O GoConqr está de visual novo! Conte para nós o que você achou.

Registre-se grátis

Flashcards

Lecture 7

Samar Castelli Inglês 102 flashcards

Flashcards do conjunto (102)

  1. 1What does ANOVA stand for?Analysis of Variance
  2. 2What is an independent groups design?When each participant is in only ONE condition or one level of the IV.
  3. 3Provide an example research question that would be tested with ANOVAIs there a statistically significant difference among the MEANS associated with four different memory improvement methods?
  4. 4What is a balanced design?A design where each group or level of the IV has the same number of participants.
  5. 5What is an independent groups design referred to as in ANOVA?Between-subjects design
    (must be hyphenated)
  6. 6Identify the IV and DV in the following research design: The statistical question: Is there a statistically significant difference among the means associated with the four different memory improvement methods? Four groups of subjects: 1. Group 1 are taught how to use the Method of Loci 2. Group 2 are given a memory-enhancing drug. 3. Group 3 are taught a technique of visual imagery. 4. Group 4 is the control group (no special treatment). 100 subjects are randomly assigned to the four groups (25 subjects per group).Between-subjects design
    I.V: Memory of memory enhancement (four levels)
    D.V: Number of words correctly recalled out of 100.
  7. 7Develop a suitable hypothesis for the studyThat degree of word recall will vary significantly according to memory improvement method.
  8. 8When do you use a lower or upper case N for sample size in APA style?Italicised uppercase N for total sample size. Italicised lower n case for subgroups.
  9. 9Pose an example research/statistical question that may be answered with ANOVAIs there a significant difference among x number of means, and, if so, what is the source of that significant difference; how has that difference come about?
  10. 10How does a statistical question related to ANOVA differ from other statistical questions?It's quite broad and sweeping. Asks if there's a difference somewhere/anywhere among four means. Second component seeks to identify the cause of the difference, if a difference is found.
  11. 11Why is the statistical question for an ANOVA worded differently?Because there are more than two groups, therefore it is necessary to frame the question broadly.
  12. 12Is this statistical test widely used?Yes, the most widely used statistical test in psychological research and when working with quantitative data.
  13. 13Is the ANOVA recent or antiquated?More recent
  14. 14What is the key statistic in ANOVA?F statistic (also called F ratio)
    Capitalise
    (because it's based on Fisher's name)
    ~italicise
  15. 15Why is ANOVA so popular?It can be used in a very wide range of experiment designs and can be tailored.
  16. 16Name some designs it can be used forIndependent groups
    1. Repeated measures
    2. Matched samples
    3. Designs involving mixtures of independent groups and repeated measures
    4. More than one independent variable can be evaluated at the same time
  17. 17Name the statistical test learnt so far and what they can be used forr = two continuous variables
    t-test: two independent means
    chi square: two categorical variables
    Paired sample t-test: two related means
  18. 18Elaborate on the flexibility of ANOVA1. Designs that have more than one IV at the same time
    2. More than one IV where one can be independent and the other repeated (at the same time)
    3. Multiple DVs
    4. Control for the influence of a nuisance variable
  19. 19What is the core statistic in ANOVA?F statistics (also called F ratio)
  20. 20If ANOVA involves questions about differences among means, why is it called analysis of VARIANCE and not analysis of MEANS?It answer the question about mean differences by looking at VARIABILITY in the data
  21. 21What are sources of variability in data?Within groups variability
    Between groups variability
  22. 22What does variability refer to?The differences we see in scores
  23. 23What are we interested in variability?Because we want to understand why individual scores differ within a group and why scores differ between groups.
    We want to understand the source of the variability
  24. 24Define within group variabilityThe difference in scores we see within each of our groups or levels of the independent variable
  25. 25Define between-groups variabilityThe variability we see across the groups
  26. 26What are the sources of within group variability?1. Individual differences (variables such as gender, age etc)
    2. Experimental error
  27. 27What is the MAIN source of of variability within groups?Individual differences
  28. 28What does experimental error refer to? Provide examplesThe random, relatively minor, mistakes that take place in an experiment that can impact results.
    Examples: human error, mistakes in data entry; systematic error.
  29. 29Can you entirely eliminate experimental error?No
  30. 30Why is the F statistic called an F ratio?Because it represents the total amount of between group variability divided by within group variability.
  31. 31The F statistic is a ratio formed as follows:Between-group variability
    ------------------------------------
    Within-group variability
    Treatment+individual differences+experimental error
    ------------------------------------------------------------
    Individual differences+experimental error (error term)
  32. 32What does 'error term' refer to?Individual differences + experimental error
  33. 33Why does the F ratio have the treatment affect on top?to highlight the treatment effect.
    The more 'effect', the larger the F statistic. For the ratio to reflect this, the treatment effect has to be on top.
  34. 34What is everything in the study apart from the treatment and what do we try and do with it?Noise. Minimise it
  35. 35How can you reduce experimental error?Examples: conducting research in controlled conditions, using consistent instructions etc
  36. 36How can you reduce individual differences?Use lab rats. Specify a group with particular characteristics (e.g., only male, particular race etc)
  37. 37What is the issue with specifying a group with particular characteristics?Sampling bias.
    Cannot generalise beyond that sample, therefore; we accept the fact that there are individual differences.
  38. 38If there is no treatment effect at all, then F =1 (nothing happening)
    Example: 1/1 = 1
  39. 39Finish this sentence... The stronger the treatment effect...The bigger the value of the F ratio.
  40. 40Can you get a negative F ratio?No
  41. 41Does the F ratio have an upper ceiling?No
  42. 42What is the null hypothesis that is tested in ANOVA?μMOL = μdrug = μimagery = μcontrol
    “In the population, there is no difference in the memory performance under the four treatment conditions, and any difference in the sample, is simply due to chance.”
  43. 43What is the alternative hypothesis, and why is it different from the alternative hypotheses in other tests?“At least one of the means is different from the rest.”
    Isn't stated like a formula as in other tests
  44. 44What do you need to do to reject the null hypothesis with ANOVA?Demonstrate, to a level of statistical significance, that one of the means is different from the others.
    NOT ALL
  45. 45How do you run an ANOVA in SPSS?1. Analyse
    2. Compare means
    3. One-way anova
    4. IV in factor
    5. DV in Dependent List
    6. Options (Descriptives, HOV, means plot)
    7. OK
  46. 46What does 'one-way ANOVA mean'?One independent variable
  47. 47Where are the core results in an ANOVA?ANOVA summary table
  48. 48What does an ANOVA summary table do?Carves up the variability in the data
  49. 49What does 'total' refer to in a ANOVA summary tableThe total amount of variability
  50. 50How is the total calculated?By adding the between group variability and the within group variability
  51. 51What is sum of squares shorthand for?*Sum of squared deviations from the mean*
    See next slide for the process to obtain this figure
  52. 52How do we calculate the total amount of variability in the data?1. Calculate grand mean (the mean of all the scores across all the groups)
    2. Subtracts each score and the grand mean
    3. Square this difference
    4. Total equals variability
  53. 53Why can we just minus each square from the grand mean to get the total variability?Positive and negatives values will cancel each other out.
  54. 54How can we get rid of negative numbers?By squaring them.
    Squaring values always get rid of negative numbers.
  55. 55What is the total sum of squares and how is it derived?1. Calculate the grand mean (the mean of all the scores across all the groups)
    2. Subtract each score from the grand mean
    3. Square the difference
    4. Add up all the squared differences (sum of squares)
    Total sum of squares = total variability
    THEN...
    Carves it out into between group variability and within group variability
  56. 56Why don't we just divide the between groups SOS into the within groups SOS to get the F ratio?The numbers between groups are different from the numbers within groups. The numbers aren't comparable.
    Example: The between groups variability has come from 4 groups but the within groups variability has come from 100 people.
  57. 57To address this problem, what do we need to do?Calculate the average between group variability and the average within group variability
  58. 58In any analysis of variance procedure, how many df are there?Two
    1. The first is based on number of groups
    (Number of groups -1)
    2. The second is based on the number of participants
    (Number of participants - the number of groups)
  59. 59What does the df value allow us to do?Calculate the average amount of between and within group variability termed the 'mean square'.
    Example: SOS of BGV =449/ df = Mean square
  60. 60Why do we need the mean square when calculating the F ratio?The numbers between groups are different from the numbers within groups and therefore ARE NOT comparable because they take into account the different group numbers. By calculating the average between-group variability and the average within-group variability, we obtain figures that are comparable and that can be used to calculate a meaningful F ratio.
  61. 61Recap: how is the F ratio calculated?Between groups mean square
    ____________________
    Within groups mean square
  62. 62Recap: how is the F ratio calculated (whole process)?1. Calculate the grand mean (the mean of all the scores across all the groups)
    2. Subtract each score from the grand mean
    3. Square the difference
    4. Add up all the squared differences (sum of squares)
    5. Divide sum of squares by df for each)
    6. Obtain
    Between groups mean square
    ____________________
    Within groups mean square
    = F ratio
    7. Calculate a sig. level based on the F ratio
  63. 63What are the two significant pieces of information related to the anova output?F ratio and significance level.
  64. 64What does a significant F ratio tell us, what does it not tell us?Initially all it tells you is that there is a significant effect somewhere among your groups, but it doesn't tell you the source of that difference. It doesn’t tell you the pattern of differences among the means of the groups that has led to that significant difference.
  65. 65Why is the F ratio called an omnibus test? (omnibus means 1)Only tells you one thing. That 'something' is going on.
  66. 66How is a an ANOVA written up?(1) Research using ANOVA is described in terms of its design:
    Differences among the four groups were evaluated using a one-way independent groups ANOVA.
    Differences among the four groups were evaluated using a single-factor between-subjects ANOVA.
  67. 67Why do you need to describe ANOVA in terms of its design?Because there are so many types of ANOVA designs that unless you're specific in how you write it up, the reader is not going to know what type of ANOVA has been done.
  68. 68Translate: (1) Differences among the four groups were evaluated using a one-way independent groups ANOVA. (2) Differences among the four groups were evaluated using a single-factor between-subjects ANOVA.(1) One-way independent groups = one IV with independent groups
    (2) One IV
  69. 69Which terns are synonymous in the write up?One-way & single factor & simple ANOVA= 1 IV
    Independent groups & Between-subjects = Independent groups
  70. 70Which terminology is traditionally used?Differences among the four groups were evaluated using a single-factor between-subjects ANOVA.
  71. 71What takes place next if we get a significant result?Post-hoc test
  72. 72What is the purpose of post-hoc tests?The process by which we evaluate which of the means is different or why we got the significant result.
  73. 73Define post-hoc testPost-hoc (latin for after):
    Tests that you carry out after you have obtained a significant overall ANOVA in order to locate the source of the significant F. Exploratory, fishing expeditions.
  74. 74When can you justify the use of a post-hoc test?Overall F must be significant in order to justify the use of post-hoc tests.
  75. 75Where are the post-hoc tests located for ANOVA?Under ANOVA command
  76. 76\What do post-hoc tests do?They basically do the same thing: test every group with every other group, in order to locate the source of your significant F statistic.
  77. 77How do post-hoc tests do this?They form every possible pair of groups that you can have and they test them.
    Example from lecture: four groups are broken down into every possible 'pairwise comparison'. So they run six mini tests in this scenario.
  78. 78Why are there so many different types of post-hoc tests?Approach the analysis through a different lens.
    The one you choose depends on your research situation (e.g., equal variances not assumed, if you want to compare control with every treatment group separately = dunnet test).
  79. 79Name some common post-hoc testsScheffé test
    Newman-Keuls test
    Tukey’s Honestly Significant Different Test
    Fisher’s Least Significant Different Test
  80. 80How many pairwise comparisons would you get for a: 4 group ANOVA 3 group ANOVA 5 groupANOVA4 = 6
    3 = 3
    5= 10
  81. 81Why are they called pairwise comparisons?Each test involves a pair of means
  82. 82Why are post-hoc tests statistically conservative (i.e., why is it hard to get a significant result with a post-hoc test?)Because our hypotheses are not a priori and because we by testing all possible mean combinations we greatly increase the likelihood that we are going to commit a Type I error as the number of groups increase
    +
    are essentially a fishing expedition based based on vague hypotheses.
    Post Hoc tests are conservative for this reason.
  83. 83What's the problem with post-hoc tests? (Fishing expedition)They're exploratory
  84. 84What does a lack of statistical power mean?It's harder to get a significant result
  85. 85What is the balance that post-hoc tests set out to achieve?Maximise your power in a slightly different statistical way. Accept that post-hoc tests are an exploratory exercise, but try to so some machinations to maximise the researcher’s ability to find a result if one is there. Therefore, it is a balancing act of accepting that it’s a fishing expedition but aiming to find a result if they’re there.
  86. 86Which is the boldest of the post-hoc tests and why?LSD (Least significant difference)
    Because it doesn't try to control for anything
  87. 87Name a conservative post-hoc testScheffe’s test
  88. 88What test strikes a balance between conservatism and statistical power?Tukey Post Hoc test
    Honest Significant Difference test
  89. 89How to run a post-hoc test1. Analyse
    2. Compare means
    3. One-way ANOVA
    4. Post-hoc
    5. Tukey
    6. OK
  90. 90How do you read a post hoc test?Search for significance by comparing the first line variable with each of the subsequent variables in each row.
  91. 91How does the result provided in the lecture (2.09) relate to the mean plot diagram?The significant result is obtained because of the difference between the two extreme values plotted on the mean plot (MOL vs control)
  92. 92What is the post hoc test telling us in this scenario?The main contributor towards the significant overall ANOVA is the big difference between the two extremes
  93. 93How do you write up the results of an ANOVA?1. MUST describe the design
    2. Simple statement describing the results
    3. Essential statistics
    STAGE 2
    Post hoc results
    1. Always specify the type of post hoc test
    2. Only need to report p-value
  94. 94Why do you need to clarify the design?Because there are so many types of ANOVA
  95. 95Breakdown the summary of the statistic for ANOVAF(3, 96) = 3.80, p = .013
    ~Italicise f & p!
    1. Two degrees of freedom (Between subjects df)
    2. Error term df (participants - groups)
    3. Value of F ratio
  96. 96What are the APA conventions in relation to ANOVAF Statistic:
    1. To two decimals places
    2. Takes a leading zero
  97. 97RECAP Degrees of freedom (figures that the F test uses to calculate its exact value): how many are there?There are two degrees of freedom for each F test.
    1. The number of groups - 1.
    2. (Number of subjects in each group - 1) x number of groups.
  98. 98RECAP Summary of Basic Overall Procedure:Overall ANOVA; if significant, follow-up with post-hoc tests (snooping the data to locate the source of the difference).
  99. 99RECAP: How do post hoc test differ despite having the same purpose?Controlling for the exploratory nature of the exercise in different ways
  100. 100How do you calculate the Grand mean?Obtain a mean of all the scores across the groups
  101. 101Who invented ANOVAFisher
  102. 102Provide an example of a write upThe main questions of interest were tested using three a priori planned comparisons. Because the homogeneity of variance assumption was met, p = .49, the analyses based on equal variance are reported. A significant difference was found between the Method of Loci condition and the other three groups combined, t(96) = 2.78, p = .006; however, no significant difference was found between the Drug condition and the two cognitive methods in combination, t(96) = 1.78, p = .08, or between the two cognitive-based methods (Imagery and Method of Loci), t(96) = 1.35, p = .18.

Informações

Descrição
FlashCards sobre Lecture 7 , criado por Samar Castelli em 25-07-2018.
Criado por
Samar Castelli
Visualizações
1
Avaliação

Fale Conosco

Sobre o que é sua mensagem?

O que você acha do novo visual?

Você gostou?