Data 101 Workshop

28
Data 101 Workshop Joshua H Morrill, PhD WILSWorld 2016

Transcript of Data 101 Workshop

Data101

Workshop

JoshuaHMorrill,PhDWILSWorld2016

Whyareyoudoingthisresearch?Whatisthedrivingneed?

Whatdoyoureallywanttoknow?HowcanIasktherightresearchquestion?

1 2 3 4

WearegoingtobeworkingthroughsomeoftheWHY-WHAT-WHAT-&HOWsofanyresearchproject

Whatareoptionsforgettinginformation?Howreliableisreliableenough?

HowdoIinterpretwhatIhave?Whataresomepitfallsforunderstandingdata?

Whyareyoudoingthisresearch?Whatisthedrivingneed?

1Whydoyoune

edto

dothis?

DoyouREALLYneedmoreinforma8on? Don’tunderes8matecom

monsense

Thecuriouscaseof:“TheExpensiveBirthdayCard”

Whatdocurrentpatronsthinkaboutexis3ngspace?

Howhaveotherlibrariesrenovated

spaces?

Whatdoyoureallywanttoknow?HowcanIasktherightresearchquestion?

2

TheNEEDshoulddriveaseriesofquestionsthatmay-ormaynot-requireadditionaldatacollection.

NEED

Mylibraryisthinkingaboutrenova8nga

space

�Whataretrendsinspaceredesign?

Whatdoyoureallywanttoknow?HowcanIasktherightresearchquestion?

2

TheNEEDshoulddriveaseriesofquestionsthatmay-ormaynot-requireadditionaldatacollection.

Mylibraryisthinkingaboutrenova8nga

space

�Whatare

trendsinspaceredesign?

�Howhave

otherlibrariesrenovatedspaces?

�Whatdo

currentpatronsthinkaboutthe

space?

•  Conferences

•  Ar8cles

•  WebSearch

•  VendorPresenta8ons

•  Callotherlibraries

•  Tours

•  Past/olderstudiesorfeedback

•  Anewstudy

Whatdoyoureallywanttoknow?HowcanIasktherightresearchquestion?

2

Assumingyouwanttodoanewstudy…Thereisadifferentsetofquestionstoconsider.

�Wh

atdocurren

tpatrons

thinkab

outthe

space?

•  Whatdopatronsthinkabouttheexis8ngspace?

ATTITUDES

BEHAVIORS•  Whatdopatronsusetheexis8ng

spacefor?

Whatareyou

preparedtoacton?

What

informa8onisholdingupyourdecision?

ASPIRATION•  Whatwouldpatronsliketobe

abletodo?

•  Arethereotherpeoplewhowouldusethespace?

Whatdoyoureallywanttoknow?HowcanIasktherightresearchquestion?

2

EXERCISETIME!

•  Whatisyourneed?

•  Doesthisreallyrequirenewresearch?

•  Whatdoyouneedtoknowthatisholdingupadecision?

3

BeforeWeDiveintoMethods…

Whatareoptionsforgettinginformation?Howreliableisreliableenough?

par·si·mo·ny /ˈpärsəˌmōnē/ Theideathatthesimplestexplana3onofaphenomenonisthebestoneMakingapointwithsimpleresearchcanbeinfinitelymorepowerfulthanthesameconclusionreachedthroughmorecomplicatedmeans.Noneedtocrackawalnutwithajackhammer!

Thereisnoperfectdataandnoerrorfreedatacollec1onprocess

Misc.BigDataInteractiveTechniques

MethodsSurveys

IF DONE WELL – Can provide a broad range of responses. Can be a good way to capture low motivation individuals in your population.

Good for tracking/ documenting of change

Really powerful when you have a comparison point for your data.

3Whatareoptionsforgettinginformation?Howreliableisreliableenough?

Misc.BigDataInteractiveTechniques

MethodsSurveys

3Whatareoptionsforgettinginformation?Howreliableisreliableenough?

43%

25%

19%

13%

VeryWell PreQyWell Nottoowell Notwellatall

Mylibraryservesmylearningandeduca3onalneeds…

Misc.BigDataInteractiveTechniques

MethodsSurveys

3Whatareoptionsforgettinginformation?Howreliableisreliableenough?

43%

25%19%

13%

49%

39%

6% 4%

VeryWell PreQyWell Nottoowell Notwellatall

Mylibraryservesmylearningandeduca3onalneeds…

PEW

Misc.BigDataInteractiveTechniques

MethodsSurveys

3Whatareoptionsforgettinginformation?Howreliableisreliableenough?

GoodSurveyQues3ons….

Should be things individuals can reliably answer. (“Please rate how efficiently the staff at your library are managed.” = BAD)

Can be 4, 5, 6 or 7 point scales. (Think about when you want to force a choice / what a mid-response means)

Are sensitive to people who are taking the survey (Empathy will get you better/ more reliable information)

Misc.BigDataInteractiveTechniques

MethodsSurveys

3Whatareoptionsforgettinginformation?Howreliableisreliableenough?

ResearchLength/Investmentforpar3cipants

ReliabilityofData

Short/Low

Long

High

Misc.BigDataInteractiveTechniques

MethodsSurveys

3Whatareoptionsforgettinginformation?Howreliableisreliableenough?

BadQues8on

Thelengthof8meIspentintheEmergencyroomwas…- Excellent- Good- Fair-Poor

Be]erQues8onThelengthof8meIspentwai8ngintheEmergencyroombeforeseeingaphysicianwas…- LongerthanIexpected- AboutwhatIexpected- FasterthanIexpected

Thelibraryhourswereeasytofindonthesite.- StronglyAgree- Agree-Indifferent-Disagree

Howeasywerethelibraryhourstofindonthesite.- Veryeasytofind- Somewhateasytofind- Somewhatdifficulttofind-Verydifficulttofind-IdidnotnoHceifthehourswereeasyordifficulttofind

Misc.BigDataInteractiveTechniques

MethodsSurveys

3Whatareoptionsforgettinginformation?Howreliableisreliableenough?

Interac8veTechniques(Interviews,FocusGroups,etc.)

As a rule of thumb - Keep things to 1 hour or less

Can get interesting information if you have participants “create” something. (IL Building Project )

Think about WHO you want to talk to and why.

Generates a LOT of information that can take time to get in a useable form.

Misc.BigDataInteractiveTechniques

MethodsSurveys

3Whatareoptionsforgettinginformation?Howreliableisreliableenough?

IllinoisCommonsBuildingProjectHadstudentsdrawtheirideallibrarystudyspace….

Books

BooksBooks

Misc.BigDataInteractiveTechniques

MethodsSurveys

3Whatareoptionsforgettinginformation?Howreliableisreliableenough?

“BigData”(datafromdatabasesothersta1s1cs)

Buzzword now – but “Big Data ≠ Error Free”

With Big data, motivations are rarely collected. Still useful, but very little “WHY” answers in big data.

Big Data requires a greater vigilance to understand limitations of the data (i.e., rural areas), and make judgments on applicability and utility.

Misc.BigDataInteractiveTechniques

MethodsSurveys

3Whatareoptionsforgettinginformation?Howreliableisreliableenough?

MISC-SEGMENTATION

A way to understand subgroups in your data

Varies in sophistication but helps fight monolithic perceptions

Can be useful, but can also leave people feeling overwhelmed

Does the segmentation pass the “smell test” for you?

Misc.BigDataInteractiveTechniques

MethodsSurveys

3Whatareoptionsforgettinginformation?Howreliableisreliableenough?

WPLCUser/Non-UserStudy(2012)

Misc.BigDataInteractiveTechniques

MethodsSurveys

3Whatareoptionsforgettinginformation?Howreliableisreliableenough?

Ambivalent+Learners+

48%$of$Sample$This segment addresses learning problems using a plan (at least they believe that they have a plan). But, mostly, they do not feel strongly about their learning. They are confident in their ability to find information, but do not enjoy studying nor do they have a need to learn. This is the largest learner segment from the sample.

Adap1ve+Learners+

26%$of$Sample$This segment exhibits a lot of characteristics of “ideal” learners (They solve problems with a plan, they are systematic, they set goals, they ask for help if they experience a problem, they enjoy studying and have a need to learn). A differentiator in this group is that there is more variance around setting specific times to study. For example, this could be a learner who studies in a hallway whenever they had some free time.

Free+Form+Learners++

13%$of$Sample$This group is not systematic in their learning, and do not solve problems with plans. But they are willing to change what they do when presented with new information (may speak to an experiential type of learner). This group also feels like they have a need to learn, but are among the least likely to set aside specific time to study.

Time+Sensi1ve+Learners+

11%$of$Sample$This segment is similar to the adaptive learners in many ways (use a plan, are systematic, etc), but they are just not quite as strong in these skills. Directionally they are identical to adaptive learners. The other key difference is that this group is the most likely to set specific times to study, and least likely to ask for assistance with a problem. This is also the smallest learner segment.

NSFDigitalResourceUserSegmentaHon(2014)

3Whatareoptionsforgettinginformation?Howreliableisreliableenough?

NSFDigitalResourceUserSegmentaHon(2014)

EXERCISETIME!

•  Whattypesofdatamightansweryourques8on?

•  Whatdoyouneedtogetthisinforma8on---andensurethereliabilityofwhatyouget?

4HowdoIinterpretwhatIhave?Whataresomepitfallsforunderstandingdata?

Small samples, Biased samples, and

unrepresentative samples.

BadSample!

Spurious associations & phantom causations…

sometimes wishful thinking.

NoVariance Overreach

AFewCommonPerilsandPitfallsofResearch

Can be a symptom of bad samples, bad

questions, or something else. Be skeptical of uniform sameness.

HowdoIinterpretwhatIhave?Whataresomepitfallsforunderstandingdata?

Small samples, Biased samples, and

unrepresentative samples.

BadSample!

“Weaskedpeoplewhocameintothelibraryanden3cedthemwithdoughnuts.”

Dependingonwhatdecisionyouaremaking…isthiswhoyouwanttoprivilegehearingfrom?Doughnut-eaters?

“Wetalkedtothreepeopletosolicitdifferentcommunityperspec3vesaboutthelibrary.”

Isthispublicrela8ons…orresearch?Canthesepeopleaccuratelyrepresentyour“community”?

“Wesentthesurveytoseveralemaillistsandpostedonfacebooktogetoursample.”

Again,thismightbeOK,butwhoareyouprivileging/whoareyoumissing?Whatarethelimitsofyourgeneralizability?

HowdoIinterpretwhatIhave?Whataresomepitfallsforunderstandingdata?

“100%ofcommunitymembersweaskedsupportedthelibrary!”

Thissopatentlydoesnotpassthe“smelltest”thatitsuggestsaflawedques8on,sampleorsimplepropaganda.

“Anaverageforpar3cipantsyieldedameanscoreof3.2ona5-pointscale.”

MeanswithoutStandardDevia8onsareMEANINGLESS.

NoVarianceCan be a symptom of

bad samples, bad questions, or something

else. Be skeptical of uniform sameness.

HowdoIinterpretwhatIhave?Whataresomepitfallsforunderstandingdata?

“Studentsbehaviorsindicatedaposi3veopinionofthenewtechnology”

Whatifpeoplearebeingpolite?Ifyouwanttoknowanopinion—ask.Donotassume.

“A$1investmentinthelibraryreturns$14backtothecommunity.”

Theseareslipperyclaims...“Sowhynotputthewholebudgetinthelibrary?A14xreturnisbeZerthangold!”

NoVarianceCan be a symptom of

bad samples, bad questions, or something

else. Be skeptical of uniform sameness.

Spurious associations & phantom causations…

sometimes wishful thinking.

Overreach

“Ourlibraryprogramsreducedunemploymentinthecommunityby12%”

Helping___numberofpeoplelookforajob–yes!ButthereareaTONoffactorsthatinfluenceunemployment..

HowdoIinterpretwhatIhave?Whataresomepitfallsforunderstandingdata?

NoVariance

Whatcanyoudo?

Cul1vateyou

rinn

ersk

ep1c

Thecuriouscaseof:“Bigdata/Li]lee-Text,&TheTime-on-Taskgang”

They don’t take things at face value.

Thevalueof[information]liesintheusingofit.

-ThomasEdison

ThankYou.

Contact

Joshua H. Morrill, PhD [email protected] 608-588-2874