Graphic Rating Scale

A graphic scale is one presented which is used to respondents visually so that they can select a position on it that best represents their desired response. Basically it is a continuous bi-polar scale with fixed points verbally anchored at either end. It can simply be a line between the two anchor points and is is a form of graphic scale which has become known as a 'visual analog scale'.

The distance from the end points of the respondent's marks is measured to provide the score for each attitudinal dimension. This is a continuously rated semantic differential scale, which provides a greater degree of precision and avoids the issues of numbers of points on the scale. It is a simple way of measuring attitudes and image perceptions, but is impractical to use with paper questionnaires. Measuring the position marked on hundreds of paper questionnaires, with possibly dozens of scales on each one is not viable for most commercial projects. This technique cannot be used with telephone interviewing.

With CAPI interviewing to a greater degree with online web-based interviewing, the continuous graphic scale is a realistic option. Respondents can drag a cursor along the line to the exact position that they want it, and that position is then automatically recorded.

When the technique is being used to measure attitudes to brands or products, more than one cursor can be used to represent different brands, or brand logos can be used in place of cursors. Then each respondent can place a number of brands along the scale, so that they are positioned relative to each other as well as to the scale ends, according to the respondent's perceptions. This is quicker for respondents than rating each brand individually, is more interesting for them when logos are used, and provides better relative measures of the attitude variation between brands.

Though data collected are continuous, the measurements will be assigned to categories and treated as interval data for analysis purposes. It is possible to have a large number of very small intervals, but the researcher must decide at what level the apparent accuracy of the data becomes spurious depends on the length of the line used, the accuracy with which respondents are able to place the cursor, and the degree of accuracy to which respondents are likely to have tried to place the cursor.

The questionnaire writer may wish to apply labels to the scale. The scale can be labelled numerically, so that one end is 0 and the other 100. The position of the cursor can then be indicated as a number between 0 and 100, which allows the respondent to place the cursor accurately.

In web-based online surveys respondents found visual analog scales (VAS) as easy to complete as rating scales using fixed points denoted by radio buttons, and that they felt that VAS scales conveyed their responses with sufficient accuracy better than with a numeric box entry. Responses obtained from VAS and the fixed-point radio buttons were similar, and the respondents found the VAS approach more interesting than the radio buttons. As maintaining the interest and involvement of respondents is one of the objectives of the questionnaire writer, the use of VAS or graphic rating scales should always be considered as an alternative to radio buttons.