Hello,
First some information about the contexts: You are able to access
every variable of every context from everywhere in the experiment.
SoPHIE's variable contexts are "only" made to keep a neat and tidy
order.
For example each participant should have a payoff variable in
which to save how much money (s)he gets from the study. If SoPHIE
would not support variable contexts, you would need variables
"payoffParticipant1", "payoffParticipant2" etc. By using PE
(Participant Everywhere) variables, you only need to have a
variable "payoff" which can have different values for different
participants, by giving the participant's label in the variable
API, you can access other participant's payoff variables. Next to
this "participant" context, the second dimension in the variable
context is the "process" context. This can be used e.g. to store
the several choices of a participant in several rounds of the same
setup. The (PSL / participant stepgroup loop) variable will can be
called "choice" and accessed by giving the participant label,
stepgroup label and stepgroup loop.
Please have a look at the API documentation:
http://www.sophie.uni-osnabrueck.de/docs/apis/latest/sophie_variable_1_0_0.html
and the wiki:
http://www.sophie.uni-osnabrueck.de/wiki/pmwiki.php?n=SoPHIE.Variables
Here you will get an overview of all possible methods of the
variable api with all parameters and default values. If you don't
give all the parameters, SoPHIE will "magically" use the current
participant / group / stepgroup / stepgroup loop.
Back to your question at hand:
You need only one step to make the choice (task 1) and a second
step to guess the other participant's value (task 2). Possibly it
makes sense to add some sync steps to make sure that all
participants have completed task 1 when continuing with task 2.
Both tasks are independent, but the result page is not. Here you
can retrieve any variables of any participants and compare them
with each other.
I would suggest that you create two participant types (A and B)
and a group structure so that each group has one "A" and one "B".
When creating a session type you would create a session type with
two groups, so you will have four participants. Then you can
prepare the matching of who will interact with whom. By having
only two participants in each group you can simply use the special
label '%partner%'
(
http://www.sophie.uni-osnabrueck.de/wiki/pmwiki.php?n=SoPHIE.API#Special_Labels)
to retrieve the partner's input value. The drawback is that in
this (simple) design, the participants will always guess their
partner's choice (A guesses B and vice versa). But I think it is
best to live with this disadvantage at the beginning. After you
are firm with this, you can go to the next level and implement a
more complex matching and design in which there are no
self-assessing pairs.
Hope this helps!
Regards,
Philipp
--
Philipp Mamat
Symbic GmbH
Persönlicher Kontakt
Telefon: +49 (0) 541 / 50 79 97-18 | Telefax: -19
Mobil: +49 (0) 176 / 800 144 65 | Mail: philip...@symbic.de
___________________________________________________________
Symbic GmbH | www.symbic.de
Im ICO | Albert-Einstein-Str. 1 | DE-49076 Osnabrück
Sitz der Gesellschaft: Marktstr. 14 | DE-38315 Hornburg
USt.-IdNr.: DE269573411
Eingetragen im Amtsgericht Braunschweig: HRB 202239
Geschäftsführer: Achim Hendriks, Philipp Mamat