Fichas
Lecture 1&2
Fichas del conjunto (26)
- 1Definition Type I error (alpha)probability of test showing statistical significance when it is not present (‘false positive’)
- 2Definition Type II error (beta)probability of test not showing statistical significance when it is present
- 3Power (1-beta)probability of test showing statistical significance when it is present
- 4Power depends on-Alpha
-Effect size
-Sample Size n - 5Dependence TechniquesOne or more variables can be identified as dependent variables and the remaining as independent variables
Choice of dependence technique depends on the number of dependent variables involved in analysis - 6Name 3 dependence methods-ANOVA
-Conjoint
-Logistic Regression - 7Interdependence TechniquesWhole set of interdependent relationships is examined
Further classified as having focus on variable or objects - 8Name 3 independence methods-factor analysis
-cluster analysis
-MDS - 9Why to conduct preliminary analysis-Get a feel for the data
-Suggest possible problems (and remedies) in next steps - 10How to conduct preliminary analysis-Univariate profiling
-Bivariate analysis - 11Two steps in conducting preliminary analysis-Detect Outliers
-Examining Missing Data - 12Tree possible causes of outliers-Procedural error
-Exceptional circumstances (cause known or unknown)
-‘Regular’ levels, yet unique in combination with other variables (bivariate en multivariate outliers) - 13Detecting Outliers with Univariate methodsHistograms, TS plots, Frequency Tables,
Mean +/- 3SD, Box Plots - 14Detecting Outliers with Bivariate methodsScatterplot, Multiple Histograms
- 154 step approach in Missing Data Analysis1: determine type of missing data (ignorable?)
2: determine Extent (%) of missing data
3: diagnose randomness of missing data (Systematic, Missing At Random (MAR), Missing
Completely At Random (MCAR)?)
4: deal with the missing data problem - 16Give the definition for the ANOVA techniqueTest whether treatments (categorical variables) lead to different levels for a (set of) metric outcome variables
- 17Assumptions for ANOVA-Independent observations
-Homoscedasticity of variances (Levene's test >0.05)
-Normal distribution (kologorov-Smirnov >0.05) - 18What if assumption of homoscedasticity rejected?-If sample size = similar across treatment groups -> robust
-Transform dependent variable (e.g. take logarithm) -> redo test
-Adjust cutoff for significance in ANOVA - 19What if assumption of normality is rejected?-Large sample -> robust
-Small sample: transform dependent variable to make distribution more symmetric - 20ORDINAL interactionMagnitude of effect
depends on the extra independent variable, but direction does not! - 21DISORDINAL interactionnot only magnitude,
but even DIRECTION of the effect
depends on the other independent variable! - 22What is the Effect SizePartial Eta-squared: “% variance explained”
.01 -> small
.06 -> moderate
.14 -> large - 23When you select the test (follow up), based on a priori expectations you usecontrast effects or planned comparisons
- 24When testing all combinations (follow up test) you can usepost hoc tests or multiple comparisons ->need for correction to avoid type-I error
- 25Downsides of missing data-reduces sample
-possibly biases outcome - 26Rule of Thump: what is the maximum of alowed missings10%
Información
- Descripción
- Introduction to Research in Marketing Fichas sobre Lecture 1&2, creado por c.l.kerkers el 15/06/2015.
- Creado por
- c.l.kerkers
- Visitas
- 2
- Valoración