Compare two designs
compare_designs( design1, design2, format = "ansi8", pager = "off", context = -1L, rmd = FALSE ) compare_design_code( design1, design2, format = "ansi256", mode = "sidebyside", pager = "off", context = -1L, rmd = FALSE ) compare_design_summaries( design1, design2, format = "ansi256", mode = "sidebyside", pager = "off", context = -1L, rmd = FALSE ) compare_design_data( design1, design2, format = "ansi256", mode = "sidebyside", pager = "off", context = -1L, rmd = FALSE ) compare_design_estimates( design1, design2, format = "ansi256", mode = "auto", pager = "off", context = -1L, rmd = FALSE ) compare_design_estimands( design1, design2, format = "ansi256", mode = "sidebyside", pager = "off", context = -1L, rmd = FALSE )
Arguments
| design1 | A design object, typically created using the + operator |
|---|---|
| design2 | A design object, typically created using the + operator |
| format | Format (in console or HTML) options from |
| pager | Pager option from |
| context | Context option from |
| rmd | Set to |
| mode | Mode options from |
Examples
design1 <- declare_population(N = 100, u = rnorm(N)) + declare_potential_outcomes(Y ~ Z + u) + declare_estimand(ATE = mean(Y_Z_1 - Y_Z_0)) + declare_sampling(n = 75) + declare_assignment(m = 50) + reveal_outcomes(Y, Z) + declare_estimator(Y ~ Z, estimand = "ATE") design2 <- declare_population(N = 200, u = rnorm(N)) + declare_potential_outcomes(Y ~ 0.5*Z + u) + declare_estimand(ATE = mean(Y_Z_1 - Y_Z_0)) + declare_sampling(n = 100) + declare_assignment(m = 25) + reveal_outcomes(Y, Z) + declare_estimator(Y ~ Z, model = lm_robust, estimand = "ATE") compare_designs(design1, design2)#>#> Research design comparison #> #> #> #> # Compare design code ----------------------------------------------------------- #> #> < design1 > design2 #> @@ 1,14 @@ @@ 1,14 @@ #> p #> opulation population #> < "declare_population(N = 100, u = r > "declare_popula #> : norm(N))" : tion(N = 200, u = rnorm(N))" #> potential #> _outcomes potential_outcomes #> < "declare_potential_outcomes(Y > "declare_potentia #> : ~ Z + u)" : l_outcomes(Y ~ 0.5 * Z + u)" #> #> ATE ATE #> "declare_estimand(ATE = mean(Y_Z_1 - "declare_estimand #> Y_Z_0))" (ATE = mean(Y_Z_1 - Y_Z_0))" #> #> sampling sampling #> < "declare_sampling > #> : (n = 75)" : "declare_sampling(n = 100)" #> a #> ssignment assignment #> < "declare_assignment > #> : (m = 50)" : "declare_assignment(m = 25)" #> #> reveal reveal #> "reveal_outcom #> es(Y, Z)" "reveal_outcomes(Y, Z)" #> #> estimator estimator #> < "declare_estimator(Y ~ Z, estimand = > "declare_estimator(Y ~ Z, model = lm_ #> : \"ATE\")" : robust, estimand = \"ATE\")" #> #> #> # Compare draw_data(design) ----------------------------------------------------- #> #> < str(design1) > str(design2) #> @@ 1,9 @@ @@ 1,9 @@ #> < 'data.frame': 75 obs. of 8 variabl > 'data.frame': 100 obs. of 8 variab #> : es: : les: #> < $ ID : chr "001" "002" > $ ID : chr "004" "005" #> : "004" "005" ... : "007" "008" ... #> < $ u : num -0.294 -0.5 > $ u : num -0.143 0.22 #> : 63 -0.143 0.224 0.404 ... : 4 -2.819 -1.315 1.696 ... #> < $ Y_Z_0 : num -0.294 -0.5 > $ Y_Z_0 : num -0.143 0.22 #> : 63 -0.143 0.224 0.404 ... : 4 -2.819 -1.315 1.696 ... #> < $ Y_Z_1 : num 0.706 0.437 > $ Y_Z_1 : num 0.357 0.724 #> : 0.857 1.224 1.404 ... : -2.319 -0.815 2.196 ... #> < $ S_inclusion_prob: num 0.75 0.75 0 > $ S_inclusion_prob: num 0.5 0.5 0.5 #> : .75 0.75 0.75 0.75 0.75 0.75 0.75 0.7 : 0.5 0.5 0.5 0.5 0.5 0.5 0.5 ... #> : 5 ... ~ #> < $ Z : int 1 1 0 0 1 1 > $ Z : int 1 0 0 0 0 1 #> : 1 0 0 0 ... : 0 1 0 0 ... #> < $ Z_cond_prob : num 0.667 0.667 > $ Z_cond_prob : num 0.25 0.75 0 #> : 0.333 0.333 0.667 ... : .75 0.75 0.75 0.25 0.75 0.25 0.75 0.7 #> ~ : 5 ... #> < $ Y : num 0.706 0.437 > $ Y : num 0.357 0.224 #> : -0.143 0.224 1.404 ... : -2.819 -1.315 1.696 ... #> #> #> # Compare draw_estimands(design) ------------------------------------------------ #> #> < design1 > design2 #> @@ 1,2 @@ @@ 1,2 @@ #> estimand_label estimand estimand_label estimand #> < 1 ATE 1 > 1 ATE 0.5 #> #> #> # Compare draw_estimates(design) ------------------------------------------------ #> #> < design1 #> > design2 #> @@ 1,4 / 1,4 @@ #> estimator_label term estimate std.error statistic p.value conf.low #> < 1 estimator Z 1.023718 0.2279354 4.491265 4.265621e-05 0.5657591 #> > 1 estimator Z 0.6078596 0.2366002 2.569142 0.01170256 0.1383341 #> conf.high df outcome estimand_label #> < 1 1.481678 49.40125 Y ATE #> > 1 1.077385 98 Y ATEcompare_design_code(design1, design2)#> < design1 > design2 #> @@ 1,14 @@ @@ 1,14 @@ #> p #> opulation population #> < "declare_population(N = 100, u = r > "declare_popula #> : norm(N))" : tion(N = 200, u = rnorm(N))" #> potential #> _outcomes potential_outcomes #> < "declare_potential_outcomes(Y > "declare_potentia #> : ~ Z + u)" : l_outcomes(Y ~ 0.5 * Z + u)" #> #> ATE ATE #> "declare_estimand(ATE = mean(Y_Z_1 - "declare_estimand #> Y_Z_0))" (ATE = mean(Y_Z_1 - Y_Z_0))" #> #> sampling sampling #> < "declare_sampling > #> : (n = 75)" : "declare_sampling(n = 100)" #> a #> ssignment assignment #> < "declare_assignment > #> : (m = 50)" : "declare_assignment(m = 25)" #> #> reveal reveal #> "reveal_outcom #> es(Y, Z)" "reveal_outcomes(Y, Z)" #> #> estimator estimator #> < "declare_estimator(Y ~ Z, estimand = > "declare_estimator(Y ~ Z, model = lm_ #> : \"ATE\")" : robust, estimand = \"ATE\")"compare_design_summaries(design1, design2)#> < design1 > design2 #> @@ 1,106 @@ @@ 1,108 @@ #> #> Design Summary Design Summary #> #> < Step 1 (population): declare_populati > Step 1 (population): declare_populati #> : on(N = 100, u = rnorm(N)) ----------- : on(N = 200, u = rnorm(N)) ----------- #> : ------ : ------ #> #> < N = 100 > N = 200 #> #> Added variable: ID Added variable: ID #> N_missing N_unique class N_missing N_unique class #> < 0 100 character > 0 200 character #> #> Added variable: u Added variable: u #> min median mean max sd N_missin min median mean max sd N_missing #> g N_unique N_unique #> < -3.04 0 -0.08 2.5 1.02 > -3.04 -0.02 0.02 2.5 0.98 0 #> : 0 100 : 200 #> #> < Step 2 (potential outcomes): declare_ > Step 2 (potential outcomes): declare_ #> : potential_outcomes(Y ~ Z + u) ------- : potential_outcomes(Y ~ 0.5 * Z + u) - #> : ------ : ------ #> #> < Formula: Y ~ Z + u > Formula: Y ~ 0.5 * Z + u #> #> Added variable: Y_Z_0 Added variable: Y_Z_0 #> min median mean max sd N_missin min median mean max sd N_missing #> g N_unique N_unique #> < -3.04 0 -0.08 2.5 1.02 > -3.04 -0.02 0.02 2.5 0.98 0 #> : 0 100 : 200 #> #> Added variable: Y_Z_1 Added variable: Y_Z_1 #> min median mean max sd N_missing min median mean max sd N_missing #> N_unique N_unique #> < -2.04 1 0.92 3.5 1.02 0 > -2.54 0.48 0.52 3 0.98 0 #> : 100 : 200 #> #> Step 3 (estimand): declare_estimand(A Step 3 (estimand): declare_estimand(A #> TE = mean(Y_Z_1 - Y_Z_0)) ----------- TE = mean(Y_Z_1 - Y_Z_0)) ----------- #> ------ ------ #> #> A single draw of the estimand: A single draw of the estimand: #> estimand_label estimand estimand_label estimand #> < ATE 1 > ATE 0.5 #> #> < Step 4 (sampling): declare_sampling(n > Step 4 (sampling): declare_sampling(n #> : = 75) ------------------------------ : = 100) ----------------------------- #> : ------ : ------ #> #> < N = 75 (25 subtracted) > N = 100 (100 subtracted) #> #> Added variable: S_inclusion_prob Added variable: S_inclusion_prob #> < 0.75 > 0.5 #> < 75 > 100 #> 1.00 1.00 #> #> Altered variable: ID Altered variable: ID #> Before: Before: #> N_missing N_unique class N_missing N_unique class #> < 0 100 character > 0 200 character #> #> After: After: #> N_missing N_unique class N_missing N_unique class #> < 0 75 character > 0 100 character #> #> Altered variable: u Altered variable: u #> Before: Before: #> min median mean max sd N_missin min median mean max sd N_missing #> g N_unique N_unique #> < -3.04 0 -0.08 2.5 1.02 > -3.04 -0.02 0.02 2.5 0.98 0 #> : 0 100 : 200 #> #> After: After: #> min median mean max sd N_missing min median mean max sd N_missing #> N_unique N_unique #> < -3.04 0.14 -0.04 2.5 1 0 > -3.04 0.12 0.01 2.5 0.98 0 #> : 75 : 100 #> #> Altered variable: Y_Z_0 Altered variable: Y_Z_0 #> Before: Before: #> min median mean max sd N_missin min median mean max sd N_missing #> g N_unique N_unique #> < -3.04 0 -0.08 2.5 1.02 > -3.04 -0.02 0.02 2.5 0.98 0 #> : 0 100 : 200 #> #> After: After: #> min median mean max sd N_missing min median mean max sd N_missing #> N_unique N_unique #> < -3.04 0.14 -0.04 2.5 1 0 > -3.04 0.12 0.01 2.5 0.98 0 #> : 75 : 100 #> #> Altered variable: Y_Z_1 Altered variable: Y_Z_1 #> Before: Before: #> min median mean max sd N_missing min median mean max sd N_missing #> N_unique N_unique #> < -2.04 1 0.92 3.5 1.02 0 > -2.54 0.48 0.52 3 0.98 0 #> : 100 : 200 #> #> After: After: #> min median mean max sd N_missing N min median mean max sd N_missing #> _unique N_unique #> < -2.04 1.14 0.96 3.5 1 0 > -2.54 0.62 0.51 3 0.98 0 #> : 75 : 100 #> #> < Step 5 (assignment): declare_assignme > Step 5 (assignment): declare_assignme #> : nt(m = 50) -------------------------- : nt(m = 25) -------------------------- #> : ------ : ------ #> #> Added variable: Z Added variable: Z #> 0 1 0 1 #> < 25 50 > 75 25 #> < 0.33 0.67 > 0.75 0.25 #> #> Added variable: Z_cond_prob Added variable: Z_cond_prob #> < 0.333333333333333 0.666666666666667 > 0.25 0.75 #> < 25 50 > 25 75 #> < 0.33 0.67 > 0.25 0.75 #> #> Step 6 (reveal): reveal_outcomes(Y, Z Step 6 (reveal): reveal_outcomes(Y, Z #> ) ----------------------------------- ) ----------------------------------- #> ------ ------ #> #> Added variable: Y Added variable: Y #> min median mean max sd N_missing min median mean max sd N_missing #> N_unique N_unique #> < -2.04 0.66 0.63 3.5 1.07 0 > -3.04 0.22 0.13 3 1.02 0 #> : 75 : 100 #> #> < Step 7 (estimator): declare_estimator > Step 7 (estimator): declare_estimator #> : (Y ~ Z, estimand = "ATE") ----------- : (Y ~ Z, model = lm_robust, estimand = #> : ------ : "ATE") #> #> Formula: Y ~ Z Formula: Y ~ Z #> ~ > #> ~ > Model: lm_robust #> #> A single draw of the estimator: A single draw of the estimator: #> estimator_label term estimate std.e estimator_label term estimate std.e #> rror statistic p.value conf.low rror statistic p.value conf.low #> < estimator Z 0.7892986 0.212 > estimator Z 0.6078596 0.236 #> : 3301 3.717319 0.0004051111 0.3657276 : 6002 2.569142 0.01170256 0.1383341 #> conf.high df outcome estimand_ conf.high df outcome estimand_label #> label ~ #> < 1.21287 69.14596 Y > 1.077385 98 Y ATE #> : ATE ~ #>compare_design_data(design1, design2)#> < str(design1) > str(design2) #> @@ 1,9 @@ @@ 1,9 @@ #> < 'data.frame': 75 obs. of 8 variabl > 'data.frame': 100 obs. of 8 variab #> : es: : les: #> < $ ID : chr "001" "002" > $ ID : chr "004" "005" #> : "003" "004" ... : "007" "008" ... #> < $ u : num -0.561 -0.6 > $ u : num -0.143 0.22 #> : 1.004 -1.237 0.19 ... : 4 -2.819 -1.315 1.696 ... #> < $ Y_Z_0 : num -0.561 -0.6 > $ Y_Z_0 : num -0.143 0.22 #> : 1.004 -1.237 0.19 ... : 4 -2.819 -1.315 1.696 ... #> < $ Y_Z_1 : num 0.439 0.4 2 > $ Y_Z_1 : num 0.357 0.724 #> : .004 -0.237 1.19 ... : -2.319 -0.815 2.196 ... #> < $ S_inclusion_prob: num 0.75 0.75 0 > $ S_inclusion_prob: num 0.5 0.5 0.5 #> : .75 0.75 0.75 0.75 0.75 0.75 0.75 0.7 : 0.5 0.5 0.5 0.5 0.5 0.5 0.5 ... #> : 5 ... ~ #> < $ Z : int 1 1 0 0 1 1 > $ Z : int 1 0 0 0 0 1 #> : 1 1 1 1 ... : 0 1 0 0 ... #> < $ Z_cond_prob : num 0.667 0.667 > $ Z_cond_prob : num 0.25 0.75 0 #> : 0.333 0.333 0.667 ... : .75 0.75 0.75 0.25 0.75 0.25 0.75 0.7 #> ~ : 5 ... #> < $ Y : num 0.439 0.4 1 > $ Y : num 0.357 0.224 #> : .004 -1.237 1.19 ... : -2.819 -1.315 1.696 ...compare_design_estimates(design1, design2)#> < design1 #> > design2 #> @@ 1,4 / 1,4 @@ #> estimator_label term estimate std.error statistic p.value conf.low #> < 1 estimator Z 1.023718 0.2279354 4.491265 4.265621e-05 0.5657591 #> > 1 estimator Z 0.6078596 0.2366002 2.569142 0.01170256 0.1383341 #> conf.high df outcome estimand_label #> < 1 1.481678 49.40125 Y ATE #> > 1 1.077385 98 Y ATEcompare_design_estimands(design1, design2)#> < design1 > design2 #> @@ 1,2 @@ @@ 1,2 @@ #> estimand_label estimand estimand_label estimand #> < 1 ATE 1 > 1 ATE 0.5