METRIC INVARIANCE OF THE MEASUREMENT MODEL UNDERLYING SYMPTOMS OF MOOD DISTURBANCE IN PATIENTS WITH TEMPORAL LOBE EPILEPSY
Abstract number :
2.363
Submission category :
Year :
2003
Submission ID :
3805
Source :
www.aesnet.org
Presentation date :
12/6/2003 12:00:00 AM
Published date :
Dec 1, 2003, 06:00 AM
Authors :
Stephen C. Bowden, Rachel E. Reilly, Fiona J. Bardenhagen, Mark J. Cook Psychology, University of Melbourne, Parkville, Victoria, Australia; Clinical Neurosciences, St. Vincent[apos]s Hospital, Fitzroy, Victoria, Australia
Controversy surrounds the question of whether there is a specific pattern of psychopathology associated with temporal lobe epilepsy (TLE) or whether the symptoms of psychological distress observed in these patients are indicative of the presence of common disorders such as depression. If symptoms of psychological distress in people with TLE are seen as psychopathology specific to the seizure disorder, then treatment and management may differ from that offered to patients with depression. Recent developments in the modeling of test scores permits a strong test of the hypothesis that the latent behaviors underlying symptoms of psychological distress, are equivalent in patients with TLE or patients with psychological distress due to other conditions.
Two patient groups were recruited, the first a sample of 187 consecutive patients with a diagnosis of TLE based on clinical neurological examination and video-telemetry. The second sample comprised a consecutive sample of 150 patients with heterogeneous neurological conditions. All patients were administered the Beck Depression Inventory (BDI). BDI item scores in both groups were subject to tests of metric invariance of the measurement model, using a mean- and covariance structure analysis for ordered-categorical data .
Results supported the inference of strict metric invariance of the measurement model across groups. For example the Tucker Lewis Index for the strict invariance model was .99. This model comprised equivalent matrices in both groups representing factor loadings, observed variable thresholds, latent factor variances and covariances, latent factor means, and scale factors. There was evidence of a group difference in the mean scores on one latent factor.
To our knowledge, this is the first report of an examination of invariance across diagnostic groups of an ordered-categorical mean- and covariance structure. From the factor-models of the BDI that have been proposed by previous researchers, a first-order model comprising three factors: negative attitude, performance difficulty, and somatic elements was found to be the best fitting model in the two samples, and overall, fit statistics suggested an excellent fit, presumably an advantage of using the more appropriate categorical data model. The pattern of strict metric invariance found for this model, suggests that the same set of latent variables, with the same metric, is measured across the two groups. This result fulfils a necessary assumption for the uncomplicated interpretation of symptoms of mood disturbance in people with TLE. This result also provides direct evidence that the latent structure of symptoms of depression, underlying scores on the BDI, are the same for people with TLE and for people with psychological distress due to other neurological conditions.
[Supported by: Australian Research Council and Australian National Health and Medical Research Council.]