The function computes predicted values, corresponding standard errors, confidence intervals, and (approximate) credibility/prediction intervals for objects of class "rma".

# S3 method for rma
predict(object, newmods, intercept, tau2.levels, gamma2.levels,
        addx=FALSE, level, digits, transf, targs, vcov=FALSE, …)

Arguments

object

an object of class "rma".

newmods

optional vector or matrix specifying the values of the moderator values for which the predicted values should be calculated. See ‘Details’.

intercept

logical specifying whether the intercept should be included when calculating the predicted values for newmods. If unspecified, the intercept is automatically added when the original model also included an intercept.

tau2.levels

vector specifying the levels of the inner factor when computing credibility/prediction intervals. Only relevant for models of class "rma.mv" (see rma.mv) and when the model includes more than a single \(\tau^2\) value. See ‘Details’.

gamma2.levels

vector specifying the levels of the inner factor when computing credibility/prediction intervals. Only relevant for models of class "rma.mv" (see rma.mv) and when the model includes more than a single \(\gamma^2\) value. See ‘Details’.

addx

logical specifying whether the values of the moderator variables should be added to the returned object. See ‘Examples’.

level

numerical value between 0 and 100 specifying the confidence and credibility/prediction interval level (if unspecified, the default is to take the value from the object).

digits

integer specifying the number of decimal places to which the printed results should be rounded (if unspecified, the default is to take the value from the object).

transf

optional argument specifying the name of a function that should be used to transform the predicted values and interval bounds (e.g., transf=exp; see also transf). If unspecified, no transformation is used.

targs

optional arguments needed by the function specified under transf.

vcov

logical specifying whether the variance-covariance matrix of the predicted values should also be returned (the default is FALSE).

other arguments.

Details

For a fixed-effects model, predict(object) returns the estimated (average) outcome in the set of studies included in the meta-analysis. This is the same as the estimated intercept in the fixed-effects model.

For a random-effects model, predict(object) returns the estimated (average) outcome in the hypothetical population of studies from which the set of studies included in the meta-analysis are assumed to be a random selection. This is the same as the estimated intercept in the random-effects model.

For models including one or more moderators, predict(object) returns the estimated (average) outcomes for values of the moderator(s) equal to those of the \(k\) studies included in the meta-analysis (i.e., the fitted values for the \(k\) studies).

For models including \(q\) moderator variables, new moderator values for \(k_{new}\) new studies can be specified by setting newmods equal to an \(k_{new} \times q\) matrix with the corresponding new moderator values. If the original model includes an intercept, then it should not be explicitly specified under newmods, as it will be added by default (unless one sets intercept=FALSE). Also, any factors in the original model get turned into the appropriate contrast variables within the rma function, so that newmods should actually include the values for the contrast variables. Examples are shown below.

For random/mixed-effects models, an approximate credibility/credible or prediction interval is also calculated (Raudenbush, 2009; Riley et al., 2011). The interval estimates where level % of the true outcomes fall in the hypothetical population of studies. Note that this interval is calculated under the assumption that the value of \(\tau^2\) is known (and not estimated). A method for calculating a credibility/prediction interval that accounts for the uncertainty in the estimate of \(\tau^2\) may be implemented in the future.

For random-effects models of class "rma.mv" (see rma.mv) with multiple \(\tau^2\) values (i.e., when the random argument includes a ~ inner | outer term and struct="HCS", struct="HAR", or struct="UN"), the function will provide credibility/prediction intervals for each level of the inner factor (since the credibility/prediction intervals differ depending on the \(\tau^2\) value). Alternatively, one can use the tau2.levels argument to specify for which level(s) the credibility/prediction interval should be provided.

If the model includes a second ~ inner | outer term with multiple \(\gamma^2\) values, credibility/prediction intervals for each combination of levels of the inner factors will be provided. Alternatively, one can use the tau2.levels and gamma2.levels arguments to specify for which level combination(s) the credibility/prediction interval should be provided.

When using the newmods argument for mixed-effects models of class "rma.mv" with multiple \(\tau^2\) (and multiple \(\gamma^2\)) values, one must use the tau2.levels (and gamma2.levels) argument to specify the levels of the inner factor(s) (i.e., a vector of length \(k_{new}\)) to obtain the appropriate credibility/prediction interval(s).

Value

An object of class "list.rma". The object is a list containing the following components:

pred

predicted value(s).

se

corresponding standard error(s).

ci.lb

lower bound of the confidence interval(s).

ci.ub

upper bound of the confidence interval(s).

cr.lb

lower bound of the credibility/prediction interval(s) (only random/mixed-effects models).

cr.ub

upper bound of the credibility/prediction interval(s) (only random/mixed-effects models).

tau2.level

the level(s) of the inner factor (only for models of class "rma.mv" with multiple \(\tau^2\) values).

gamma2.level

the level(s) of the inner factor (only for models of class "rma.mv" with multiple \(\gamma^2\) values).

X

the moderator value(s) used to calculate the predicted values (only when addx=TRUE).

some additional elements/values.

If vcov=TRUE, then the returned object is a list with the first element equal to the one as described above and the second element equal to the variance-covariance matrix of the predicted values. The "list.rma" object is formatted and printed with print.list.rma.

Note

The predicted values are based only on the fixed effects of the model. Best linear unbiased predictions (BLUPs) that combine the fitted values based on the fixed effects and the estimated contributions of the random effects can be obtained with blup.rma.uni (currently only for objects of class "rma.uni").

When using the transf option, the transformation is applied to the predicted values and the corresponding interval bounds. The standard errors are then set equal to NA and are omitted from the printed output. Also, vcov=TRUE is ignored when using the transf option.

Confidence and credibility/prediction intervals are calculated based on the critical values from a standard normal distribution (i.e., \(\pm 1.96\) for level=95). When the model was fitted with the Knapp and Hartung (2003) method (i.e., test="knha" in the rma.uni function) or with test="t" in the rma.glmm and rma.mv functions, then the t-distribution with \(k-p\) degrees of freedom is used.

References

Hedges, L. V., & Olkin, I. (1985). Statistical methods for meta-analysis. San Diego, CA: Academic Press.

Raudenbush, S. W. (2009). Analyzing effect sizes: Random effects models. In H. Cooper, L. V. Hedges, & J. C. Valentine (Eds.), The handbook of research synthesis and meta-analysis (2nd ed., pp. 295--315). New York: Russell Sage Foundation.

Riley, R. D., Higgins, J. P. T., & Deeks, J. J. (2011). Interpretation of random effects meta-analyses. British Medical Journal, 342, d549.

Viechtbauer, W. (2010). Conducting meta-analyses in R with the metafor package. Journal of Statistical Software, 36(3), 1--48. http://www.jstatsoft.org/v36/i03/.

See also

Examples

### meta-analysis of the log risk ratios using a random-effects model res <- rma(measure="RR", ai=tpos, bi=tneg, ci=cpos, di=cneg, data=dat.bcg) ### average risk ratio with 95% CI predict(res, transf=exp)
#> #> pred ci.lb ci.ub cr.lb cr.ub #> 0.4894 0.3441 0.6962 0.1546 1.5490 #>
### mixed-effects model with absolute latitude as a moderator res <- rma(measure="RR", ai=tpos, bi=tneg, ci=cpos, di=cneg, mods = ~ ablat, data=dat.bcg) ### predicted average risk ratios for given absolute latitude values predict(res, transf=exp, addx=TRUE)
#> #> pred ci.lb ci.ub cr.lb cr.ub X.intrcpt X.ablat #> 1 0.3574 0.2714 0.4705 0.1947 0.6560 1 44 #> 2 0.2595 0.1749 0.3848 0.1328 0.5070 1 55 #> 3 0.3788 0.2927 0.4901 0.2079 0.6900 1 42 #> 4 0.2831 0.1977 0.4054 0.1478 0.5422 1 52 #> 5 0.8809 0.6321 1.2276 0.4667 1.6625 1 13 #> 6 0.3574 0.2714 0.4705 0.1947 0.6560 1 44 #> 7 0.7397 0.5639 0.9704 0.4036 1.3557 1 19 #> 8 0.8809 0.6321 1.2276 0.4667 1.6625 1 13 #> 9 0.5861 0.4716 0.7284 0.3270 1.0505 1 27 #> 10 0.3788 0.2927 0.4901 0.2079 0.6900 1 42 #> 11 0.7616 0.5752 1.0083 0.4138 1.4016 1 18 #> 12 0.4922 0.3989 0.6073 0.2753 0.8799 1 33 #> 13 0.4922 0.3989 0.6073 0.2753 0.8799 1 33 #>
### predicted average risk ratios for 10-60 degrees absolute latitude predict(res, newmods=c(10, 20, 30, 40, 50, 60), transf=exp)
#> #> pred ci.lb ci.ub cr.lb cr.ub #> 1 0.9612 0.6668 1.3857 0.5000 1.8478 #> 2 0.7185 0.5526 0.9343 0.3936 1.3117 #> 3 0.5371 0.4355 0.6623 0.3005 0.9600 #> 4 0.4015 0.3151 0.5115 0.2218 0.7266 #> 5 0.3001 0.2144 0.4201 0.1586 0.5678 #> 6 0.2243 0.1423 0.3538 0.1105 0.4552 #>
### mixed-effects model with two moderators (absolute latitude and publication year) res <- rma(measure="RR", ai=tpos, bi=tneg, ci=cpos, di=cneg, mods = ~ ablat + year, data=dat.bcg) ### predicted average risk ratios for 10 and 60 degrees latitude in 1950 and 1980 predict(res, newmods=cbind(c(10,60,10,60),c(1950,1950,1980,1980)), transf=exp, addx=TRUE)
#> #> pred ci.lb ci.ub cr.lb cr.ub X.intrcpt X.ablat X.year #> 1 0.8995 0.3689 2.1933 0.2981 2.7146 1 10 1950 #> 2 0.2217 0.1278 0.3847 0.0944 0.5208 1 60 1950 #> 3 0.9525 0.6199 1.4637 0.4361 2.0802 1 10 1980 #> 4 0.2348 0.1005 0.5481 0.0805 0.6843 1 60 1980 #>
### mixed-effects model with two moderators (one of which is a factor) res <- rma(measure="RR", ai=tpos, bi=tneg, ci=cpos, di=cneg, mods = ~ ablat + factor(alloc), data=dat.bcg) ### examine how the factor was actually coded for the studies in the dataset predict(res, addx=TRUE)
#> #> pred se ci.lb ci.ub cr.lb cr.ub X.intrcpt X.ablat #> 1 -1.1744 0.2137 -1.5932 -0.7557 -2.0293 -0.3196 1 44 #> 2 -1.4745 0.2742 -2.0119 -0.9370 -2.3933 -0.5556 1 55 #> 3 -1.1199 0.2061 -1.5238 -0.7159 -1.9676 -0.2722 1 42 #> 4 -1.3926 0.2552 -1.8928 -0.8925 -2.2902 -0.4951 1 52 #> 5 -0.0614 0.3336 -0.7152 0.5924 -1.0528 0.9300 1 13 #> 6 -0.9069 0.3176 -1.5295 -0.2844 -1.8780 0.0642 1 44 #> 7 -0.4925 0.2338 -0.9508 -0.0343 -1.3674 0.3823 1 19 #> 8 -0.3289 0.2694 -0.8568 0.1991 -1.2422 0.5845 1 13 #> 9 -0.7107 0.2007 -1.1040 -0.3175 -1.5534 0.1320 1 27 #> 10 -0.7939 0.2667 -1.3166 -0.2712 -1.7042 0.1164 1 42 #> 11 -0.1393 0.2654 -0.6594 0.3808 -1.0481 0.7695 1 18 #> 12 -0.5484 0.2438 -1.0263 -0.0706 -1.4337 0.3369 1 33 #> 13 -0.5484 0.2438 -1.0263 -0.0706 -1.4337 0.3369 1 33 #> X.factor.alloc.random X.factor.alloc.systematic #> 1 1 0 #> 2 1 0 #> 3 1 0 #> 4 1 0 #> 5 0 0 #> 6 0 0 #> 7 1 0 #> 8 1 0 #> 9 1 0 #> 10 0 1 #> 11 0 1 #> 12 0 1 #> 13 0 1 #>
### predictd average risk ratios at 30 degrees for the three factor levels ### note: the contrast (dummy) variables need to specified explicitly here predict(res, newmods=c(30, 0, 0), addx=TRUE) # for alternate allocation
#> #> pred se ci.lb ci.ub cr.lb cr.ub X.intrcpt X.ablat #> -0.5251 0.2923 -1.0980 0.0478 -1.4651 0.4149 1 30 #> X.factor.alloc.random X.factor.alloc.systematic #> 0 0 #>
predict(res, newmods=c(30, 1, 0), addx=TRUE) # for random allocation
#> #> pred se ci.lb ci.ub cr.lb cr.ub X.intrcpt X.ablat #> -0.7926 0.1941 -1.1729 -0.4122 -1.6293 0.0442 1 30 #> X.factor.alloc.random X.factor.alloc.systematic #> 1 0 #>
predict(res, newmods=c(30, 0, 1), addx=TRUE) # for systematic allocation
#> #> pred se ci.lb ci.ub cr.lb cr.ub X.intrcpt X.ablat #> -0.4666 0.2420 -0.9410 0.0078 -1.3501 0.4168 1 30 #> X.factor.alloc.random X.factor.alloc.systematic #> 0 1 #>
### can also use named vector with arbitrary order and abbreviated variable names predict(res, newmods=c(sys=0, ran=0, abl=30))
#> #> pred se ci.lb ci.ub cr.lb cr.ub #> -0.5251 0.2923 -1.0980 0.0478 -1.4651 0.4149 #>
predict(res, newmods=c(sys=0, ran=1, abl=30))
#> #> pred se ci.lb ci.ub cr.lb cr.ub #> -0.7926 0.1941 -1.1729 -0.4122 -1.6293 0.0442 #>
predict(res, newmods=c(sys=1, ran=0, abl=30))
#> #> pred se ci.lb ci.ub cr.lb cr.ub #> -0.4666 0.2420 -0.9410 0.0078 -1.3501 0.4168 #>