GoConqr has a new look! Let us know what you think.

Sign up free

Flashcards

Lecture 5

Samar Castelli English 102 cards

Cards in this deck (102)

  1. 1T-testsT-tests
  2. 2What is the most commonly used and reported statistical test?Pearson's correlation
  3. 3What is the basic question that all t-tests ask?Is there a significant difference between two means?
  4. 4Write up an example null hypothesis for a t-testμmales = μfemales (t= 0)
    (italicised)
  5. 5A null hypothesis is a statement about...What's going on in the population
  6. 6What idea do we need to build into a null hypothesis?Sampling error.
    If we do see a difference in our sample, it's just a reflection of sampling error.
  7. 7How do you report p-values in APA format?You should report exact p values. However, if they are small, they should be presented to two decimal places.
    e.g., p = .97. & p = .001
  8. 8Paired Samples t-testPaired Samples t-test
  9. 9How is the data set up differently in an paired sample t-test compared to an independent samples t-test?When setting up data files for a paired samples t-test there is no column for the IV. Instead, this IV is ‘encoded’ by the columns representing the two conditions under which the dependent variable was measured.
  10. 10How does a paired samples t-test differ from an independent samples t –testCompare means over the entire sample. The paired sample t-test doesn’t actually breakdown the comparison by another variable.
  11. 11What is the technical term for a person in a data file?Case.
    A case can be a school, household, person etc.
  12. 12What are the features of a paired samples design?1. Each case, across the whole sample, gives you multiple data points and you want to compare those data points.
    2. The paired sample t-test doesn’t actually breakdown the comparison by another variable.
  13. 13What are some other names for as paired samples design?Paired samples design
    Repeated measures design
    Matched samples design
    Within subjects design
    Within samples design
  14. 14What question does an independent t-test ask?Is there a statistically significant difference between two sample means taken from a categorical/grouping/nominal variable in your sample?
  15. 15What question does a single sample t-test ask?Is there a statistically significant difference between your sample mean and a known comparative population or large-sample value?
  16. 16What question does a paired sample t-test ask?Is there a significant difference between the means associated with a pair of variables.
  17. 17What is the formal definition of a paired samples design?Designs where each case (e.g., a person) gives you data on every level of the independent variable
  18. 18Give an example of a repeated measures designEach case in a sample get their blood pressure measured three times during the day (on every level of the independent variable).
  19. 19How can we determine the IV and IV levels in SPSS in a repeated measures design?IV would be the grouping variable that defines the columns of variables that we are interested in.
    Example: Blood pressure example.
    The IV is time, and the levels are morning, noon, and evening
  20. 20Why are repeated measure designs statistically powerful?1. Remove a source of error – individual differences between groups
    2. Need fewer participants for a large amount of data
    3. Very statistically sensitive (powerful). More likely to get you a significant result.
  21. 21What is a strong design?Very statistically sensitive (powerful). More likely to get you a significant result.
  22. 22Why do you need fewer participants for a repeated measures design?Because you're getting multiple data points from each individual
  23. 23Why does removing sources of error make your test more sensitive?Because it allows you to maximise the difference between groups and minimise error variability
  24. 24What source of error does a repeated measures design remove?Individual differences between groups
    (or individual differences across your groups).
  25. 25Why do repeated measures designs make a test more sensitive by removing the error of individual differences between groups?1. You have the same individuals in each group or in each of the levels of the IV.
    2. You can't have individual differences between the groups because they're the same people.
    3. Therefore you remove a source of error, you make a test more sensitive.
  26. 26What's the problem with repeated measures design and what does it mean?Carry-over, practice, order effects
    Definition: Your participation at an earlier stage in a study influences your response at a later stage in the study.
    Therefore, some studies are precluded from this sort of design.
  27. 27What are some examples of studies that would not suit a repeated measures design?Weight loss: involvement at an earlier stage is going to impact involvement at a later stage.
    Learning: Engagement in method no. 1 impacts performance on method no.2.
  28. 28Name a very common variable that easily accommodates a repeated measures designTime or phase
    Example: you have a study where you want to track a group of people over time (e.g., longitudinal design or pre-test post-test). The IV is time.
  29. 29Can you carry-over, practice order effects?Yes
    For example, using the common alternative of a matched sample or randomised blocks design.
  30. 30What is a matched samples or randomised block designWith a randomised block design, the experimenter divides subjects into subgroups called blocks, such that the variability within blocks is less than the variability between blocks. Then, subjects within each block are randomly assigned to treatment conditions.
  31. 31Why is randomised block (or matched samples) design useful?Mimics a repeated measures design but with different people
  32. 32What are some limitations of the randomised block design?1. Matched sample is not a perfect replication of having the same person
    2. Impacted by sample size
    3. May have people you can't match
  33. 33When is a randomised block design usually used?Often used when you can't do a proper repeated measures design but want to control for some crucial variables
  34. 34Reframe the following into a statistical question: Are fathers more educated than mothers?Is there a statistically significant difference in the mean number of years of education for mothers versus the mean number of years of formal education for fathers?
  35. 35What is the IV and DV in this scenario?IV= parents (levels are mother and father)
    DV= number of years of formal education
  36. 36Describe the steps to carry out a paired sample t-test for this research question1. Analyse
    2. Compare means
    3. Paired samples t-test
    4. Move two variables into the right hand side
    5. Ok
  37. 37How do you produce a graph for a repeated measures design?1. Graph
    2. Legacy dialogues
    3. Bar
    4. Summarises of separate variables
    5. Define
    6. Move two variables of interest into bars represent
    7. Ok
  38. 38What change do you make when producing a graph for a repeated measures design vs an independent measures t-test?Select 'summaries of separate variables' as opposed to 'summarises of groups of cases' as you are comparing separate variables and not groups.
  39. 39What is the null hypothesis for the research question: Is there a statistically significant difference in the mean number of years of education for mothers versus the mean number of years of formal education for fathers?Null hypothesis: μFather = μMother (t = 0)
    (and any difference that we see in the sample just reflects sampling error)
  40. 40What is the alternative hypothesis for the research question: Is there a statistically significant difference in the mean number of years of education for mothers versus the mean number of years of formal education for fathers?μFather ≠ μMother
  41. 41Why are the descriptives different when generated independently than when generated for a paired samples t-test?T-test only includes cases that have provided information on all variables of interest
  42. 42Do correlations have anything to do with mean differences?No
  43. 43In what way does the SPSS output from the paired samples t-test procedure differ from that of the independent samples t-test?1. Only one test provided
    2. Includes correlations
  44. 44How do you calculate df for a paired samples t-testN-1 (only one group)
  45. 45How do you write up the result for a paired samples t-test? (using research question provided)A paired-samples t-test revealed no significant difference between the number of years of formal education for mothers and fathers, t(973) = 0.07, p = .94.
  46. 46What is the APA rule regarding leading zeros?If a statistic can have a value greater than 1, it can include a leading 0. If a statistic cannot have a value greater than , it does not.
    So...
    T values can have a value greater than 1, it takes a leading zero e.g., (0.)07
    P values can't, so they are simply .94
  47. 47When writing up a t-score, do you have to report the negative sign?No
    Negative values are simply the result of a substraction in the wrong order (smallest sample minus biggest sample). Doesn't matter how it's reported.
  48. 48What question does a single-sample (one-way) t-test ask?Used to test your sample mean against a comparison value.
    Example: Is there a statistically significant difference between your sample mean and a known comparative population or large-sample value?
  49. 49Where does the comparison value come from in a single sample t-test?(1) An established normative score
    (2) Another researcher’s result
    (3) A known population parameter
  50. 50Provide some examples whereby a single sample t-test would be appropriateExample: mean IQ of a class sample compared to the known population mean.
    Example: mean of a clinical sample on a Beck inventory (15) compared to the normative value on the questionnaire of the population (12).
    Example: collected some data on a measure and compare it to a figure published in a journal article.
  51. 51Provide an example relate to the GSS dataset of the use of a single sample t-testIs the average number of children reported in the 1991 GSS sample significantly different from the expected population norm of two children per household?
  52. 52How do you run a single sample t-test?1. Analyse
    2. Compare means
    3. One sample t-test
    4. Enter in test variable
    5. Enter in known value
    6. Ok
  53. 53How is the null stated in a single sample t-test? (not comparing two groups, comparing with a known value)Ho: μnumber of children = 2 (t = 0)
    (and any deviation that we see in the sample is just a reflection of sampling error).
  54. 54Write up the results of a single sample t-testA single-sample t-test revealed a significant difference between the sample mean for number of respondents’ children (M = 1.90, SD = 1.77) compared against the hypothesised population value of 2, t(1508) = 2.17, p = .03.
  55. 55What determines whether you get a positive or negative t-score in a single sample t-test?Whether the sample mean is more or less than the comparison value
    More = +
    Less = -
  56. 56Does the sign of a t-score impact significance?No, negative scores can be just as significant as positive scores
  57. 57How do you write up mean and SD in APA style?(M = 1.90, SD = 1.77)
    Capitalised and Italicised!~
  58. 58How does a single sample write differ from other t-tests?Includes M and SD
  59. 59Pearson's CorrelationPearson's Correlation
  60. 60What represents person's correlation?r
    (italicised)
  61. 61What are some names that person's correlation is known by?1. Pearson’s product-moment correlation
    2. Pearson’s correlation
    3. Pearson’s r
    4. Correlation
    5. r
  62. 62What's the difference between correlations and t-tests?Whereas a t-test measures a difference between two means (differences),
    r measures the strength and direction of the LINEAR relationship between two continuous variables (looks for similarities).
  63. 63What is the fundamental question addressed by Pearson’s correlation?Is there are statistically significant relationship between two continuous/scale variables?
  64. 64What are three aspects of a relationship that are examined by Pearson’s correlation?Strength: What is the size/magnitude of the relationship?
    Direction: Is the relationship positive or negative?
    Linear: Straight line
  65. 65What does r measure?The strength and direction of the linear relationship between two continuous/scale variables.
  66. 66Break this down: The strength and direction of the linear relationship between two continuous/scale variables.(1) Strength and direction: The value of the statistic tell us about the strength an direction
    (2) Linear: Straight line relationship
    (3) Continuous/scale variables: are not categorical. Must be continuous e.g., height and weight
  67. 67Are there variations of r can be used to measure the strength and direction of the relationship between non-scale (i.e., categorical) variables,Yes, but not covered in the course.
  68. 68Pearson’s correlation only investigates the linear component of a relationship between two continuous or scale variables. What does this mean?r only measures/assesses the straight line (i.e., linear component of the relationship).
  69. 69What are some examples of non-linear relationships?Learning curve (curvy linear or quadratic relationship) and relationship between stress and performance (inverted u).
    Pearson’s correlation calculates these but provides a meaningless number as it is not designed to measure these.
  70. 70What's the relationship between stress and performance?Inverted U
    1. If you're not stressed enough, performance is bad.
    2. Stressed enough results in good performance.
    3. High stress reduces performance.
  71. 71Convert the following research question into a statistical question: What is the nature of the relationship between years of schooling and job prestige?Statistical question: Is there a significant correlation between years of schooling and job prestige?
  72. 72Write up the previous statistical question as a nullρ = 0.
    In the population, the correlation between years of schooling and job prestige = 0, and any deviation that we see from that in our sample is just a reflection of sampling error: ρ = 0.
  73. 73What does this represent? ρGreek letter rho which is the value of the correlation in the population.
  74. 74What graphical tool do you use to explore correlationsScatterplot or Scattergram
  75. 75What does a scatterplot do?Plots relationship between two continuous or scale variables.
    Allows us to look at characteristics, shape, relationships etc
  76. 76How many people are represented per dot on a scatterplot?Usually one, but can be more than one with identical scores in a large sample.
  77. 77PRACTICE READING SCATTERPLOT (pg 3 example)Indicates:
    1. General relationship between higher education and higher schooling
    2. A lot of variability (occupational prestige varies from lowest to highest after yr 12)
    3. Shows clusters of people finishing after high school and after their bachelor
    4. Suggests that there may be more factors associate with OP than just education.
  78. 78How do you create a scatterplot in SPSS?1. Graphs
    2. Legacy dialogues
    3. Scatter or dot
    4. Simple
    5. Define
    6. Enter variables
    DV goes on Y axis
  79. 79Define a positive correlationHigh scores on education are associated with high job prestige
    AND
    low scores on education are associated with low job prestige, but relationship is far from perfect. Both in same direction
    [High/High] & [Low/Low]
  80. 80What are the steps to perform a correlation in SPSS?1. Analyse
    2. Correlate
    3. Bivariate
    4. Select
    5. Enter two variables
    6. Ok
  81. 81What are the three essential statistics that you get from correlation output?1. Pearson correlation: value of r
    2. Significance level
    3. Sample size
  82. 82Provide an example of a write up for a correlationThere was a significant correlation between educational level and job prestige, r(N =1415) = .52, p < .001.
  83. 83How many decimal places are correlations presented to?2
  84. 84What are the guidelines for writing significance levels?p=.000 IS NEVER CORRECT
    (SPSS just rounds to 3 decimal places)
    Must note it as less than:
    p < .001.
    This is the only time you write a sig. love with a less than, all other occasions you use an =
  85. 85Features of r:(1) It must be between +1 and -1
    (2) Negative correlations are just as important as positive correlations
    (3) If you square r, you get R2, which is an important statistic that tells you the level of overlapping/shared variance.
    (4) If there is no correlation, then r = 0.
    (5) ONLY assesses the linear component of the relationship.
  86. 86What happens is the relationship is non-linear?r will be calculated, but it will be mis-leading. Hence the need to always draw scatterplots.
    For example, with an inverted U, something is going on, but the statistic will indicate that r=0 as the two sides cancel each other out.
  87. 87What does R2 mean?Important statistic that tells you the level of overlapping/shared variance (what two things have in common)
    Can be obtained by squaring the value of r. This result can the be expressed as a percentage and can be thought of as...
    49% of the variability in weight is directly associated with changes in height.
  88. 88What does R2 of 0.27 between education and occupational prestige?There's 27% shared variance between education and occupation prestige. This indicates that only 27% worth of your education is associated with occupational prestige. Therefore, other factors contribute to occupational prestige.
  89. 89If R2 was 1, what would this mean?Perfect overlap, one variable completely explains another.
  90. 90What is a negative relationship/correlation:As scores on one variable go up, scores on the other variable go down. High scores on one variable are associated with low scores on the other variable.
    [High/ Low] & [Low/ High]
  91. 91Provide an example of a negative correlationCorrelation between depression and quality of life is negative. As depression increases, quality of life decreases. As scores on depression go up, scores on quality of life go down.
  92. 92What is a correlational matrix?Running every possible correlation from a set of variables
  93. 93How do you run a correlation matrix in SPSS? (to get very possible variable)1. Analyse
    2. Correlate
    3. Bivariate
    4. Add all variables wanted
    5. Ok
  94. 94Why would you get a perfect correlation when running a correlational matrixShowing you the value of something correlated with itself which is a perfect correlation
  95. 95What is the direction of a positive correlation on a scatterplotBottom to top
    Left to Right
  96. 96What is the direction of a positive correlation on a scatterplotTop to Bottom
    Left to right
  97. 97How to read the strength of a correlation NB: strength is not affected by directionTrivial = < .1
    Small = 0.1
    Moderate = 0.3
    Large = 0.5
    Very large= 0.7
    Nearly perfect = 0.9
    Perfect = 1
  98. 98What do we have to be careful with correlations?1. Spurious correlations.
    Correlations exist but are meaningless.
    2. Not interpreting in context e.g., two markers mark a multiple choice test and get .9. Though it's high, it's low in context.
  99. 99Which statistics are written with leading zeroes and which are not?1. t values can have a value greater than 1, it takes a leading zero e.g., 0.07
    2. significance levels can't, so they are simply .94
    3. Correlations also can’t so do not have leading zeros.
  100. 100How can you use a scatterplot to learn more about a personHover over the entry, retrieve the case file , and return to the data set to flesh out the story.
  101. 101List statistics italicised in APAt
    p
    SD
    M
    r
    R2
    Also…
    N
    n
    df
  102. 102T score write up (N or df)Include df unlike correlations

Information

Description
Flashcards on Lecture 5, created by Samar Castelli on 05/07/2018.
Created by
Samar Castelli
Views
3
Rating

Feedback

What is it about?

What do you think of the new look?

How do you like it?