BOLD Educational Software
Choosing the Correct Statistical Analysis

Visit our Web Store
Return to B.A.R.R.C.

 
Choosing the correct Statistical Analysis
What you need to know

Before selecting the correct statistical analysis, you need to

  1. Have a clearly identified problem statement
  2. Have a clearly identified purpose statement
  3. Have research questions that address your purpose
  4. Testable hypotheses, and
  5. Operationalized variables
Testable hypotheses

A hypothesis is the researcher's prediction, derived from theory or speculation, about how two or more measured variables are related to one another.

Testable hypotheses often state
H1: There is a significant difference in (continuous variable) between (categorical variable)

or

H1: There is a significant relationship between (continuous variable) and (continuous variable).

These are the two most basic types of hypotheses. Review quantitative research studies for examples of many more types of testable hypotheses.

Operationalized variables

Operationalizing your variable means that you know how you are going to measure it. For example, we operationalize 'gender' by asking a person if they are male or female.

Variable level

Is your variable

Nominal level (one or more mutually exclusive categories)

Ordinal level (ranked data where the differences in value are not equal - categories are mutually exclusive and exhaustive)

Interval level (there are meaningful amounts of differences between the data values but there is no absolute zero)

Ratio (equal distances between data points (e.g., between 1 and 2, 2 and 3, etc.) and there is an absolute zero

Categorical/Continuous

Nominal level variables are always categorical level

Ordinal level variables are usually considered categorical, but if you sum a series of Likert-type ordinal level questions, you will end up with a continuous variable

Interval/ratio level variables are always continuous variables

Steps to choosing your statistical analysis

Follow this link to a presentation that will ask you a series of questions. Each answer will lead you closer to a recommended data analysis. Before you begin, you must know

  1. How many dependent (outcome) variables will you have? This is the variable you are trying to predict
  2. What type of outcome (dependent variable) are you measuring (categorical or continuous)?
  3. How many predictor (independent variable) variables will you have?
  4. What type of predictor variable are you measuring (categorical or continuous)?
  5. If you are using categorical variables, how many categories (groups) will you have - one, or two or more?
  6. If you are using a categorical variable as your independent variable, will you be using independent samples, paired samples, or a single sample (compared to a hypothesized mean)?
  7. Finally, you will need to know whether you expect your data to meet the assumptions for parametric statistics (for example, are the data normally distributed and was the sample randomly selected).

 

Need help with research methodology, survey design, data analysis, or editing?
Contact us at editing@bold-ed.net

Copyright BOLD Educational Software 2014
by Diane M. Dusick, Ph.D.
All Rights Reserved