graphic rating scale

A type of rating scale which presents a continuum in the form of a straight line, along which a theoretically infinite number of ratings are possible. The implicit rationale for using graphic rating scales is that they make it easier to detect fine shades of differences in attitudes, to quantify responses to a possible question, one measures the physical distance between the left extreme position and the response position on the line; the greater the distance, the more favorable is the attitude. Even less refined rating scales, however, are typically labeled "graphic" rating scales as long as they contain a straight line signifying an underlying continuum.
 
Several graphic rating scales are illustrated in the figure below. In each case, graphics have been used to help define stems and anchors representing the characteristic to be assessed.
 
The first scale in the figure uses three conventions for anchors:
 
(1) a straight horizontal line marked off in equal appearing intervals;
 
(2) a series of numbers going from 1 to 5, again reinforcing the notion of order and equal intervals; and
 
(3) a set of adjectives describing various levels of the characteristic being measured.
 



 
The scale in the first case is a Likert scale.
 



 
The respondent is asked to indicate the degree of agreement or disagreement with the statement by indicating with a mark somewhere along the 5-point scale (3, 7, and 9-point scales are also used). Level of satisfaction is inferred from the respondent's degree of agreement or disagreement about the adequacy of the pay on his or her job. A 1 would indicate the most extreme disagreement with the statement indicating, presumably, extreme dissatisfaction. A 5, in contrast, would indicate the most extreme agreement with the statement, indicating, presumably, the highest level of satisfaction. Sometimes the numbers are omitted as anchors and are simply used afterwards to score a person's response.
 
The second scale replaces the graphic line with a horizontally arranged set of boxes. The advantage of this technique is that it forces the person to choose one of five numbers rather than an almost infinite variety of gradations along the scale in the first example.
 
The third scale is a semantic differential scale, which sets up polar opposite descriptions or adjectives defining opposite ends of some continuum. The respondent uses the equal intervals along the scale between these two extremes to place the person or thing being evaluated.
 
A frequent question in scale construction concerns the number of anchors (points along the scales) to give the respondent. Conceivably, one could provide any number between 2 and ∞. The specific number of scale points to provide is a judgment that requires the scale designer to consider two tradeoffs:
 
(1) If too few points are provided, the respondent will be unable to make the distinctions among objects being assessed that he or she might want to.
 
(2) If too many points are provided, the respondent will become confused and errors of judgment will occur. In point of fact, the vast majority of rating scales in use employ between 5 and 9 points. Fewer than 5 probably does not provide sufficient room for making distinctions, and more than 9 will probably confuse many respondents. The incremental information gained from each additional scale category above 10 or 11 is minimal. An important aspect in selecting the number of rating scale categories is deciding whether the number should be odd or even. Like the broader number decision, the odd versus even decision hinges on certain considerations. If a neutral or indifferent scale response is a possibility, an odd number of categories is required. On the other hand, if the researcher wants to force a scale response or believes no neutral or indifferent response exists, the scale should contain an even number of categories.
 
A potential difficulty in using graphic rating scales is that coding and data analysis will require a substantial amount of time since one has to first measure physical distances on the scale for each respondent. An even more serious drawback is that respondents may be incapable of even mentally perceiving fine shades of differences in attitudes, let alone accurately translating their perceptions into measurable physical distances.
 
In other words, while graphic rating scales are capable of facilitating precise attitude measurements in theory, whether they can be meaningfully used in practice is arguable. In contrast, itemized rating scales have a set of distinct response categories; any suggestion of an attitude continuum underlying the categories is implicit. They essentially take the form of the multiple-category questions. While less refined than a graphic rating scales, itemized rating scales should be easier to respond to ? and more meaningful from the respondent's perspective, with coding and analysis of the raw data being somewhat less laborious.