There are two forms of intra-individual variability (IIV) in the literature: inconsistency, or variability on one task across many time points, and dispersion, or variability across many tasks at one time point. These forms of IIV are often lumped together into one construct, but there is limited evidence supporting this practice, as few studies have examined the relationship between these measures. Additionally, it is not clear how stable these constructs are over time. Therefore, the goal of the present study was (1) to explore the relationship between (a) inconsistency and dispersion and (b) IIV and mean performance, and (2) to determine whether these relationships are stable over a one- to two-year follow-up interval.

A total of 123 community-dwelling older adults (Mage=69.5, Meducation=15.6 years) from an archival database completed the Push-Turn-Taptap task to assess inconsistency and the Delis-Kaplan Executive Functioning System (D-KEFS) to assess dispersion. These measures were selected because both are highly executive, thereby allowing us to hold the cognitive domain constant across forms of IIV. Dispersion was calculated by taking the standard deviation of the executive conditions of four D-KEFS subtests (Verbal Fluency, Design Fluency, Trail Making, and Color Word Interference). Follow-up data were collected one to two and a half years after baseline. Bivariate and partial correlations controlling for time to follow-up were examined.

Inconsistency and dispersion were not correlated at baseline but were weakly correlated at follow-up (r=.281, p=.012). Additionally, both forms of IIV were moderately correlated with themselves across the follow-up interval (inconsistency: r=.450, p<.001; dispersion: r=.448, p<.001). The partial correlations were nearly identical to bivariate correlations.

Additionally, inconsistency was correlated with poorer mean executive functioning (EF) performance on both the PTT (baseline: r=.281, p<.001, follow-up: r=.435, p<.001) and D-KEFS (baseline: r= -.270, p=.003, follow-up: r=-.573, p<.001). In contrast, dispersion was correlated with mean EF performance only on the D-KEFS at baseline (r= -.292, p<.001) but with both measures at follow-up (PTT: r=.232, p=.039; D-KEFS: r= -.378, p<.001). When controlling for follow-up interval, inconsistency was no longer correlated with baseline mean PTT performance, but all other relationships remained the same (i.e., dispersion and inconsistency displayed the same pattern of correlations with mean EF).

Although inconsistency and dispersion are both forms of IIV, they are weakly related. In other words, although they may have shared mechanisms, these two methods of measuring IIV likely represent different constructs. In the future, authors should take care to specify the form of IIV being investigated in their publications rather than referring to either form as IIV generally. Additionally, both forms of IIV are weakly to moderately correlated with mean EF performance, indicating that IIV is related to but separate from mean-level performance. Interestingly, IIV and mean performance were more strongly correlated at follow-up, which may be suggestive of incipient cognitive decline. Lastly, it seems that both inconsistency and dispersion are somewhat stable across a one- to two-year follow-up interval, suggesting that IIV may be a trait-level construct to some extent. However, IIV may also be influenced by state-level contextual factors, and more research examining the stability of and contributors to IIV is necessary.