Detection of a "faked" strength task effort in volunteers using a computerized exercise testing system

The objective of this study was to develop an experimental method to separa te a "faked" strength effort from a "best" effort in volunteers. Thirty-fou r pain-free volunteers (18 males, 16 females) performed a shoulder press an d pull-down on an isokinetic computerized exercise testing system (CETS), g iving a best effort followed by a faked effort. Two months later, a randoml y selected subgroup (6 males) repeated the experiment to test the predictiv e validity of the derived variables. In the statistical analysis, best effo rts were first compared with fake efforts by paired t test for 80 CETS vari ables for males and females separately. Variables showing a strong differen ce between the best and faked effort were then selected for further analysi s. In the second step of the analysis, the method of multiple correlations (r(2) method) was used to reduce the number of redundant GETS variables to five in both the male and female groups. In the third step, a stepwise disc riminant analysis was used to select predictor variables for the male and f emale groups. For the variables selected by the discriminant analysis for b oth males and females, sensitivities and specificities were calculated. Fin ally, the developed discriminant formula was used in the predictive validit y part of the study to determine the sensitivities and specificities of the developed method. The discriminant analysis selected the following GETS va riables for male and female groups, respectively: duty cycle down, work wei ght/down, peak value up (males); and average power up, 40% repetition down, duty cycle up (females). For males, using their three variables, the discr iminant function classified 77.14% of the efforts correctly with 88.9% sens itivity and 64.7% specificity. For females, using their three variables, th e discriminant function classified 90.63% of the efforts correctly with 100 % sensitivity and 81.3% specificity. In the predictive validity group, the discriminant function classified 75% of the efforts correctly with 83.3% se nsitivity and 66.7% specificity. This pilot study indicates that the method developed here may be useful in the experimental study for the discriminat ion between faked and best efforts on this isokinetic GETS machine. Future studies using this method will need to involve a larger number of volunteer s.