I, Me, Mine: Identity, Disclosure,

1
I, Me, MineIdentity, Disclosure, the Self

• Michael C. Habib
• JOMC 391.3
• November 1, 2005

2
Can You See the Real Me?Activation and
Expression of the True Self on the
InternetBargh, J.A., McKenna, K.Y.A.,
Fitzsimmons, G.M. (2002)

• Public persona vs. unconscious self (Jung, 1953)
• Actual self vs. true self (Rogers, 1951)
• Actual self Expression of self in social
  interactions.
• True self Complete psychological existence of
  self.
• IF The internet is a place where individuals
  can explore alternative, yet currently existing,
  conceptions of self (Turkle, 1995).
• THEN People will use the internet to express
  their true selves.

3
Internet facilitates self-expression

• Anonymity
• Free of social expectations and constraints.
• Reduction of risks and social sanctions.
• Costs of revealing taboo things about oneself
  disappear outside of ones established social
  sphere.
• Leads to self-disclosure
• Strangers on a train (Ruben, 1975)
• Then friendship
• Disclosing ones true self leads to feelings
  of empathy and understanding between oneself and
  the other.
• Being able to disclose ones true self through
  CMC might then lead to close relationships.

4
If validation of ones true self is desirable,
then people will naturally use the internet to
express their true selves to others.

• H1a Ones conceptions of his/her true self
  should be more accessible cognitively during an
  initial interaction with a stranger over CMC than
  during an initial FtF interaction.
• H1b In the initial FtF interaction, the actual
  self concept will be more accessible than in the
  CMC interaction.
• H2a One will better present their true self
  in the initial CMC interaction than in a FtF,
  thus causing the CMC partner to form an
  impression closer to the that of ones true
  self than ones actual self.
• H2b Following an initial positive impression,
  one will like the CMC partner more than in an
  initial FtF interaction.

5
Experiment 1 Methodology

• Speeded self-judgment, reaction time to
• Me/Not-Me (Markus, 1977)
• 35 words chosen from normative likability ratings
  (Anderson, 1968) and 10 words from participant
  generated list of actual self and true self
  identifiers.
• N46 (16 same sex pairs and 7 cross-sex pairs)

6
Results and analysis

• 2 (interaction mode CMC or FtF)
• x 2 (interaction length 5 or 15 min.)
• x 2 (self-concept actual vs. true)
• x 2 (participated within pair)
• repeated measures ANOVA
• w/ first two as between-participants and
• second two as within-participants.

7
Significant interactions
H1a H2a confirmed
8
Experiment 2

• Purpose Was the anticipation of a CMC vs. FtF
  interaction or the interaction itself cause for
  the results from experiment one?
• Method
• Participants perform Me/Not-Me with anticipation
  of, but no actual interaction (time of
  anticipation remained the same as in Exp. 1)
• Control group was added who anticipated nothing.
• Results
• 3 (anticipated interaction type) x 2
  (self-concept)
• No significant effect accept that actual self
  was more accessible then true self across all
  conditions.
• H1a and H1b re-confirmed

9
Experiment 3 Method

• H2a and H2b should hold true because
• Greater ease of expressing ones true self in
  CMC vs. FtF.
• On account of the lack of traditional impression
  forming cues, one will also be freer to imagine
  an idealized version of the other.
• N40 (20 male, 20 female)
• Cross-sex pairs meet for 40min
• Actual self measure, true self measure, ideal
  partner measure, ideal friend measure, and liking
  of partner measure.

10
Self-presentation measures
Support for H2a
11
Liking and projection of ideals

• CMC partners liked each other significantly more
  than in FtF condition.
• H2b supported for friendship but not romance.
  Projected ideals of friend in CMC but not FtF
  condition.
• No differences on projection between CMC and FtF
  independent of initial liking.

12
Questions

• In an on-line social sphere that is a 'usual
  social sphere' do the findings of this study
  hold? I would suggest that they do not. Will the
  presentation of a true self over CMC give way to
  a presentation of the actual-self over time? (TJ)
  
• For example Facebook (Mike)
• What is the topic for those interactions in the
  experiments. The sensitivity of the topic might
  influence the result (sensitive topic, more
  actual self non-sensitive topic, more real
  self). (Cong)
• Might participants have guided the conversation
  themselves based on their increased comfort
  levels. However, I too wish they had covered
  this (Mike).

13
Another question

• It is not just that people are willing to present
  their true selves, but we need to examine the
  opposite why is it that we are not willing
  accept people's true selves in ftf communication?
  There must be something that we do that prevents
  other people from feeling comfortable to be
  themselves.
• When I teach listening skills to executives, I
  teach them to stop talking and stop thinking
  about what they will say next and start truly
  listening to the other person and what they have
  to say.
• Maybe another experiment could include a
  listening skills session in both cmc and ftf
  environments to see if there is a difference in
  the way people both transmit and receive the true
  selves of others. (Karen)

14
Impression Formation in Computer-Mediated
Communication RevisitedAn Analysis of the
Breadth and Intensity of ImpressionsHancock,
J.T., Dunham, P.J. (2001)

• Cues filtered-out (CFO) theories
• CMC lacks necessary cues for developing personal
  relations and clear impressions.
• Social Identification/Deindividuation (SIDE)
  model
• Lack of cues leads to increased reliance on
  existing cues for impression formation.
• Impressions reflect social categorization
  techniques such as stereotyping.
• Social information-processing (SIP) theory
• Because of the limited cues, impression forming
  is slower in CMC than FtF, but levels out over
  time.

15
Hyperpersonal model

• Like in SIDE, stereotyping based off limited
  cues.
• But, people also participate in
  self-presentation to control what cues they send
  out.
• Unlike CFO, people can put additional resources
  into word choice, content, and self-presentation.

16
Hypotheses

• H1 After an initial interaction, CMC
  participants will confidently rate partners on
  fewer characteristics than FtF participants will.
  (CFO and Hyperpersonal)
• H2a CMC participants will make more intense
  attributions about the characteristics they rate
  than FtF participants. (Hyperpersonal)
• H2b CMC participants will give relatively
  neutral ratings. (CFO)

17
Experiment

• N84 randomly paired to complete task
• NEO-Five Factor Inventory
• Neuroticism, Extraversion, Openness,
  Agreeableness, and Conscientiousness.
• 60 items, 12 for each trait.
• Six choices including neutral and cannot make
  judgment.
• Number answered ? Breadth index
• Deviation from neutral point ? Intensity index

18
Results H1 Supported

• 2 (communicative environment) x 5 (trait) mixed
  General Linear Model for Breadth and again for
  Intensity.
• CMC participants answered significantly fewer
  questions per trait.
• Differences b/w CMC and FtF existed for
  Extraversion, Agreeableness, Neuroticism (more
  visual traits?).
• But not for Openness and Conscientiousness (less
  visible traits?)
• CMC answered 59 of items while FtF answered
  74.6

19
H2a Supported, but not H2b

• Intensity was significantly greater in CMC
  condition.
• Interaction was consistent for all traits.

20
What next?

• Change over time? (to support the rest of the
  hyperpersonal model and SIP).
• If given reason and time, would participants
  develop methods to overcome the lack of cues?
• Would this intensify self-presented traits, or
  would intensity level out over time?

21
Questions

• Hancock and Dunham all but beg for follow up
  study I propose a study that looks at
  impression formation over time to test the
  hyperpersonal theories assertion that initial,
  stereotypical and extreme impressions of
  conversation partners will settle over time,
  becoming more realistic as cues are introduced.
  (TJ)
• We might be able to tell more in a text-based
  environment than in ftf because we control what
  the other person finds out. In ftf environment,
  they might not get a good grasp of who we are in
  a first meeting--just like dating, takes time to
  get to know a person. What if we did this study
  over time, to find out how their judgments
  changed after getting to know and work with
  another person? (Karen)

22
More Questions

• Impression formation in CMC revisited? The
  authors instructed participants in CMC conditions
  not to reveal their names, gender, or age. I am
  wondering why the authors did not let
  participants to reveal their gender. Since gender
  is obvious in the face-to-face condition, why
  they intentionally create this difference across
  the CMC and FTF conditions?
• The authors stated that they did this to
  optimize deindividuation, a future study could
  look at how individual cues effect outcomes.
  Age, sex, and location is usually the first thing
  asked when meeting someone in a chat room.

23
e-PerceptionsPersonality Impressions Based on
Personal WebsitesVazire, S., Gosling, S.D. (2004)

• Two ways personality is manifested in physical
  environment
• Identity claims
• Behavioral residue
• Impressions gained through these have been shown
  to converge with what people are actually like.
• These two mechanisms need to be analyzed
  separately.
• Personal web sites are chosen as examples of
  virtual space consisting of almost entirely
  identity claims.

24
Research Questions on Personal Websites.

• Consensus Do they provide a coherent,
  interpretable message to readers.
• Likely yes, because they are highly structured
  and constructed of symbols holding shared
  meanings.
• Accuracy Do they convey their message
  accurately.
• Likely yes, because people form accurate
  impressions quickly and easily given little
  information and these sites have lots of
  information.
• Impression Management Do they convey an overly
  positive message?
• Specifically, do site authors try to present
  their idealized self instead of their normal self?

25
Method

• 11 reviewers made recorded personality
  impressions for 89 websites.
• Accuracy was determined by comparison with an
  Accuracy index combining self-report and
  informants (friends).
• Ideal-self was collected from site authors.
• Five-factor model used (44 factor Big Item
  Inventory).

26
Results

• Question 1 Consistency measures were positive
  and significant for all 5 factors.
• Openness and Extraversion displayed the
  strongest consistency, and Emotional Stability
  the weakest.
• Question 2 Accuracy correlations were also all
  positive and significant.
• A number of possible influencing factors were
  analyzed and only gender had any effect. Gender
  significantly accounted for ratings of
  Agreeableness and Openness. This was in addition
  to the Accuracy correlation for those factors.
• Question 3 Extraversion and Agreeableness
  showed effects of impression formation.

27
(No Transcript)
28
(No Transcript)
29
I didnt understand why this was particularly
relevant.
30
Questions

• Can we use the same examination of corporate
  websites to determine the self-concept or image
  the firm is 1) intending to convey and 2) is
  observed to convey? This is important for both
  profit and nonprofit firms who are trying to
  portray themselves in a certain way, but may not
  understand that e-perceptions are being made
  about them by what they reveal or do not reveal.
  (Karen)
• the researcher might consider categorizing the
  personal web sites further into more specific
  types. For example, many academic personal pages
  fit the definition on p. 125 but do not have much
  to tell about the authors personality while many
  creative sites do. (Shirong, Sanghee also
  expressed concern about this)
• What if they analyzed profiles from social
  networking sites instead? These have the express
  purpose of presenting oneself to the world.
  Blogs on personal issues also would work well,
  but analysis might be more difficult. (Mike)

31
Dont Blame the ComputerWhen Self-Disclosure
Moderates the Self-Service BiasMoon, Y. (2003)

• Internal vs. External (Heider, 1958)
• Is the self responsible or some environmental
  factor?
• Self-serving bias
• When there is a positive outcome, external
  factors are responsible.
• When there is a negative outcome, the self gets
  the credit.

32
Why the Self-Serving Bias?

• Possibly an information processing deficiency?
• However, most evidence points to the human need
  for ego enhancement.
• How does this translate in an HCI situation?

33
Interpersonal Contextsan Exception to
Self-Serving bias

• When individuals partake in intimate
  self-disclosure (sharing of high-risk
  information) with one another, self-serving bias
  is diminished.
• Intimate self-disclosure ? attraction
• Attraction ? including the other within ones
  concept of self and thus protect their ego as
  well.
• Given the media equation, does this behavior
  extend in to HCI scenarios?

34
Hypotheses

• H1 Users will be more likely to blame the
  computer when there is a negative outcome than
  when there is a positive outcome.
• H2a After having engaged in self-disclosure
  with a computer, users are more likely to credit
  the computer when there is a positive outcome,
  than when they have not previously disclosed
  information to the computer.
• H2b After having engaged in self-disclosure
  with a computer, users are less likely to blame
  the computer credit when there is a negative
  outcome, than when they have not previously
  disclosed information to the computer.
• H3 The relation between self-disclosure and
  attributional tendencies predicted in H2a and H2b
  will be partially mediated by the level of
  attraction users feel toward the computer.

35
Experiment 1 Method

• All users participate in intimate
  self-disclosure, then filler task, then some
  perform the purchasing task on either the
  original computer or a different computer.
• Disclosure-reciprocity norms Computer shared
  information about itself, asked gradually more
  intimate questions,
• 2 (computer type same or different)
• x 2 (outcome positive or negative)
• between subjects full factorial.
• Told successful completion of task would lead to
  10 dollar reward.

36
Measures and Results

• Attribution 1 2 questions, 10 pt scale,
  who/what was more responsibleself or computer?
• Attribution 2 2 questions, 10 pt scale, the
  computer contributed heavily agree or disagree?
• H1, H2a and H2b Supported by both Attribution 1
  and 2.

37
Positive Outcome
38
Experiment 2 Method

• All participants use the same computer in Tasks 1
  and 3.
• Three types of disclosure
• RECIP (disclosure-reciprocity)
• NoRECIP (random order, Computer does not
  disclose)
• LG (same as NoRECIP but using same number of
  words as RECIP)
• User supposed to disclose more to RECIP which
  will mediate affection and attributional
  tendencies.

39
Results of Experiment 1 Replicated

• Manipulation worked Disclosure was
  significantly higher for RECIP and H1 and H2 were
  further supported.

40
H3 Supported

• Liking (attraction) measured after Task 1
• Self-disclosure had a significant effect on
  liking, with participants liking the computer
  more in the RECIP condition.
• The effect of self-disclosure on Attributions
  then re-tested with liking as a covariate.
• The effect of Liking on Attributions 1 and 2 was
  significant in both the positive and negative
  conditions.
• ? Thus attraction partially mediated
  attributional tendencies.

41
Possible Applications

• Traditionally, it is assumed that users only like
  significant self-disclosure if they are to get
  some functionality in return.
• Furthermore, it is traditionally assumed that
  having disclosed information, users have higher
  expectations for the system.
• However, this study suggests that the act of
  self-disclosure can, in some cases, increase both
  attraction and praise and decrease criticism of a
  system.
• Will this hold true in the real world with
  privacy and confidentiality concerns, etc.
• Attraction caused by other factors may accomplish
  the same goals.

42
Questions

• The alternate prediction is that consumers who
  engage in self-disclosure will be more forgiving
  in their expectations of performance This causes
  me to wonder what motivates users to divulge
  intimate personal information. Are there a
  variety of reasons? If a user is not expecting a
  pay-off from a commercial site what compels them
  to provide personal information?
• Might it be the True Self effect? (Mike)
• Recommendation systems like Amazon.com come to
  mind. People put tons of personal information in
  the system. Does that act of self-disclosure
  endear people to the site? (Mike)

43
Questions

• I would be interested in seeing the computers
  answers to the 15 questions and their
  computer-disclosures. From whose perspective
  the experimenters typed in the computers
  answers? From a machines only or a personalized
  AI/another human being? If the former, people
  treat machines as machines but unconsciously give
  it some human-like characteristicsit becomes an
  HCI question. If the latter, people actually
  treat machines as other human beings whose name
  are represented as computersit becomes a CMC
  question, to a certain degree. (Shirong)