Table 3 The final model of bi-directional stepwise regression analyses for L1 proficiency using the original dataset.

From: Contributions of common genetic variants to specific languages and to when a language is learned

Predictor

Gene

Major allele

Estimate

Confidence intervals

Uncorrected p

FDR corrected p

Partial eta squared

∆R2

(Intercept)

  

7.49

4.51–10.47

1.13e−06

   

Gender

  

0.38

0.15–0.61

0.001*‡

0.012*‡

0.026

0.022

Family SES

  

− 0.01

− 0.01 to − 0.00

0.008*‡

0.028*‡

0.017

0.014

rs6980093

FOXP2

A

0.12

− 0.03 to 0.26

0.110

0.121

0.006

0.003

rs1800497

DRD2

G

0.21

0.06–0.35

0.005*‡

0.025*‡

0.020

0.016

rs3765502

DCDC2

T

− 0.65

− 1.38 to 0.08

0.082

0.121

0.007

0.004

rs6940827

DCDC2

G

− 0.87

− 1.62 to − 0.13

0.022*‡

0.049*‡

0.013

0.009

rs2255526

DIP2A

A

0.15

− 0.03 to 0.34

0.106

0.121

0.006

0.004

rs6803202

ROBO1

C

− 0.11

− 0.25 to 0.04

0.162

0.162

0.005

0.002

rs9722

S100B

G

0.15

− 0.01 to 0.31

0.074

0.121

0.008

0.005

rs1087266

DCDC2

A

− 0.63

− 1.38 to 0.11

0.094

0.121

0.007

0.004

rs6456593

DCDC2

C

− 0.20

− 0.36 to − 0.04

0.012*‡

0.033*‡

0.015

0.012

  1. Gender, family SES, and three SNPs (rs1800497, rs6940827, rs6456593) are independently associated with L1 HKDSE grades. The original model included gender (Female = 1; Male = 0), music training (Yes = 1; No = 0), family SES, and the 28 hypothesized SNPs. * indicates p < 0.05 (uncorrected); ‡represents significant associations after FDR corrections for multiple comparisons.
  2. Observations: 421.
  3. R2/R2 adjusted: 0.119/0.096, p = 2.215e−07.
  4. Significant values are in bold.