GoConqr hat ein neues Design! Sag uns, wie es dir gefällt.

Kostenlos registrieren

Karteikarten

ANOVA, regression and hypothesis testing definitions

Grace Dimmock Englisch 52 Karten

Karten in diesem Stapel (52)

  1. 1Type 1 ErrorReject a true null hypothesis
  2. 2Type 2 ErrorFailure to reject a false null hypothesis
    e.g. you found that there is no difference, yet you conclude that there is.
  3. 3Null HypothesisAll things are equal
  4. 4One-tailed hypothesisPredicts the direction of the difference
  5. 5Two-tailed hypothesisPredicts a difference but not the direction of the difference
  6. 6One way ANOVAUsed when there's only one IV. Testing for differences between more than two groups
  7. 7Two-way ANOVACompares the mean differences between groups that have been split on two independent variables (called factors)
  8. 8Factorial ANOVAUsed when the research question asks for the influence of two or more independent variables on one dependent variable.
  9. 9BPS CODE OF ETHICSRESPECT, COMPETENCE, RESPONSIBILITY, INTERITY
  10. 10COHEN'S THREE EFFECT SIZESSMALL, MEDIUM, LARGE
  11. 11Test of internal consistencyCronbach's Alpha
  12. 12Test for independence of errorsDurbin Watson
  13. 13ANOVA statistical abbreviationF
  14. 14Pearson's Correlation abbreviationr
  15. 15A problem with the relationship between predictor variables will be indicated by an r value greater than or equal to?r=.8
  16. 16Why do we need to consider effect size?To establish the size and implications of the difference
  17. 17If it is <.05...It is not spherical (we need greenhouse geisser correction)
    It does not meet the homogeneity of variance assumption
    It does not meet the normality assumption
  18. 18Sig 2-tailed statistical abbreviationp
  19. 19Mixed factorial ANOVAWhere our experiments have one or more independent measures IV and one or more repeated measures IV
  20. 20What is the HO?Null hypothesis
  21. 21When can we reject the HO?When we can conclude that there is a significant difference/change/effect
  22. 22If there are two IV's with 3 levels each, what kind of ANOVA is necessary?3X3 ANOVA
  23. 23Enter method regressionThe default option, where all predictors are forced into model to see if significant
  24. 24Hierarchal regressionAllows the researcher to enter predictors in a specific order. If a predictor is found as non-significant you must remove it and run the analysis again
  25. 25Stepwise regressionSPSS finds the best predictor and enters it into the model, and continues to do so. Using this methods ensures model only includes significant predictors
  26. 26Hierarchical stepwise regressionWhen you enter the predictor you know is important first, and then enter the other predictors stepwise due to not knowing their importance
  27. 27When are post hoc tests suitable?You only need to do post hoc tests on variables which have three or more levels.
  28. 28Most important table when reporting ANOVA results?ANOVA Table (Tests of Between-Subjects Effects)
  29. 29When reporting results of an ANOVA, where should you look?You need the degrees of freedom from the variable line and error lines. Then the F and p values from the variable line
  30. 30For factorial ANOVA's we report a what effect?A significant MAIN effect
  31. 31For one way ANOVA's, we report what effect?A significant effect
  32. 32How do we report the results of an ANOVA?There was a significant main effect of type of aftershave on perceived attractiveness, F(3, 32) = 78.52, p <.001.
    And repeat again for the other IV, after post hoc comparisons.
  33. 33In ANOVA, when main effect is significant, we immediately report what?Tukey Post Hoc comparisons
    E.g. Tukey post hoc comparisons revealed that there was a significant difference in attractiveness between all aftershaves tested. Mean perceived attractiveness was highest in the Thrust condition (mean = 82.70), followed by Lust (mean = 65.40), Eau de Beer (mean = 56.00) and Dogs Breath (mean = 41.20) (p<.05 in each case).
  34. 34If p = <.000 report is as what?<.001
  35. 35Simple regression is what?Predict scores on a criterion variable from a single predictor variable
  36. 36Multiple regression is what?Allows us to analyse the contribution of several predictor variables as opposed to just one
  37. 37Regression analysis is concerned with what?The relationships between the variables, not mean differences.
  38. 38How to report a regression analysis?Report whether predictors are significant, providing the t and p value for each predictor from the final model in the coefficients table.
    Then state the direction, positive or negative for all predictors on criterion variable.
    E.g. Bank balance (t = 12.26, p <.001), looks (t = 12.14, p <.001) and personality (t = 4.57, p <.001) significantly predicted number of encounters. All predictors were positively associated with number of encounters, so that a higher bank balance, better looks and a more disagreeable personality were associated with a higher number of romantic encounters.
  39. 39How to report a regression analysis cont...?Report the results of the ANOVA for the final model: Degrees of freedom from regression + residuals and then the F and p value. Conclude with the R2 from Model 3 in the Model Summary, to percentage of variance.
    E.g. The final model was significant, F(3, 196) = 129.75, p <.001, and explained 66.5 per cent of the variance in encounters.
  40. 40What assumption do we look at for repeated measures designs?The assumption of sphericity.
  41. 41If p =<..05 in Mauchly's Test of Sphericity what can we conclude?The test is significant so the assumption of sphericity is violated
  42. 42A violation of homogeneity of variance only occurs if?p= <.05
  43. 43How to report correlationsLook at the N (r) value are report this first e.g.
    r (40)
    then report the stat from pearson's correlation and then the sig 2 tailed.
    e.g. r(40) = .766, p= 0.31
  44. 44Why would we consider the effect size of any significant differences in a statistical test?As a small difference may be misleading
  45. 45When can you assume equal variance?When p is greater than .05
  46. 46Cut off value for VIF10
  47. 47Cut off point for tolerance level.2
  48. 48When is there an issue with multicollinearityValues of VIF that exceed .10 are often regarded as indicating multicollinearity
  49. 49What to look at when checking for normal distribution?Shapiro Wilk
  50. 50What statistic is used to test homogeneity of variance?Levene's statistic
  51. 51When do we know data is normal?If the test is significant then the data is not-normal/violates normality assumption.
  52. 52What would we class as a strong correlation? What would be class as a weak correlation?Correlations above 0.8 are relatively strong.
    Correlations below 0.2 are considered weak.

Informationen

Beschreibung
Karteikarten am Ohne Titel, erstellt von Grace Dimmock am 12/05/2019.
Erstellt von
Grace Dimmock
Aufrufe
3
Bewertung

Feedback

Worum geht es?

Wie gefällt dir das neue Design?

Dein Eindruck