SlideShare a Scribd company logo
Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix
Making Sense of Statistics in HCI
Part 2 – Doing It
if not p then what
Alan Dix
http://alandix.com/statistics/
Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix
probing the unknown
quantified uncertainty
accepting unknowing
and common sense.
Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix
recall … the job of statistics
Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix
conditionals and likelihood
probability road will be busy
vs probability busy given 4am on Sunday
probability die is a 6
vs prob a 6 given it is 4 or bigger
conditional probability
is probability given some knowledge
Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix
likelihood
likelihood is
probability of measurement
given unknown things in the world
(conditional probability)
e.g. prob. of six heads given coin is fair = 1/64
… but is it fair?
Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix
counterfactual
statistics turns this round
from likelihood of measurement
probabilistic knowledge given the unknown world
to uncertain knowledge of the unknown world
based on measurement
… but in various different ways
Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix
statistical reasoning
Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix
types of statistics
• hypothesis testing (the dreaded p!)
– robust but confusing
• confidence intervals
– powerful but underused
• Bayesian stats
– mathematically clean but fragile
– issues of independence
Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix
Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix
hypothesis testing
the ubiquitous p
Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix
significance test
hypotheses:
H1 – what want to show
H0 – null hypothesis (to disprove)
idea (when experiment/study successful)
if H0 were true
then observed effect is very unlikely
therefore H1 is (likely to be) true
Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix
5% significance level?
it says:
if H0 were true
then probability observed effect
happening by chance
is less than 1 in 20 (5%)
so H0 is unlikely to be true
and H1 is likely to be true
Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix
does not say
8 probability of H0 is < 1 in 20
8 probability of H1 is > 0.95
8 effect is large or important
i.e. significant in the real sense
Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix
all it says
4 if H0 were true ...
... effect is unlikely
( prob. < 1 in 20)
Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix
non-significant result
can NEVER reason:
non-significant result  H1 false
things are the same
all you can say is:
H1 is not statistically proven
Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix
Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix
confidence intervals
bounds of uncertainty
Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix
proving equality
non-significant – not proved different
real difference may always be smaller
than experimental error
 can never prove equality
but can put bounds on inequality
Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix
confidence interval
bound on true value – same theory as p values
e.g. mean of data is 0.3
95% confidence interval is [–0.7,1.3]
says if the real value not in the range [–0.7,1.3]
probability of seeing observed data
is less than 5%
Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix
counterfactuals
95% confidence interval is [–0.7,1.3]
does not say:
there is 95% probability that the
real mean is in the range [–0.7,1.3]
it either is or it isn’t!
all it says:
probability of seeing the observed data
if real value outside the range is less than 5%
Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix
proven … what?
H0: no difference (real mean is zero)
experimental result: mean is 0.3
significance test: n.s. at 5% – so what?
95% confidence interval: [–0.7,1.3]
? is 1.3 is an important difference
Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix
… and don’t forget …
you still need to say
what test/distribution – e.g. Student’s T
how many – degrees of freedom
it is still uncertain
the real value could be outside the interval
Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix
Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix
Bayesian statistics
putting a number on it
Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix
traditional statistics
probability of having antennae:
if Martian = 1
if human = 0.001
hypotheses:
H0: no Martians in the High Street
H1: the Martians have landed
! you meet someone with antennae
reject null hypothesis p ≤ 0.1%
call the
Men in Black!
Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix
Bayesian thinking
prior probability of meeting:
a Martian = 0.000 001
a human = 0.999 999
probability of having antennae:
Martian = 1; human = 0.001
! you meet someone with antennae
posterior probability of being:
a Martian ≈ 0.001
a human ≈ 0.999
what you think
before evidence
Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix
Bayesian inference
combine prior
with likelihood
prior
distribution
posterior
Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix
Bayesian issues
how do you get the prior?
sometimes doesn't matter too much!
traditional stats rather like uniform prior
handling multiple evidence
can re-apply iteratively
problems with interactions
internecine warfare
traditionalists and Bayesians often fight ;)
Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix
Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix
philosophical
differences
probing the unknown
Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix
philosophical stance
? what do you know about the world
traditional statistics
– assume nothing!
– reason from unknowledge
Bayesian statistics
– 'prior' probabilities
– reason from guess-timates
Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix
uncertain and unknown
assumptions:
traditional – NO knowledge of real value
Bayesian – precise probability distribution
in reality … sort of half know
traditional – choice of acceptable p
Bayesian – ‘spread’ distributions (uniform, Cauchy)
ideally should do
sensitivity analysis
… but no one does 
Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix
Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix
issues
from “what is worse”
to really significant
Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix
issues
likelihood is conditional probability
what to do with 'non-sig'.
priors for experiments? finding or verifying?
significance vs effect size vs power
idea of 'worse' for measures
Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix
example: playing cards
run of cards
badly shuffled?
Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix
example: playing cards (2)
probability three cards make a decreasing run:
– first card needs to be 3 or bigger: 11/13
– second card exactly the next one down: 1/51
– third card exactly one down again: 1/50
= 11/33150 approx 1 in 3000 – p <0.001
any card
except ace
or 2
next card
down
next
again
Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix
example: playing cards (3)
but equally surprised if …
– either direction
– 15 positions for the first card
= 15x2x11/33150
approx 1 in 100
and then …
how often do I play?
Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix
dangers
if it can go wrong
then with 95% confidence …
Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix
dangers
avoiding cherry picking
multiple tests, multiple stats
outliers, post-hoc hypotheses
non-independent effects
e.g. fat and sugar
correlated features
e.g. weight, diet and exercise
including trying
both traditional
and Bayes!
the same UI but
‘not’ consistent
Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix
Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix
so which is it?
to p or not to p
Bayes forward?
Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix
the statistical crisis?
mostly well known problems with known
solutions
maybe more people doing stats with less
training?
some of the papers about it use flawed stats!
Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix
on balance (my advice!)
traditional
more conservative – if done properly
Bayes
many of the same problems as trad. plus a few more …
needs expertise
for most things stick with trad.
… but always confidence intervals as well as p
for special things go for Bayes
if you know prior, decision making with costs, internal algorithms
Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix
Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix
a quick experiment
… but are they fair
Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix
calculations – six coins
given coin is fair:
probability six heads = 1/26 = 1/64
probability six tails = 1/26 = 1/64
probability either = 2/64 ~ 3%
H0 – coin is fair
H1 – coin is not-fair
likelihood ( HHHHHH or TTTTTT | H0 ) < 5%
Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix
your experiment
toss 6 coins
record how many heads or tails
if HHHHHH or TTTTTTT
you can reject H0 with p< 5%
repeat exp. 2 more times with rest of coins
Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix

More Related Content

Similar to Making Sense of Statistics in HCI: part 2 - doing it

D. Mayo: Philosophy of Statistics & the Replication Crisis in Science
D. Mayo: Philosophy of Statistics & the Replication Crisis in ScienceD. Mayo: Philosophy of Statistics & the Replication Crisis in Science
D. Mayo: Philosophy of Statistics & the Replication Crisis in Science
jemille6
 
Meeting #1 Slides Phil 6334/Econ 6614 SP2019
Meeting #1 Slides Phil 6334/Econ 6614 SP2019Meeting #1 Slides Phil 6334/Econ 6614 SP2019
Meeting #1 Slides Phil 6334/Econ 6614 SP2019
jemille6
 
April 3 2014 slides mayo
April 3 2014 slides mayoApril 3 2014 slides mayo
April 3 2014 slides mayo
jemille6
 
P values and the art of herding cats
P values  and the art of herding catsP values  and the art of herding cats
P values and the art of herding cats
Stephen Senn
 
Probing with Severity: Beyond Bayesian Probabilism and Frequentist Performance
Probing with Severity: Beyond Bayesian Probabilism and Frequentist PerformanceProbing with Severity: Beyond Bayesian Probabilism and Frequentist Performance
Probing with Severity: Beyond Bayesian Probabilism and Frequentist Performance
jemille6
 
P value wars
P value warsP value wars
P value wars
Stephen Senn
 
Phil 6334 Mayo slides Day 1
Phil 6334 Mayo slides Day 1Phil 6334 Mayo slides Day 1
Phil 6334 Mayo slides Day 1
jemille6
 
The replication crisis: are P-values the problem and are Bayes factors the so...
The replication crisis: are P-values the problem and are Bayes factors the so...The replication crisis: are P-values the problem and are Bayes factors the so...
The replication crisis: are P-values the problem and are Bayes factors the so...
StephenSenn2
 
Crisis of confidence, p-hacking and the future of psychology
Crisis of confidence, p-hacking and the future of psychologyCrisis of confidence, p-hacking and the future of psychology
Crisis of confidence, p-hacking and the future of psychology
Matti Heino
 
The replication crisis: are P-values the problem and are Bayes factors the so...
The replication crisis: are P-values the problem and are Bayes factors the so...The replication crisis: are P-values the problem and are Bayes factors the so...
The replication crisis: are P-values the problem and are Bayes factors the so...
jemille6
 
Severe Testing: The Key to Error Correction
Severe Testing: The Key to Error CorrectionSevere Testing: The Key to Error Correction
Severe Testing: The Key to Error Correction
jemille6
 
Philosophy of Science and Philosophy of Statistics
Philosophy of Science and Philosophy of StatisticsPhilosophy of Science and Philosophy of Statistics
Philosophy of Science and Philosophy of Statistics
jemille6
 
Probability Homework Help
Probability Homework Help Probability Homework Help
Probability Homework Help
Statistics Assignment Help
 
P-Value "Reforms": Fixing Science or Threat to Replication and Falsification
P-Value "Reforms": Fixing Science or Threat to Replication and FalsificationP-Value "Reforms": Fixing Science or Threat to Replication and Falsification
P-Value "Reforms": Fixing Science or Threat to Replication and Falsification
jemille6
 
D. Mayo: Philosophical Interventions in the Statistics Wars
D. Mayo: Philosophical Interventions in the Statistics WarsD. Mayo: Philosophical Interventions in the Statistics Wars
D. Mayo: Philosophical Interventions in the Statistics Wars
jemille6
 
Mayod@psa 21(na)
Mayod@psa 21(na)Mayod@psa 21(na)
Mayod@psa 21(na)
DeborahMayo4
 
Paradoxes and Fallacies - Resolving some well-known puzzles with Bayesian net...
Paradoxes and Fallacies - Resolving some well-known puzzles with Bayesian net...Paradoxes and Fallacies - Resolving some well-known puzzles with Bayesian net...
Paradoxes and Fallacies - Resolving some well-known puzzles with Bayesian net...
Bayesia USA
 
Applied statistics part 2
Applied statistics  part 2Applied statistics  part 2
Frequentist inference only seems easy By John Mount
Frequentist inference only seems easy By John MountFrequentist inference only seems easy By John Mount
Frequentist inference only seems easy By John Mount
Chester Chen
 
D. G. Mayo Columbia slides for Workshop on Probability &Learning
D. G. Mayo Columbia slides for Workshop on Probability &LearningD. G. Mayo Columbia slides for Workshop on Probability &Learning
D. G. Mayo Columbia slides for Workshop on Probability &Learning
jemille6
 

Similar to Making Sense of Statistics in HCI: part 2 - doing it (20)

D. Mayo: Philosophy of Statistics & the Replication Crisis in Science
D. Mayo: Philosophy of Statistics & the Replication Crisis in ScienceD. Mayo: Philosophy of Statistics & the Replication Crisis in Science
D. Mayo: Philosophy of Statistics & the Replication Crisis in Science
 
Meeting #1 Slides Phil 6334/Econ 6614 SP2019
Meeting #1 Slides Phil 6334/Econ 6614 SP2019Meeting #1 Slides Phil 6334/Econ 6614 SP2019
Meeting #1 Slides Phil 6334/Econ 6614 SP2019
 
April 3 2014 slides mayo
April 3 2014 slides mayoApril 3 2014 slides mayo
April 3 2014 slides mayo
 
P values and the art of herding cats
P values  and the art of herding catsP values  and the art of herding cats
P values and the art of herding cats
 
Probing with Severity: Beyond Bayesian Probabilism and Frequentist Performance
Probing with Severity: Beyond Bayesian Probabilism and Frequentist PerformanceProbing with Severity: Beyond Bayesian Probabilism and Frequentist Performance
Probing with Severity: Beyond Bayesian Probabilism and Frequentist Performance
 
P value wars
P value warsP value wars
P value wars
 
Phil 6334 Mayo slides Day 1
Phil 6334 Mayo slides Day 1Phil 6334 Mayo slides Day 1
Phil 6334 Mayo slides Day 1
 
The replication crisis: are P-values the problem and are Bayes factors the so...
The replication crisis: are P-values the problem and are Bayes factors the so...The replication crisis: are P-values the problem and are Bayes factors the so...
The replication crisis: are P-values the problem and are Bayes factors the so...
 
Crisis of confidence, p-hacking and the future of psychology
Crisis of confidence, p-hacking and the future of psychologyCrisis of confidence, p-hacking and the future of psychology
Crisis of confidence, p-hacking and the future of psychology
 
The replication crisis: are P-values the problem and are Bayes factors the so...
The replication crisis: are P-values the problem and are Bayes factors the so...The replication crisis: are P-values the problem and are Bayes factors the so...
The replication crisis: are P-values the problem and are Bayes factors the so...
 
Severe Testing: The Key to Error Correction
Severe Testing: The Key to Error CorrectionSevere Testing: The Key to Error Correction
Severe Testing: The Key to Error Correction
 
Philosophy of Science and Philosophy of Statistics
Philosophy of Science and Philosophy of StatisticsPhilosophy of Science and Philosophy of Statistics
Philosophy of Science and Philosophy of Statistics
 
Probability Homework Help
Probability Homework Help Probability Homework Help
Probability Homework Help
 
P-Value "Reforms": Fixing Science or Threat to Replication and Falsification
P-Value "Reforms": Fixing Science or Threat to Replication and FalsificationP-Value "Reforms": Fixing Science or Threat to Replication and Falsification
P-Value "Reforms": Fixing Science or Threat to Replication and Falsification
 
D. Mayo: Philosophical Interventions in the Statistics Wars
D. Mayo: Philosophical Interventions in the Statistics WarsD. Mayo: Philosophical Interventions in the Statistics Wars
D. Mayo: Philosophical Interventions in the Statistics Wars
 
Mayod@psa 21(na)
Mayod@psa 21(na)Mayod@psa 21(na)
Mayod@psa 21(na)
 
Paradoxes and Fallacies - Resolving some well-known puzzles with Bayesian net...
Paradoxes and Fallacies - Resolving some well-known puzzles with Bayesian net...Paradoxes and Fallacies - Resolving some well-known puzzles with Bayesian net...
Paradoxes and Fallacies - Resolving some well-known puzzles with Bayesian net...
 
Applied statistics part 2
Applied statistics  part 2Applied statistics  part 2
Applied statistics part 2
 
Frequentist inference only seems easy By John Mount
Frequentist inference only seems easy By John MountFrequentist inference only seems easy By John Mount
Frequentist inference only seems easy By John Mount
 
D. G. Mayo Columbia slides for Workshop on Probability &Learning
D. G. Mayo Columbia slides for Workshop on Probability &LearningD. G. Mayo Columbia slides for Workshop on Probability &Learning
D. G. Mayo Columbia slides for Workshop on Probability &Learning
 

More from Alan Dix

AI for HCI – could this be a better title if I’d asked ChatGPT
AI for HCI – could this be a better title if I’d asked ChatGPTAI for HCI – could this be a better title if I’d asked ChatGPT
AI for HCI – could this be a better title if I’d asked ChatGPT
Alan Dix
 
Just Counting – a tool ecosystem for personal numeric information
Just Counting – a tool ecosystem for personal numeric informationJust Counting – a tool ecosystem for personal numeric information
Just Counting – a tool ecosystem for personal numeric information
Alan Dix
 
A flexible QR-code infrastructure for heritage
A flexible QR-code infrastructure for heritageA flexible QR-code infrastructure for heritage
A flexible QR-code infrastructure for heritage
Alan Dix
 
Epistemic Interaction - tuning interfaces to provide information for AI support
Epistemic Interaction - tuning interfaces to provide information for AI supportEpistemic Interaction - tuning interfaces to provide information for AI support
Epistemic Interaction - tuning interfaces to provide information for AI support
Alan Dix
 
AI and the Humanities – provocations – The Arts, Humanities & Responsible AI...
AI and the Humanities – provocations  – The Arts, Humanities & Responsible AI...AI and the Humanities – provocations  – The Arts, Humanities & Responsible AI...
AI and the Humanities – provocations – The Arts, Humanities & Responsible AI...
Alan Dix
 
Can AI do good? at 'offtheCanvas' India HCI prelude
Can AI do good? at 'offtheCanvas' India HCI preludeCan AI do good? at 'offtheCanvas' India HCI prelude
Can AI do good? at 'offtheCanvas' India HCI prelude
Alan Dix
 
CDT Away Day Talk: Qualitative–Quantitative reasoning and lightweight numbers
CDT Away Day Talk: Qualitative–Quantitative reasoning and lightweight numbersCDT Away Day Talk: Qualitative–Quantitative reasoning and lightweight numbers
CDT Away Day Talk: Qualitative–Quantitative reasoning and lightweight numbers
Alan Dix
 
Swan(sea) Song – personal research during my six years at Swansea ... and bey...
Swan(sea) Song – personal research during my six years at Swansea ... and bey...Swan(sea) Song – personal research during my six years at Swansea ... and bey...
Swan(sea) Song – personal research during my six years at Swansea ... and bey...
Alan Dix
 
From Family Reminiscence to Scholarly Archive .
From Family Reminiscence to Scholarly Archive .From Family Reminiscence to Scholarly Archive .
From Family Reminiscence to Scholarly Archive .
Alan Dix
 
Human-Centred Artificial Intelligence – Malta 2024
Human-Centred Artificial Intelligence – Malta 2024Human-Centred Artificial Intelligence – Malta 2024
Human-Centred Artificial Intelligence – Malta 2024
Alan Dix
 
The future of UX design support tools - talk Paris March 2024
The future of UX design support tools - talk Paris March 2024The future of UX design support tools - talk Paris March 2024
The future of UX design support tools - talk Paris March 2024
Alan Dix
 
Qualitative–Quantitative reasoning and lightweight numbers
Qualitative–Quantitative reasoning and lightweight numbersQualitative–Quantitative reasoning and lightweight numbers
Qualitative–Quantitative reasoning and lightweight numbers
Alan Dix
 
Invited talk at Diversifying Knowledge Production in HCI
Invited talk at Diversifying Knowledge Production in HCIInvited talk at Diversifying Knowledge Production in HCI
Invited talk at Diversifying Knowledge Production in HCI
Alan Dix
 
Exceptional Experiences for Everyone
Exceptional Experiences for EveryoneExceptional Experiences for Everyone
Exceptional Experiences for Everyone
Alan Dix
 
Inclusivity and AI: opportunity or threat
Inclusivity and AI: opportunity or threatInclusivity and AI: opportunity or threat
Inclusivity and AI: opportunity or threat
Alan Dix
 
Hidden Figures architectural challenges to expose parameters lost in code
Hidden Figures architectural challenges to expose parameters lost in codeHidden Figures architectural challenges to expose parameters lost in code
Hidden Figures architectural challenges to expose parameters lost in code
Alan Dix
 
ChatGPT, Culture and Creativity simulacrum and alterity
ChatGPT, Culture and Creativity simulacrum and alterityChatGPT, Culture and Creativity simulacrum and alterity
ChatGPT, Culture and Creativity simulacrum and alterity
Alan Dix
 
Why pandemics and climate change are hard to understand and make decision mak...
Why pandemics and climate change are hard to understand and make decision mak...Why pandemics and climate change are hard to understand and make decision mak...
Why pandemics and climate change are hard to understand and make decision mak...
Alan Dix
 
Beyond the Wireframe: tools to design, analyse and prototype physical devices
Beyond the Wireframe: tools to design, analyse and prototype physical devicesBeyond the Wireframe: tools to design, analyse and prototype physical devices
Beyond the Wireframe: tools to design, analyse and prototype physical devices
Alan Dix
 
Forever Cyborgs – a long view on physical-digital interaction
Forever Cyborgs – a long view on physical-digital interactionForever Cyborgs – a long view on physical-digital interaction
Forever Cyborgs – a long view on physical-digital interaction
Alan Dix
 

More from Alan Dix (20)

AI for HCI – could this be a better title if I’d asked ChatGPT
AI for HCI – could this be a better title if I’d asked ChatGPTAI for HCI – could this be a better title if I’d asked ChatGPT
AI for HCI – could this be a better title if I’d asked ChatGPT
 
Just Counting – a tool ecosystem for personal numeric information
Just Counting – a tool ecosystem for personal numeric informationJust Counting – a tool ecosystem for personal numeric information
Just Counting – a tool ecosystem for personal numeric information
 
A flexible QR-code infrastructure for heritage
A flexible QR-code infrastructure for heritageA flexible QR-code infrastructure for heritage
A flexible QR-code infrastructure for heritage
 
Epistemic Interaction - tuning interfaces to provide information for AI support
Epistemic Interaction - tuning interfaces to provide information for AI supportEpistemic Interaction - tuning interfaces to provide information for AI support
Epistemic Interaction - tuning interfaces to provide information for AI support
 
AI and the Humanities – provocations – The Arts, Humanities & Responsible AI...
AI and the Humanities – provocations  – The Arts, Humanities & Responsible AI...AI and the Humanities – provocations  – The Arts, Humanities & Responsible AI...
AI and the Humanities – provocations – The Arts, Humanities & Responsible AI...
 
Can AI do good? at 'offtheCanvas' India HCI prelude
Can AI do good? at 'offtheCanvas' India HCI preludeCan AI do good? at 'offtheCanvas' India HCI prelude
Can AI do good? at 'offtheCanvas' India HCI prelude
 
CDT Away Day Talk: Qualitative–Quantitative reasoning and lightweight numbers
CDT Away Day Talk: Qualitative–Quantitative reasoning and lightweight numbersCDT Away Day Talk: Qualitative–Quantitative reasoning and lightweight numbers
CDT Away Day Talk: Qualitative–Quantitative reasoning and lightweight numbers
 
Swan(sea) Song – personal research during my six years at Swansea ... and bey...
Swan(sea) Song – personal research during my six years at Swansea ... and bey...Swan(sea) Song – personal research during my six years at Swansea ... and bey...
Swan(sea) Song – personal research during my six years at Swansea ... and bey...
 
From Family Reminiscence to Scholarly Archive .
From Family Reminiscence to Scholarly Archive .From Family Reminiscence to Scholarly Archive .
From Family Reminiscence to Scholarly Archive .
 
Human-Centred Artificial Intelligence – Malta 2024
Human-Centred Artificial Intelligence – Malta 2024Human-Centred Artificial Intelligence – Malta 2024
Human-Centred Artificial Intelligence – Malta 2024
 
The future of UX design support tools - talk Paris March 2024
The future of UX design support tools - talk Paris March 2024The future of UX design support tools - talk Paris March 2024
The future of UX design support tools - talk Paris March 2024
 
Qualitative–Quantitative reasoning and lightweight numbers
Qualitative–Quantitative reasoning and lightweight numbersQualitative–Quantitative reasoning and lightweight numbers
Qualitative–Quantitative reasoning and lightweight numbers
 
Invited talk at Diversifying Knowledge Production in HCI
Invited talk at Diversifying Knowledge Production in HCIInvited talk at Diversifying Knowledge Production in HCI
Invited talk at Diversifying Knowledge Production in HCI
 
Exceptional Experiences for Everyone
Exceptional Experiences for EveryoneExceptional Experiences for Everyone
Exceptional Experiences for Everyone
 
Inclusivity and AI: opportunity or threat
Inclusivity and AI: opportunity or threatInclusivity and AI: opportunity or threat
Inclusivity and AI: opportunity or threat
 
Hidden Figures architectural challenges to expose parameters lost in code
Hidden Figures architectural challenges to expose parameters lost in codeHidden Figures architectural challenges to expose parameters lost in code
Hidden Figures architectural challenges to expose parameters lost in code
 
ChatGPT, Culture and Creativity simulacrum and alterity
ChatGPT, Culture and Creativity simulacrum and alterityChatGPT, Culture and Creativity simulacrum and alterity
ChatGPT, Culture and Creativity simulacrum and alterity
 
Why pandemics and climate change are hard to understand and make decision mak...
Why pandemics and climate change are hard to understand and make decision mak...Why pandemics and climate change are hard to understand and make decision mak...
Why pandemics and climate change are hard to understand and make decision mak...
 
Beyond the Wireframe: tools to design, analyse and prototype physical devices
Beyond the Wireframe: tools to design, analyse and prototype physical devicesBeyond the Wireframe: tools to design, analyse and prototype physical devices
Beyond the Wireframe: tools to design, analyse and prototype physical devices
 
Forever Cyborgs – a long view on physical-digital interaction
Forever Cyborgs – a long view on physical-digital interactionForever Cyborgs – a long view on physical-digital interaction
Forever Cyborgs – a long view on physical-digital interaction
 

Recently uploaded

原版一比一利兹贝克特大学毕业证(LeedsBeckett毕业证书)如何办理
原版一比一利兹贝克特大学毕业证(LeedsBeckett毕业证书)如何办理原版一比一利兹贝克特大学毕业证(LeedsBeckett毕业证书)如何办理
原版一比一利兹贝克特大学毕业证(LeedsBeckett毕业证书)如何办理
wyddcwye1
 
DSSML24_tspann_CodelessGenerativeAIPipelines
DSSML24_tspann_CodelessGenerativeAIPipelinesDSSML24_tspann_CodelessGenerativeAIPipelines
DSSML24_tspann_CodelessGenerativeAIPipelines
Timothy Spann
 
Orchestrating the Future: Navigating Today's Data Workflow Challenges with Ai...
Orchestrating the Future: Navigating Today's Data Workflow Challenges with Ai...Orchestrating the Future: Navigating Today's Data Workflow Challenges with Ai...
Orchestrating the Future: Navigating Today's Data Workflow Challenges with Ai...
Kaxil Naik
 
End-to-end pipeline agility - Berlin Buzzwords 2024
End-to-end pipeline agility - Berlin Buzzwords 2024End-to-end pipeline agility - Berlin Buzzwords 2024
End-to-end pipeline agility - Berlin Buzzwords 2024
Lars Albertsson
 
在线办理(英国UCA毕业证书)创意艺术大学毕业证在读证明一模一样
在线办理(英国UCA毕业证书)创意艺术大学毕业证在读证明一模一样在线办理(英国UCA毕业证书)创意艺术大学毕业证在读证明一模一样
在线办理(英国UCA毕业证书)创意艺术大学毕业证在读证明一模一样
v7oacc3l
 
原版制作(unimelb毕业证书)墨尔本大学毕业证Offer一模一样
原版制作(unimelb毕业证书)墨尔本大学毕业证Offer一模一样原版制作(unimelb毕业证书)墨尔本大学毕业证Offer一模一样
原版制作(unimelb毕业证书)墨尔本大学毕业证Offer一模一样
ihavuls
 
writing report business partner b1+ .pdf
writing report business partner b1+ .pdfwriting report business partner b1+ .pdf
writing report business partner b1+ .pdf
VyNguyen709676
 
原版一比一多伦多大学毕业证(UofT毕业证书)如何办理
原版一比一多伦多大学毕业证(UofT毕业证书)如何办理原版一比一多伦多大学毕业证(UofT毕业证书)如何办理
原版一比一多伦多大学毕业证(UofT毕业证书)如何办理
mkkikqvo
 
Beyond the Basics of A/B Tests: Highly Innovative Experimentation Tactics You...
Beyond the Basics of A/B Tests: Highly Innovative Experimentation Tactics You...Beyond the Basics of A/B Tests: Highly Innovative Experimentation Tactics You...
Beyond the Basics of A/B Tests: Highly Innovative Experimentation Tactics You...
Aggregage
 
Predictably Improve Your B2B Tech Company's Performance by Leveraging Data
Predictably Improve Your B2B Tech Company's Performance by Leveraging DataPredictably Improve Your B2B Tech Company's Performance by Leveraging Data
Predictably Improve Your B2B Tech Company's Performance by Leveraging Data
Kiwi Creative
 
STATATHON: Unleashing the Power of Statistics in a 48-Hour Knowledge Extravag...
STATATHON: Unleashing the Power of Statistics in a 48-Hour Knowledge Extravag...STATATHON: Unleashing the Power of Statistics in a 48-Hour Knowledge Extravag...
STATATHON: Unleashing the Power of Statistics in a 48-Hour Knowledge Extravag...
sameer shah
 
Population Growth in Bataan: The effects of population growth around rural pl...
Population Growth in Bataan: The effects of population growth around rural pl...Population Growth in Bataan: The effects of population growth around rural pl...
Population Growth in Bataan: The effects of population growth around rural pl...
Bill641377
 
一比一原版(Unimelb毕业证书)墨尔本大学毕业证如何办理
一比一原版(Unimelb毕业证书)墨尔本大学毕业证如何办理一比一原版(Unimelb毕业证书)墨尔本大学毕业证如何办理
一比一原版(Unimelb毕业证书)墨尔本大学毕业证如何办理
xclpvhuk
 
一比一原版(UCSF文凭证书)旧金山分校毕业证如何办理
一比一原版(UCSF文凭证书)旧金山分校毕业证如何办理一比一原版(UCSF文凭证书)旧金山分校毕业证如何办理
一比一原版(UCSF文凭证书)旧金山分校毕业证如何办理
nuttdpt
 
4th Modern Marketing Reckoner by MMA Global India & Group M: 60+ experts on W...
4th Modern Marketing Reckoner by MMA Global India & Group M: 60+ experts on W...4th Modern Marketing Reckoner by MMA Global India & Group M: 60+ experts on W...
4th Modern Marketing Reckoner by MMA Global India & Group M: 60+ experts on W...
Social Samosa
 
Global Situational Awareness of A.I. and where its headed
Global Situational Awareness of A.I. and where its headedGlobal Situational Awareness of A.I. and where its headed
Global Situational Awareness of A.I. and where its headed
vikram sood
 
A presentation that explain the Power BI Licensing
A presentation that explain the Power BI LicensingA presentation that explain the Power BI Licensing
A presentation that explain the Power BI Licensing
AlessioFois2
 
Analysis insight about a Flyball dog competition team's performance
Analysis insight about a Flyball dog competition team's performanceAnalysis insight about a Flyball dog competition team's performance
Analysis insight about a Flyball dog competition team's performance
roli9797
 
Udemy_2024_Global_Learning_Skills_Trends_Report (1).pdf
Udemy_2024_Global_Learning_Skills_Trends_Report (1).pdfUdemy_2024_Global_Learning_Skills_Trends_Report (1).pdf
Udemy_2024_Global_Learning_Skills_Trends_Report (1).pdf
Fernanda Palhano
 
Experts live - Improving user adoption with AI
Experts live - Improving user adoption with AIExperts live - Improving user adoption with AI
Experts live - Improving user adoption with AI
jitskeb
 

Recently uploaded (20)

原版一比一利兹贝克特大学毕业证(LeedsBeckett毕业证书)如何办理
原版一比一利兹贝克特大学毕业证(LeedsBeckett毕业证书)如何办理原版一比一利兹贝克特大学毕业证(LeedsBeckett毕业证书)如何办理
原版一比一利兹贝克特大学毕业证(LeedsBeckett毕业证书)如何办理
 
DSSML24_tspann_CodelessGenerativeAIPipelines
DSSML24_tspann_CodelessGenerativeAIPipelinesDSSML24_tspann_CodelessGenerativeAIPipelines
DSSML24_tspann_CodelessGenerativeAIPipelines
 
Orchestrating the Future: Navigating Today's Data Workflow Challenges with Ai...
Orchestrating the Future: Navigating Today's Data Workflow Challenges with Ai...Orchestrating the Future: Navigating Today's Data Workflow Challenges with Ai...
Orchestrating the Future: Navigating Today's Data Workflow Challenges with Ai...
 
End-to-end pipeline agility - Berlin Buzzwords 2024
End-to-end pipeline agility - Berlin Buzzwords 2024End-to-end pipeline agility - Berlin Buzzwords 2024
End-to-end pipeline agility - Berlin Buzzwords 2024
 
在线办理(英国UCA毕业证书)创意艺术大学毕业证在读证明一模一样
在线办理(英国UCA毕业证书)创意艺术大学毕业证在读证明一模一样在线办理(英国UCA毕业证书)创意艺术大学毕业证在读证明一模一样
在线办理(英国UCA毕业证书)创意艺术大学毕业证在读证明一模一样
 
原版制作(unimelb毕业证书)墨尔本大学毕业证Offer一模一样
原版制作(unimelb毕业证书)墨尔本大学毕业证Offer一模一样原版制作(unimelb毕业证书)墨尔本大学毕业证Offer一模一样
原版制作(unimelb毕业证书)墨尔本大学毕业证Offer一模一样
 
writing report business partner b1+ .pdf
writing report business partner b1+ .pdfwriting report business partner b1+ .pdf
writing report business partner b1+ .pdf
 
原版一比一多伦多大学毕业证(UofT毕业证书)如何办理
原版一比一多伦多大学毕业证(UofT毕业证书)如何办理原版一比一多伦多大学毕业证(UofT毕业证书)如何办理
原版一比一多伦多大学毕业证(UofT毕业证书)如何办理
 
Beyond the Basics of A/B Tests: Highly Innovative Experimentation Tactics You...
Beyond the Basics of A/B Tests: Highly Innovative Experimentation Tactics You...Beyond the Basics of A/B Tests: Highly Innovative Experimentation Tactics You...
Beyond the Basics of A/B Tests: Highly Innovative Experimentation Tactics You...
 
Predictably Improve Your B2B Tech Company's Performance by Leveraging Data
Predictably Improve Your B2B Tech Company's Performance by Leveraging DataPredictably Improve Your B2B Tech Company's Performance by Leveraging Data
Predictably Improve Your B2B Tech Company's Performance by Leveraging Data
 
STATATHON: Unleashing the Power of Statistics in a 48-Hour Knowledge Extravag...
STATATHON: Unleashing the Power of Statistics in a 48-Hour Knowledge Extravag...STATATHON: Unleashing the Power of Statistics in a 48-Hour Knowledge Extravag...
STATATHON: Unleashing the Power of Statistics in a 48-Hour Knowledge Extravag...
 
Population Growth in Bataan: The effects of population growth around rural pl...
Population Growth in Bataan: The effects of population growth around rural pl...Population Growth in Bataan: The effects of population growth around rural pl...
Population Growth in Bataan: The effects of population growth around rural pl...
 
一比一原版(Unimelb毕业证书)墨尔本大学毕业证如何办理
一比一原版(Unimelb毕业证书)墨尔本大学毕业证如何办理一比一原版(Unimelb毕业证书)墨尔本大学毕业证如何办理
一比一原版(Unimelb毕业证书)墨尔本大学毕业证如何办理
 
一比一原版(UCSF文凭证书)旧金山分校毕业证如何办理
一比一原版(UCSF文凭证书)旧金山分校毕业证如何办理一比一原版(UCSF文凭证书)旧金山分校毕业证如何办理
一比一原版(UCSF文凭证书)旧金山分校毕业证如何办理
 
4th Modern Marketing Reckoner by MMA Global India & Group M: 60+ experts on W...
4th Modern Marketing Reckoner by MMA Global India & Group M: 60+ experts on W...4th Modern Marketing Reckoner by MMA Global India & Group M: 60+ experts on W...
4th Modern Marketing Reckoner by MMA Global India & Group M: 60+ experts on W...
 
Global Situational Awareness of A.I. and where its headed
Global Situational Awareness of A.I. and where its headedGlobal Situational Awareness of A.I. and where its headed
Global Situational Awareness of A.I. and where its headed
 
A presentation that explain the Power BI Licensing
A presentation that explain the Power BI LicensingA presentation that explain the Power BI Licensing
A presentation that explain the Power BI Licensing
 
Analysis insight about a Flyball dog competition team's performance
Analysis insight about a Flyball dog competition team's performanceAnalysis insight about a Flyball dog competition team's performance
Analysis insight about a Flyball dog competition team's performance
 
Udemy_2024_Global_Learning_Skills_Trends_Report (1).pdf
Udemy_2024_Global_Learning_Skills_Trends_Report (1).pdfUdemy_2024_Global_Learning_Skills_Trends_Report (1).pdf
Udemy_2024_Global_Learning_Skills_Trends_Report (1).pdf
 
Experts live - Improving user adoption with AI
Experts live - Improving user adoption with AIExperts live - Improving user adoption with AI
Experts live - Improving user adoption with AI
 

Making Sense of Statistics in HCI: part 2 - doing it

  • 1. Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix Making Sense of Statistics in HCI Part 2 – Doing It if not p then what Alan Dix http://alandix.com/statistics/
  • 2. Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix probing the unknown quantified uncertainty accepting unknowing and common sense.
  • 3. Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix recall … the job of statistics
  • 4. Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix conditionals and likelihood probability road will be busy vs probability busy given 4am on Sunday probability die is a 6 vs prob a 6 given it is 4 or bigger conditional probability is probability given some knowledge
  • 5. Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix likelihood likelihood is probability of measurement given unknown things in the world (conditional probability) e.g. prob. of six heads given coin is fair = 1/64 … but is it fair?
  • 6. Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix counterfactual statistics turns this round from likelihood of measurement probabilistic knowledge given the unknown world to uncertain knowledge of the unknown world based on measurement … but in various different ways
  • 7. Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix statistical reasoning
  • 8. Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix types of statistics • hypothesis testing (the dreaded p!) – robust but confusing • confidence intervals – powerful but underused • Bayesian stats – mathematically clean but fragile – issues of independence
  • 9. Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix
  • 10. Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix hypothesis testing the ubiquitous p
  • 11. Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix significance test hypotheses: H1 – what want to show H0 – null hypothesis (to disprove) idea (when experiment/study successful) if H0 were true then observed effect is very unlikely therefore H1 is (likely to be) true
  • 12. Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix 5% significance level? it says: if H0 were true then probability observed effect happening by chance is less than 1 in 20 (5%) so H0 is unlikely to be true and H1 is likely to be true
  • 13. Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix does not say 8 probability of H0 is < 1 in 20 8 probability of H1 is > 0.95 8 effect is large or important i.e. significant in the real sense
  • 14. Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix all it says 4 if H0 were true ... ... effect is unlikely ( prob. < 1 in 20)
  • 15. Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix non-significant result can NEVER reason: non-significant result  H1 false things are the same all you can say is: H1 is not statistically proven
  • 16. Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix
  • 17. Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix confidence intervals bounds of uncertainty
  • 18. Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix proving equality non-significant – not proved different real difference may always be smaller than experimental error  can never prove equality but can put bounds on inequality
  • 19. Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix confidence interval bound on true value – same theory as p values e.g. mean of data is 0.3 95% confidence interval is [–0.7,1.3] says if the real value not in the range [–0.7,1.3] probability of seeing observed data is less than 5%
  • 20. Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix counterfactuals 95% confidence interval is [–0.7,1.3] does not say: there is 95% probability that the real mean is in the range [–0.7,1.3] it either is or it isn’t! all it says: probability of seeing the observed data if real value outside the range is less than 5%
  • 21. Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix proven … what? H0: no difference (real mean is zero) experimental result: mean is 0.3 significance test: n.s. at 5% – so what? 95% confidence interval: [–0.7,1.3] ? is 1.3 is an important difference
  • 22. Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix … and don’t forget … you still need to say what test/distribution – e.g. Student’s T how many – degrees of freedom it is still uncertain the real value could be outside the interval
  • 23. Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix
  • 24. Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix Bayesian statistics putting a number on it
  • 25. Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix traditional statistics probability of having antennae: if Martian = 1 if human = 0.001 hypotheses: H0: no Martians in the High Street H1: the Martians have landed ! you meet someone with antennae reject null hypothesis p ≤ 0.1% call the Men in Black!
  • 26. Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix Bayesian thinking prior probability of meeting: a Martian = 0.000 001 a human = 0.999 999 probability of having antennae: Martian = 1; human = 0.001 ! you meet someone with antennae posterior probability of being: a Martian ≈ 0.001 a human ≈ 0.999 what you think before evidence
  • 27. Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix Bayesian inference combine prior with likelihood prior distribution posterior
  • 28. Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix Bayesian issues how do you get the prior? sometimes doesn't matter too much! traditional stats rather like uniform prior handling multiple evidence can re-apply iteratively problems with interactions internecine warfare traditionalists and Bayesians often fight ;)
  • 29. Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix
  • 30. Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix philosophical differences probing the unknown
  • 31. Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix philosophical stance ? what do you know about the world traditional statistics – assume nothing! – reason from unknowledge Bayesian statistics – 'prior' probabilities – reason from guess-timates
  • 32. Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix uncertain and unknown assumptions: traditional – NO knowledge of real value Bayesian – precise probability distribution in reality … sort of half know traditional – choice of acceptable p Bayesian – ‘spread’ distributions (uniform, Cauchy) ideally should do sensitivity analysis … but no one does 
  • 33. Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix
  • 34. Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix issues from “what is worse” to really significant
  • 35. Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix issues likelihood is conditional probability what to do with 'non-sig'. priors for experiments? finding or verifying? significance vs effect size vs power idea of 'worse' for measures
  • 36. Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix example: playing cards run of cards badly shuffled?
  • 37. Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix example: playing cards (2) probability three cards make a decreasing run: – first card needs to be 3 or bigger: 11/13 – second card exactly the next one down: 1/51 – third card exactly one down again: 1/50 = 11/33150 approx 1 in 3000 – p <0.001 any card except ace or 2 next card down next again
  • 38. Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix example: playing cards (3) but equally surprised if … – either direction – 15 positions for the first card = 15x2x11/33150 approx 1 in 100 and then … how often do I play?
  • 39. Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix dangers if it can go wrong then with 95% confidence …
  • 40. Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix dangers avoiding cherry picking multiple tests, multiple stats outliers, post-hoc hypotheses non-independent effects e.g. fat and sugar correlated features e.g. weight, diet and exercise including trying both traditional and Bayes! the same UI but ‘not’ consistent
  • 41. Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix
  • 42. Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix so which is it? to p or not to p Bayes forward?
  • 43. Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix the statistical crisis? mostly well known problems with known solutions maybe more people doing stats with less training? some of the papers about it use flawed stats!
  • 44. Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix on balance (my advice!) traditional more conservative – if done properly Bayes many of the same problems as trad. plus a few more … needs expertise for most things stick with trad. … but always confidence intervals as well as p for special things go for Bayes if you know prior, decision making with costs, internal algorithms
  • 45. Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix
  • 46. Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix a quick experiment … but are they fair
  • 47. Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix calculations – six coins given coin is fair: probability six heads = 1/26 = 1/64 probability six tails = 1/26 = 1/64 probability either = 2/64 ~ 3% H0 – coin is fair H1 – coin is not-fair likelihood ( HHHHHH or TTTTTT | H0 ) < 5%
  • 48. Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix your experiment toss 6 coins record how many heads or tails if HHHHHH or TTTTTTT you can reject H0 with p< 5% repeat exp. 2 more times with rest of coins
  • 49. Making Sense of Statistics in HCI: From P to Bayes and Beyond – Alan Dix