# How do I interpret significant results in meta-regression?

                         estimate      se     tval    pval    ci.lb    ci.ub
intrcpt                     0.6952  0.0796   8.7313  <.0001   0.5232   0.8672  ***
participantsyounger kids   -0.4279  0.1501  -2.8513  0.0136  -0.7521  -0.1037    *

---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1


I ran a meta regression for my meta-analysis. I divided the participants into younger kids and older kids and the results showed that both groups are significant as the p-value is less than 0.05. How should I report this data?

If I understood it correctly, the intrcpt value is for the older kids group. Is this correct?

• If you specified it as you claim, then yes, the intercept is the base value for the older kids. For the younger kids the estimate is intrcpt + participantsyounger kids. – user2974951 Jul 11 at 6:14
• My interpretation is simply that you estimates correctly and beyond random variability the effect of your study (I guess a treatment) in older subjects... This is what the intercept tells you... – Joe_74 Jul 11 at 6:46
• @user2974951Thank you for your reply! How about the p-value in older kids group? Where can I find this information? Thanks! – M. L Jul 12 at 8:02

Because you coded older and younger kids as a factor, lm automatically created dummy variables. The first level of your factor is older kids, this is your reference level, and is estimated in the intercept term intrcpt, which is equal to 0.6952.

The effect of the younger kids (the second level) is estimated in the participantsyounger kids term, however, this is estimated as the difference from the reference level. To get the estimate of the younger kids you do intrcpt + participantsyounger kids.

• Thank you! How about the p-value in the older kids group? Where can I find this information? – M. L Jul 12 at 8:00
• @M.L The p-values do not need to be added, they are in the right place already, so the p-value for the older kids is <.0001 while for the younger it is 0.0136. – user2974951 Jul 12 at 8:02
• Thank you! So does it mean that both groups are statistically significant? – M. L Jul 12 at 8:14
• @M.L Yes, the older kids estimate is significantly different from zero, while the younger kids are significantly different from the older kids. – user2974951 Jul 12 at 8:15
• Thanks! So the p-value for younger kids here is not the p-value across studies; instead, it is the comparison with the intrcpt group, is it correct? – M. L Jul 12 at 8:39

The intercept value is not for the older kids group. Further,the results indcate that age (older Versus younger kids) acts as a moderator variable for the effect-size of your study. Tt is because t statistic is significant at alpha = .05

• I am afraid you are mistaken. Since the OP has declared participants as a factor the intrcpt is indeed the estimate for older children. – mdewey Jul 11 at 10:10
• If it is so, it is not a case of meta-regression. How is it possible to examine the role of a categorical variable with a single category. Either the rresults are fake or OP has not checked or copied the result/output correctly. To me it is the Latter case. – Subhash C. Davar Jul 11 at 10:29
• @Subhash C. Davar I also believe you are viewing the results incorrectly – Joe_74 Jul 12 at 10:21
• If I understood it correctly, the intrcpt value is for the older kids group. Is this correct? OP is looking for an answer ? The OP has not declared participants as a factor. – Subhash C. Davar Jul 12 at 13:14
• And a mere declaration does not mean that intercept reflects the particular variable. In a meta-regression, y = a +bX + lamda Y. Please read meta- regression to understand it comprehesively. – Subhash C. Davar Jul 12 at 13:23