home *** CD-ROM | disk | FTP | other *** search
- What Analysis Should You Use?
- @1,You want to get descriptive statistics of SINGLE variable(s)
- @2,You want to get DESCRIPTIVE statistics of two RELATED variables
- @3,You want to COMPARE two variables, Independent or Paired
- @4,You want to COMPARE more than two variables, Independent or Related
- @5,You want to examine ASSOCIATION between two variables
- @6,You want to examine ASSOCIATION between more than two variables
- @7,Definitions of Terms Used
- ##1
- DESCRIPTIVE STATISTICS & GRAPHS PROCEDURES TO USE
- ═══════════════════════════════ ════════════════════
- ┌─ Data is ─────> Mean, S.D., Box Plot, 5 number summary
- │ Normal Histogram, Conf. Interval
- │ (Stat Module, B, C, & E)
- │
- │─ Data not ────> Median, Box Plot
- │ Normal Histogram, 5 number summary
- One Sample ───│ (Stat Module, B & E)
- │─ Data is
- │ Categorical──> Frequencies, Pictogram
- │ (Crosstabs Module, B)
- │
- └─ Observations─> Time Series Plot
- Over Time (Stat Module, option G)
-
- ##2
- DESCRIPTIVE STATISTICS & GRAPHS PROCEDURES TO USE
- ═══════════════════════════════ ════════════════════
- ┌─ Data are─────> Pearson's Corr. Coeff. &
- │ Normal X─Y Scatterplot
- │ (Stat Module, option F &
- │ Regression Module option B & D)
- │
- Two Samples────│─ Data not─────> Spearmans Corr. Coeff. &
- (Related) │ Normal X─Y Scatterplot
- │ (Stat Module, option F &
- │ Regression Module, option D)
- │
- └─ Data are─────> Crosstabulations and
- Qualitative 3─D Bar Chart
- (Crosstabs Module,
- options D & E)
-
- ##3
- COMPARISON TESTS ─ TWO SAMPLES TEST TO USE
- ════════════════════════════════ ═════════════════════
- ┌─ Data are─────> Paired t─test
- │ Normal (t─test & ANOVA Module,
- │ Option C)
- │───Samples───│─ Data not ────> Freidmans Test
- │ Related │ Normal (Non-Parametrics Module
- │ │ Option C)
- │ └─ Data are
- │ Dichotomous──> McNemar's test
- Two Samples ──│ (Crosstabs Module,
- │ Option F)
- │ ┌─ Data are─────> Ind. Group t─test
- │ │ Normal (t─test, ANOVA Module,
- │ │ option B)
- │ │
- │──Samples────│─ Data not─────> Mann─Whitney U test
- Independent│ Normal (Non-Parametrics Module,
- │ Option B)
- │
- └─ Data are─────> Chi─Square (Homogeniety)
- Qualitative (Crosstabs Module,
- option D)
- ##4
- COMPARING MORE THAN TWO SAMPLES TEST TO USE
- ═════════════════════════════════════════ ═════════════════════
-
- ┌─ Data are─────> Repeated Measures ANOVA
- │ Normal (t─test & ANOVA Module,
- │ Option C)
- │
- ┌─Samples───│─ Data not ────> Friedman ANOVA
- │ Related │ Normal (Non-Parametrics Module,
- │ │ Option C)
- │ └─ Data are
- │ Dichotomous──> Cochran's Q test
- │ (Non-Parametrics Module,
- More than │ Option D)
- Two Samples ──│ ┌─ Data are─────> Independent Group ANOVA
- │ │ Normal (t─test & ANOVA Module,
- │ │ Option B)
- │ │
- └─Samples───│─ Data not─────> Kruskal─Wallis
- Independent│ Normal (Non-Parametrics Module,
- │ Option B)
- │
- └─ Data are─────> Chi─Square Test
- Qualitative (Crosstabs Module,
- Option D)
- ##5
-
- TESTING ASSOCIATION BETWEEN TWO VARIABLES PROCEDURE TO USE
- ═════════════════════════════════════════ ═════════════════
-
- ┌─ Data are─────> Pearson Correlation
- │ Normal Simple Linear Regression
- │ (Regression Module
- │ Option B or D)
- │
- Two Samples Related────│─ Data not ────> Spearman Correlation
- │ Normal (Regression Module,
- │ option D)
- │─ Data are
- │ Qualitative──> Chi-Square (Independence)
- │ (Crosstabs Module,
- │ Option D)
- └─ Data mixed────> Spearman Correlation
- Normal, Not (Regression Module,
- Normal option D)
- ##6
-
-
- MORE THAN TWO ASSOCIATED VARIABLES PROCEDURE TO USE
- ═════════════════════════════════════ ═════════════════
-
- ┌─ Data are─────> Multiple Regression
- │ Normal (Regression Module,
- │ Option C)
- │
- More than 2 Samples ──│─ Data not─────> Kendall partial rank─
- Related │ Normal correlation
- │ (N.A.)
- │
- └─ Data are─────> Discriminant Analysis
- Qualitative (N.A.)
-
-
-
- ##7
- DEFINITIONS
-
- NORMAL refers to data that are well approximated by a normal (Gaussian)
- distribution.
-
- NOT NORMAL refers to quantative data that are not normally distributed.
-
- CATEGORICAL refers to nominal data, such as male/female or brown/blue/black.
-
- QUANTITATIVE refers to data that are numeric such as height, batting average,
- number of people per household, etc.
-
- QUALITATIVE refers to data that describe attributes such as hair color,
- socioeconomic class, sex, etc.
-
- ASSOCIATED refers to variables where knowledge of one helps predict the
- other.
-
- INDEPENDENT refers to variables where knowledge of one does not help predict
- others. Usually, samples from unrelated populations.
- (continued)
- ##8
- DEFINITIONS
- (Continued)
-
- RELATED refers to samples where multiple measures are taken on the same or
- related entities. For example, before after weights for a diet, or heights of
- twins.
-
- DICHOTOMOUS refers to data that are categorical and can take on only one of
- two possible states. For example, yes,/no or on/off. VARIABLE refers to the
- observed measure, such as height, hair color, etc.
-