RE: st: RE: simulation: finding the winner, and generating unique values

[Kieran McCaul <kieran.mccaul@uwa.edu.au>]

...

I think you have the variables created in the wrong order; you need to create the volunteers and then assign the questions.
The following creates 20 records, 4 per volunteer.
Then it assigns the questions, 1 to 4, in random order to each volunteer.

I don't see how this simulates anything however, because the answer - correct - is assigned independently, so there will be no association between the order of the questions and the correct answer.

clear
set seed 12358
set obs 20

gen volunteer = mod(_n,5) + 1

sort volunteer

by volunteer: gen byte question = _n

gen rnumb = uniform()
sort volunteer rnumb

gen byte correct = uniform()>=0.5








-----Original Message-----
From: owner-statalist@hsphsun2.harvard.edu [mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of Michael McCulloch
Sent: Saturday, 25 May 2013 11:10 PM
To: statalist@hsphsun2.harvard.edu
Subject: Re: st: RE: simulation: finding the winner, and generating unique values

Thank you Kieran,
What you suggest is exactly what I intended. This is a simulation based on an educational testing project, in which a sequence of questions (denoted by variable A), are posed to five volunteers (denoted by variable B) in sequence (hence needing 1-5 for each level of A). And, they will each get the question correct, or not (hence 0 or 1 for C). What you suggest is perfect!

I have two other questions on which to ask your coding advice:
1. The sequence in which volunteers answer the question varies. Thus, I wish to randomly assign the sort order of 1-5, within each level of A.

2. Each volunteer will encounter many questions within a day's work, and I wish to test whether being the first to address a question has an effect on success. Thus, I wish to count, (and test if possible) the total number of correct answers each volunteer gets in a day's work, and whether who goes first (or second, and so on...) influences performance.

Any further advice would be much appreciated, as I appreciate what you've already suggested.

Best wishes,
Michael McCulloch

--
Pine Street Foundation, since 1989
124 Pine Street | San Anselmo | California | 94960-2674  
P: (415) 407-1357 | F: (206) 338-2391 | http://www.PineStreetFoundation.org


On May 24, 2013, at 8:40 PM, Kieran McCaul <kieran.mccaul@uwa.edu.au> wrote:

> ...
> 
> I'm not quite sure what the intent of the simulation is.
> 
> 1. How can I simulate B so that only unique values of 1-5 are created?
> 
> I don't understand what you mean here, particularly given that the code you use is under a comment "* simulate variable: B (values 1-5 sampled without replacement) ".
> 
> There are 20 observations, 5 per level of "a".  The only way that "b" can be assigned the values 1 to 5 without replacement is by assigning them within "a".
> 
> I don't think this is what you are trying to do, but if it is, the following will do that:
> 
> 
> clear
> set seed 12358
> * simulate A
> set obs 20
> 
> gen byte a = mod(_n,4) + 1
> 
> sort a
> by a:gen byte b=_n
> 
> gen byte c = uniform()>=0.5
> 
> 
> 
> 
> 
> -----Original Message-----
> From: owner-statalist@hsphsun2.harvard.edu [mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of Michael McCulloch
> Sent: Saturday, 25 May 2013 3:31 AM
> To: statalist@hsphsun2.harvard.edu
> Subject: st: simulation: finding the winner, and generating unique values
> 
> Greetings!
> 
> * I have three questions related to the following data simulation.
> * 1. How can I simulate B so that only unique values of 1-5 are created?
> * 2. Is there a more elegant way to code simulation of variable B? 
> * 3. How can I identify which variable B had the highest total count of C, within levels of A?
> 
> clear
> set seed 12358
> * simulate A
> set obs 20        //create 20 observations
> gen    a=1 in 1/5
> replace a=2 in 6/10
> replace a=3 in 11/15
> replace a=4 in 16/20
> 
> * simulate variable: B (values 1-5 sampled without replacement) gen b=.
> replace b=(1 + int(5*runiform())) if a==1    
> replace b=(1 + int(5*runiform())) if a==2    
> replace b=(1 + int(5*runiform())) if a==3    
> replace b=(1 + int(5*runiform())) if a==4    
> 
> * simulate variable: C
> gen c1=uniform()    //Generate n (equal chance of 0-1)
>    gen c=.
>    replace c=0 if c1<0.5
>    replace c=1 if c1>=0.5
>    drop c1
> list, noobs
> 
> 
> 
> Best wishes,
> Michael McCulloch, LAc MPH PhD
> 
> --
> Pine Street Foundation, since 1989
> 124 Pine Street | San Anselmo | California | 94960-2674
> P: (415) 407-1357 | F: (206) 338-2391 | http://www.PineStreetFoundation.org
> 
> 
> *
> *   For searches and help try:
> *   http://www.stata.com/help.cgi?search
> *   http://www.stata.com/support/faqs/resources/statalist-faq/
> *   http://www.ats.ucla.edu/stat/stata/
> 
> *
> *   For searches and help try:
> *   http://www.stata.com/help.cgi?search
> *   http://www.stata.com/support/faqs/resources/statalist-faq/
> *   http://www.ats.ucla.edu/stat/stata/

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/