1) Measures of Central Tendency:
mean
median
mode

2) Measure of dispersion

variance
standard deviation
 

3) BIVARIATE RELATIONSHIPS

When you are relating two NOMINAL variables

measure of association:

Lambda

test of statistical significance:

chi-square

When you are relating two ORDINAL variables

measure of association:

Gamma
(or when N is small Kendall's Tau B)

test of statistical significance:

chi-square OR
z-score of Gamma

When you are relating two variables, one ordinal and one nominal:

measure of association:

Gamma
(or when N is small Kendall's Tau A or B)

test of statistical significance:
chi-square OR
z-score of Gamma
 

4) When you are using interval level data, and multiple variables, use regression (OLS- ordinary least squares)

sign and value of coefficient: tells you the direction of the relationship (positive or inverse) and how much a one-unit change in independent variable causes a change in the dependent variable (e.g. however you have measured your dependent variable-- dollars, percentages, etc)

t-score and p value (in SPSS "Sign T"):
tells you if the effect of the independent variable on the dependent is statistically significant (not due to chance)

Standard error of each variable: If it is large relative to the coefficient then this means that if you took a "critical" observation out then your results would change alot.

R-square: tells you how much of the variance in the dependent variable you have explained (i.e. How much of the explanation have you accounted for)

F-ratio and prob (p-value): tells you if the model as a whole is statistically significant.
 

5) When you are using a dichotomous dependent variable (i.e. coded as 0 or 1) then use in SPSS logistic regression. The test of significance is called the WALD test.

6) When you are using a ordinal dependent variable and multiple independent variables then use LOGIT regression. Test of significance is T-test.