Class-4 (UNK)

Class - 4

Observation – Non verbal communication. It is a quantitative research. (Similarly mystery shopping is a quantitative research).

What can be observed:

1. Physical actions / Human Behavior – (Shoppers movement pattern in stores) - Whether the customer uses a shopping list (planned) or picks up product in an impulse manner. In case respondent is followed in the store, both respondent’s and retail outlet’s owner permission is necessary. If not following the respondent, then only permission of retail outlet owner is sufficient. (Whether the customer is picking up things at / below / above the eye contact level). Pepsi / Coke are always at the exit as the people can take it in chilled form and can drink it immediately after leaving the shop (product still remains cold). Similarly Dairy is also placed closer to exit for the same reason. 1000 Dhirams a month per SKU from Pepsi / coke is typically charged by stores (SKU – Stock keeping unit) if they are placed at prime position(s) in stores.

2. Verbal behavior – How are people speaking to each other (Example – Statements made by airline travelers who wait in line). What happens to people’s reaction when the queue is broken / followed.

3. Expressive behavior – (Facial expressions, tone of voice, and other forms of body language)

4. Spatial relations and locations – (How close visitors at an art museum stand to paintings) – Sales staff should always be in close proximity with Arab customer (minimal distance between them) as the customer feels that he /she is taken care of (Can I Help you / Follow you / do … for you). Opposite to that, Sales person should stand at a distance when dealing with westerns or Asians. 1^st sign of interest to be observed in the customer – First thing customer picks up in store.

5. Temporal patterns – (how long fast food customers wait for their order to be served) – Mood swings to be observed at different times of the day – peak time, lean time, weekdays, weekends.

6. Physical objects – (What brand name items are stored in consumers pantries) – Checking the pantries to ensure whether a specific brand liked by the respondent is stored with him / her or not. To check, what all one consumes in the whole day, garbage bins are checked to ensure answer matches with what people answered.

7. Verbal and pictorial records – (Bar codes on product packages) – Lots of information is stored in bar-codes like source of product, kind of people who bought these products, demographics of such people. In west, it is observed in depth. In middle east, it is limited upto whether the product is in stock or not.

Mystery shopping – 3^rd party does it. To check Service standards are adhered to. A person who poses as a customer comes and visits the outlet / bank and then evaluates the sales staff / service on various service level parameters. Purpose is not to fire but to train the staff.

Body odor, how to communicate – Informal but direct expression is better than any other method of communication.

2 kinds of observation – Direct and Indirect observation. In direct observation, the party / person being observed knows that he / she is being observed.

Errors in Research

(How to be aware of errors and how to avoid them)

Margin of error on your data can be 1% (Stringent), 5% (acceptable) or 10% (Lenient).

Survey Error

· Random Sample Error

· Systematic Error

§ Respondent Error

ü Non Response Error

ü Response Bias

Ø Acquiescence Bias

Ø Extremity Bias

Ø Interviewer Bias

Ø Auspices Bias

Ø Social Desirability Bias

§ Administrative Error

ü Data Processing Error

ü Sample Selection Error

ü Interviewer Error

ü Interviewer Cheating

2 types of errors –

· Random sampling Error – A statistical fluctuation that occurs because of change variation in the elements selected for the sample.

· Systematic error / bias – It results from some imperfect aspect of the research design or from a mistake in the execution of the research.

RSE has to be live with whereas systematic error needs to be lessened or avoided.

Systematic Error –

· Administrative Error

· Respondent Error

o Non Response Error – Those who fail to respond / cooperate. Respondent not a home / not picking his / her mobile. Respondent leaves the interview in between due to some personal reason.

o Response Bias – A bias that occurs when respondents tend to answer questions in a certain direction that consciously or unconsciously misrepresents the truth.

If your questionnaire should take 30 minutes, should you say 5 minutes or ball-park figure of 20 minutes – Chose the latter option, as Interviewer might stop answering questionnaire after 5 minutes and all the efforts / time spent will go waste. Always give a ball park time (20 min or more for a 30 min interview), but never give a narrow time (5 / 10 / 15 min for a 30 minute interview). It has to made sure that interviewer gives you full quota of time required to completely answer the questionnaire.

Response Bias -

· Acquiescence bias – A category of response bias that results because some individuals tend to agree with all questions or to concur with a particular position. Check / ensure the answer by challenging the respondent that the response given by them is actually yes. Even after challenging, the response remains yes, then the interviewer has to live with that response.

· Extremity bias – A category of response bias that results because response styles vary from person to person; some individuals tend to use extremes when responding to questions. One can have some questions, but not all questions answered to extremities.

· Interviewer bias – MOST COMMON ERROR. A response bias that occurs because the presence of the interviewer influences the answer. Created verbally or by expressions. Example – Interviewer asks - What’s your income, Respondent says - 9500 Dhirams. Interviewer says – That’s it! (expression of disgust shown to respondent by interviewer so that response changes to above 10,000 Dhirams)

· Auspices bias – Bias in the response of subjects caused by the respondents being influenced by the organization conducting the study. Example – RM professor asking students to give independent feedback of RM subject in his presence (No one will give a negative feedback). So, in case one is doing a study on chocolates for Nestle, then they should never disclose the name Nestle while asking questions because respondent may have a perception about Nestle which will influence his / her answer towards answering general chocolate questions which do not involve Nestle alone. Only in customer satisfaction survey, name of the brand / client should be revealed (I am conducting a survey on SP Jain and require your feedback on it).

· Social desirability bias – Bias in responses caused by respondent’s desire either conscious or unconscious to gain prestige or appear in a different social role. Hence, never ask socially loaded questions.

Administrative Error -

· Data processing Error – Example – Year of birth of respondent is found to be 1875, which is wrong. It should be 1975.

· Sample selection Error – Example - Interview to be done for teens, but actually done for adults

· Interviewer Error – Example – In a satisfaction survey, interviewer asks about satisfaction for a product on a 10 point scale and the respondent just says satisfied with product. Interviewer can’t mark any point till the interviewer says it. Example 2 – gone to interview people above age of 25, respondent is 24 years, Interview is not conducted though incorrectly the age is circled as 25.

· Interviewer Cheating – Interviewer fills up the question on their own without asking any respondent. It results in lot of erratic movement of data.

Sampling Terminology (Stage 3)

Sampling types

Ø Non – Probabilistic – All the sample entities do not have equal chance / probability to get selected in the sample

· Convenience – Here, sample is selected based on the technique which is most convenient to the researcher

· Judgment / Experience – Example -

§ Researcher wants to interview people in a building with monthly income > 10,000 AED. The parking lot of the building has Kia, Hyundai and Corolla. Since, all these vehicles are low end in the middle-east region, researcher will not go into the building for interview

§ In case one wants to interview westerns’ in Dubai, interviewer will go either to Jumeriah or Sheikh Zayed road (would never go to knife road)

· Quota (Most popular method) – Interviewer selects the sample based on certain quotas. Quotas are pre-agreed / pre-decided on which interviewer is going to select the sample. Example – 60% population in UAE is from Indian sub continent. Hence, Quota sampling method will ensure that sample share of Indian sub continent respondents is 60% for current survey so that sample is representative of the universe.

· Snowball / Referral – 1^st respondent is selected purely in random manner. After the interview, the respondent is asked if he knows people of the same profile like his so that those people can be interviewed and so on. It is typically done for niche products (focused / targeted). Example – To interview people with investable assets of $200,000 and above in Dubai (Niche sample).

Ø Probabilistic – The sample entities have equal chance of getting selected into the sample.

· Simple Random – Pick up the person in purely random manner. Use random number table to choose the respondent (Even the building chosen is using random number table). One can use excel also to generate random numbers. Example – To interview 1 child in a house below 10 years. Random numbers are 2,5,7…….... Go to each house and see how many children are eligible and give them any number. Then from first house, pick up the 2^nd numbered child, from second house, pick up 5^th child and so on. In case, the second house has less than 5 children who are eligible for the interview, then skip that house.

· Systematic – It is selecting every N^th entity / respondent in the sample. Example – Interviewing every 2^nd or 5^th child in the family.

· Stratified - Break the universe / sample into stratas of Age / Sex / Work experience (Years) etc and choose sample from a particular strata, which fulfills the research objective.

· Cluster / Geographic Sampling – Example – Want to interview people residing in UAE, will only go to 4 geographic states – Dubai, Abu Dhabi, Alain, Sharjah; need not go to other emirates as 70% of population of UAE resides in these states.

· Multistage – Sampling done for many countries simultaneously. Also, sample of the 1^st stage becomes universe of the 2^nd stage. Example – Researcher goes door knocking to 3 houses for a food product test. Respondent 1 tastes the product and gives feedback. Respondent 2 is not available at his residence. Respondent 3 didn’t participate in the survey at all. Hence, next time, the researcher will only go to respondent 1. Hence, sample of 1^st stage of interview (which is respondent 1 from all the three) has now become the universe for second stage – where researcher only approaches respondent one).

Ø Which is the most popular method –

· Example – Select the sample of 100 people from UAE population based on nationality.

UAE Nationals	Expact Arabs	Expact Asians	Westners
15%	20%	60%	5%

· Hence, quota sampling / Stratified quota sampling is most popular as it is representative of the universe. Stratified will ensure that sample is selected from a particular strata and quota will ensure that the sample from the strata is representing the universe. (If only stratified is followed, then it may happen that strata of nationality is chosen, but when sample is selected, Westners % may increase to 50% instead of 5%)

Ø What determines the sample size –

· What is the unit of analysis / How researcher is going to analyze the data / What breaks are required by researcher to display to the client

· Minimum sample size requirement for each break

· Example –

§ Unit of Analysis – Nationality like Westners etc.

§ Minimum sample size = 30 (Conventional)

Nationality	UAE Nationals	Expact Arabs	Expact Asians	Westners	Total
Actual % distribution	15%	20%	60%	5%	100%
Minimum Sample Size	30	30	30	30	120
Actual Sample Size (Fulfilling Quota criterion)	90	120	360	30	600
For sample size of 200	30	40	120	10 (but westners are under-represented)	200

Boost the westners segment separately by 20 so that it reaches 30 (minimum sample size) for complete analysis. But the total will still remain 200 and booster will only add to westners and not total. This is useful in case cost per interview has to lowered ($70 per interview). Costing will be for 200 random sample and 20 booster sample (Random will cost $70, then booster will cost $90).

Example 2 – To find out people who drink only black tea (tea without milk) - 11%-12% (for 1000 people interviewed in UAE). All telephonic interview ($10 per interview cost).

Since, trend / number for the universe of above scenario was not available, hence, process of LISTING is adopted. (One has to list / determine / to find out penetration of what researcher is finding out in the universe.

Stages in the Selection of a Sample –

1) Define the target population / operationally defining the sample –

a. Workaholic people – Who works at least 12 or more hours in a day.

b. Serious car buyers – Has a license to drive; Has bought a car in last 1 year / has a car / planning to buy a car in next 1 year; How to finance the car buying; Have the buyer visited any car showroom; Have the buyer test driven the car.

2) Select a sampling frame

3) Determine if a probability or non-probability sampling method will be chosen

4) Plan procedure for selecting sampling units

5) Determine sample size

6) Select actual sampling units

7) Conduct fieldwork

*Operational definition – Target population. It has implication on the questionnaire in the following way – Screener question has to be asked upfront on the questionnaire. The sample size is dependent on the unit of analysis and the minimum sample size (which is 30). So, it is a bottom up approach.