Statistical analysis
Descriptive statistics were calculated for all variables. Multiple
linear regressions were performed with total and subscale scores of
the WFIRS-P, self-report SDQ and parent SDQ as the dependent
variables. Predictor variables included sex, age, gaming time, nongaming
computer time, television time and CGAS addiction score.
Regression coefficients were calculated showing the associations
between predictor variables and youth functioning as shown by the
SDQs and WFIRS-P. IAT score was subsequently substituted for theCGAS score in the regressions to check for concurrence between
the two scales. Missing values on the SDQ were imputed as per the
SDQ scoring protocol (www.sdqinfo.org). Missing WFIRS-P, IAT and
CGAS items were handled the same way. Subjects were dropped for
a specific regression if they were missing >2 subscale items, except
for the WFIRS-P ‘self’ subscale which contained only three items
and, therefore, all responses were required. This protocol resulted in
dropping one subject for the self-report SDQ and parent SDQ regressions,
and four subjects for the WFIRS-P regressions. STATA version
9.1 (Statacorp, USA) was used for the statistical analysis. Statistical
significance was set at P
Statistical analysis
Descriptive statistics were calculated for all variables. Multiple
linear regressions were performed with total and subscale scores of
the WFIRS-P, self-report SDQ and parent SDQ as the dependent
variables. Predictor variables included sex, age, gaming time, nongaming
computer time, television time and CGAS addiction score.
Regression coefficients were calculated showing the associations
between predictor variables and youth functioning as shown by the
SDQs and WFIRS-P. IAT score was subsequently substituted for theCGAS score in the regressions to check for concurrence between
the two scales. Missing values on the SDQ were imputed as per the
SDQ scoring protocol (www.sdqinfo.org). Missing WFIRS-P, IAT and
CGAS items were handled the same way. Subjects were dropped for
a specific regression if they were missing >2 subscale items, except
for the WFIRS-P ‘self’ subscale which contained only three items
and, therefore, all responses were required. This protocol resulted in
dropping one subject for the self-report SDQ and parent SDQ regressions,
and four subjects for the WFIRS-P regressions. STATA version
9.1 (Statacorp, USA) was used for the statistical analysis. Statistical
significance was set at P<0.05.
การแปล กรุณารอสักครู่..