Research study

overview
This study analyses whether the acceptance of smart speakers depends on the personality of a person.
With the OCEAN model and its dimension openness towards new technologies, a Wizard of Oz experiment was conducted. 

role
User research

timeline
March - July 2018
  
motivation
Google's voice assistant Google Duplex uses artificial intelligence to enable people to have a natural conversation with computers, as they would with each other. But a study from the year preceding my research conducted that only 8 % of Germans own a smart virtual assistant.
The need for stimulation and security and their balance is different for everyone. 
A possible correlation between these preferences and the acceptance of smart speakers needs to be examined.
01
Analysis
related work
"Potential attackers do not have to be inside the home; they could easily yell through a window or door or leave a voice message in order to communicate with Alexa." – Haack et al.

"As virtual assistants become more intelligent and the IVA ecosystem of services and devices expands, there's a growing need to understand the security and privacy threats from this emerging technology." – Chung et al.
  
research questions
To what extent does the personality dimension 'openness' according to the OCEAN model, influence the openness towards, and acceptance of smart speakers?
hypotheses

h1 The dimensions of personality, specifically the value associated with Openness, have a positive influence on one's acceptance of smart speakers.

h2 The higher an individual's value of Openness, the more accepting they are of smart speakers.

h3 Smart speakers with a natural tone of voice are generally more accepted than ones using robotic voices.
02
Study setup
Participants
Because the study is partly a between-subject design, the recruiting of participants was done beforehand and online with a questionnaire collecting data of all dimensions of personality according to the OCEAN model.
  
big five personality test
●  9 open
●  9 non-open

demographics
●  7 female, 11 male
●  72 % aged 18 - 25
●  ​​​​​​​61 % students
measures
test 1: affinity for technology interaction
measures an individual's likelihood to actively engage with technology

test 2: technology acceptance model
●  Intention to Use
●  Perceived Usefulness
●  Perceived Ease of Use
●  Subjective Norm

test 3: attrakdiff
●  Hedonic Quality
●  Pragmatic Quality
●  ​​​​​​​Attractiveness
procedure
All participants filled out questionnaires before, afterwards and in between the individual interaction methods.
The questionnaires included self constructed questions and standardized psychological tests to each interaction method and were used to assess the user's acceptance of smart speakers and experiences during the interactions.
Lastly, a short semi-structured interview was conducted with every participant, asking them about any additional remarks they might have.
  
3 runs with randomized interaction methods
●  smart speaker with computer generated voice
●  smart speaker natural voice
●  without smart speaker
4 tasks per run
●  switching the light on
●  finding out the capital of a given country
●  adding an 'Apple iPhone 7 in black' to the shopping cart
●  setting a timer for 10 minutes
wizard of oz
Wizard of Oz is a research method where participants interact with a system that they believe to be autonomous, but in reality, is controlled by an unseen human operator in the next room.
In reality the smart speaker was only a portable speaker, which was connected to a smartphone via Bluetooth. This phone had a mobile connection to the wizard room, so the wizard could listen to the instructions of the participants and play the right response form the pre-recorded voice snippets.
A second researcher was behind a partition wall to switch the light on and off.
Researcher 1 was with the participant in the study room to overlook the experiment, as well as help the interviewee in case of any misunderstandings, questions or uncertainties.
  
03
Evaluation
results
self constructed questionnaire
●  17 % currently own a smart speaker (N = 18)
interviews
affinity for technology interaction
●  no significant differences between open and non-open personality types
⟶  openness and technology acceptance do no correlate

technology acceptance model
●  Subjective Norm is rated higher for traditional interaction by both groups
⟶  high social pressure to not use speakers

●  Perceived Ease of Use of computer-generated voice is rated significantly higher by non-open

●  no more significant differences
⟶  open and non-open personalities do not accept smart speakers differently

attrakdiff
●  no significant difference between open and non-open
⟶  personality does not have an influence on the acceptance of smart speakers

●  Hedonic Quality is judged higher for computer-generated and natural human voice than traditional interaction by both groups
⟶  regardless of personality, speakers are more fun to use
  
conclusion
The results showed no significant correlation between 'openness' and the acceptance of smart speakers.
It could not definitively be concluded whether users prefer a natural or a robotic sounding voice.
future work
Because of the small sample size and the controlled study environment, further research needs to be done.

●  long term study
●  higher sample size
●  the use of smart speakers in everyday life
↑Back to Top