNOWN | t-Test |
| QUANTITATIVE data, comparing 3 or MORE groups | F-Test (ANOVA) |
Report Types · Content & Format · Oral Presentation · Research Agencies · Data Analysis Software · IMRaD
A research report is a written document that describes the research process and findings, meant to be read by the audience for whom the research was conducted. It is the final and most important output of any research activity.
Even brilliant research is WASTED if the report is poorly written. A well-written report translates findings into actionable business insights.
Written for expert audiences — academics, researchers. Very detailed, includes all statistical analyses, formulas. Length: 100–500 pages.
Example: PhD dissertation, peer-reviewed journal article
Written for non-experts — general public, company employees. Simple language, lots of visuals, focuses on key findings. Length: 5–20 pages.
Example: Annual customer satisfaction summary
Written for management/decision-makers. Concise, action-oriented, focuses on recommendations. Technical details in appendices. Length: 15–50 pages.
Example: Consulting firm report to Wipro
Written DURING the research process to update clients on progress. Does not contain final conclusions.
Example: Monthly report submitted to NASSCOM during 6-month study
Most detailed and rigorous. Written to fulfil academic requirements. Must follow strict institutional guidelines.
Example: MBA dissertation, M.Phil. or PhD thesis
Good Title Example: "Impact of Social Media Advertising on Purchase Intention Among Young Adults in Mumbai (2024)" — Specifies: DV (purchase intention), IV (social media advertising), population (young adults), location (Mumbai), time (2024).
Bad Title: "A Study of Social Media and Shopping." — Too vague!
Background of study, problem statement, research objectives, significance, scope and limitations, chapter plan.
Survey of existing knowledge. Must critically ANALYZE (not just summarize). Identify contradictions, gaps. Establish theoretical framework. Justify why current research is needed.
Research design, population and sample, data collection instruments, variables, data analysis methods, reliability and validity, limitations.
Core chapter! Demographic analysis, analysis of each objective, tables and figures, statistical tests (with calculated values, critical values, decisions), interpretation in plain language.
Findings = What was found (factual). Conclusions = What it means (interpretation). Recommendations = What to do (actionable). Scope for further research.
APA Format Example:
Cooper, D.R., & Schindler, P.S. (2019). Business Research Methods (13th ed.). McGraw-Hill Education.
Beri, G.C. (2018). Marketing Research (6th ed.). Tata McGraw-Hill Education.
| Formatting Element | Recommended Standard |
|---|---|
| Paper Size | A4 (210mm × 297mm) |
| Margins | Top & Bottom: 1 inch | Left: 1.5 inches (for binding) | Right: 1 inch |
| Font (Body) | Times New Roman 12pt or Arial 11pt |
| Font (Headings) | H1: 16–18pt Bold | H2: 14–16pt Bold | H3: 12–14pt Bold |
| Line Spacing | 1.5 lines for body text; single spacing for tables, footnotes, references |
| Paragraph Alignment | Justified (aligned on both left and right margins) |
| Page Numbering | Roman numerals (i, ii, iii) for preliminary pages; Arabic (1, 2, 3) from Introduction chapter |
| Tables & Figures | Numbered sequentially (Table 4.1, Figure 4.2); Title ABOVE table; Source BELOW; Caption BELOW figure |
Greet audience, introduce yourself, state the title and purpose of the research.
Clearly explain what problem the research addresses and why it matters.
Briefly explain how research was conducted — sample, data collection, analysis methods.
Present most important results with visual aids — charts, tables, infographics.
What do findings mean? What should be done?
Answer questions from audience. Be prepared for challenging questions. If you don't know, say so honestly.
| Chart Type | Best Used For | Example |
|---|---|---|
| Bar Chart | Comparing categories (most common) | Sales comparison across 5 regions |
| Pie Chart | Showing proportions of a whole | Market share of Tata, Hyundai, Maruti |
| Line Graph | Showing trends over time | Monthly sales Jan to Dec |
| Histogram | Frequency distribution of continuous data | Distribution of exam scores |
| Scatter Plot | Relationship between two variables | Price vs demand, advertising vs sales |
| Box Plot | Showing spread and outliers | Salary range across departments |
Research agencies are specialized organizations that have expertise, tools, and trained personnel to design and execute research projects on behalf of clients.
Complete services: problem definition, questionnaire design, data collection, analysis, report writing.
Examples: Nielsen, IMRB International (Kantar)
Focus on specific activities or sectors. Data collection only, data analysis only, or sector-specific (healthcare, automotive, financial).
Examples: IQVIA India (healthcare), CRISIL (financial)
Conduct continuous research and sell results to multiple subscribers. Cost-effective — many clients share the cost.
Examples: NASSCOM annual IT reports, Nielsen Retail Measurement
Conduct surveys exclusively through digital platforms. Maintain large online panels.
Examples: SurveyMonkey, Qualtrics, YouthSight India
Large-scale official research for policy-making.
NSSO, NCAER, CSO, RBI research wing, NITI Aayog
| Software | Cost | Ease of Use | Power | Best For | Level |
|---|---|---|---|---|---|
| SPSS (IBM) | Paid | Easy (GUI) | High | Social science / business research | MBA / PhD |
| Microsoft Excel | Paid (MS) | Very Easy | Medium | Business analysis, assignments | MBA / UG |
| R Software | Free | Difficult (code) | Very High | Academic research, economics | PhD |
| Python | Free | Moderate | Very High | Big Data, AI/ML research | Tech / PhD |
| SAS | Very Expensive | Moderate | Very High | Banking, pharmaceutical, large datasets | Corporate |
| Google Sheets/Forms | Free | Very Easy | Low | Small surveys, minor projects | Minor Project |
IMRaD stands for Introduction, Methods, Results, and Discussion. It is used by journals like Harvard Business Review, Journal of Marketing Research, and Indian Journal of Management Studies. Research papers: typically 4,000–8,000 words.
Research problem, background, key prior work, research gap, objectives/hypotheses. Answers: What? Why does it matter? What was done before? What gap does this fill?
How research was conducted. Research design, sample, instruments, procedure, analysis methods. Written in past tense, third person/passive voice.
Actual findings WITHOUT interpretation. Tables, graphs, statistical outputs. Must include: test statistic value, df, p-value, effect size.
Interpretation of results, comparison with previous studies, implications, limitations. Answers: What do results mean? How do they relate to previous research? Practical implications?
| Common Mistake | How to Avoid It |
|---|---|
| Vague research problem | State specifically — who, what, where, when, measurable variable |
| Too many objectives | Limit to 3–5 specific, measurable objectives |
| Poor literature review | Critically analyze — don't just summarize. Identify gaps. |
| Methodology not justified | Always explain WHY you chose each method |
| Data without interpretation | After every table/test, write what it means in plain language |
| Missing citations | Every borrowed idea/fact/statistic must be cited |
| Overgeneralization | Clearly state scope and limitations of your study |
| Conclusions = Findings | Findings = WHAT | Conclusions = WHAT IT MEANS | Recommendations = WHAT TO DO |
| Informal language | Use formal, academic language. No contractions (don't, can't), no slang. |
MBA 202 — Research Methodology
P.R. Pote Patil College of Engineering & Management, Amravati | Vikas Bhusare | Sem II 2025–26
Reference Books: Brayman & Bell · G.C. Beri · Naval Bajpai · S.L. Gupta & H. Gupta · Donald Cooper & Pamela Schindler