Issue with If and If Else in correlation with tables

Issue with If and If Else in correlation with tables Aug 5, 2022 11:53:12 GMT

Quote

Post by drjonathankim on Aug 5, 2022 11:53:12 GMT

Hi all,

I have run into an issue with using the 'if' and 'if else' commands to define what a table contains, and I was hoping that someone would be able to provide insight. What I want to do is, using the 'var in' command, bring a per-participant numeric value into an experiment and use this to define which table of stimuli an individual will see. The 'var in' side is working perfectly, the issue I'm having is specifically in how to define a table with 'if' and 'if else' statements. It seems that the code wishes for if else statements to either show participants a specific thing or to be modifiers after a table, task, or block is set. As an example for tables, if I had the code

table 1
if &stimulus == 1
one two
fi

then it would take this as being a table composed of four, two, and one column, and will therefore throw up the error 'ERROR: table_unequal_rows in line: 3'. Similarly, while a simple 'Include tableX.txt' command works, using 'if else' statements to include a specific table returns the same error.

As an example for tasks, if I code the tables separately and then had the task code

task 1
if &stimulus == 1
table test1
else
if &stimulus == 2
table test2
fi

then the inclusion of multiple lines referencing a table throws up the error 'ERROR: multiple_tables in line: 6'.

And as an example for blocks, if I code the tasks separately and then had the block code

block 1
tasklist
if &stimulus == 1
task1 1
else
if &stimulus == 2
task2 1
fi
end

then it sees 'if' as the task name and throws up the error 'ERROR: non_existing_task_in_block in line: 3... the task 'if' has never been defined as a task.' If instead I had the code

block instructions
if &stimulus == 1
tasklist
task1 1
else
if &stimulus == 2
tasklist
task2 1
fi
end

then it doesn't recognise the 'if' statement, returning 'ERROR: unknown_statement in line: 2'. There is always the option to instead make a huge number of experiments that differ only in which table is attached, but I would like to avoid that if possible so as to not have my UI filled with a ton of experiments that each were only used once. Would anyone have any insight into how to deal with this?

Issue with If and If Else in correlation with tables Aug 23, 2022 8:09:53 GMT

Quote

Post by drjonathankim on Aug 23, 2022 8:09:53 GMT

As an update on this topic, if someone was trying to do something similar to what we were attempting, Dr. Stoet reminded us of the ability to use a 'mega-table' and use a global variable set by participant to define what rows should be shown to the participant. We used the counterbalancing variable, psy_group, to do this, although this was a little complex since the survey code doesn't seem to properly register 'psy_group' as a variable unless you redefine it as itself with the code. We also set the counterbalancing 'number' to be the maximum amount of participants we wanted, set each 'group size' to 1, and created a mega-table that had all of the pseudorandomised per-participant tables organised into a single mega-table. Our redefining psy_group as itself was as follows

l: psy_group
 t: set
  - sum $psy_group

which, in order to not interfere with psy_group, we followed with the code

l: Part_Nr      
t: set
- calc ($psy_group *1)

Then in the experiment, under options we included the code

options
  var in Part_Nr

Then, in a block where we were presenting sixty independent table lines per participant, we included the code

block example
  set &local1 expression (&Part_Nr -1) * 60 +1 #the '60' is affected by the number of trials; if 10 trials in your experimental block, it would be '*10 +1'
  tasklist
   test

And finally, in the task that the block referred to, we included the code

task test
  table megatable
   tablerow &local1
  #task-related code
  set &local1 increase 1

It is worth noting that with this solution, since the 'local1' variable is set at the start of a block based off the per-participant variable, you can use the same code with multiple experimental blocks that rely on the same table. It's also worth noting that this approach doesn't allow for the presentation of the rows in a random order, so randomisation (or pseudorandomisation) has to occur during the creation of your mega-table.