Survey design

Response codes

All questions use a 7-point Likert-type scale. A Likert scale is one that allows people to give varying degrees of positive or negative response to a question or statement. A 7-point scale is generally considered to be best practice; though some surveys use 5 point scales right through to 11 point scales.

In our experience using more response options can cause some confusion amongst respondents as they ponder whether this particular item deserves a six or a seven. In contrast a scale shorter than 7 does not allow for enough differentiation. The choice of 7 response options is considered to be a good balance between these two tensions. There is also debate amongst academics as to whether it is better to have an even or odd number of response codes options. Some researchers like to be able to categorise responses as positive or negative and so don’t want a ‘middle’ code option that an odd number inevitably produces. Other researchers believe that it is best practice to give respondents the opportunity to indicate a middle position, because this may best reflect how someone actually feels. This is the argument that persuades us to use a scale with an odd number.

Response anchors

All the questions use ‘anchors’ at each end of the 7 point Likert-type scale. In this survey we use lots of different anchors. For the question, “In general would you say that you feel happy when you are at work”, we used “not at all happy” at one end of the scale and “extremely happy” at the other.

Anchoring the extreme ends of the scale rather than naming each specific number on the scale is considered to be best practice, especially when doing international studies.  For practical considerations, the language required to name each response option on a scale can become clumsy: extremely happy, very happy, somewhat happy etc. These problems become exaggerated when translating into different languages as words like ‘somewhat’ have different ‘strengths’ or meanings.

The use of anchors also allows more flexibility as the negative anchor can be worded as either a strong negative “extremely dissatisfied” or an absence “not at all happy”. We predominantly use the ‘absence’ anchors as they tend to stretch responses across the codes options more. However, for some specific questions which already have extensive academic usage with strong negative anchors, such as the question we use about job satisfaction, we have chosen to stick with the norm.

It is also worth noting that we have deliberately stayed away from agree-disagree statements. The reason for this is that research has suggested that agree-disagree scales suffer from what is called ‘acquiescence bias’ with some respondents not liking to ‘disagree’ with any statement. Having more item-specific anchors – that vary from question to question – also increases the reliability of the data gathered.