Four standard deviations were not obtainable from the primary publications. These were imputed using multiple imputation.
Meta-regresison on ability to delay gratification over time.
metareg delaymin age year, wsse(se)
Meta-regression Number of obs = 31
REML estimate of between-study variance tau2 = 4.787
% residual variation due to heterogeneity I-squared_res = 90.12%
Proportion of between-study variance explained Adj R-squared = 82.01%
Joint test for all covariates Model F(2,28) = 49.21
With Knapp-Hartung modification Prob > F = 0.0000
------------------------------------------------------------------------------
delaymin | Coef. Std. Err. t P>|t| [95% Conf. Interval]
-------------+----------------------------------------------------------------
age | 3.234519 .3418233 9.46 0.000 2.534326 3.934712
year | .1044717 .0282488 3.70 0.001 .0466066 .1623369
_cons | -217.1422 56.73186 -3.83 0.001 -333.3522 -100.9323
------------------------------------------------------------------------------
Meta-regression without outliers
metareg delaymin age year if resid < 1.9601 & resid > -1.9601, wsse(se)
Meta-regression Number of obs = 29
REML estimate of between-study variance tau2 = 3.567
% residual variation due to heterogeneity I-squared_res = 89.89%
Proportion of between-study variance explained Adj R-squared = 79.68%
Joint test for all covariates Model F(2,26) = 38.55
With Knapp-Hartung modification Prob > F = 0.0000
------------------------------------------------------------------------------
delaymin | Coef. Std. Err. t P>|t| [95% Conf. Interval]
-------------+----------------------------------------------------------------
age | 2.894399 .3631089 7.97 0.000 2.148018 3.64078
year | .1121949 .0256243 4.38 0.000 .0595234 .1648663
_cons | -230.9541 51.50464 -4.48 0.000 -336.8234 -125.0848
------------------------------------------------------------------------------
Change in variance over time
regress delaysdmin year
Source | SS df MS Number of obs = 31
-------------+------------------------------ F( 1, 29) = 0.01
Model | .063759671 1 .063759671 Prob > F = 0.9232
Residual | 195.753417 29 6.75011784 R-squared = 0.0003
-------------+------------------------------ Adj R-squared = -0.0341
Total | 195.817177 30 6.52723924 Root MSE = 2.5981
------------------------------------------------------------------------------
delaysdmin | Coef. Std. Err. t P>|t| [95% Conf. Interval]
-------------+----------------------------------------------------------------
year | .0027032 .0278138 0.10 0.923 -.0541825 .0595889
_cons | .4489772 55.66103 0.01 0.994 -113.3906 114.2886
------------------------------------------------------------------------------
Controlling for possible changes in sample size
regress delaysdmin year n
Source | SS df MS Number of obs = 31
-------------+------------------------------ F( 2, 28) = 0.30
Model | 4.15402876 2 2.07701438 Prob > F = 0.7407
Residual | 191.663148 28 6.84511244 R-squared = 0.0212
-------------+------------------------------ Adj R-squared = -0.0487
Total | 195.817177 30 6.52723924 Root MSE = 2.6163
------------------------------------------------------------------------------
delaysdmin | Coef. Std. Err. t P>|t| [95% Conf. Interval]
-------------+----------------------------------------------------------------
year | .0038114 .0280455 0.14 0.893 -.0536373 .0612601
n | -.0026145 .0033822 -0.77 0.446 -.0095427 .0043137
_cons | -1.593728 56.11358 -0.03 0.978 -116.5372 113.3497
------------------------------------------------------------------------------
