Yes, that is a 25 cent title for an equally difficult task to undertake. Breaking apart the core differences and assumptions in using both of these methods in your survey questionnaires is a difficult but important task. It’s necessary to ask the right questions before implementing thousands of dollars worth of change in a company based on survey results.
When we are looking at these two models, we should first have a quick review of their features. The Likert Model, developed by Rensis Likert in 1932, attempts to put a quantitative value to an emotional feeling the respondent may have on an statement. The scale, typically in 5 or 7 intervals, allows the respondent to specify a neutrality well as extremes of agreement to particular statement. The neutrality is central to the benefit of this model. Here is a sample of a Likert quesiton.
On the other side is the semantic differential, developed by Charles Osgood. The concept for this type of question came from the deeper understanding of how well we physiologically understand and express our feelings. The basis is the use of adjective pairs, which we fully understand apart from our experiences, applied to those experiences or feelings. The strength of this approach is that we do not have to be an expert interpreting our feelings, but giving objective feedback based on additives we know connecting the emotional to existing understanding. Here is a sample of this type of question.
Likert’s highs and lows
The first strength of the Likert is the qualitative measurement. Since the mean has a neutral value, the overall measurement of these scores can present as a positive or negative number. In addition, this resulting number can be easily reviewed for its extreme and significance. Knowing a 1.0 is the extreme result and a .5 is a median, scores that present above the median can show a significance towards being a strong success. The quantitative result also allows you to compare the result to similar questions easily since the numeric scale is easily compared.
The down side of the Likert is the inability to quantify anything rather than agreement to a question. We are asking the respondent to respond to our question, rather than express their feeling or emotion. The phrasing of the question becomes incredibly important to make sure the right question is asked to gain consensus. If you enter into leading questions, you might not get the true feedback you are looking for. But this concern is generally negated by good survey practitioners.
What we love about the semantic scale is the absence of any prejudgement about the topic in question. In contrast to the Likert, you do not have to take a position on the question to gain an agreement level, you simply ask the respondent for their response without any pre-judgement compared to two different value extremes. In addition you can gain the same emphasis on extremes that you can from a Likert type question by using various values between the two choices. The issue we have with the semantic is partially its strength in that you can not ask directly about a specific metric point. The format does not allow you to find out if “customer service answers my calls in less than five minutes”, it can only answer similarly “my customer service calls are.. good/bad, fast/slow,etc.
Ask the right question
In evaluating these two approaches, it comes down to what you are trying to ask. The value of both question methods can bring some major insight into your survey methods. Having a real understanding of the two methods can allow you to best position your survey questions to gain increasing insights into your customer behaviors, feelings, and affinity for your brand. Typically for us, we’ll start with the semantic model to see if we can get enough information and if not, we go to Likert model. It’s very easy to setup Likert questions to solidify a hypothesis, so that is why we don’t start there. From the concerns raised from the semantic question evaluation, it can help us better frame the Likert-type question. In either case, we want the questions to be presented in a way that allows the data to surprise us. That is the real value of surveys.