A scale is a set of levels or degrees that are used to measure something. It is important to know how to properly use a scale, as it can affect the results of a study.
While many studies have incorporated expert judges into the scale development process, few have taken into account opinions from the target population. Future studies of scale development should incorporate both of these approaches.
Likert-type scales
The use of Likert scales in surveys is a common method for collecting data. However, it is important to choose the correct response scale for your survey question in order to gather accurate research data. For example, the Net Promoter Score (NPS) question, which measures customer satisfaction and loyalty, uses an 11-point scale that ranges from 0 to 10. It is also important to consider how you will analyze your data. You can use descriptive statistics to summarize your results in numeric or visual form, such as the median or mode, or you can analyze them at the interval level.
When designing a Likert scale, it is important to keep in mind that the answer options must have two extreme positions and an intermediate position that allows respondents to express their ambivalence. The use of neutral midpoints can help avoid acquiescence bias, which occurs when respondents agree with positive statements unconsciously. The neutral midpoint is also useful for avoiding false agreement or disagreement results.
Scales with multiple items
Historically, researchers have used multiple-item scales to measure constructs such as customer satisfaction and loyalty. This practice was recommended by Jum Nunnally in his classic text Psychometric Theory, which advocates the use of multi-item instruments because measurement error is averaged out when the results are summed up. However, recent research has challenged the assumption that multi-item scales are a good way to measure constructs.
In the first step, usually referred to as the item generation phase, the researcher must identify items that capture the desired construct and are free of bias. This step also involves assessing the content validity of the final set of items, which can be accomplished by obtaining opinions from experts and target population members. However, this review found that only a small minority of studies sought the opinion of the target population to assess the validity of their final scales (Uzunboylu and Ozdamli 2011).
Another limitation of multi-item scales is the risk of social desirability response bias. This can occur when the questions are posed in a manner that makes it easy for respondents to give answers that please their investigators. This can lead to invalid or biased data, which undermines the validity of the scale.
Scales with single items
Despite the fact that scales with single items are often considered psychometrically suspect, some have demonstrated acceptable reliability (Cronbach’s alpha) in comparison to multi-item measures. Moreover, they are often easier to understand for respondents. This may be a key reason why single item measures have gained popularity in marketing research.
The first step in developing a scale is the development of an initial pool of items. This can be done using both deductive and inductive approaches. It is important to develop a preliminary pool that is at least twice as large as the final scale. This will allow you to eliminate items that are tangential or unrelated to your target construct.
This step also includes evaluation of content validity. Ideally, this should include both expert judges and target population judges. However, this can be difficult to do when resources are limited. A possible alternative is to evaluate the scale on a cross-sectional sample. This will give you a more general sense of how well the scale performs in a real-world setting.
Scales with scale factors
Scale factors are similar to ratios and proportions, but are used to adjust the size of a model or drawing. A scale factor is a multiplier that enlarges or reduces an image while preserving its proportions. For example, a garden’s length of 20 meters can be reduced to 4 meters in a brochure while maintaining the original proportions. Scales with scale factors are important for pre-algebra and geometry students, as they play an essential role in comparing maps, models, and drawings.
To ensure that a scale is valid, it should be tested on multiple samples. The ideal procedure involves a cross-sectional or longitudinal sample. For both types of samples, the sample size should be large enough to allow for statistically significant results and to minimize variance among items. It should also be large enough to permit the use of item reduction procedures and to assess content validity (115). The most common methods of assessing content validity include expert and target population judges.