Notes
Quantitative Research (Oct 11, 2006)
- Syllogysm
If <antecedent>, then <consequent>
Antecedent: Cause
Consequent: Effect
Converse: Change antecedent and consequent
Adverse: Negate antecedent and consequent
Equivocal: open to two or more interpretations
A hypothetical syllogysm
"If it rains, sue carries and umbrella"
| | Antecedent | Consequent |
| Affirm | OK,
Sue carries | EQUIVOCAL
May not be raining |
| Deny | EQUIVOCAL
Sue may be eccentric | OK
Not raining |
Converse: If sue carries umbrella, it rains.
Adverse: If it does not rain, sue will not carry umbrella
Quantitative Research (Oct 18, 2006)
- Types of variance
Total variance
- we calcualte from the entire sample (both group combinedvariance). i.e.: 42.5
Explained variance: between groups
- sum of two treatment group variance seaparately and add. i.e.: 22.5=(11.25 11.25)/1 (where 1=DOF)
Unexplained variance (error): within groups
- remaining variance. i.e.: 2.5=(42.5-22.5)/8 (where 8=DOF)
Quantitative Research (Nov 1, 2006)
- Analysis of variance
If null hypothesis is three equalities H0: X1bar=X2bar=X3bar
- usually there are 2 variables for a hypothesis IV and DV
- but more than one levels of one variable X1, X2, X3
If there are two IVs then ANOVA is called "two-way anova"
If there are three IVs then ANOVA is called "three-way anova"
But we usually have one DV and it is called "univariate analysis"
ANOVA table is also called "source table"
There are two sources of variation: 1. between groups and 2. within groups
Mean square: Sum of Squares/DOF
Post-Hoc t-test:
Anova does not say everything about three level null hypothesis
We need post hoc t-test to understand every relationships within null hypothesis
Quantitative Research (Nov 8, 2006)
- Crosstabs (Contingency Tables)
ASN 6: Quantitative results usually follow these steps in papers. So follow these steps in HW:
1. Purpose (RQ) and H0
2. Display results
3. Reject H0? at what level?
4. Interpret and characterize results
All quantitative analysis test null hypothesis but their difference comes from the data structure which requires different analysis.
Bonus: graph in excel
ASN6 SPSS:
- analyze/descriptive statistics/crosstabs
- put DV as the row, and IV as column
- goto cells (press cells button) select column in percentages, make sure that observed counts already checked and continue
- click the statistics button select chi-squared (chi squared test is the simplest one: it tests the null hypothesis) and continue
- you get the first output for homework
Quantitative Research (Nov 15, 2006)
- Measurement scales
1. Nominal: categories only (i.e.: male, female)
Order
2. Ordinal: difference between orders are not included (i.e.: first, second, third...)
Intervals: Same distances
3. Interval scale: an ordinal scale intervals between observations are same (i.e.: likert scale)
Notion of absolute/natural zero
4. Ratio scale
Quantitative Research (Nov 29, 2006)
- Correlation and Regression
Correlation Measures:
- Direction of relationship (sign)
- Scatter (goodness of fit)
Correlation does not measure "rate of change"
Regression measures "rate of change" on top of what correlation does
So...
Regression measures:
- Direction of relationship (sign)
- Scatter (goodness of fit)
- Rate of change