Dear all,
For my thesis, I ran a regression where the independent variable consists of a three way interaction term of 3 dummy variables.
For my moderating effect, I hypothesized an inverted u-shape effect on the relationship between the dependent and independent variables.
I'm doing a diff-in-diff analysis with panel data and a contol & treated group.
I ran the following command for my main effect:
xtreg google_citations_yr2 i.after_2013_2##i.pledge_nonpledge1##i.google_nong oogle2 backward_citations claims_tot_1 nr_inventors_1 i.section_id i.date_1, re vce(robust)
and the following for the moderation:
xtreg google_citations_yr2 i.after_2013_2##i.pledge_nonpledge1##i.google_nong oogle2##c.duration_quadratic backward_citations claims_tot_1 nr_inventors_1 i.section_id i.date_1, re vce(robust)
xtreg google_citations_yr2 i.after_2013_2##i.pledge_nonpledge1##i.google_nong oogle2##c.duration backward_citations claims_tot_1 nr_inventors_1 i.section_id i.date_1, re vce(robust)
For the moderating effect (duration), I first ran the regression including the quadratic term of the moderator and later ran the same regression but with the linear moderating variable.
When I saw the results I had some trouble interpreting them:
4-way interaction with quadratic term:
Coefficient: -2.30e-08
Std, Error: 7.85e-09
z-value: -2.93
p-value: 0.003 95%
conf. interval: -3.84e-08 -7.65e-09
4-way interaction with linear term:
Coefficient:-.0002353
Std, Error: .0000727
z-value: :-3.23
p-value: 0.001
95% conf. interval: -.0003779 -.0000927
I was hoping to see one of the two moderator coefficients being positive and the other negative. However, since both are negative I do not understand how to interpret this, and what the shape of this relationship is.
I really hope someone can tell me what this means for my moderator hypothesis. If additional information is required, please let me know.
For my thesis, I ran a regression where the independent variable consists of a three way interaction term of 3 dummy variables.
For my moderating effect, I hypothesized an inverted u-shape effect on the relationship between the dependent and independent variables.
I'm doing a diff-in-diff analysis with panel data and a contol & treated group.
I ran the following command for my main effect:
xtreg google_citations_yr2 i.after_2013_2##i.pledge_nonpledge1##i.google_nong oogle2 backward_citations claims_tot_1 nr_inventors_1 i.section_id i.date_1, re vce(robust)
and the following for the moderation:
xtreg google_citations_yr2 i.after_2013_2##i.pledge_nonpledge1##i.google_nong oogle2##c.duration_quadratic backward_citations claims_tot_1 nr_inventors_1 i.section_id i.date_1, re vce(robust)
xtreg google_citations_yr2 i.after_2013_2##i.pledge_nonpledge1##i.google_nong oogle2##c.duration backward_citations claims_tot_1 nr_inventors_1 i.section_id i.date_1, re vce(robust)
For the moderating effect (duration), I first ran the regression including the quadratic term of the moderator and later ran the same regression but with the linear moderating variable.
When I saw the results I had some trouble interpreting them:
4-way interaction with quadratic term:
Coefficient: -2.30e-08
Std, Error: 7.85e-09
z-value: -2.93
p-value: 0.003 95%
conf. interval: -3.84e-08 -7.65e-09
4-way interaction with linear term:
Coefficient:-.0002353
Std, Error: .0000727
z-value: :-3.23
p-value: 0.001
95% conf. interval: -.0003779 -.0000927
I was hoping to see one of the two moderator coefficients being positive and the other negative. However, since both are negative I do not understand how to interpret this, and what the shape of this relationship is.
I really hope someone can tell me what this means for my moderator hypothesis. If additional information is required, please let me know.
