7. How to Measure Attitudes, Behaviour and Traits

(Textbook Page 255)

This textbook unit provides an overview of the scales that are most frequently used in marketing research and principles in asking different categories of questions. 

How to Measure Attitudes (P. 256)

• Attitudes exist in consumers’ minds and cannot be directly observed. We depend on consumers to tell us what they believe.  We generally need to know if a consumer holds a given attitude (existence), and if so, how strongly (intensity). 
• Measurement of attitudes should be valid (measure what we intended to measure), reliable (give consistent measurements across similar studies) and sensitive (can detect the shade or intensity of meaning). 

Here is a list of the types of scales used to measure attitudes: 

• Nominal scaling is ‘naming’, telling us only which response go the highest number of votes. 
• Dichotomous scales record the existence of an attitude (yes/no).
• Ordinal or Ranking scales give us the order of preference but not the intensity.
• Interval or Rating scales can provide intensity by capturing the distance between scale points. 
• Semantic scales use words instead of numbers to measure the strength of an attitude.
• Likert-type scales generally use 5-points to measure the strength of attitudes (e.g. as agreement/disagreement with a set of statements)
• Semantic Differential scales name only the end-points and useful when we have 7 or more points (can’t name them all).  If anchored by different words on each end they are called ‘bipolar’. 
• Graphic rating scales use pictures as scale points (e.g. hapy to sad faces) 
• Constant sum method asks respondents to divide a given number of points between all alternatives, forcing them to think about their preferences more carefully.   Chip allocation is similar, offering 11 chips to be allocated so there will never be a tie.

Key issues discussed in this textbook unit include: 

• Keeping rating scales consistent over time in trackers
• The relative merits of a 5, 7, 10 or 11 point scale
• Data collected on one scale cannot easily be mapped onto another scale
• Small screens on mobile devices mean sliders are often used instead of scales
• Which way scales should run – ascending or descending 
• How the meaning of scales might differ from one person to the next

How to Assess Behaviour (P.278)

The reliability of behavioural measurement is central to marketing research.  What are consumers doing?  Behavioural measures can be asked as: 

• A rating question (e.g. very frequently, somewhat frequently, somewhat infrequently, very infrequently).  Consider that “frequent” could mean different things to different respondents, making it more of an attitudinal measure – how they think of their behaviour.
• A latent variable question (e.g. on average, how often in a typical month) – assumes a ‘characteristic’ rate, but different respondents may calculate the ‘average’ differently.  Still more of an attitudinal question, but easier to quantify.
• A manifest variable question (e.g. how many times last month) – a true behavioural question based on a specific time frame, but can be impacted by seasonality or special occasions. 

Lifestyle Measurement (P. 281)

We often need to dig deeper into motivations than we can by asking an attitudinal question followed by a ‘why’.

• Psychographics refers to a combination of self-reported behaviour, personality traits and attitudes. 
• Grouping together consumers with similar attitudes on several measures, using a technique known as cluster analysis is called market segmentation.  
• The questions and attitude statements should pertain specifically to the product or service category being investigated

Demographic Traits (P.270)

Personal questions not needed to screen respondents are generally asked at the end.   A few principles for sensitive questions (e.g. income):

• using categories rather than asking for an exact answer 
• provide very low and very high options so most don’t end up in the lowest or the highest 
• ask for household rather than personal income (higher and better reflects spend power)
• using as few categories as possible 
• allow respondents to refuse to answer