ADHD is indeed typically associated with increased RT variability. Often this co-occurs with a longer mean RT as compared to controls. Yet, ADHD-control differences in both the mean RT and its variability may depend on the nature of the task used. In more simple, less complex, long and boring tasks ADHD children often show a longer RT and increased RT variability. Yet, when ADHD children are fully engaged, that is, when the task is difficult, complex, short, or intrinsically salient to the child, ADHD-control differences in RT and accuracy are often not found. So, the question is what kind of task you precisely used. Were there any ADHD-control differences in the mean RT? Also, ADHD children are highly sensitive to external control. For example, were the ADHD children tested alone or was one adult person like one of the researchers present during task administration? In the latter case, it has been found that the task performance of ADHD children can hardly be differentiated from that of controls. Finally, it has been suggested that ADHD children, when equal or better task performance have been found, utilize extra effort to compensate for their cognitive dysfunction. In this case, the smaller RT variability seen in the ADHD group may be interpreted in terms of increased compensatory effort.   

Those diagnosed with ADHD can show an increased tendency for false alarms. Perhaps there is something explanatory to be found in Barkley's notion of a behavioural inhibition system.

ADHD is indeed typically associated with increased RT variability. Often this co-occurs with a longer mean RT as compared to controls. Yet, ADHD-control differences in both the mean RT and its variability may depend on the nature of the task used. In more simple, less complex, long and boring tasks ADHD children often show a longer RT and increased RT variability. Yet, when ADHD children are fully engaged, that is, when the task is difficult, complex, short, or intrinsically salient to the child, ADHD-control differences in RT and accuracy are often not found. So, the question is what kind of task you precisely used. Were there any ADHD-control differences in the mean RT? Also, ADHD children are highly sensitive to external control. For example, were the ADHD children tested alone or was one adult person like one of the researchers present during task administration? In the latter case, it has been found that the task performance of ADHD children can hardly be differentiated from that of controls. Finally, it has been suggested that ADHD children, when equal or better task performance have been found, utilize extra effort to compensate for their cognitive dysfunction. In this case, the smaller RT variability seen in the ADHD group may be interpreted in terms of increased compensatory effort.   

Thanks Odin. I should have specified that I have significantly increased RTV in the ADHD groups compared with two groups without ADHD. There is no effect of ADHD on RT, and I used the coefficient of variation measure of RTV to control for RT effects anyway. These findings are within two robust cognitive control tasks and external control was minimised and comparable with previous work in cognitive control in ADHD. I like the idea that these individuals with ADHD could be increasing effort to compensate for their difficulties, but I'm not sure this can hold given the increased RTV (and other impairments in cognitive control) that they exhibit. 

Do those with ADHD get it "wrong" when they have the short RT? Do those with ADHD take a longer time on trials when they get it right?

Thanks Tom. I only analysed RT on correct correct trials, so the normal RTs and increased RTV occur when the participants with ADHD get it right. 

Ok, well I think that scrubs any explanation related to false alarms and participants not being able to inhibit incorrect responses thus performing faster on some trials. How about the idea that their "sustained attention" drifts from a momentary state of attentivity when responses are fast into an attentional lapse when responses are slower with more fluidity than in normals? If such a view were valid, trials with fast reaction times would cluster together for shorter periods in those with ADHD than normals. Similarly, normals might show more clustering have longer yet less severe lapses in cognitive control and longer periods of attentivity.

I absolutely agree, Tom, thanks. I'm not sure how that can explain those individuals with the worst ADHD symptoms having the best (lowest) reaction time variability though. That's the bit I'm really confused about

Sounds like your data absolutely disagree though! Imagining your data somewhat, revealing may be to plot individually trial-by-trial the long-term dynamics of RT and whether the response was wrong or right. Such individual dynamics of attentivity and lapses may say more about what is disordered in cognitive control in the syndrome than the individual mean reaction times.

Hi Lizzie,

You mentioned you're using the Coefficient of Variation (Mrt/SDrt  I would imagine). There's good evidence that that method for calculating RT variability is problematic (though still associated with ADHD). I'd recommend using the ex-gaussian approach to parse out which part of the RT distribution is associated with ADHD severity. Informally, my colleagues and I have looked at severity in several of our ADHD samples using this method and observed the positive correlations you're expecting. For some group-based analyses check out Kofler, Rapport, Sarver et al (2013; meta analytic review on 319 ADHD rtv studies; Clinical Psych Review) and Kofler et al 2014 (RTV as a neurocognitive indicator of ADHD; Neuropsychology)...attached.

As an potentially interesting aside, our group has found a couple associations among lower effect size associations between ADHD and various other variables (e.g., classroom inattention) when more stringent ADHD inclusion methods are used. This might be at play depending on how well-characterized your sample is, perhaps. We've hypothesized that studies using more strict criteria have more homogenous individuals within them, which while showing more variability as a group, essentially increases the variability of the effect size denominator value, which in turn provides lower effect sizes.  Less sure how that might play out when examining adhd severity in a correlation. However, are your collapsing all three groups (ADHD + group B + group C) or just looking at the ADHD group. One thought might be that if your other groups have other clinically related difficulties (perhaps comorbid?) that the negative effect might reflect that they are more impaired because of comorbid features that are additive effects on attention rather than ADHD severity, per se. Without additional details, this is speculation on my part though. 

Wish you well in your endeavor!

Article Reaction time variability in ADHD: A meta-analytic review of...

Article Working Memory and Intraindividual Variability as Neurocogni...

Hi Dustin. This is very helpful, thank you. I wasn't aware of your group's 2014 paper - I look forward to reading it. I am indeed collapsing across a comorbid (ADHD plus tics) group and an ADHD group and within these individuals, I get the negative correlations. If I include controls and another non-ADHD clinical group, I get the expected positive correlation. Hopefully with an ex-Gaussian approach I will be able to disentangle this!

Sounds very interesting! Feel free to pass it along once its published. Best of luck in disentangling the effects.-Dustin

These three articles might be of help.  

http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3053341/

http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3461120/

http://www.hindawi.com/journals/apsy/2014/502685/

Thanks Candace; very helpful. 

Hi. A couple of other relevant sources you might want to check out include:

Tamm, L., Epstein, J. N., Denton, C. A., Vaughn, A. J., Peugh, J., & Willcutt, E. G. (2014). Reaction time variability associated with reading skills in poor readers with ADHD. Journal Of The International Neuropsychological Society, 20, 292-301. doi:10.1017/S1355617713001495

Adamo, N., Di Martino, A., Esu, L., Petkova, E., Johnson, K., Kelly, S., & ... Zuddas, A. (2014). Increased response-time variability across different cognitive tasks in children with ADHD. Journal Of Attention Disorders, 18, 434-446. doi:10.1177/1087054712439419

As highlighted in other responses, comorbidity may account for some of the sample differences you are finding, as could the consideration of specific subtests or scores of neurocognitive performance. Somewhat notorious inconsistency in the latter.

Thanks Will. I'm actually friends with one of the authors of the second paper and have discussed this strange finding of mine with her! I will look forward to reading the Tamm paper - I haven't read this one yet.