SPSSFinal Exam Flashcards
Bivariate - cat, cat - graph?
clustered, stacked bar graph or pie chart
Bivariate - cat. scale - Graph?
Boxplot
Bivariate - scale scale - graph
Scatter
Bivariate - cat. cat. - Table
Crosstabs
Bivariate - cat. scale - Table
Descriptive
Bivariate - scale scale - Table
Correlations
Bivariate - cat. cat. - Stat?
Pearson Chi-square test
Bivariate - cat. scale - Stat?
T-test -> compare 2 means
ANOVA -> more than 2
Bivariate - scale scale - Stat?
T-test for corr=0
when agglomerative hierarchical cluster?
- we want to know about the agglomeration process (similarities and diff. between groups).
- Limited cases.
- Homogeneous set of data - all scale or categorical.
How hierarchical cluster?
> Analyze > Classify > Hierarchical Cluster > var. in ‘Variable’ > ind. var. in ‘Case label variable > Statistics: Proximity Matrix, Agglomeration schedule if you want a specific # Clusters: Cluster Membership > Plots: Dendogram, Icicle=None > Method: Centroid Clustering, Squared Euclidian Distance if var. are diff in size - Transform > Standardize
To explore groups > Save clustership > Analyze > Compare Means > Means > clust. var. in ‘Dependent List’ > ind. var. in ‘Independent List’
To know stat. sig. > ANOVA
How bar graphs?
> Graphs > Legacy Dialogs > Bar > clustered or stacked > categorical v. in ‘Category axis’ > v. diff. section in ‘Def. stacks by’ > numerical v. in ‘Columns’
how pie chart?
> Graphs > Legacy Dialogs > Pie > summ. for groups > Define
How boxplot?
> Graphs > Legacy Dialogs > Boxplot > scale v. in ‘Variable’ >
How scatter?
> Graphs > Legacy Dialogs > Scatter/Dot > indep. v. in ‘Row’ > dep. v. in ‘Column’
How crosstabs?
> Analyze > Descriptive analytics > Crosstabs > x var. as ‘Row’ & y var. as ‘Column’ > Cells, Percentages ‘Row’
How descriptive?
> Analyze > Descriptive analytics > Explore > scale/dep./split var. as ‘Dependent list’ & nominal var. as ‘Factor list’
How correlations?
> Analyze > Correlate > Bivariate > two var. in Variables
How pearson chi-square
Null Hypothesis = Independency between x and y
> Analyze > Descriptive analytics > Crosstabs > x var. as ‘Row’ & y var. as ‘Column’ > Statistics > Chi-square
> p-value/sig. is the risk taken if rejecting Null is True.
→ sig. <= 5% reject Null / sig. >= 5% accept Null
Statistical relationship → reject Null
How t-test?
Null Hypothesis = No relationship between x and y
> Analyze > Compare means > Independent-Samples t-test or Paired Sample t-test (if obs. can belong to two samples) > scale var. as ‘Test variables’ & cat. var. as ‘Grouping variable’ > Define groups > define the two groups of cat. var.
> t-value → statistic value > df → degrees of freedom > Sig. (2-tailed) → if <0.05 indicates significant difference.
how anova?
Null Hypothesis = No relationship between x and y
> Analyze > Compare means > One-way ANOVA > scale var. as ‘Dependent List’ & cat. var. as ‘Factor’
> F-value → statistic value > df → degrees of freedom > Sig. → if <0.05 indicates significant difference.
How t-test corr=0?
> Analyze > Correlate > Bivariate > scale var. in dialog box > check Pearson & Two-tailed
Sig. (2-tailed) → if <0.05 indicates significant difference from 0
When two-step cluster?
Used when…
Not interested in agglomeration process but only in the final groups obtained.
Large number of cases/indiv.
how two-step cluster?
> Analyze > Classify > Two-Step Cluster W > var. in ‘Categorical variables’ & ‘Continuous variables’ > Distance Measure : Euclidian
Analyze: Cluster quality, click on the column(s), change order of importance
Improving cluster > Change settings > Option > Specify fixed > Save cluster membership val.
How regression?
> Analyze > Regression > Linear > var. of interest in ‘Dependent’ & explanatory. cand. var. in ‘Independent(s)’
Regression what is R?
> R → multiple corr. coef. betw. obs. and pred. val. of dep. var.
Regression - what is R squared?
> R-squared → ↑ = better fit
Regression - what is standard error estimate?
> Stand. Err. Estim. → avg. dist. obs. from reg. line
Regression - what is F?
> F → overall fit?
Regression - what is sig.
> Sig. or p-value → if p<5%, stat. significant = ind. var. predict dep. var
Regression - what is Beta?
> Beta → Strength & direction of rel. betw. ind. & dep. var.
