Validity can be defined as the degree to which a test or test item measures what it is intended to measure and is the most important characteristic of any specific test (Baechle and Earle 2008). Without validity, test results have no meaning. Measures of basic physical characteristics of an individual (e.g., height and weight) are relatively easy to validate. The validity of metrics to be used during exercise performance, especially perceived exertion scales, can be more difficult to establish. Therefore, there are two main types of validation experiments used to confirm the validity of a perceived exertion scale: concurrent validity and construct validity.

Concurrent validity is the extent to which test scores are associated with those of other accepted tests when both measures are obtained along a common stimulus range. In the case of a scale that measures RPE, this definition refers to the accuracy of the metric to measure perceived exertion across one’s entire physiological range as exercise intensity is systematically increased from low to high levels. A test of concurrent validity involves a statistical calculation of the relation between a criterion variable (the stimulus) and the concurrent variable (the response). This statistical paradigm often uses a Pearson correlation calculation that yields “r” values referred to as validity coefficients (Baechle and Earle 2008). To establish concurrent validity of a perceived exertion scale, it is expected that the concurrent variable, RPE, increases concurrently with increases in a physical and/or physiological criterion variable as intensity of exercise increases (Robertson 2004). A statistically significant concurrent validity coefficient indicates a strong relation between the concurrent and criterion variables, often resulting in high r values greater than 0.70.

The theoretical framework underlying the assessment of concurrent validity of a perceived exertion scale is derived from the basic tenets of Borg’s Effort Continua and Range Models. There are three main effort continua: performance, physiological, and perceptual. An increase in exercise performance, usually denoted as increasing intensity, results in corresponding and interdependent increases in both physiological and perceptual responses. Exercise intensity can be measured as minute per mile pace or PO for aerobic exercise and as absolute weight lifted or %1RM for resistance exercise. Physiological responses are underlying processes that an individual subjectively monitors during exercise to ultimately mediate their RPE response. HR and VO₂ are respiratory-metabolic exertional mediators that are commonly measured during exercise serving as physiological indicators of exercise intensity. Physiological and perceptual responses display a functional interdependence. As such, the model predicts that perceptual responses will increase in correspondence with physiological responses throughout the individual’s entire exercise intensity range, from a very low to a maximal level. In addition, the lowest RPE value matches the lowest exercise intensity and the highest RPE value matches maximal exercise intensity. This holds true whether exercise intensity is expressed in physical units, such as PO, or using a physiological variable such as HR or VO₂. In this context, it is the goal of scale anchoring procedures to set the low and high anchor points on an RPE scale, linking them to very low and maximal exercise intensities. Once it is known that an individual conforms to the model following scale anchoring, a concurrent validation experiment can be used to measure the physiological and perceptual effort continua across the full range of possible exercise performance intensities.

Construct validity can be defined as the ability of a test to represent the underlying construct, or the theory developed to organize and explain aspects of existing knowledge and observations (Baechle and Earle 2008). For perceived exertion scales, construct validity is tested by comparing RPE measured using a newly developed scale for which validity has yet to be established with the RPE derived from a perceived exertion scale having well-established construct validity. In this paradigm, it is expected that both the new (i.e., conditional) scale and the criterion scale have demonstrated a high level of concurrent validity. Traditionally, construct validity of a perceived exertion scale is statistically determined by correlating RPE measured using the conditional metric with RPE measured using the 15-category Borg Scale (i.e., the criterion metric).

Concurrent and construct validity of a perceived exertion scale can be tested simultaneously using perceptual estimation test protocols. An estimation protocol is a GXT during which an individual estimates RPE during each exercise stage. Using commonly employed procedures for determining maximal aerobic or resistance exercise capacity, an estimation protocol allows an individual to rate RPE from a very low exercise intensity to maximal exercise intensity. For example, the Bruce treadmill protocol for the determination of VO₂max employs incremental stages of walking and running exercise. For resistance exercise, variations on 1RM or multiple-RM procedures are used. These procedures must include measurements of physiological exertional mediators (e.g., HR, VO₂) and the recording of physical markers of exercise intensity (e.g., PO, weight lifted, %1RM) necessary for the determination of concurrent validity. As such, a concurrent and construct validation experiment only requires that RPE be measured using the 15-category Borg Scale and the RPE scale for which validity is sought.

It is important to note that the scale anchoring procedures should be presented separately from and prior to the estimation test protocol used for a scale validation experiment. For the concurrent/construct validation experiment to be valid, it must be known that the individual’s RPE responses conform to Borg’s Range Model. Individuals who have experience using perceived exertion scales and have participated in exercise anchoring procedures in the past are more likely to provide RPE responses that conform to the prediction of Borg’s Range Model. However, as discussed previously, some individuals are perceptual outliers, either overestimating or underestimating the RPE response. Such responses usually occur upon initial exposure to RPE assessment and prior to administration of full memory and exercise anchoring procedures. When possible, youth and sedentary adult subjects should undergo both the memory and exercise procedures for perceived exertion scale anchoring. However, even active adults have been known to be perceptual outliers and as such can benefit from additional anchoring and practice in estimating RPE prior to undertaking the actual exercise trial.

Experiments to validate perceived exertion scales can employ both undifferentiated and differentiated RPE. A differentiated RPE is linked to a specific anatomical region of the body. Differentiated RPE specific to the leg muscles (RPE-L) can be measured during cycle ergometry and treadmill exercise. The RPE-L reflects peripheral exertional signals resulting from localized metabolic acidosis, blood glucose level, muscle glycogen content, and muscle blood flow (Robertson 2004). Differentiated RPE specific to the chest and breathing (RPE-C) can be measured during any aerobic activity. The RPE-C reflects respiratory-metabolic exertional mediators such as V _E and total body VO₂. Differentiated RPE rated during resistance exercises are usually specific to the active muscle mass (RPE-AM). Undifferentiated RPE is a measure of the overall body (RPE-O) exertional level. It is formed by integrating the various exertional signals arising from the composite of anatomical regions involved in the exercise task. Many investigations have asked subjects to rate RPE-O only, but important information can be derived by also measuring differentiated exertional ratings.

Numerous investigations have established concurrent validity of various iterations of the OMNI Perceived Exertion Scale using mode-specific estimation protocols. Experiments have included male and female children and adults performing a wide variety of exercise modalities: cycle ergometry, treadmill walking and running, stepping exercise, elliptical ergometry, and resistance exercise. High validity coefficients were reported for male and female children and adults during cycle ergometry and treadmill exercise, with r values ranging from 0.67 to 0.99 for the associations between RPE and HR or VO₂ (Balasekaran et al. 2012; Robertson 2004; Robertson et al. 2000; Utter et al. 2004). High validity coefficients were found in a sample of male and female children and a sample of adult females performing load-incremented stepping exercise. In these stepping experiments, the relation between RPE and VO₂ exhibited r values ranging from 0.87 to 0.96. The relation between RPE and HR exhibited r values ranging from 0.81 to 0.95 (Krause et al. 2012; Robertson et al. 2005b). High validity coefficients were found in adult males and females performing elliptical ergometry. The relation between RPE and VO₂ exhibited r values ranging from 0.93 to 0.95, while the relation between RPE and HR exhibited r values ranging from 0.95 to 0.97 (Mays et al. 2010). High validity coefficients were reported during biceps curl and knee extension exercises, with r values ranging from 0.72 to 0.91 for the association between RPE and total weight lifted in both children and adults (Robertson et al. 2003, 2005a). In addition, an r = 0.87 was determined for the association between RPE-AM and blood lactic acid concentration in adults; providing evidence for lactacidemia as a physiological exertional mediator for resistance exercise (Robertson et al. 2003).

Concurrent validation has been tested and confirmed for other perceived exertion scales as well, such as the Children’s Effort Rating Table (CERT). CERT, a 10-category scale ranging from 1 to 10, was developed specifically for children to be easily understood with verbal descriptors positioned at each numerical category. CERT, however, does not include pictorial descriptors as does the OMNI Scale. Concurrent validation of CERT has been examined for various youth populations performing stepping and cycle ergometer exercise. Validity coefficients for the relation between RPE and HR ranged from r = 0.73 to 0.99 during stepping exercise (Williams et al. 1994) and from r = 0.70 to 0.97 for cycle ergometry (Eston et al. 1994; Lamb 1995; Leung et al. 2002). In addition, investigations determined the relations between RPE measured by the CERT and both PO and VO₂ for cycle ergometer exercise. The relation between RPE and power output exhibited r values ranging from 0.70 to 0.98 (Eston et al. 1994; Lamb 1995; Leung et al. 2002). The relation between RPE and VO₂ exhibited r values ranging from 0.85 to 0.91 (Leung et al. 2002).

HR and VO₂ are the most commonly used physiological criterion variables to demonstrate concurrent scale validity for aerobic exercise modalities. They are the most widely used because they increase as a positive function of increases in exercise intensity. However, other physiological criterion variables have been used to study concurrent scale validity during aerobic exercise. Investigations have correlated OMNI Scale RPE with %VO₂max, %HRmax, pulmonary ventilation (V _E), respiratory rate (RR), the respiratory exchange ratio (RER), and the V _E to VO₂ ratio (V _E · VO₂ ⁻¹). All of these physiological variables are expected to increase concurrently with increases in aerobic exercise intensity, demonstrating either linear or polynomial growth functions. High validity coefficients for OMNI Scale responses were found for adolescent girls performing treadmill exercise, as evidenced by the relation between RPE and %HRmax (r = 0.86) and the relation between RPE and %VO₂max (r = 0.89) (Pfeiffer et al. 2002). Moderate validity coefficients ranging from r = 0.33 to 0.43 were shown between RPE with %VO₂max, V _E, RR, and V _E · VO₂ ⁻¹ for children performing treadmill exercise (Utter et al. 2002). Another study involving children performing treadmill exercise found high validity coefficients ranging from r = 0.71 to 0.81 for the relation between RPE with %VO₂max, V _E, RR, and RER using the Spanish version of the Children’s OMNI-Walk/Run RPE Scale (Suminski et al. 2008). In addition, high validity coefficients ranging from r = 0.67 to 0.88 were found for the relation between RPE with %VO₂max, V _E, RR, and RER where data were determined for adults performing treadmill exercise (Utter et al. 2004) (Tables 6.1 and 6.2).