Collecting Data with Free Software

1. Collecting Data with Free Software Henrik Singmann

2. Free Software (in the GNU sense) • Free Software isfirstof all not onlyfreeofcharge but givesyouthefreedomto do whatyouwantwithyoursoftware: • Freedom 0: The freedom to run the program for any purpose. • Freedom 1: The freedom to study how the program works, and change it to make it do what you wish. • Freedom 2: The freedom to redistribute copies so you can help your neighbor. • Freedom 3: The freedom to improve the program, and release your improvements (and modified versions in general) to the public, so that the whole community benefits.

3. Data Collection • Responses from a limited numberofparticipantsto a limited numberofitems • Items constructedtoanswertheresearchquestions • Experimental research: participantsand/oritemsarerandomlyassigned • Items sharesimilarities: e.g., item orreponseformat, topic, layout… • Further informationfromparticipants: e..g., demographicinformation, priorknowledgeregardingtheresearchquestion, … • Variouswaystocollect such data: • Paper Questionnaires • Web Based Experiments • Computerized Lab Experiments

4. Data Collection • Responses from a limited numberofparticipantsto a limited numberofitems • Items constructedtoanswertheresearchquestions • Experimental research: participantsand/oritemsarerandomlyassigned • Items sharesimilarities: e.g., item orreponseformat, topic, layout… • Further informationfromparticipants: e..g., demographicinformation, priorknowledgeregardingtheresearchquestion, … • Variouswaystocollect such data: • Paper Questionnaires • Web Based Experiments • Computerized Lab Experiments

5. General Structureof an Experiment Most experiments follow the following general structure: • Greeting, introduction and instruction to the experiment and task • Main task: repeatedly working on similar items • Demographics, debriefing, … Depending on the design, part 2 may consist of multiple blocks or different tasks intermixed with additional instructions.

6. Different typesof Items Two "types" ofstimuli/items: • Items whereexactpresentationandexacttimingisimportant. e.g.: • responsetimes, eyetracking, orsimilardependent variables • primingresearch (presentingstimuliverybrief) • psychophysicalresearch (randomdots) • Items thatcontain a lotof (semantic) informationandtheresponseshouldreflectdeliberatecognitiveprocesses.

7. Different typesof Items Two "types" ofstimuli/items: • Items whereexactpresentationandexacttimingisimportant. e.g.: • responsetimes, eyetracking, orsimilardependent variables • priming Research (presentingstimuliverybrief) • psychophysicalresearch (randomdots) • Items thatcontain a lotof (semantic) informationandtheresponseshouldreflectdeliberatecognitiveprocesses.

8. Example Experiment

9. PsychoPy • Free softwarefor "the presentation of stimuli and collection of data for a wide range of neuroscience, psychology and psychophysics experiments". • Mainly developed for presenting "type 1" stimuli, but I will show how to easily present "type 2" stimuli. • PsychoPy uses the Python programming language: a free "general-purpose, interpreted high-level programming language whose design philosophy emphasizes code readability" (wikipedia)

10. PsychoPy: Example "Type 1" Stimuli

11. PsychoPy: Interfaces PsychoPyhastwo different interfaces: • The builderview: Build an experimentusing a graphicaluserinterface (i.e., via clickandplay) • The coderview: Build an experimentusing Python code. PsychoPyisverygood in controllingthemaintaskof an experiment but not so comfortablewhenitcomesaboutinstructionscreensandcollectingdemographicdata.

12. The Builder View

13. The Builder View • Goodforitemswhichconsistof a smallnumberofsingleelements (i.e., not toomuchtext). • The builderviewcanbeusedtoeasilydevelopclassicalpsychological "type 1" paradigms such as e.g., theStrooptask.

15. Goodfordoingwhateveryouwantto do andis not possiblewithothermethods. • We will learnhowtousethecoderviewtocontrolthestructureoftheexperiment, collectthedata, presentthestimuli (usingadd-on PsyTML), and do all necessary administrative stuff. • We will design the ("type 2") items in htmlusing an external easy touse WYSIWYG editor.

16. Designing "Type 2" Stimuli

17. htmlForm Elements

18. PsyTML • add-on forPsychoPydevelopedbyme • presentshtmlpageandcollectsthestatusof form elements after clickingtheSubmitbutton • htmlformsmaycontainany type ofhtml form element • data will behanded back to Python • See: https://github.com/singmann/psytml

19. Creating HTML Pages • KompoZeris a free WYSIWYG htmleditor (http://www.kompozer.net/) • Layouting in htmlisdoneusingtableswithborder = 0 Steps in creating a htmlpage: • Create a 1 celltablewithwidthof e.g., 800 pixelcentrally. • Create a form insidethetableusingmethod "get". • Place all elementsinsidethe form (headline, text, form elements, …) • Usetable (andtablesinsidetables) fornicelayouting. • Add a "Submit" button.

20. Creating HTML forms • html form elementshave a nameand a value • The nameisthe variable nameandthevaluethevalueofthis variable in Python • Ofradiobuttonswiththe same nameonlyonecanbeselected • Form elementsotherthanradiobuttonsshouldhaveuniquenames

21. A briefintroductionto HTML • Insteadofusingthe WYSIWYG editor, itis easy tochangethings in thehtmlcode. • htmlisplaintextaccompaniedwithmarkup tags. • markup tags areenclosedbwteen< >: <tag> • Usuallyyouhave an openning tag and a closing tag: • <tag> opens • </tag> closes • htmlmaylookdifficult in thebeginning, but is easy tolearn.

22. A barebonehtmlpage • A htmldocumentissplitintotwoparts: headandbody • The headcontainsmetainformationandthe title • The bodycontainswhatisvisible. <html> <head> <title>Hello HTML</title> </head> <body> <p>Hello World!</p> </body> </html>

23. Importanthtml Tags • a html form usesisenclosedby<form> and</form>(forPsyTMLmethod="get"ismandatory) • most form elementsarecreatedwith<input …> • <tr> and</tr> create a tablecolumn, <td> and</td> a cell. • <br> adds a linebreak/newline • <p> </p> encloses a paragraphoftext. • <div …> </div> isusedfortextformatting

24. Advancedhtml • General formatting (backgroundandothercolor, font type, fontsize, …) isdoneusingcascading style sheets (CSS). Add e.g. totheheader:<link rel="stylesheet" type="text/css" href="design.css"> • Tomakebuttonpressesmandatory, Javascriptisnecessary. (See theexamplesofPsyTML) • PsyTMLshouldallow Flash andotherplugins (thisbehaviorisuntested)

25. Summaryhtml • htmlformsare a simple waytoconstructinstructionscreensor "type 2" items. • htmltutorialscanbefound all overthe web, goodplacestostartare e.g.: www.w3schools.comorwww.boogiejack.com/html_tutorials.html(in German: selfhtml.org) • KompoZeris a freecomfortableeditor (but seemsunstablesometimeswhenopeningmorethanonefile)

26. Summaryhtml Files • Eachscreenand item is a singlehtmlfile(onecouldreplacecertaincontent, but …) • Instructionsandotherscreens also needtocontain a form and a Submitbutton (otherwisePsyTMLdoes not knowwhentheparticipantisdone). • Items are all ofthe same structureandcontain EXACTLY the same form elementswiththe same names. • <input … type = hidden> canbeusedtocodethe item.

27. TransformingtheExample Experiment into an own Experiment

28. The Example Experiment • PsyTMLcontainstheexampleexperiment in thefolderexamples/psychopy/ • The mainfoldercontains: • .pyfilescontainingthe Python scripts • id.lst andmake.idLst.R: forcontrollingwhichisthenextparticipant • README files • thehtmlfoldercontains: • thehtmlfilesrepresenting all screens (instruction, items, …) • thecssfilescontaininggraphicalparameters • jsval.js: a javascriptlibrarycontrollingrequireditems • an imagesfolderwithsomeimages • The datafoldercontainsthedata after runningtheexperiment

29. The exampleexperiment • PsyTMLcontainstheexampleexperiment in thefolderexamples/psychopy/ • The mainfoldercontains: • .pyfilescontainingthe Python scripts • id.lst andmake.idLst.R: forcontrollingwhichisthenextparticipant • README files • thehtmlfoldercontains: • thehtmlfilesrepresenting all screens (instruction, items, …) • thecssfilescontaininggraphicalparameters • jsval.js: a javascriptlibrarycontrollingrequireditems • an imagesfolderwithsomeimages • The datafoldercontainsthedata after runningtheexperiment

30. Buildingyourown Experiment I • Write instructionscreensand end screens. • Develop a templateofthe item in themaintask. • Makeasmanycopiesofthisfileasyouneed it. • Exchange thecontentandadaptsettheappropriatehiddenhtmlformstoidentifyeach item. • Feed thehtmlfilestoPsychoPy.

31. Buildingyourown Experiment I • Write instructionscreensand end screens. • Develop a templateofthe item in themaintask. • Makeasmanycopiesofthisfileasyouneed it. • Exchange thecontentandadaptsettheappropriatehiddenhtmlformstoidentifyeach item. • Feed thehtmlfilestoPsychoPy.

33. A verybriefIntroductionto Python I • Python has a shell (>>>) so onecaninteractwith Python directly. • Variables aredynamicallytyped (i.e., type of variables does not needtobedeclared):>>> x = 3>>> type(x)<type 'int'> • Whitespaceindentationdelimitblocks (insteadof{}):>>> if (x > 0):... print("Yes")... else:... print("no")... Yes • four spaces (and not tab) represent the indentation to bind blocks of code together.

34. A verybriefIntroductionto Python II Data types: • Numbers:>>> x = y = z = 0 # Zero x, y andz>>> x0>>> y0>>> z0>>> tax = 12.5 / 100>>> price = 100.50>>> price * tax12.5625 • Strings( use ' ' or " "):>>> word = 'Help' + 'A' >>> word 'HelpA'>>> word[0]'H'>>> word[2:4]'lp' • Unicode Strings (u' '):>>> y = u'Rüdiger'>>> yu'R\xfcdiger'

35. A verybriefIntroductionto Python III • Lists: A list of comma-separated values between square brackets, that do not have to have the same type:>>> a = ['spam', u'eggs', 100, 1234]>>> a['spam', u'eggs', 100, 1234] • Like string indices, list indices start at 0, and lists can be sliced, concatenated and so on:>>> a[1:-1][u'eggs', 100]>>> a[1:][u'eggs', 100, 1234]>>> a[2] = a[2] + 23>>> a['spam', u'eggs', 123, 1234]

36. A verybriefIntroductionto Python IV • Dictionaries: An unorderedsetofkey: valuepairsenclosedbybraces{}.>>> tel = {'jack': 4098, 'sape': 4139} >>> tel['guido'] = 4127 >>> tel{'sape': 4139, 'guido': 4127, 'jack': 4098} >>> tel['jack'] 4098

37. A verybriefIntroductionto Python V • Python containsloops (repeatedoperations):>>> y = 0>>> for x in xrange(5):... y = y + x... >>> y10 • Importantkeywords / elements in Python are: • import: importslibrary • if / else / for: controlflow • . : seperatesfunctionsfromlibraries; methods/elementsfromobjects • # : afterwordscomes a comment

38. Summary Python • Run a script in PsychoPybypressing • Terminate a scriptbypressing • PsychoPycontains a Python shell (>>>): • Work throughparts 3, 4, and 5 of "The Python Tutorial": http://docs.python.org/tutorial/index.html • The importantdatatypesarestrings (" "), unicodestrings (u" "), lists ( [ ] ), anddictionaries ( { } ) withtheircorrespondingmethods. • Python runs a scriptfromthefirstlinetothe end.

39. The exampleexperiment • PsyTMLcontainstheexampleexperiment in thefolderexamples/psychopy/ • The mainfoldercontains: • .pyfilescontainingthe Python scripts • id.lst andmake.idLst.R: forcontrollingwhichisthenextparticipant • README files • thehtmlfoldercontains: • thehtmlfilesrepresenting all screens (instruction, items, …) • thecssfilescontaininggraphicalparameters • jsval.js: a javascriptlibrarycontrollingrequireditems • an imagesfolderwithsomeimages • The datafoldercontainsthedata after runningtheexperiment

40. The exampleexperiment • PsyTMLcontainstheexampleexperiment in thefolderexamples/psychopy/ • The mainfoldercontains: • .pyfilescontainingthe Python scripts • id.lst andmake.idLst.R: forcontrollingwhichisthenextparticipant • README files • PsychoPy1.py containsthecodeoftheexampleexperiment. • PsyTML.py containsthecodethatdisplaysthehtmlpages. • helper.py containssomehelperfunctions (e.g., forwritingthedata).

41. PsychoPy1.py Besidesnecessarytechnicalstuff, thefilecontains: • Parameters fortheexperiment such asnameandpathofthedatafolder. • List ofitemsforthemaintask (splitforthetwoconditions) • Presentationofinstructions, itemsand end screen. • Functionswritingthedata.

42. Data Examples

43. Howdoesthe Data looklike? • Internallysaved in a listofdictionnarieswiththe same keys (dictionarykeysarethenamesofthehtmlforms) • helper.writeTrials() writesdatalisttothedatafolder in a singlefile per participant (filenamecontainsexperimentnameandid) seperatingvaluesbyargumentsep (defaultseperatoristab). • Eachtrialoccupiesonerow in thedatafile. • theheaderargumentdefineswhich variables shouldbewritten (andthe order ofthe variables.)

44. Demographic Data • Demographicdataof all participantsiswrittento a singlefile. • Eachrowcorrespondstooneparticipantandeachnewparticipantissimplyappendedtothefile. • Comments oftheparticipantsareaddedto a commentsfolder.

45. Data Examples

46. HowtodetermineParticipant ID? • id.lst is a textfilecontainingparticipantnumberandcondition, one per row • * indicatesthattheidhasbeenused • helper.getIdFile()triesto open id.lst andreturnsthenextidandconditionandmarksitwith * • Whenrunningexperiments on multiple machinesthisfileisbeststored in a networkfolder • make.idList.Rcontains an R scriptforcreatingfileswithrandomizedconditions

47. Buildingyourown Experiment II After creatingthedesiredhtmlfiles: • Create a PsychoPy Monitor andsetthecorrectscreensize (http://psychopy.org/general/monitors.html) • Change experimentnameandsetcorrectfolders. • Replacethereferencestothehtmlfileswithyourownhtmlfiles. • Add callstoviewPsyTMLfor additional screens, e.g.: intro.viewPsyTML("html/intro2.html") • Set thedata_headerlisttomatchyour variable names (i.e., html form names). • Create yourown id.lst

48. Summary • PsychoPyandPsyTMLprovide a frameworkforrunningexperimentswith "type 2" items: • Create thescreensanditemsusinghtmlforms. • Python andPsychoPyareusedforrunningtheexperiment, PsyTMLdisplaystheformsandgathersthedata. • Learning basichtmland Python isprobably a goodinvestment.