A Comparison between Two Rating Scales for Perceived Exertion

Thirtytwo subjects,16 men and 16 women, participated in an experiment to compare two commonly used rating scales for perception of exertion, viz. the Borg RPE scale and the Borg CR10 scale (see, e.g., Borg, 1998). One group of 8 men and 8 women used each scale. Workloads were increased every minute, with 15 W for men, and with 10 W for women, to a voluntary maximum. With a basic perceptual noise constant in the power function, group exponents of n=0.67 for men and n=0.93 for women, with the RPE scale, and of n=1.2 for men and n=0.85 for women, with the CR10 scale, were obtained. The mean exponent of n=0.76 for the RPE scale, was significantly less than 1.0, and the mean exponent of n=1.05 for the CR10 scale, was significantly less than 1.5. If this is an underestimation of the exponent, a work-test with one minutes increase of workloads may be risky, e.g., for cardiac patients, and lacking ecological validity. Estimations of individual working capacity from the two scales gave significant correlations with a heart rate based measure, W170, of .778 (RPE) and .587 (CR10). A transformation equation for the two scales was also determined, RPE = 6 + 2.8(CR10 0.3)0.79, supporting the theoretical relationship. Fundamental determinations in psychophysics concern descriptions of the relation between physical stimulus (S) and its perceived magnitude (R). Several techniques have been developed to accomplish this with the highest possible precision and the best possible control, of, e.g., influences of instruction, experimental design, rating behavior, context effects, etc. (See, e.g, Gescheider, 1997). Mathematically, obtained psychophysical functions are typically described by power functions. In the simple power function (preferred by Stevens), R = cSn (Eq. 1), c is a measure constant, and n is the exponent. Ekmans work (1959) and later also the work of Borg (1962) suggested the need of additional constants in the equation. A still more general formulation of the S-R-functions was proposed by Borg (1961, 1962), and also by Mountcastle, Poggio and Werner (1963): R = a + c(S-b)n (Eq. 2), where the constants a and b can describe the absolute threshold or the starting point of the function. Stevens ratio scaling methods worked well for describing relative growth functions (see, e.g., Stevens, 1975), but allowed no natural level identification, or interindividual comparisons. On the other hand, this is something that language gives us in every day communication. Wellknown intensity expressions, such as adjectives and adverbs, have fairly precise intersubjective agreement. This is probably also the reason why ordinal rating scales with 5-7 verbal categories have remained popular despite their limits for statistical treatment of data. Borg’s range model (Borg, 1961, 1962, 1998), with the assumption that the subjective range from a minimal to a maximal intensity perception may be set equal for all persons, provided the theory needed for the development of a series of verbally level-anchored scales, the most well-known being the RPE scale (for Ratings of Perceived Exertion, Borg, 1970, 1998) and the CR10 scale (a general intensity Category scale with Ratio properties with numbers from 0 to 10, Borg, 1982, slightly modified, Borg, 1998), see Figure 1. The RPE scale was constructed to give data that grow linearly with Stimulus intensity, heart rates (HR) and oxygen consumption for steady state aerobic work on bicycle ergometer (4 to 6 minutes). Because of its construction it can be said to give interval data (with regard to aerobic demands). To underscore this, the scale starts with the number 6 (and not zero), and the number range from 6 to 20 roughly corresponds to a HR range from 60 to 200 beats x minute–1 in healthy people, about 30 40 years old. The linear growth function of RPE data has been confirmed in several studies (see, e.g., Borg, 1998, Noble and Robertsson, 1996) On the CR10 scale the verbal anchors and numbers are placed congruently to render ratio data that mimic what is obtained by magnitude estimation. (For a thorough description of the scales, see Borg, 1998; Borg and Borg, 2001 (in press)).