Class-3 (UNK)

Class - 3

Types of Scales used

Broadly there are 4 kinds of scale –

1. Nominal Scale – Every scale in research is a nominal scale. Sex = Male – 1, Female -2, is a nominal scale.

2. Ordinal Scale – It ranks order in categories like excellent, good, fair, poor etc. It is also a numerical scale. Measures the order of the object or phenomenon (Excellent-1 / Very Good-2 / Good-3 / Bad-4 / Poor-5)

3. Interval Scale – Measures distance between 2 points / values and range. Example: Under which age category you fall under (26-31 years / 31-35 years / 36-40 years )

4. Ratio Scale – Rate importance between zero to 100 (How many times an impact of something, need base as zero, which is a ratio scale). Please rate importance 0-100, indicating how important is this to you (Ratio Scale). Please rate importance 1-100, indicating how important is this to you (Interval Scale).

5. Likert Scale: It is an odd point scale / most common scale. Used as a 5 point scale, except in a regular satisfaction survey where it is 4 point scale to force the respondent to give an answer / opinion for better action-ability on results (to avoid the center point where most respondents tend to select to express neutral satisfaction). Respondents indicate their own attitude by checking how strongly they agree or disagree with statements. Response alternatives are – Strongly Agree, Agree, Uncertain, Disagree, strongly Disagree.

5 point scale should not be used when opinion are being taken (Satisfaction survey), else most of the respondents tend to be biased towards neutral opinion. Hence, satisfaction survey is always a 4 point scale.

6. Constant Sum Scale

7. Graphic Rating Scale (For Visual communication)

8. Semantic differential scale: It is a bipolar 7 point scale (Vertically, ranking is done and horizontally rating is done), trying to get response of the person at the 2 extremes to best understand what he/she thinks about that question being asked.

Example: The following example illustrates the rating of a restaurant by a single respondent based on 3 parameters – Rating of 7 being the Best and 1 being the Worst.

Rating is done for a single parameter for the restaurant.

Ranking of Parameters - Users may be asked to rank which parameter is most important to them.

Rating Parameters of a Restaurant	1	2	3	4	5	6	7
Variety of Food						Y
Value for Money					Y
Location							Y

While preparing the questionnaire, the need to know questions should be avoided. All open ended and closed ended questions needs to be recorded, analyzed and presented to the client in a quantitative research.

Properties of a Scale

Scale should be reliable and valid and hence, accurate

ü Reliable – Error Free scale / research

ü Valid – Answers the objective of research (Hence, automatically it is also error free)

Whatever is reliable need not be valid (as research done may be error free but still might not be answering the objective), but whatever is valid must be reliable (Anything answering the objective is automatically error free).

Reliability + Validity = Accuracy

Hence, validity and accuracy is same and used synonymously in the industry.

Attitude measuring process

· Ranking – Ranking is relative. It requires that the respondent rank order a small number of objects in overall performance on the basis of some characteristic or stimulus. Example - Rate top 5. Ranking is vertical.

· Rating – It asks the respondent to estimate the magnitude of a characteristic or quality that an object possesses. The respondent’s position on the scale(s) where he or she would rate an object. Rating is absolute - measures magnitude of object’s characteristic. Rating is horizontal.

· Sorting – Classifying into groups based on certain characteristics.

· Choice – Only one answer is chosen out of several alternatives

…………………………………………………………………………………………………………………………………………………………….......

SCALES in detail

Semantic Differential Scale –

A series of 7 point bi-polar rating scale. Bipolar adjectives such as ‘good’ & ‘bad’ anchoring both ends (or poles) of the scale.

Example – Rating of a restaurant based on following parameters / factors –

· Value for Money (Poor-1 to Excellent-7)

· Location (Inconvenient-1 to Convenient-7 location)

· Variety of food (Narrow Variety -1 to Wide Variety - 7 of food)

Also, relative rating of the parameters / factors needs to be done.

	Restaurant 1	Restaurant 2	Restaurant 3	Total
Value for Money (50%)	5	4	4	=5*0.5+4*0.5+4*0.5 = 7
Location (30%)	5	2	3	=5
Variety of Food (20%)	6	3	3	=6

Hence, Restaurant 1 is doing best as per the respondent.

SDC works well with image kind of question / to understand perception about 2-4 competing brands.

Which restaurant has to work harder – The one who is at rank 1 (the lead brand) has to work hardest to maintain his / her lead position.

Constant Sum Scale –

Sum of all responses given by a respondent is either 10 or 100 (easier for respondent to think on the scale of 10 or 100). (Objective - How much importance a respondent gives towards a particular brand. Problem occurs when the criterion of rating is not specified upfront to the respondent).

Distribute 100 points in the following 5 subjects

Respondent	RM	FA	QT	CF	ME	Total (Always 100)
1	40	10	10	20	20	100
2	20	40	10	5	5	100
3	30	10	50	5	5	100
4	50	20	10	5	5	100
5	40	30	10	10	10	100
Total	180	110	90	45	45	500

Parameter thought over for high rating of a subject –

· Effort taken by the faculty (for the subject to sink in with the respondent)

· Liking for the subject by the respondent

· Utility of the course in work / job

Hence, does not specify the criterion to break the 100 point for a respondent. Discourage respondents to give equal rating / points to all subjects or very close points, which will not show any result in regression analysis.

Also, not more than 5 brands else it becomes too much to judge by the respondent. Also, either a 10 or a 100 scale should be used for designing this.

Graphic Rating Scale –

It is used if one wants to get a pictorial representation of something like Mood / behavior / feelings. It is usually used in restaurants to get certain kind of feedbacks in slightly relaxed manner / used with children.

Unbalanced Scale – It has more responses distributed at one end of the scale. No need of a center scale as people tends to make a neutral view if provided with an option to choose and the survey does not show a view to be considered.

Questionnaire design is the 2^nd stage i.e research design. The questionnaire should be valid (answering the objective).

Please ensure that codes have been assigned to each option while designing the questionnaire. Hence, all scales are numeric and use numbers for coding purposes later on.

All types of scales are popular as long as they are valid.

The major decisions in questionnaire design -

What should be asked – All valid questions which answer the objective of the research should be asked. All need to know questions should be asked and nice to know questions should be avoided.

Phrasing of question –

· Open ended questions – No options given.

· Closed ended question / Fixed alternative questions – Options can be many to choose from, but answer to be chosen from those alternative(s) only.

§ Others option does not make the question closed ended, instead it shows that majority of answers expected from respondents will be in fixed options, only some may chose the other option for answering. (Others should be less than 10%, else put it as close ended option)

Both are on the questionnaire and questionnaire is quantitative. Hence, both are quantitative because they can be coded, measured & analyzed (coding is done in stage 5 – Data processing and analysis). Both open ended and close ended questions are recorded, analyzed, measured and presented.

· Discussion guide – Many people eat chocolate because…..

· Questionnaire – 30% people eat chocolate because……

Avoids on the questionnaire (should not be on the instrument) –

§ Avoid complexity – Use simple conversational language. Example – “How did you interact with the customer service officer of XYZ bank”. Instead of using officer, use the word representative, staff etc as people outside the bank will get confused in understanding the rank (officer). “Of all the chocolate brands you can think of, what brands are you salient of”. (Use the word aware / think instead of salient, which is pure marketing word). “Record, response under TOM”. (Use the word first mentioned instead of TOM – Top of Mind).

§ Avoid leading and loaded question –

Ø Leading question –

§ All intelligent elite professionals eat Kellogg’s. Do you?

§ All dumbos carry as credit card. Do you?

§ All intelligent people do their MBA from SP Jain. Where would you like to do your MBA from?

Ø Loaded question (socially loaded) – Ask from a 3^rd party context and not directly. Best way to discuss is with qualitative research / discussion guide in Personal interview as it is sensitive to the respondent.

§ Is smoking cigarette injurious to health? (No one will say NO to it)

§ One should never drink and drive!

§ One must never take dowry!

§ One must never cheat in exams!

§ If one does not drink at a party, he is considered Sissy (one drinks for social acceptance) – Majority said Yes

§ Avoid ambiguity (be as specific as possible) – Please record your monthly household income (not specifying the currency; whether rounding off is required and if Yes, till what decimal place)

§ Avoid double-barreled items – 2 or more questions asked in the same question. “Do you like the food and the ambience of this restaurant”

§ Avoid making assumptions – How’s your wife (assuming that person is married)

§ Avoid burdensome questions – “Of all the liquid you drank yesterday, how many liters of water did you drink”. “Of all the money spent on the groceries in a month, how much you spend on a chocolate colored biscuits”.

Question Sequence –

Order Bias -

Funnel Technique – Ask all general / category (chewing gum) questions first and then ask brand (Rigleys) specific questions. For example – In a ice cream brand survey, ask ice-cream related questions first (taste, cost) and then ask about various brand related questions for ice-cream.

Filter Bias – Example - If consumer follows cricket, go to set 1 questionnaire, else go to Set 2 questionnaire.

Gender is not asked, but is recorded.

Pretesting of questionnaire / Piloting the questionnaire – One need to test it with actual respondents to ensure questionnaire is understood correctly; there are no errors / fatigue levels in the questionnaire. After pre-testing, respondents in the field are asked these questions. It is a trial run with a group of respondents to iron out fundamental problems in the instructions of survey design.

Example – Did pretesting of 10% for a sample size of 100, ie 10, and there is no change in the response. No what will be the next course of action

Ø Do the rest 90 interviews

Ø Take a fresh sample of 100 and again do the above exercise

Do the former (90 additional) as cost will be less and sample consisted of actual respondents.

Difference between filter question and screener question – In case answer to screener question is No, Interview is stopped / terminated, while in filter question, it is redirected to a different set of question (filter means one take the respondent and direct them to different part of questionnaire).

Question asked for a survey for particular income group – We are conducting an interview for people having monthly household salary above 10,000 Dhirams. Please tell us by looking at this card, to which salary range u fall in.

One can have many screening questions. All screener questions no matter how sentimental it is, will be asked upfront.

Ask every question from the questionnaire should be asked even the sample is a referral sample (I know something about the sample, which answers few questions of my questionnaire). Reason – Questionnaire is structured, every question should be asked in the same order / sequence as given in the questionnaire following all instructions.

For interviewing 10,000 and above; what should be done with 9500 salaried respondent. Round it off to 10,000 or give the instructions carefully.

Written permission of the respondent needs to be taken in case his / her name is disclosed to client. In B2B interviews, always get the visiting card of the client signed by him / her.

Number of questions in a questionnaire – To be decided based on the following criterions are met -

Ø Interviewer has the questionnaire and asked question to respondents and interviewer will circle the response.

Ø Interview length is not more than 20 -35 minutes for face to face and 15 minutes for telephonic interview.

Ø Questionnaire is valid.

Awareness Question –

Question - When you think of chocolates, what are the brands of chocolate you are aware / think of?

Answer –

1^st brand mentioned – Cadbury (Circle 3 in First mentioned column) = 1^st chocolate brand mentioned by the respondent

2^nd, 3^rd, 4^th ……. Brand mentioned – Galaxy, Kit-Kat, Tiffany (Circle 1, 4, 8 in Other Mentions) = Will come in “Other mentions”

Aided = Brand which when reminded to the customer, is remembered by the customer

Un-aided = Brand even when reminded to customer is not known to him. Needs to build awareness for the brand.

BRAND NAME	1st MENTIONED	OTHER MENTIONED	AIDED
Galaxy	1	1	1
Galaxy Jewels	2	2	2
Cadburys	3	3	3
Kit-Kat	4	4	4
Nestle Milky Bar	5	5	5
Lindt	6	6	6
Ferrero	7	7	7
Tiffany	8	8	8
Others (Specify)	9	9	9

Hence, all the brands listed here will be covered either in 1^st two or will be covered finally in Aided or unaided. In case, respondent mentions a name not in the list, then it should be noted and Others (Specify) should be circled in either 1^st mentioned or Other mentioned (Aided and Un-aided won’t ask Others – Specify question)

How aided questions are asked – Show the respondent the brand photo cards also to specify which brand one is referring to.

On a spontaneous note, a respondent is aware of atleast 9 brands of chocolates. When prompted, he / she is aware of another 10 brands. So, in totality, respondent is aware of around 20 brands chocolates. It indicates highly knowledgeable consumer / customer, highly competitive market, well advertised / positioned brands. Total Brands of chocolates that exist = 80 (Like Cadburys Fruit & nit, Cadburys Hazle nut are separate brands).

Establishing relationship between mother brand and its variant - Cadburys in India was generic name for chocolates like Xerox is to photo-copy or drinking water to Bisleri. Cadburys did a survey to find out how many people were able to recall Cadburys dairy milk in particular Vs how many people are recalling any Cadbury. So, under 1^st mentioned, respondent says Cadburys, then they are prompted by interviewer – which Cadbury (80% said Dairy Milk).

Alphabetize the brand names while designing the awareness question to avoid any order bias.

Companies are interested in the top 3 brands maximum & do not prompt for more than that.

Brand Awareness importance – How close is the brand to the customer, whether customer can recall on its own or needs prompting, Whether brand is recalled as 1^st or subsequent.

Usage question comes after awareness – Same style, only difference is it consumed rather than it is just being aware of.

How to ask Brand Image question –

Brand 1 Brand 2 Brand 3 ……………….. Brand 7 (None)

· Has Attractive packaging

· Is Available everywhere

· Is Reasonably prices

User can select Single / multiple / no option. Set of Image statements are referred to as Image battery. Here, image battery has 3 statements.

Why coding should be uniform in all questionnaires prepared for different purposes (like Cadburys should always get code-1 irrespective of it being present in any question / questionnaire)- Everything on the questionnaire is structured which means that when asked / analyzed, a particular option is always coded same irrespective of it being present in any question (uniformity maintained across respondents to get consistent answers asked in the questionnaire). Can’t code Cadburys as 6 by one person and 1 by other. Hence, codes should be fixed before the questions are designed.

Satisfaction Scale – 10 point scale is preferred over 5 point as data is more dispersed and hence, more sensitive, which is internationally acceptable. How high / low can one can be rated for a particular characteristic. NA option is for question which user cannot answer as he / she doesn’t know (Please refer DON’T KNOW article for further details) or has not used the product / service. It helps to remove these respondents from sample as they are not the target audience. In case a screener question is present in the questionnaire, then NA is not required to be present as an option.

Positioning cues – Current perception / Image / How is brand seen as today