In order for any tool to be effective

In order for any tool to be effective, it must actually measure what it intends to accurately and reliably. One question many researchers have with quantitative EEG is how reliable the measurements are as there are many factors that can alter the data, such as regions of the brain being poorly connected with the device (Budzynski et al., 2009). When working with quantitative EEG it is important to investigate and understand its reliability to know if it is actually measuring what it intends to consistently.
One type of reliability that has been studied extensively with quantitative EEG is test-retest reliability. In one study, researchers were able to demonstrate high test-retest reliability during two different types of balance tests, specifically finding that the alpha band measurements had excellent reliability with ICC scores of 0.88 to 0.98 (Collado-Mateo, Adsuar, Olivares, Cano-Plasencia, & Gusi, 2015). This study also used an advanced wireless EEG device with dry electrodes to collect the data. This is significant as it demonstrates that portable devices that will increase the applicability of this tool can take measurements that are reliable. Another study (Corsi-Cabrera, Galindo-Vilchis, del-Rio-Portilla, Arce, & Ramos-Loyo, 2006) investigated the within subject reliability of quantitative EEG readings of 6 adult females over 9 months and found promising results of multiple correlation coefficients being higher than 0.89, indicating that there was a high level of reliability. While this study showed some good results for EEG reliability, it should be noted that the sample size was very small with only 6 subjects.
While these studies show that quantitative EEG is very reliable, there are some studies that also show that it can be unreliable at times. One study (Ratti, Waninger, Berka, Ruffini, & Verma, 2017) performed test-retest analysis on the measurements found on a consumer EEG system and found that measurements were either somewhat reliable or had relatively low reliability. This study shows that some caution should be used when performing quantitative EEG with a tool that is meant for the general consumer, and it is recommended that a medical grade system be used during clinical assessment and research (Ratti et al., 2017).
Another aspect of quantitative EEG that needs to be considered is the validity of the coherence, which is a measure of the stability of phase differences between two time series (Thatcher, 2010). Coherence is not a direct measurement of something like time or speed, but one of reliability as it detects changes in the phases over time. Due to this, EEG coherence is dependent on both the strength of connections and the number of connections and can be calculated by multiplying the two variables together. One method of evaluating the validity of EEG coherence is to test it in situations against noise to calculate a linear relationship between the magnitude of the coherence against the magnitude of noise (Thatcher, 2010). Based off of the linear relationship determined from the simulation it can demonstrate how much noise is being detected. If the simulation is detecting more noise, than the coherence will be lower. If no linear relationship is found the coherence is considered invalid.
Normative databases taken from quantitative EEG measurements are a tool that can be used by a clinician to aid in their diagnosing of a patient based on the measurements of their EEG and the characteristics that they present. In order for these databases to be effective their validity must be determined as well. This can be accomplished by performing cross-validation of an EEG database to create a normal distribution. Once this is done, predictive validity and content validity can be calculated on the normative database. An example of predictive validity with quantitative EEG is its ability to predict a person’s cognitive function by comparing IQ scores to their EEG measurements (Thatcher, 2010).