The purpose of this study was to investigate the concurrent validity of the MPSA by measuring COP variables and comparing these with a validated reference standard. The ICCs of the postural control measures were equal to 0.60 for the average velocity and 0.14 for the sway area between the force plates. Results of the reliability study of the MPSA also reported higher agreement for the average velocity and a lower agreement for sway area . The ICC is a ratio estimate and there is no widely agreed upon thresholds for identifying an acceptable level of agreement with ICC reporting. As a result, there are various interpretations of ICC values available in the literature. Landis and Koch  suggested the following qualitative approach to ICC interpretation: 0.00-0.20 slight, 0.21-0.40 fair, 0.41-0.60 moderate, 0.61-0.80 substantial and 0.81-1.00 almost perfect. Alternatively, Portney and Watkins  suggest that values over 0.75 indicate good agreement, and values below 0.75 are indicative of poor to moderate agreement. The highest level of agreement identified in the current study was ICC = 0.60. Therefore, these results represent evidence that the MPSA does not possess sufficient concurrent validity.
The high standard error of measurement values of both variables especially the sway area, when compared to the mean values obtained by the two force plates, indicates a lack of consistency of measurement between the force plates. Furthermore, our results demonstrated significant differences between the COP measurements obtained from the force plates for the COP average velocity and sway area.
Comparison of standard deviations to means of sway area captured by the two force plates shows that the large standard deviation of MPSA sway area indicates a high degree of random error in the MPSA data. Additionally, the large mean difference between MPSA and Accugait sway area might be a sign of a systematic error in the MPSA when assessing postural control.
Visual inspection of the Bland-Altman plot (Figure 2) of average velocity shows a funnel effect, meaning there is more variability when the magnitude of the average velocity is greater. Our results represent excessive variation between the force plates in higher magnitudes of average velocity and less variation in lower magnitudes of average velocity. Differences between the measures of COP velocity between the force plates depend on the magnitude of the measurement. The excessive variations in higher magnitudes of average velocity indicate that data captured by the MPSA contain errors for measuring higher COP velocities.
The bias estimates for COP velocity were small but statistically different from zero. Statistically significant bias was found for the average velocity variable where the MPSA values underestimate the criterion in most of the cases with an estimate of 1.7 and a 95% confidence interval of 1.1 - 2.2. One possible explanation for this is that data captured by the Accugait was filtered with a 5 Hz cut off frequency whereas data obtained by the MPSA was filtered with a fixed 0.5 Hz cut off frequency. This may have resulted in the MPSA eliminating true signal. The lower cut off frequency causes removing higher proportion of true signals.
The plot of the sway area (Figure 3) shows a systematic trend. With lower magnitudes of sway area the differences between the two force plates are relatively low as compared to higher magnitudes, but the greater the sway area the bigger the difference between the two force plate measures. Moreover, the statistically significant bias of sway area showed that the MPSA overestimated sway area. Overall, larger measures of average velocity and sway area result in increased systematic and random error in MPSA.
While the MPSA previously demonstrated acceptable reliability , its validity is not satisfactory; therefore, it cannot be considered a replacement of a known valid force plate for the assessment of postural control. In addition, results obtained by the MPSA on clinical populations should be treated with caution. Clinical populations are potentially less stable and show higher magnitudes of average velocity and sway area [17, 18] and the MPSA was incapable of measuring accurate data with higher magnitudes of average velocities and sway areas. Additionally, to reduce measurement error, we used a mean of 5 trials, which is unlikely to be the case with clinical use given the time constraints of clinical practice. Clinicians should consider all the above mentioned issues when assessing postural control using the MPSA.
Although we cannot rely on the MPSA for assessment of postural control, it may possibly be used in obtaining qualitative estimates of postural control, for instance as a biofeedback training tool and to enhance motivation level of patients with balance defects. Having said that, we suggest clinicians intending to use a force plate purchase a reliable and valid instrument.
Limitations exist within this study. The study sample consisted of healthy individuals and the findings may or may not generalize to clinical populations. This study did not attempt to assess the validity of the MPSA in different testing positions such as eyes closed, single limb standing and narrow stance (feet together) although we hypothesis that this would not enhance the MPSA performance.
Future research could be done to assess the validity of the MPSA in different balance testing positions and clinical populations. Estimates of reliability and validity should be known prior to using any type of force plates.
In conclusion, postural control parameters cannot be validly measured in clinical or research settings using the MPSA. The MPSA is a lower cost force plate with a low-technology design and easy to use software but it did not fulfil the criteria to be regarded as a valid force plate for clinical use.