Millisecond Forums

Thanks for clarification!

Latency measurement starts with the 1st (relevant) display frame Inquisit can sync to. /ontrialbegin logic is executed before that; hence the result of the calculation in expressions.timespentonpage will be slightly larger. Whether you analyze latency or the expression's result is largely a matter of preference, IMO. Both will be highly correlated.

Thank you Dave. This code works perfectly! 

One more question: in my data files I have two variables: latency and expressions.timespentonpage. They show similar results but there's always a few ms difference that usually varies between 10 and 100ms, e.g. latency: 3438 vs. expressions.timespentonpage: 3506, or latency:49226 vs.expressions.timespentonpage:49260. Where's that difference coming from? And which variable should I analyze in this case?

Side note for the future novice users: make sure that you use /finishlabel command not /nextlabel one (this one doesn't work here).

You cannot really do this using *instruction* pages (i.e., <page> or <htmlpage> elements displayed via /pre- and /postinstructions attributes). With <surveypage>s, you can do something like this:

<block instructionsblock>
/ trials = [1=page1; 2=page2;]
</block>

<surveypage page1>
/ caption = "Page 1"
/ ontrialbegin = [values.t_start=script.elapsedtime;]
/ ontrialend = [values.t_end=script.elapsedtime;]
/ showpagenumbers = false
/ finishlabel = "Next"
</surveypage>

<surveypage page2>
/ caption = "Page 2"
/ ontrialbegin = [values.t_start=script.elapsedtime;]
/ ontrialend = [values.t_end=script.elapsedtime;]
/ showpagenumbers = false
/ finishlabel = "Next"
</surveypage>

<values>
/ t_start = 0
/ t_end = 0
</values>

<expressions>
/ timespentonpage = (values.t_end-values.t_start)
</expressions>

<data>
/ columns = [date time subject blocknum blockcode trialnum trialcode response latency correct expressions.timespentonpage]
/ separatefiles = true
</data>

Note: You need to run your surveypages via a <block>.