Bayes 2010

Embed Size (px)

Citation preview

  • 7/28/2019 Bayes 2010

    1/57

    PHILOSOPHY OF SCIENCE:

    Bayesian inference

    Zoltn Dienes

    Thomas Bayes1702-1761

  • 7/28/2019 Bayes 2010

    2/57

    Subjective probability:

    Personal conviction in an opinionto which a number is assignedthat obeys the axioms of probability.

    Probabilities reside in the mind of the individual not the external

    world.

    Your subjective probability/conviction in a proposition is just up to

    you

    but you must revise your probability in the light of data in waysconsistent with the axioms of probability.

  • 7/28/2019 Bayes 2010

    3/57

    Personal probabilities:

    Which of the following alternatives would you prefer? If yourchoice turns out to be true I will pay you 10 pounds:

    1. It will rain tomorrow in Brighton2. I have a bag with one blue chip and one red chip. I

    randomly pull out one chip and it is red.

  • 7/28/2019 Bayes 2010

    4/57

    Personal probabilities:

    Which of the following alternatives would you prefer? If yourchoice turns out to be true I will pay you 10 pounds:

    1. It will rain tomorrow in Brighton2. I have a bag with one blue chip and one red chip. I

    randomly pull out one and it is red.

    If you chose 2, your p(it will rain) is less than .5

    If you chose 1, your p(it will rain) is greater than .5

  • 7/28/2019 Bayes 2010

    5/57

    Assume you think it more likely than not that it will rain.

    Now which do you chose? I will give you 10 pounds if your chosen

    statement turns out correct.

    1. It will rain tomorrow in Brighton.

    2. I have a bag with 3 red chips and 1 blue. I randomly pick a chip

    and it is red.

  • 7/28/2019 Bayes 2010

    6/57

    Assume you think it more likely than not that it will rain.

    Now which do you chose? I will give you 10 pounds if your chosen

    statement turns out correct.

    1. It will rain tomorrow in Brighton.

    2. I have a bag with 3 red chips and 1 blue. I randomly pick a chip

    and it is red.

    If you chose 2, your p(it will rain) is less than .75 (but more than .5)

    If you chose 1, your p(it will rain) is more than .75.

  • 7/28/2019 Bayes 2010

    7/57

    By imagining a bag with differing numbers of red and blue

    chips, you can make your personal probability as precise as

    you like.

  • 7/28/2019 Bayes 2010

    8/57

    This is a notion of probability that applies to the truth of theories

    (Rememberobjective probability does not apply to theories)

    So that means we can answer questions about p(H)the probability

    of a hypothesis being trueand also p(H|D)the probability of ahypothesis given data (which we cannot do on the Neyman-Pearson

    approach).

  • 7/28/2019 Bayes 2010

    9/57

    Consider a theory you might be testing in your project.

    What is your personal probability that the theory is true?

  • 7/28/2019 Bayes 2010

    10/57

    ODDS in favour of a theory = P(theory is true)/P(theory is false)

    For example, for the theory you may be testing extroverts have

    high cortical arousal

    Then you might think (its completely up to you)

    P(theory is true) = 0.5

    It follows you think that

    P(theory is false) = 1P(theory is true) = 0.5

    So odds in favour of theory = 0.5/0.5 = 1

    Even odds

  • 7/28/2019 Bayes 2010

    11/57

    If you think

    P(theory is true) = 0.8

    Then you must think

    P(theory is false) = 0.2

    So your odds in favour of the theory are 0.8/0.2 = 4 (4 to 1)

    What are your odds in favour of the theory you are testing in your

    project?

  • 7/28/2019 Bayes 2010

    12/57

    Odds before you have collected your data are called yourprior odds

    Experimental results tell you by how much to increase your odds (the

    Bayes Factor, B)

    Odds after collecting your data are called yourposterior odds

    Posterior odds = B * Prior odds

    For example, a Bayesian analysis might lead to a B of 5.What would your posterior odds for your project hypothesis be?

  • 7/28/2019 Bayes 2010

    13/57

    You can convert back to probabilities with the formula:

    P (Theory is true) = odds/(odds + 1)

    So if your prior odds had been 1 and the Bayes factor 5

    Then posterior odds = 5

    Your posterior probability of your theory being true = 5/6 = .83

    Not a black and white decision like significance testing (conclude

    one thing if p = .048 and another if p = .052)

  • 7/28/2019 Bayes 2010

    14/57

    While your prior odds are subjective

    ALL assumptions going into B are explicit, and can be argued aboutuntil - in principle - agreed on by everyone

    Then B is to that degree objective

    Contrast Neyman Pearsone.g. nobody might know what the actual

    stopping rule was. Assumptions are not necessarily explicit. People

    can cheat in this way with Neyman Pearson one cant in Bayes.

  • 7/28/2019 Bayes 2010

    15/57

    If B is greater than 1 then the data supported your experimental

    hypothesis over the null

    If B is less than 1, then the data supported the null hypothesis over

    the experimental one

    If B = about 1, experiment was not sensitive.

    (Automatically get a notion of sensitivity;

    contrast: just relying on p values in significance testing.)

  • 7/28/2019 Bayes 2010

    16/57

    For each experiment you run, just keep updating your

    odds/personal probability by multiplying by each new Bayes

    factor

    No need for p values at all!

    No need for power calculations!No need for critical values of t-tests!

    And as we will see:

    No need for post hoc tests!No need to plan number of subjects in advance!

  • 7/28/2019 Bayes 2010

    17/57

    To know which theory data support need to know what the

    theories predict

    The null is normally the prediction of e.g. no difference

    Need to decide what difference or range of differences areconsistent with ones theory

    Difficult -but forces one to think clearly about ones theory.

    And this is what is required for power calculations, so

    psychologists should be doing it anyway. Bayes forces one to.

  • 7/28/2019 Bayes 2010

    18/57

    Contrasts with Neyman-Pearson:

    1. A) A non-significant result can mean one should decrease onesconfidence in the null hypothesis.

  • 7/28/2019 Bayes 2010

    19/57

    0 5 10 15 20

    SE

    Difference between conditions

    Prediction of

    nullMean of data

    A theory predicts a difference between conditions, but a null

    result is obtained.

    Should one be less confident in the theory relative to the null hypothesis?

  • 7/28/2019 Bayes 2010

    20/57

    What does the theory predict?

  • 7/28/2019 Bayes 2010

    21/57

    0 5 10 15 20

    SE

    Difference between conditions

    What does the theory predict?

    Prediction of

    null

    Mean of data

    If theory predicts a difference of

    10 units, should slightly

    increase confidence in theoryover null

  • 7/28/2019 Bayes 2010

    22/57

    0 5 10 15 20

    SE

    Difference between conditions

    What does the theory predict?

    Prediction of

    null

    Mean of data

    But if theory predicts a

    difference of 20 units, data

    should decrease confidence intheory over null

  • 7/28/2019 Bayes 2010

    23/57

    NB It is simplistic to think the theory predicts only one

    valuebut Bayes can deal with a range of values

    (it typically does)

  • 7/28/2019 Bayes 2010

    24/57

    Contrasts with Neyman-Pearson:

    1. A) A non-significant result can mean one should decrease onesconfidence in the null.

    Conversely

    B) A significant result (i.e. accepting the experimental

    hypothesis on Neyman Pearson) can mean one should increase

    ones confidence in the null

  • 7/28/2019 Bayes 2010

    25/57

    0 5 10 15 20

    SE

    Difference between conditions

    Prediction of null

    Data mean

    A theory predicts a difference between conditions, and one is

    obtained. Should one be more confident in the theory relative to

    the null?

  • 7/28/2019 Bayes 2010

    26/57

    What does the theory predict?

  • 7/28/2019 Bayes 2010

    27/57

    0 5 10 15 20

    SE

    Difference between

    conditions

    Prediction of

    null

    Data

    mean If theory predicts e.g.

    20 or more, then

    significant result is

    evidence against theory

  • 7/28/2019 Bayes 2010

    28/57

    More normally it is because the theory is vague that a significant

    result leads to less confidence in the theory

    i.e. many of the possible predictions of the theory are far from the

    data

  • 7/28/2019 Bayes 2010

    29/57

    i.e. because the theory makes no precise prediction at all!

    The fact the null makes a precise prediction makes it more

    falsifiable, and hence preferredthis CAN mean one should

    be more confident in null

    Bayesian inference would encourage scientists to make more

    falsifiable theories; significance testing (without power)

    encourages theories of low falsifiability

  • 7/28/2019 Bayes 2010

    30/57

    2. On Neyman Pearson, it matters whether you formulated your

    hypothesis before or after looking at the data.

    Post hoc vs planned comparisons

    Predictions made in advance of rather than before looking at the

    data are treated differently

    Bayesian inference: It does not matter what day of the week you

    thought of your theory

    The evidence for your theory is just as strong regardless of its timing

  • 7/28/2019 Bayes 2010

    31/57

    3. On Neyman Pearson standardlyyou should plan in advance how

    many subjects you will run.

    If you just miss out on a significant result you cant just run 10 moresubjects and test again.

    You cannot run until you get a significant result.

    Bayes: It does not matter when you decide to stop running subjects.

    You can always run more subjects if you think it will help.

  • 7/28/2019 Bayes 2010

    32/57

    4. On Neyman Pearsonyou must correct for how many tests you

    conduct in total.

    For example, if you ran 100 correlations and 4 were just significant,researchers would not try to interpret those significant results.

    On Bayes, it does not matter how many other statistical hypotheses you

    investigated. All that matters is the data relevant to each hypothesis

    under investigation.

  • 7/28/2019 Bayes 2010

    33/57

    The strengths of Bayesian analyses are also its weaknesses:

    1) P-values do not require specifying a predicted effect size so

    significance testing is simple. Bayes requires saying what the

    theory predictsthere may be argument over what effect size a

    theory predicts. Using p-values side steps this argument.

  • 7/28/2019 Bayes 2010

    34/57

    The strengths of Bayesian analyses are also its weaknesses:

    1) P-values do not require specifying a predicted effect size so is

    simple. Bayes requires saying what the theory predictsthere

    may be argument over what effect size a theory predicts.

    BUT isnt this just the sort of argument the field needs?

    If someone disagrees they can argue their casein Bayes all

    assumptions must be explicit.

    In Neyman Pearson some important aspects are hidden.

    Neyman Pearson done properly requires specifying effect size.

  • 7/28/2019 Bayes 2010

    35/57

    The strengths of Bayesian analyses are also its weaknesses:

    2) Bayesian proceduresbecause they are not concerned with

    long term frequencies - are not guaranteed to control errorprobabilities (Type I, type II).

  • 7/28/2019 Bayes 2010

    36/57

    The strengths of Bayesian analyses are also its weaknesses:

    2) Bayesian proceduresbecause they are not concerned with

    long term frequencies - are not guaranteed to control errorprobabilities (Type I, type II).

    Which is more important to youto use a procedure with knownlong term error rates or to know the degree of support for your

    theory (the amount by which you should change your

    conviction in a theory)?

  • 7/28/2019 Bayes 2010

    37/57

    To calculate a Bayes factor must decide what range of differences is

    consistent with the theory

    1)Uniform distribution

    2)Half normal

    3)Normal

  • 7/28/2019 Bayes 2010

    38/57

    Probability density

    Difference between conditions

    0 5 10 20

    Example: The theory predicts a difference will be in one

    direction.

    Minimally informative prior, other than difference is

    positive:

    Maximum difference

    allowed

  • 7/28/2019 Bayes 2010

    39/57

    Seems more plausible to think the larger effects are less likely than

    the smaller ones:

    0

    Probability

    density

    Difference between conditions

    But how to scale the rate of drop?

  • 7/28/2019 Bayes 2010

    40/57

    Similar sorts of effects as those predicted in the past have been on

    the order of a 5% difference between conditions in classification

    accuracy.

    0 5

    Implies: Smaller effects more likely than bigger ones; effects

    bigger than 10% very unlikely

    Probability

    density

    Difference between conditions

  • 7/28/2019 Bayes 2010

    41/57

    Probability density

    Difference between conditions

    0 5 10

  • 7/28/2019 Bayes 2010

    42/57

    To calculate Bayes factor in a t-test situation

    Need same information from the data as for a t-test:

    Mean difference, Mdiff

    SE of difference, SEdiff

  • 7/28/2019 Bayes 2010

    43/57

    To calculate Bayes factor in a t-test situation

    Need same information from the data as for a t-test:

    Mean difference, Mdiff

    SE of difference, SEdiff

    Note: t = Mdiff / SEdiff

    So if a paper gives Mdiff and t

    For any of between- or within-subjects or single-sample t-test you

    know

    SEdiff = Mdiff / t

    NB: F(1,x) = t2(x)

  • 7/28/2019 Bayes 2010

    44/57

    To calculate a Bayes factor:

    1) Google Zoltan Dienes

    2) First site to come up is the right one:

    http://www.lifesci.sussex.ac.uk/home/Zoltan_Dienes/

    3) Click on link to book

    4) Click on link to Chapter Four

    5) Scroll down and click on Click here to calculate your Bayes

    factor!

    http://www.lifesci.sussex.ac.uk/home/Zoltan_Dienes/http://www.lifesci.sussex.ac.uk/home/Zoltan_Dienes/
  • 7/28/2019 Bayes 2010

    45/57

  • 7/28/2019 Bayes 2010

    46/57

    E l O t k h l k bi

  • 7/28/2019 Bayes 2010

    47/57

    Probability density

    Difference between conditions

    0 5 10 15

    Example: On a task where people make binary

    classifications about 65% correct (i.e. 15% above chance)

    we load working memory to see if it interferes

    difference = control conditionmemory load condition

    Maximum difference

    expected

  • 7/28/2019 Bayes 2010

    48/57

  • 7/28/2019 Bayes 2010

    49/57

    Jeffreys, 1961: Bayes factors more than 3 or less than a third are substantial

  • 7/28/2019 Bayes 2010

    50/57

  • 7/28/2019 Bayes 2010

    51/57

    Similar sorts of effects as those predicted in the past have been on

    the order of a 5% difference between conditions in classification

    accuracy.

    0 5

    Implies: Smaller effects more likely than bigger ones; effects

    bigger than 10% very unlikely

    Probability

    density

    Difference between conditions

  • 7/28/2019 Bayes 2010

    52/57

  • 7/28/2019 Bayes 2010

    53/57

  • 7/28/2019 Bayes 2010

    54/57

    Probability density

    Difference between conditions

    0 5 10

  • 7/28/2019 Bayes 2010

    55/57

  • 7/28/2019 Bayes 2010

    56/57

  • 7/28/2019 Bayes 2010

    57/57

    Assignment:

    12) What was the mean difference obtained in the study?

    13) What was the standard error of this difference?

    14) Extending your answer in (7), specify a probability distribution

    for the difference expected by the theory and justify it

    15) What is the Bayes factor in favour of the theory over the nullhypothesis?

    16) What does this Bayes factor tell you that the t-test does not?