The first hypothesis was there would be significant differences between the groups on whether J--- would be found guilty. We found that people were les slikely to find J--- guilty if the performed the action because of low self-contorl caused by TBI versus any other reason. probit guilty i.tbi Iteration 0: log likelihood = -129.53534 Iteration 1: log likelihood = -118.22468 Iteration 2: log likelihood = -117.89032 Iteration 3: log likelihood = -117.88882 Iteration 4: log likelihood = -117.88882 Probit regression Number of obs = 376 LR chi2(3) = 23.29 Prob > chi2 = 0.0000 Log likelihood = -117.88882 Pseudo R2 = 0.0899 ------------------------------------------------------------------------------ guilty | Coef. Std. Err. z P>|z| [95% Conf. Interval] -------------+---------------------------------------------------------------- tbi | 1 | -1.167782 .2885333 -4.05 0.000 -1.733297 -.6022674 2 | -.4296939 .3171776 -1.35 0.175 -1.051351 .1919628 3 | -.5136041 .3131396 -1.64 0.101 -1.127346 .1001382 | _cons | 1.867329 .2519018 7.41 0.000 1.37361 2.361047 ------------------------------------------------------------------------------ Afterwards, we tested whether people would recommend the 'harsher' punishment of prison over probation: We found that participants overwhelmingly recommended prison over probation. In cases of low self-contorl from TBI, they were significnatly less likely to suggest prison. probit prison i.tbi Iteration 0: log likelihood = -245.47736 Iteration 1: log likelihood = -229.83839 Iteration 2: log likelihood = -229.75616 Iteration 3: log likelihood = -229.75616 Probit regression Number of obs = 376 LR chi2(3) = 31.44 Prob > chi2 = 0.0000 Log likelihood = -229.75616 Pseudo R2 = 0.0640 ------------------------------------------------------------------------------ prison | Coef. Std. Err. z P>|z| [95% Conf. Interval] -------------+---------------------------------------------------------------- tbi | 1 | -1.029969 .2052034 -5.02 0.000 -1.43216 -.6277775 2 | -.1388954 .1826613 -0.76 0.447 -.496905 .2191142 3 | -.171668 .183941 -0.93 0.351 -.5321857 .1888497 | _cons | -.0646488 .1273515 -0.51 0.612 -.3142531 .1849554 ------------------------------------------------------------------------------ These results were the same whether 'guilty' was in the model. probit prison i.tbi i.guilty Iteration 0: log likelihood = -245.47736 Iteration 1: log likelihood = -225.79176 Iteration 2: log likelihood = -225.49487 Iteration 3: log likelihood = -225.49399 Iteration 4: log likelihood = -225.49399 Probit regression Number of obs = 376 LR chi2(4) = 39.97 Prob > chi2 = 0.0000 Log likelihood = -225.49399 Pseudo R2 = 0.0814 ------------------------------------------------------------------------------ prison | Coef. Std. Err. z P>|z| [95% Conf. Interval] -------------+---------------------------------------------------------------- tbi | 1 | -.9167714 .2105063 -4.36 0.000 -1.329356 -.5041866 2 | -.1052928 .1839191 -0.57 0.567 -.4657676 .2551819 3 | -.1336701 .1856696 -0.72 0.472 -.4975757 .2302356 | guilty | Yes | .7958777 .2924033 2.72 0.006 .2227779 1.368978 _cons | -.8401287 .3136956 -2.68 0.007 -1.454961 -.2252967 ------------------------------------------------------------------------------ Finally, we tested whether people would recommend longer sentences. This will be done by running an ANOVA on sentence length, with HSD and Bonferroni-corrected post-hoc tests. There was no difference in the duration of prison/probation sentences. Tests of Between-Subjects Effects Dependent Variable: You recommended prison, this assault and battery is punishable by between 0-60 months in prison....-months in prison Source Type III Sum of Squares df Mean Square F Sig. Corrected Model 1392.659a 3 464.220 1.297 .275 Intercept 391544.199 1 391544.199 1093.873 .000 tbi 1392.659 3 464.220 1.297 .275 Error 133154.745 372 357.943 Total 525998.000 376 Corrected Total 134547.404 375 a R Squared = .010 (Adjusted R Squared = .002) To test for differences in participant's beliefs about effort and the 'true self', we ran a path analysis on the two variables, allowing them to correlate, using WLSMV estimator due to the fact that they are ordered categorical variables, testing for whether there are differences between the three groups. STDYX Standardization Two-Tailed Estimate S.E. Est./S.E. P-Value SELFCONT ON TBIGROUP 0.001 0.061 0.016 0.987 GENESGROUP 0.078 0.063 1.233 0.217 BORNGROUP -0.027 0.061 -0.437 0.662 TRUESELF ON TBIGROUP 0.349 0.054 6.460 0.000 GENESGROUP 0.083 0.061 1.366 0.172 BORNGROUP 0.037 0.062 0.598 0.550 TBIGROUP WITH GENESGROUP 0.000 0.000 0.000 1.000 BORNGROUP 0.000 0.000 0.000 1.000 GENESGRO WITH BORNGROUP 0.000 0.000 0.000 1.000 TRUESELF WITH SELFCONTRO 0.248 0.057 4.341 0.000 Means TBIGROUP 0.581 0.035 16.827 0.000 GENESGROUP 0.565 0.034 16.453 0.000 BORNGROUP 0.573 0.034 16.642 0.000 Intercepts SELFCONTRO 1.935 0.119 16.266 0.000 TRUESELF 2.175 0.160 13.560 0.000 Finally, as a test of the 'meatbag model', we ran an ANOVA on the total dualism score of the 5 items with an ANOVA on dualism, with HSD and Bonferroni-corrected post-hoc tests. This showed no differences in participants' belief about dualism as a result of the manipulation. Tests of Between-Subjects Effects Dependent Variable: dualism Source Type III Sum of Squares df Mean Square F Sig. Corrected Model 55.615a 3 18.538 .342 .795 Intercept 210127.640 1 210127.640 3877.179 .000 tbi 55.615 3 18.538 .342 .795 Error 20160.915 372 54.196 Total 230361.000 376 Corrected Total 20216.529 375 a R Squared = .003 (Adjusted R Squared = -.005)