FAQ

Report 18 Downloads 366 Views
 

Affdex  FAQ   Dashboard  Components     What  are  the  traces  on  the  dashboard  showing  me?   The  dashboard  shows  traces  for  7  measures  for  all  respondents  with  usable  data  for  that  viewing  of  the   ad  (1st  or  2nd).    Dashboard  metrics  are  aggregates  of  all  individual  sessions.       The  Affdex  traces  include  two  categories  of  emotion  metrics:  dimensions  of  emotion  and  discrete   emotions:  Dimensions  of  emotion  are  used  to  characterize  the  participants  emotional  responses  while   discrete  emotions  are  used  to  describer  the  specific  emotional  state.     The  dimensions  of  emotion  that  Affdex  measures  includes:   • Valence  –  A  measure  of  the  positive  (or  negative)  nature  of  the  participant’s  experience   with  the  content.  If  participants  tend  to  be  displaying  more  positive  emotions  Valence   increases  on  the  positive  side  or  if  more  negative  emotions  are  expressed  Valence  decreases   toward  the  negative  side.   • Attention  –  A  measure  of  the  participant’s  attention  to  the  screen,  using  the  orientation  of   the  face  to  assess  if  they  are  looking  directly  at  the  screen  or  if  they  are  distracted  (turning   away)  while  viewing  content.     • Expressiveness  –  A  measure  of  how  emotionally  engaging  content  is,  computed  by   accumulating  the  frequency  and  intensity  of  the  discrete  emotions  including  smile,   dislike/disgust,  surprise  and  concentration/frown.  Unlike  valence,  expressiveness  is   independent  of  the  positive  or  negative  aspect  of  the  facial  expressions.     The  discrete  emotion  measures  includes:   • Smile  –  The  degree  to  which  the  participant  is  displaying  a  smile.  The  smile  classifier  looks  at   the  full  face  rather  than  just  the  mouth/lip  area,  incorporating  other  facial  cues,  like  the   eyes,  to  accurately  indicate  a  true  smile.  Smirks  and  grimaces  are  trained  not  to  trigger  our   smile  classifier.     • Concentration/Frown  –  The  degree  to  which  the  participant  is  frowning  (displaying  a  brow   furrow)  in  the  absence  of  a  Dislike/Disgust  classifier  response.  Concentration/Frown   responses  during  ad  test  are  most  often  negative  in  valence.  They  are  usually  attributable  to   mental  effort,  focus,  or  confusion.     • Surprise  –  The  degree  to  which  the  participant  is  showing  a  face  of  surprise,  indicated  by   raised  eyebrows.   • Dislike/Disgust  –  The  degree  to  which  the  participant  is  showing  expressions  of  dislike  or   even  disgust.  These  expressions  include  nose  wrinkles  and/or  grimaces.               1    

 

Confidential,  Proprietary  to  Affectiva                                                            

  What  does  the    Indicator  mean?    On  May  6,  2013  Affectiva  released  an  upgrade  to  our  dashboard  metrics.  The  alert  indicates  which   dashboards/summary  metrics  pages  are  “old”  or  delivered  prior  to  this  upgrade.  This  is  important  to   note,  as  with  the  significant  updates  to  the  new  metrics  the  data  is  not  comparable  between  “old”  and   “new”  dashboards  &  summary  metrics.     What  is  the  ‘Bubble’  on  the  moving  trace?     The  number  in  the  bubble  indicates  the  average  moment  by  moment  degree  to  which  respondents   are  displaying  a  particular  emotion  (smiling,  surprised  etc.),  taking  into  account  both  frequency  (e.g.   the  number  of  the  sample  displaying  a  smile)  and  intensity  (e.g.  the  strength  of  smiles).    Therefore,  a   peak  in  the  line  for  smile  could  represent  a  lot  of  the  sample  smiling  a  small  amount,  or  a  small   number  of  the  sample  smiling  broadly.   The  theoretical  range  for  the  discrete  emotion  measures    (smile,  surprise,  concentration/frown  &   dislike/disgust)  is  0-­‐100  but  in  reality  the  vast  majority  of  scores  typically  fall  between  0  and  20.   For  the  dimensions  of  emotion  the  ranges  can  vary  as  follows:   • • •

For  Attention  -­‐  The  theoretical  range  for  these  scores  is  0-­‐100  but  in  reality  the  vast  majority   of  scores  typically  fall  between  70  and  100.   For  expressiveness-­‐  the  theoretical  range  for  these  scores  is  0-­‐100  but  in  reality  the  vast   majority  of  scores  typically  fall  between  0  and  30.   For  Valence  -­‐  The  theoretical  range  for  these  scores  is  -­‐100  to  +100  but  in  reality  the  vast   majority  of  scores  typically  fall  between  -­‐10  to  +20.    

  How  can  I  view  different  segments  of  the  data?   The  dashboard  data  can  be  segmented  by  using  combinations  of  the  follow  controls:   • Ad  Selector:  allows  viewing  data  for  the  ads  tested  in  the  project   o Clicking  the  arrows  to  the  left/right  of  the  ad  will  allow  you  to  navigate  through  the  ads   in  the  project.   o Hovering  over  the  ad  selector  will  display  a  downward  arrow  that  when  selected  will   display  all  of  the  ads  in  the  project  for  selection.   •  New  Select  Exposure:  allows  viewing  of  the  data  of  one  of  more  exposures.  An  exposure  is   defined  as  the  number  of  times  the  participant  viewed  the  media,  so  in  most  cases  there  will  be   one  or  two  of  these.  This  control  will  only  display  on  the  dashboard  if  more  than  one  exposure   was  collected  in  the  project.   o To  select  exposures,  click  on  the  control  and  check  the  box(s)  to  choose  which   exposures  to  view  on  the  dashboard.  

              2    

   

Confidential,  Proprietary  to  Affectiva                                                            

  o

o

o

A  minimum  of  one  and  up  to  three  exposures  can  be  viewed  on  the  dashboard  at  one   time.  You  can  select  to  look  at  each  exposure  individually  or  combine  exposures  for   comparison.   The  first  exposure  is  selected  by  default,  which  means  that  the  other  controls  on  the   dashboard  (participant  group,  question  selector)  are  referencing  exposure  one.  To  view   data  specific  to  the  second  or  third  exposure  select  them  individually.     Each  exposure  is  represented  by  a  different  line  style  on  the  dashboard:   !

Exposure  1  –  Solid  colored  line  

!

Exposure  2  –  Dotted  colored  line  

   

! Exposure  3  –  Dashed  colored  line     Viewing  Segment  Breaks  and  Participant  Groups  –  When  viewing  segment  breaks  and/or   changing  the  participant  groups  in  view  the  exposure  selector  will  persist  and  show  the   traces  for  each  exposure  selected.   o Known  Limitation  with  Bookmarks:  When  viewing  multiple  exposures  the  available   bookmarks  will  only  be  displayed  for  the  1st  exposure.  To  see  bookmarks  on  subsequent   exposures  you  will  need  to  view  them  individually     Participant  Groups:  allows  viewing  data  specific  to  the  following  groups:   o All  Participants  -­‐  includes  data  for  all  survey  participants  that  have  usable  data.   o Most  Expressive  Participants  -­‐  includes  data  for  participants  that  were  emotive  in  any  of   our  captured  emotions.     o Most  Expressive  in  Selected  Emotion  -­‐  includes  data  for  participants  that  were  emotive   in  the  selected  emotion  (i.e.  selecting  this  participant  group  and  the  “Smile”  measure   will  only  show  data  for  participants  that  smiled  during  the  ad).     *Note:  this  participant  group  is  not  applicable  when  viewing  data  for  Valence,  Attention   and  Expressiveness,  thus  when  these  combinations  are  chosen  no  data  tracks  will   display  and  the  user  will  be  prompted  to  change  their  selections.     Survey  Questions:  allows  viewing  data  for  specific  data  survey  questions.  The  questions   displayed  are  determined  prior  to  beginning  a  study  and  generally  consists  of  a  standard  set   (age,  gender,  brand  appeal  etc.)  and  occasionally  some  client-­‐specific  custom  questions.   Emotions  Metrics:  clicking  on  any  emotion  will  show  data  specific  to  the  participant’s  reaction  in   the  selected  emotion.  (See  details  in  “What  are  the  traces  on  the  dashboard  showing  me  “   question  above)   Settings>  Suppress  Below  –  data  traces  for  samples  below  10  participants  are  suppressed  from   the  dashboard  display.  Without  an  adequate  number  of  respondents,  the  ability  to  break  down   the  data  and  make  meaningful  conclusions  is  compromised.  Data  that  has  been  suppressed  is   annotated  with  (Low  Data  Sample).     o









            3    

 

Confidential,  Proprietary  to  Affectiva                                                            

  How  do  I  use  the  Bookmark  Feature?   This  dashboard  feature  allows  you  to  save  certain  views  of  the  Facial  Coding  data,  making  it  easy  to  re-­‐ visit  and  navigate  between  these  views  when  analyzing  and  presenting  your  data.    The  Bookmarks   feature  is  turned  on  by  default,  allowing  you  to  view  and  add  bookmarks  right  away.     You  may  temporarily  turn  off  the  Bookmarks  feature,  but  it  will  reset  to  On  each  time  you  return  to  the   dashboard.   • To  turn  the  Bookmark  feature  on  or  off:   o Go  to  the  ‘settings’  menu  by  clicking  on  the  gear  icon.   o Click  the  ‘bookmarks’  setting  to  on/off     With  the  Bookmarks  feature,  you  can:   • View  any  bookmark  that  might  have  been  added  by  the  Affectiva  team  to  highlight  possible   points  of  interest  in  the  data  lines.       o To  view  these  bookmarks,  select  them  from  the  drop-­‐down  list  or  click  the  pin  icon   directly   • Create  new  bookmarks:   o Select  the  view  you  want  to  bookmark  –  the  relevant  ad,  exposure,  metric,  Link  question   split  and  time     o Right  click,  choose  a  bookmark  title  and  description  and  click  save   • Delete  bookmarks  you  have  created:   o Select  the  relevant  bookmark  from  the  drop-­‐down  list   o Click  on  the  pin  for  that  bookmark  which  shows  on  the  data  line   o Click  ‘delete’  in  the  box   • Hide  bookmarks  create  by  others:   o Left  click  the  ‘hide  bookmark’  button   • Unhide  bookmarks  created  by  others:   o Go  to  the  ‘settings’  menu  by  clicking  on  the  gear  icon   o Click  the  ‘hidden  bookmarks’  setting  to  ‘show’   o Select  the  bookmark  you  desire  to  unhide  from  the  drop-­‐down  list     o Click  the  “unhide  bookmark”  link.     o Turning  the  ‘hidden  bookmarks’  setting  back  to  “hide”  will  hide  all  other  hidden   bookmarks    

Interpreting  results   How  should  valence  be  interpreted?  What  is  the  meaning  for  us?  -­‐  Valence  is  a  measure  of  perceived   positivity  or  negativity  to  the  stimulus.  It  is  a  derived  metric  blending  elements  of  Smile  and   Concentration/Frown,  yet  is  measured  separately;  i.e.,  Affdex  is  trained  to  detect  Valence  discretely.   Beyond  providing  an  assessment  of  positive  and  negative  emotional  impact,  variability  in  Valence  curves   indicates  content  that  contains  emotional  ebbs  and  flows  (usually  a  desirable  aspect  of  ads).  Further,   ads  that  produce  a  Valence  curve  that  climbs  "up  and  to  the  right",  especially  through  brand  reveal,  are   indicative  of  impactful  ads.   Are  higher  degrees  related  to  the  intensity  of  the  emotion?  -­‐    The  trace  is  a  representation  of  the   degree  to  which  of  a  particular  expression  occurring  at  that  moment.  It  is  influenced  both  by  the   4    

 

Confidential,  Proprietary  to  Affectiva                                                            

  existence  of  the  expression  and  by  intensity.    For  example  a  point  on  the  Smile  trace  of,  50  means  either   that  each  respondent  reflected  at  that  point  is  smiling  with  a  degree  of  exactly  50-­‐-­‐or,  more  likely,  that   some  respondents  are  smiling  at  a  higher  degree  (some  lower),  but  all  averaging  at  50.  

Summary  Metrics   How  do  I  use  the  Summary  Metrics  page?   Delivered  with  each  project’s  dashboard  is  a  view  of  projects  summary  metrics  for  quick  comparison   including  how  they  compare  to  the  Affectiva  Ad  Testing  Norms.  These  are  accessible  from  the  project  list   page  in  the  portal  by  clicking  the    icon.   The  metrics  included  in  the  Summary  Metrics  tab  help  you  to  answer  different  questions  about  the  facial   expressions  included  in  the  dashboard,  for  example  using  smile:   •

Mean  is  the  average  of  the  aggregate  line  for  the  metric.   •



Variability  is  a  measure  of  the  deviation  of  the  aggregated  data  line  for  the  metric.   •



Questions  you  can  consider:  Does  the  ad  elicit  a  lot  of  smiles?    Does  it  bring  about  more   or  less  smiles  than  other  ads  in  the  region?  

Questions  you  can  consider:  How  much  do  smiles  change  during  the  course  of  the  ad?   Higher  levels  of  variability  would  suggest  that  viewers  are  responding  to  the  ad  by   smiling  at  different  points  rather  than  simply  smiling  throughout.  

Max/Min  is  the  highest  point  of  the  aggregate  line  for  the  metric   •

You  can  use  max  and  min  values  to  work  out  the  range  over  which  the  smile  varies  

How  do  my  projects  summary  metrics  compare  to  the  Affectiva  Ad  Testing  Norms?   Metrics  for  Mean,  Variability,  Maximum  (Max)  &  Minimum  (Min)  are  compared  against  the  Affectiva   norm  data  for  the  region  the  ad  was  tested  in.  You  can  determine  which  region  your  ads  are  being   compared  against  by  looking  in  the  top  left  corner  of  the  metrics  page.  The  market  to  region  mapping  as   deployed  in  our  product  today  is  below.  This  mapping  is  updated  periodically  through  the  calendar  year,   as  we  acquire  more  data.  The  goal  is  to  get  each  market  to  have  its  own  norm  for  comparison,  needing   approximately  30  projects  to  get  there.             5    

   

Confidential,  Proprietary  to  Affectiva                                                            

  View  when  comparing  to  the  regional  norm:  

      View  when  comparing  to  a  market  norm:  

  Norms  are  also  calculated  separately  according  to  project  type:  online  or  offline.    In  the  example  below,   you  can  see  that  the  norms  for  an  online  project  from  the  Mexico  market  were  calculated  based  on  a   comparison  to  144  ads  from  other  online  projects  in  that  market.    

  The  norms  for  an  offline  project  from  the  Mexico  market  were  calculated  based  on  a  comparison  to  85   ads  from  other  offline  projects  in  that  market.  

    The  ad  testing  norm  appears  to  the  right  of  the  ad  metric  norm  in  parentheses,  with  an  arrow  indicating   whether  the  ad  is  significantly  above  or  below  the  norm.    As  noted  above,  the  norm  values  displayed  on   the  summary  metrics  page  are  calculated  separately  by  online  or  offline  project  type.     NOTE:  the  metrics  for  max,  min  and  %  expressive  do  not  display  the  arrow  indicators  as  are  they  are   not  subject  to  significance  testing.  Project  values  are  displayed  against  the  ad  norms  only.  

6    

 

Confidential,  Proprietary  to  Affectiva                                                            

 

    Current  Market  to  Region  mapping:   Country   Angola   Cameroon   Ghana   Kenya   Madagascar   Mauritius   Mozambique   Nigeria   Senegal   South  Africa  -­‐  LSMA  (7-­‐10)  -­‐  F2F   South  Africa  -­‐  LSMB  (5-­‐6)  -­‐  F2F   Tanzania   Uganda   Zambia   Zimbabwe   7    

Mapped  to  Region   AMAP_Africa   AMAP_Africa   AMAP_Africa   AMAP_Africa   AMAP_Africa   AMAP_Africa   AMAP_Africa   AMAP_Africa   AMAP_Africa   AMAP_Africa   AMAP_Africa   AMAP_Africa   AMAP_Africa   AMAP_Africa   AMAP_Africa    

Confidential,  Proprietary  to  Affectiva                                                            

  China  -­‐  Unspecified   Hong  Kong   China  -­‐  Beijing   China  -­‐  Secondary  Cities  Central  and  West   China  -­‐  Secondary  Cities  East   China  -­‐  Secondary  Cities  North   China  -­‐  Secondary  Cities  South  and  Guangzhou   China  -­‐  Shanghai   China  -­‐  Tertiary  Cities   Taiwan   Algeria   Egypt   Iran   Kuwait   Lebanon   Morocco   Saudi  Arabia   Syria   Tunisia   United  Arab  Emirates   Australia   New  Zealand   Indonesia   Japan   Korea   Malaysia   Malaysia  -­‐  Chinese  sample   Malaysia  -­‐  Malay  sample   Philippines   Singapore   Thailand   Vietnam   Vietnam  -­‐  Hanoi   Vietnam  -­‐  Ho  Chi  Minh   Myanmar   India  -­‐  Unspecified   Bangladesh   India  -­‐  East   India  -­‐  North   India  -­‐  Rural   8    

AMAP_Eastern-­‐Asia   AMAP_Eastern-­‐Asia   AMAP_Eastern-­‐Asia   AMAP_Eastern-­‐Asia   AMAP_Eastern-­‐Asia   AMAP_Eastern-­‐Asia   AMAP_Eastern-­‐Asia   AMAP_Eastern-­‐Asia   AMAP_Eastern-­‐Asia   AMAP_Eastern-­‐Asia   AMAP_Middle-­‐East   AMAP_Middle-­‐East   AMAP_Middle-­‐East   AMAP_Middle-­‐East   AMAP_Middle-­‐East   AMAP_Middle-­‐East   AMAP_Middle-­‐East   AMAP_Middle-­‐East   AMAP_Middle-­‐East   AMAP_Middle-­‐East   AMAP_Oceania   AMAP_Oceania   AMAP_South-­‐Eastern-­‐Asia   AMAP_South-­‐Eastern-­‐Asia   AMAP_South-­‐Eastern-­‐Asia   AMAP_South-­‐Eastern-­‐Asia   AMAP_South-­‐Eastern-­‐Asia   AMAP_South-­‐Eastern-­‐Asia   AMAP_South-­‐Eastern-­‐Asia   AMAP_South-­‐Eastern-­‐Asia   AMAP_South-­‐Eastern-­‐Asia   AMAP_South-­‐Eastern-­‐Asia   AMAP_South-­‐Eastern-­‐Asia   AMAP_South-­‐Eastern-­‐Asia   AMAP_South-­‐Eastern-­‐Asia   AMAP_Southern-­‐Asia   AMAP_Southern-­‐Asia   AMAP_Southern-­‐Asia   AMAP_Southern-­‐Asia   AMAP_Southern-­‐Asia    

Confidential,  Proprietary  to  Affectiva                                                            

  India  -­‐  South   India  -­‐  West   Nepal   Pakistan   Sri  Lanka   Denmark  -­‐  mixed  methodology   Germany   France   Italy   Netherlands   Israel   Poland   Turkey   United  Kingdom   Albania   Austria   Azerbaijan   Belgium   Belgium  -­‐  North  (Flemish)   Belgium  -­‐  South  (French)   Bosnia  And  Herzegovina   Bulgaria   Croatia   Cyprus   Czech  Republic   FYR  Macedonia   Finland  -­‐  mixed  methodology   Greece   Hungary   Iceland   Ireland   Kazakhstan   Latvia   Lithuania   Luxembourg   Malta   Norway  -­‐  mixed  methodology   Portugal   Romania   Russia   9    

AMAP_Southern-­‐Asia   AMAP_Southern-­‐Asia   AMAP_Southern-­‐Asia   AMAP_Southern-­‐Asia   AMAP_Southern-­‐Asia   Europe   Europe   Europe   Europe   Europe   Europe   Europe   Europe   Europe   Europe   Europe   Europe   Europe   Europe   Europe   Europe   Europe   Europe   Europe   Europe   Europe   Europe   Europe   Europe   Europe   Europe   Europe   Europe   Europe   Europe   Europe   Europe   Europe   Europe   Europe    

Confidential,  Proprietary  to  Affectiva                                                            

  Serbia   Slovakia   Slovenia   Spain   Sweden  -­‐  mixed  methodology   Switzerland   Ukraine   Brazil  -­‐  Unspecified   Mexico   Argentina   Bolivia   Brazil  (CAWI  Only)   Brazil  -­‐  North  East   Brazil  -­‐  Rio  de  Janeiro   Brazil  -­‐  Sao  Paulo   Brazil  -­‐  South   Chile   Colombia   Costa  Rica   Dominican  Republic   Ecuador   El  Salvador   Guatemala   Honduras   Jamaica   Nicaragua   Panama   Paraguay   Peru   Puerto  Rico   Trinidad   Uruguay   Venezuela   USA  -­‐  English   Canada  -­‐  English  Speaking   Canada  -­‐  French  Speaking   USA  -­‐  Acculturated  Hispanics   USA  -­‐  African  American   USA  -­‐  Bicultural  Hispanics   USA  -­‐  Spanish     10    

Europe   Europe   Europe   Europe   Europe   Europe   Europe   LatAm   LatAm   LatAm   LatAm   LatAm   LatAm   LatAm   LatAm   LatAm   LatAm   LatAm   LatAm   LatAm   LatAm   LatAm   LatAm   LatAm   LatAm   LatAm   LatAm   LatAm   LatAm   LatAm   LatAm   LatAm   LatAm   NA   NA   NA   NA   NA   NA   NA    

Confidential,  Proprietary  to  Affectiva