Comparing Inferential and Univariate Statistics

Inferential and Unvariate Statistics go hand in hand.

Analyses start with univariates on all your variables. Well, at least analyses should start that way.....

1. Univariate Analysis describe a variable in a sample or population: mean, median, mode, std. dev, range. This enables you to describe a variable's distribution, which enables you to describe how a group of people act, think, feel, etc..

2. Determine whether a sample represents a population.

3. With the mean and a standard deviation we can graph a variable's distribution without knowing all the data points.

4. Provide baseline data.

5. We can use central tendency and dispersion to determine if a variable's distribution approximates a normal distribution and if we can, thus, calculate various inferential statistics, which assume a normal distribution, on that variable. If not, we may be able to transform them into a normal approximation.

6. Univariate analyses guide inferential analyses:

• Determines where to focus inferential analysis.

• Determines what type of inferential analysis to do.

Why Do Inferential Analysis:

Enables us to use a sample to determine if a relationship between two or more variables exists in a population. I.e., making an inference

Bivariate and multivariate analyses

Inferential Analysis = crosstabs (proportion tests), means tests, correlations, regression.

Inferential analysis is done via hypothesis testing.

Need to choose which inferential analysis to do based on level of measurement of independent and dependent variable(s).

IV	DV	Analysis	Distribution Assumptions
categorical	categorical	crosstabs	Few
categorical	continuous	means tests	DV Normal
continuous	continuous	correlation	IV and DV Normal