For psychologists, attitude testing is our bread and butter. Psychological measurement, evaluation and development are the air we breathe. No psychologist can be called a psychologist without proper training in measurement and assessment. Similarly, in advertising, consumer research is becoming ever more important as insights-based data begin to direct marketing and communications planning. Hence, consumer research lies within the domain of psychological testing.
My experience at a recent agency showed me the “dark side” of market research. I was utterly appalled when I saw the items in the survey. Here are 3 major mistakes that every researcher should avoid.
Assuming attitude and personality as the same construct.
At first glance, I saw that attitude related items are lumped together with personality related items. Why is this wrong? Attitude and personality measure two different things. Attitudes refer to a person’s preference towards objects, things or people. Attitude can be measured objectively via the ABC model which refers to a person’s thinking, feelings and behaviour towards a particular object. On the other hand, personality refers to psychological traits that influence a person’s thoughts, feelings and behaviours. In short, attitudes are preferences while personality refers to traits of a person.
Here are some examples to help illustrate the difference in items that measure attitude and personality respectively.
I love my children, measures feeling
I believe that law should be implemented, measures thinking, belief or perception.
I buy brands that give back to society, measures purchasing behaviour.
I tend to worry a lot, is to measure emotional stability.
I am generally trusting others, measures agreeableness.
I am curious about different things, measures openness to experience.
Assuming that attitude and personality are the same constructs leads to a distorted understanding of the consumers. To avoid this error, be sure to operationally define each construct. The safest way is to use another researcher’s definition of Costa and McCrae when defining personality and using the ABC model when defining attitude.
Using only positively worded items.
The second mistake that I caught in the survey was that ALL items are positively worded. As opposed to negatively worded items, positive items are those without the words “do not”, “am not”, “dislike” etc. For instance, a positively worded item is “I find advertisements entertaining” while a negatively worded item is “I do not find advertisements entertaining”.
The danger with only using positive items is obvious. It leads to response bias. People have this natural instinct to impress people, ironically even when taking a test! Test takers usually have a habit of going autopilot and agreeing to most of the statements leaving you with high 4s and 5s. A quick analysis of the items shows that most participants (90% of the time) strongly agree/agree with the statements. Mixing the survey up with positively and negatively worded items keep the answers reliable, test takers are less likely to breeze through the assessment by agreeing with all the statements.
Trusting only face validity.
Here is a situation: Hassan is looking for a wife. He assesses two potential wives using a photograph given by their families. He then decides merely on the photograph that he wants to marry potential wife A instead of potential wife B. Hassan has made a mistake. His only criteria to choose a wife is attractiveness and those who are already married will know that having only one indicator is a recipe for disaster.
The same principle applies to consumer research. An item that looks like it fits the bill does not mean that it is valid and reliable. For example, consider this item to be measuring one’s attitude towards religion: “Religion is the most important thing in my life”. What is wrong with this statement? First of all, does it measure thinking, feeling or behavioural aspect of attitude? Secondly, does it merely measure one’s perception or actual state of life? Make all the assumptions you want. But, the only way to be certain is to use statistical tests such as factor analysis and Cronbach alpha to make decisions with regards to the validity and reliability of items.
Here’s a pro tip: Construct your items based on a theoretical framework and run a pilot test. A pilot test is, sort of, like a test run where a small number of participants answer the questions you construct. Then, a researcher analyzes the answers and arrives at conclusions regarding what the items measure, what items need to revised or eliminated and so on. Once you attain a desirable state of accuracy, then you can be more confident in pitching the data and ideas to clients.
Testing is becoming more and more important in the advertising world and what better way than to start using consultants and experts in psychology to accurately measure consumers and buyers.
Which mistake did you unintentionally or intentionally commit? Try applying these simple solutions to achieve better results for you and your client.