COMBI >> Scales >> Cog-Log >> Properties


Tom Novack, Ph.D.
University of Alabama at Birmingham




Novack, T. (2004). The Cognitive Log. The Center for Outcome Measurement in Brain Injury.
combi/coglog ( accessed ).





Cog-Log Properties

Normative Sample: The Cog-Log was presented to 83 individuals without acquired brain injury (Alderson & Novack, 2003). Individuals without known neurological insult received average total Cog-Log scores of 28 (± 2), and mean individual item scores were greater than or equal to 2.4. Demographic variables of age, education, and gender did not predict total or individual item scores (p>.05). Stepwise discriminant analyses on a sample of 82 brain injured patients and 82 normal controls matched for age, education, and gender revealed that a cut-off score of 25 correctly classified 88.4% of individuals in their respective groups.

Reliability: A sample of 150 individuals with acquired brain injury were examined with the Cog-Log. Most of the sample (80%) had sustained moderate to severe TBI, with the remainder including patients with CVA and anoxia. For detailed information about this sample and the properties of the scale please read Alderson & Novack (2003). Internal consistency analysis (Cronbach’s alpha) was conducted for the total Cog-Log score. Inter-rater reliability estimates (Spearman’s rho) were calculated using a subset of 19 patients (75 total observations), who did not differ in age, education, injury type, or injury severity (p > .05) from the sample as a whole. Interrater reliability coefficients for each of the 10 Cog-Log items ranged from .749 (Go/No-go task) to 1.0 (Time Estimation). Standard errors of measurement are no more than .10 for single item scores, which range from 0 to 3.

The overall internal consistency of the 10-item Cog-Log is .778, with a standard error of measurement of .53. Factor analysis of the Cog-Log using principal components extraction (N = 150) revealed a unitary factor (Eigenvalue = 3.48). Although all scale items loaded on this factor, those items involving delayed recall of verbal information and recitation of months backwards demonstrated the highest loadings, suggesting a stronger contribution of complex working memory and long-term recall to this unitary factor.

Concurrent Validity: Multiple regression evaluation of individual Cog-Log items revealed a number of significant relationships to well-established standardized neuropsychological measures of verbal memory, attention/working memory, and executive functions assessed on the same day the Cog-Log was administered (Alderson & Novack, 2003). In order to account for shared variance among Cog-Log items, a stepwise multiple regression evaluation was utilized. No a priori hypotheses were made about the entry order of Cog-Log items. Age, education, and estimated IQ were also entered as independent factors to account for individual differences that are also known to affect raw score performance on standardized cognitive measures. Immediate story recall (WMS-R) was significantly predicted by the Cog-Log item assessing immediate address recall (p<.001). Delayed story recall (WMS-R) was predicted by Cog-Log items assessing delayed address recall (p<.001) and months reversed (p=.003). Cog-Log items assessing immediate and delayed address recall, orientation to date, and reversal of months were significantly associated with performance on standardized list learning measures. Performance on the initial list learning trial of the RAVLT was significantly predicted by Cog-Log immediate address recall (p=.004). Cog-Log items assessing delayed address recall (p<.001) and orientation to date (p=.001) predicted performance on the final RAVLT learning trial. Delayed address recall also predicted delayed recall of RAVLT list items (p<.001). Cog-Log items assessing immediate address recall (p<.001) and time estimation (p=.003) predicted forward digit span (WAIS-R), while Cog-Log months backward (p<.001) predicted backwards digit span (WAIS-R). Finally, performance on part B of the Reitan Trail Making Test (TMTB) was predicted by the Cog-Log item months backward (p<.001).

Predicting Outcome: In a sample of 50 individuals participating in acute rehabilitation following moderate to severe TBI, the lowest Cog-Log score during rehabilitation was found to predict cognitive outcome at one year after injury (Lee, LeGalbo, Baños, & Novack, 2004). The Cog-Log was found to contribute significantly to the prediction of outcome in three of six neuropsychological domains (attention, executive functioning, and visuomotor-visuospatial abilities) after controlling for demographics and injury severity. The companion measure to the Cog-Log, the Orientation Log, was found to predict memory functioning at one year post-injury. The neuropsychological tests employed were those being used by the TBI Model System program. The results underscore the potential usefulness of using both the O-Log and Cog-Log. The results of the study do not suggest that the Cog-Log can be considered a substitute for a neuropsychological examination. Rather, the study confirms that the Cog-Log is providing a general measure of cognitive ability.



Copyright © 1998-2012
Home | Background | Scales | Survey | Newsletter


NIDRR Logo A project funded by the National Institute on Disability and Rehabilitation Research.