{smcl}
{* v 1.0.3 Chunsen Wu 10August2019 generate/simulate child variables for a given DAG (directed acyclic graph)}{...}
{cmd:help child}{right: ({browse "http://medical-statistics.dk/MSDS/epi/dag/dag.html":Directed acyclic graph (DAG) in Epidemiology})}
{hline}
{title:Title}
{p 4 4 2}{hi:child} {hline 2} generate/simulate a binary {it:child-variable} for a given DAG (directed acyclic graph)
{title:Syntax}
{p 8 17 2}
{cmd:child}
{it:childvar parent1var parent2var ... parent20var}
{ifin}
[{cmd:,} {it:options}
]
{synoptset 21 tabbed}{...}
{synopthdr}
{synoptline}
{syntab :Options}
{synopt :{opt base:risk(#)}}The risk of the {it:child-variable} among the baseline/unexposed group and the default value is 0.05{p_end}
{synopt :{opt p1or(odds ratio)}}Causal odds ratio from the {it:parent 1 variable} to the {it:child-variable} and the default value is 5{p_end}
{synopt :{opt p2or(odds ratio)}}Causal odds ratio from the {it:parent 2 variable} to the {it:child-variable} and the default value is 5{p_end}
{synopt :{opt ...}}{p_end}
{synopt :{opt p20or(odds ratio)}}Causal odds ratio from the {it:parent 20 variable} to the {it:child-variable} and the default value is 5{p_end}
{synoptline}
{title:Description}
{pstd}
Combining command {helpb ancestor}, {helpb child}, and {helpb childc} can simulate a dataset comprising binary and continous variables for a given DAG.
{helpb ancestor} is a command creating {it:ancestor-variables} for a given DAG.
{helpb child} is a command creating {it:child-variable} for a given DAG.
{helpb childc} is a command creating normally distributed continous {it:child-variable} for a given DAG.
{p_end}
{pstd}
At an once, the command {helpb child} can create only one {it:child variable} whose {it:parents-variables} can be up to 20. By default, the risk of the {it:child-variable} is 0.05 among the baseline/unexposed group.
{title:Options}
{phang}
{opt baserisk} specifies the risk of the {it:child variable} among the baseline/unexposed group. The default value is 0.05
{phang}
{opt p1or} specifies the causal odds ratio from the first {it:parent variable} to the {it:child variable}. The default value is 5
{phang}
{opt p2or} - {opt p20or} specifies the causal odds ratios for the second {it:parent variable} - the 20th {it:parent-variable} to the {it:child-variable}, respectively. The default odds ratios are 5
{title:Examples}
{pstd}
To simulate a dataset comprising variables in a DAG (E <- C -> O), suppose:
{phang}1. The {it:ancesor-variable} C is a binary variable and prevalence of the C is 0.05.{p_end}
{phang}2. The {it:child-variable} E is a binary variable and the risk of E among the baseline (C=0) is 0.05{p_end}
{phang}3. The causal odds ratio for C -> E is 5 {p_end}
{phang}4. The {it:child-variable} O is a binary variable and the risk of O among the baseline (C=0 & E=0) is 0.05{p_end}
{phang}5. The causal odds ratio for E -> O is 1 and for C -> O is 5, respectively{p_end}
{phang}6. We wish the population size is 10000{p_end}
{phang}{stata "clear": .clear} {p_end}
{phang}{stata "set seed 150": .set seed 150} {p_end}
{phang}{stata "ancestor C, pre1(0.05) popu(10000)": .ancestor C, pre1(0.05) popu(10000)} {p_end}
{phang}{stata "child E C, baserisk(0.05) p1or(5)": .child E C, baserisk(0.05) p1or(5)} {p_end}
{phang}{stata "child O E C, baserisk(0.05) p1or(1) p2or(5)": .child O E C, baserisk(0.05) p1or(1) p2or(5)} {p_end}
{phang}Crude analysis (without adjust for the common cause variable C){p_end}
{phang}{stata "logistic O i.E": .logistic O i.E} {p_end}
{phang}Adjusted for the common cause variable C{p_end}
{phang}{stata "logistic O i.E i.C": .logistic O i.E i.C} {p_end}
{title:More examples} click on {browse "http://medical-statistics.dk/MSDS/epi/dag/dag.html":her}
{title:Acknowledgment}
{pstd}
Jørn Olsen provided encouragement.
{title:References}
{phang}Sander Greenland, Judea Pearl, and James M. Robins 1999.{p_end}
{phang}{browse "https://www.ncbi.nlm.nih.gov/pubmed/9888278":Causal diagrams for epidemiologic research.}
{phang}Judea Pearl{p_end}
{phang}Causality:{it: Models, Reasoning and Inference.} Cambridge University Press, New York, NY, USA, 2nd edition, 2009.
{phang}Lash Kenneth J. Rothman, Sander Greenland, and Timothy L. Lash.{p_end}
{phang}The chapter 12: {bf:Causal Diagrams} in the Epidemiology textbook of {it: Modern Epidemiology}. Wolters Kluwer, 2008.
{phang}Miguel A. Hernán, James M. Robins{p_end}
{phang}The chapter 6: {bf:Graphical representation of causal effects} of {browse "https://www.hsph.harvard.edu/miguel-hernan/causal-inference-book/":Causal Inference Book.}
{title:Author}
{pstd}
Chunsen Wu, the University of Southern Denmark; Odense University Hospital, Denmark{break}
{browse cwu@health.sdu.dk}{break}
{browse chunsen.wu@rsyd.dk}
{title:Also see}
{p 7 14 2}
Help: {helpb ancestor}, {helpb childc}
{p_end}