Erratum to: Behav Genet DOI 10.1007/s10519-011-9509-7

The original publication of the article includes confidence intervals for the parameter estimates that are too narrow. The authors wish to correct this and the parameter estimates with the correct 95% confidence intervals (in parentheses) are given below.

Alcohol consumption:

$$ F1: \ y = \mathop {0.289}\limits_{(-0.029 -0.606)} * \,\exp (\mathop {0.205}\limits_{(-0.379 -0.788)} *\, D_{neg} ) + \mathop {0.0105}\limits_{(-0.0161 -0.0371)} *\, \ln (1 + \;\mathop {100}\limits_{(-2378 -2577)} *\, D_{pos} ) + \mathop {0.237}\limits_{(-0.100 -0.575)} $$

Smoking:

$$ F3: \ y = \mathop {0.508}\limits_{( - 2.079 - 3.096)} *\, \exp (\mathop {0.049}\limits_{( - 0.265 - 0.363)} *\, D_{neg} ) - \mathop {0.000188}\limits_{(0.000067{-}0.000308)} *\, (D_{pos} - \mathop {0.0}\limits_{(-13.8 - 13.8)} )^{2} + \mathop {0.117}\limits_{( - 2.482 - 2.715)} $$

Exercise:

$$ F3: \ y = \mathop {0.093}\limits_{( - 0.080 - 0.266)} *\, \exp (\mathop {0.274}\limits_{(- 1.091 - 1.639)} *\, D_{neg} ) + \mathop {0.000203}\limits_{(0.000085 - 0.000322)} *\, (D_{pos} - \mathop {11.67}\limits_{(4.71{-} 18.63)} )^{2} + \mathop {0.188}\limits_{(0.015-0.361)} $$

The errors are due to an error in the weighting procedure in the nonlinear regression analysis we used. The parameter estimates remain practically unchanged. Even if some parameter estimates in the original models are no longer significant, testing a simpler model, specifying only a linear convergence before marriage/cohabitation, shows that the main conclusions remain unchanged: There are significant spouse correlations even before the spouses met, there is significant convergence during the years before marriage or cohabitation, and concordance/discordance after marriage is moderate. Details are shown at http://www.fhi.no/artikler/?id=96795