Topic 01
Attitude Measurement
+
Attitude = learned predisposition to respond favourably or unfavourably to an object or person.
🔥 ABC Model — Always Say in Exam!
A = Affective = FEELING (How do you feel?)
B = Behavioural = ACTION (What will you do?)
C = Cognitive = BELIEF (What do you know/believe?)
B = Behavioural = ACTION (What will you do?)
C = Cognitive = BELIEF (What do you know/believe?)
Why Measure Attitudes?
Attitudes predict buying behaviour · Help product development · Guide advertising design · Aid brand positioning · Identify market segments
Measurement
Assigning numbers to observations so they can be analyzed. Converts feelings/opinions into data. Example: Instead of "customers are happy" → "satisfaction score = 7.8/10."
📌 Swiggy ABC Model
C — I KNOW Swiggy delivers in 30 minutes. A — I FEEL happy using Swiggy because it saves time. B — I ORDER on Swiggy 3 times a week. All 3 together = my complete ATTITUDE towards Swiggy!
Topic 02 — VERY IMPORTANT
4 Types of Measurement Scales
+
4 levels from simplest (Nominal) to most powerful (Ratio) — each allows different statistical analysis.
🔥 NOIR Memory Trick
N — Nominal (Names) · O — Ordinal (Order) · I — Interval (Equal Intervals, No true zero) · R — Ratio (Real zero + Ratios possible)
NOIR = Black in French = 4 shades from simple to powerful!
NOIR = Black in French = 4 shades from simple to powerful!
| Scale | Key Property | Example | Stats Allowed |
|---|---|---|---|
| NOMINAL | Names/Labels only. No ranking, no arithmetic. Numbers are just labels. | Gender (M=1, F=2). City codes. Dhoni's jersey no.7 ≠ less value than no.11! | Mode, Frequency, % |
| ORDINAL | Can rank/order. But gap between ranks is NOT equal or known. | Hotel ratings: Excellent > Good > Average. Race: 1st, 2nd, 3rd — but don't know by how much! | Median, Percentile, Rank correlation |
| INTERVAL | Equal intervals. NO true zero (zero ≠ absence). Can calculate differences, NOT ratios. | Temperature: 0°C ≠ no temperature. Likert 1–5 scale. 40°C is NOT twice as hot as 20°C. | Mean, SD, Correlation, t-test, ANOVA |
| RATIO | Equal intervals + TRUE ZERO. Can calculate ratios. Most powerful scale. | Sales: Rs.200cr = 2× Rs.100cr. Age. Income. Weight. Rs.0 = truly no sales! | ALL statistics including geometric mean and ratios |
📌 All 4 in One Smartphone Survey
Nominal → What is your gender? | Ordinal → Rank these 5 brands 1st to 5th | Interval → Rate satisfaction 1 to 5 on Likert scale | Ratio → How much did you spend on your last phone in rupees? All 4 scales in ONE survey!
Topic 03
Scaling Techniques
+
Tools to convert attitudes and opinions into measurable numbers for analysis.
Likert Scale ⭐ Most Popular
1–5 agreement scale: 1=Strongly Disagree to 5=Strongly Agree. Also called Summated Rating Scale. Easy to understand, allows mean score. Example: "I am satisfied with my work environment." Avg = 3.6 = moderately satisfied.
Semantic Differential Scale
Bipolar OPPOSITE adjectives on a 7-point scale. Good←→Bad, Fast←→Slow. Used for brand image. Example: Rate Paytm vs Google Pay on Fast/Slow, Trustworthy/Untrustworthy.
Stapel Scale
Single adjective (not bipolar), scale from −5 to +5. No zero. Positive = applies, Negative = doesn't apply. Easier for telephone surveys. Example: Rate Jio on AFFORDABLE: +3 means quite affordable.
Graphic Rating Scale
Mark on a continuous line or choose emoji/icon. Example: Airport happy-face buttons. Mumbai airport got 50,000 responses in one month!
Paired Comparison Scale
Choose preferred item from pairs. Easy — just compare two at a time. Formula: n(n-1)/2 pairs. Example: Coca-Cola vs Pepsi → choose one. Coca-Cola vs Thums Up → choose one.
Constant Sum Scale
Distribute 100 points among attributes by importance. Total MUST = 100. Shows relative importance. Example: Car buying — Safety=35, Mileage=25, Price=20, Brand=12, Design=8. Safety is top priority!
Topic 04
Questionnaire Design
+
Most widely used data collection tool. A planned set of questions to collect research data.
Types of Questions
Open-ended: Free answer in own words. Rich qualitative data. "What do you like most about our service?"
Closed-ended: Fixed options. Dichotomous (Yes/No), Multiple choice, Rating scale, Checklist.
Filter questions: Route respondents. "Do you own a car? YES → Q5, NO → Skip to Q10"
Closed-ended: Fixed options. Dichotomous (Yes/No), Multiple choice, Rating scale, Checklist.
Filter questions: Route respondents. "Do you own a car? YES → Q5, NO → Skip to Q10"
Key Design Principles (Brayman & Bell)
Clarity · Single Focus (no double-barrelled Qs) · No Leading Questions · Logical Flow (general → specific) · Sensitive Qs at end · Keep it short · Always PRE-TEST!
🔥 9 Steps in Questionnaire Design
1.Specify info needed → 2.Type of questionnaire → 3.Content of questions → 4.Form of response → 5.Wording → 6.Sequence → 7.Physical layout → 8.Pre-test → 9.Final questionnaire
📌 Pizza Company Example
Cover: "Takes 3 minutes, all confidential." Filter: "Ordered pizza online in last 3 months?" Main Q: "Rate your experience 1–5." Open Q: "What problems did you face?" After 500 responses: 60% gave low ratings for delivery time. Solution: Hire more delivery staff!
Topic 05
Primary Data Collection Methods
+
5 ways to collect fresh primary data: Observation, Interview, Schedule, Case Study, FGD.
1. Observation Method
Watch and record behaviour WITHOUT asking questions. Real behaviour captured — not what people SAY they do. Cannot observe attitudes. Example: Supermarket CCTV shows 80% turn right at entrance but display is on the left. Move display → sales +30%!
2. Interview Method
Personal: Best quality, most expensive. Can show visual aids, observe non-verbal cues.
Telephone: Faster, cheaper, wider reach. No visual aids.
Online/Email: Very cheap, global reach. Low response rate. No identity verification.
Telephone: Faster, cheaper, wider reach. No visual aids.
Online/Email: Very cheap, global reach. Low response rate. No identity verification.
3. Schedule Method
Like a questionnaire BUT filled by the INTERVIEWER, not respondent. Used for illiterate populations, rural surveys, complex questions. Government census uses schedules!
4. Case Study Method
In-depth investigation of a single unit (person, org, event) over time. Uses multiple data sources: interviews, documents, observations. Cannot generalize to whole population. Example: Amul success story research.
5. Focus Group Discussion (FGD)
6–12 people discuss a topic under a moderator's guidance. Rich qualitative data. Group dynamics spark ideas. Small sample — cannot generalize. Example: Lays tested 5 new chip flavours across 3 cities → launched regional flavours!
📌 LIC Insurance Interview Example
LIC wanted to know why youth (22–30) don't buy life insurance. Personal interviews revealed: "I don't think about death at this age" · "Too expensive" · "I trust FDs more." These rich insights helped LIC design a new product targeted at young professionals!
Topic 06
Secondary Data Sources
+
Existing data collected by someone else. Quick, cheap, but may not perfectly fit your needs.
Internal Secondary Data
From within your OWN organization: Sales records · Customer databases · CRM data · Previous research reports · Financial statements · Employee records
External Secondary Data
From OUTSIDE: Govt (Census, RBI, SEBI) · Industry reports (NASSCOM, CII, FICCI) · Academic journals · World Bank · IMF · Economic Times · Google Scholar · Statista
Advantages vs Disadvantages
Advantages: Saves time and money · Large volume available · Good for background research.
Disadvantages: May not match needs · May be outdated · Accuracy unknown · Different definitions may be used.
Disadvantages: May not match needs · May be outdated · Accuracy unknown · Different definitions may be used.
📌 EdTech Startup Example
A startup wants to enter EdTech. No budget for primary research yet. So they use: Ministry of Education reports (student enrollment) + NASSCOM EdTech report + news about Byju's and Unacademy + academic papers on online learning. Secondary research first → then primary research later!
Topic 07 — VERY IMPORTANT
Data Processing — 4 Steps
+
Converting raw, messy collected data into clean, organized, analysis-ready information.
🔥 ECTAB Memory Trick
E — Editing · C — Coding · T — (Classifica)tion · A — (Tabul)ation · B — Base for Analysis (Done!)
Say it loud: E-C-T-A-B!
Say it loud: E-C-T-A-B!
E
Editing — Check collected data for errors, inconsistencies, omissions. Field editing = done by interviewer right after collection. Office editing = done at research office. Check: Completeness · Legibility · Consistency · Accuracy (Age=150 is clearly wrong!) · Uniformity.
C
Coding — Assign NUMERICAL CODES to responses for computer analysis. Especially important for open-ended questions. Read all answers → identify themes → assign code numbers. Create a codebook.
T
Classification — Group data into classes/categories based on common characteristics. One-way (by 1 attribute) · Two-way (by 2 attributes) · Manifold (3+ attributes). Also: Geographical, Chronological, Qualitative, Quantitative classification.
A
Tabulation — Arrange classified data in TABLE format (rows + columns). Simple table (1 variable), Cross/Contingency table (2 variables), Complex table (3+ variables). Good table must have: Table number · Title · Column/Row headings · Body · Footnotes · Source.
📌 WhatsApp Coding Example
Open-ended Q: "Why do you prefer WhatsApp?" Answers: "Easy to use" · "All friends use it" · "Free calls" · "Good for groups" · "Simple interface." After coding: 1=Ease of use, 2=Popularity, 3=Free services, 4=Group features. Now "Easy to use" AND "Simple interface" both = Code 1. Unstructured text becomes analyzable data!
📌 Monthly Sales Tabulation
Raw: "Jan North Rs.45L, South Rs.38L, East Rs.22L, West Rs.51L…" Very confusing! After tabulation with regions as rows and months as columns → suddenly you can SEE West is always highest! Decision: Increase West region marketing budget!