Non-parametric Statistics

1
Non-parametric Statistics

• EDRS 5305
• Educational Research Statistics

2
When to Use

• Non-parametric statistics are used when
• data are not normal (i.e., not normally
  distributed)
• assumptions inherent in parametric statistics are
  not met

3
Chi-Square

• A non-parametric technique used to analyze the
  results of research in which categorical data are
  present.
• A most versatile and popular non-parametric
  method.

4
Chi-Square

• Different types of chi-square tests but all deal
  with frequencies, proportions, or percentages.
• Types are chi-square goodness-of-fit test
  chi-square test of homogeneity of proportions
  and a chi-square test of independence.

5
Requirements for a Chi-square

• Frequency data must be used (e.g., counts of
  persons or events)
• Expected value in any cell should never be less
  than 5
• Sum of observed frequencies must equal the sum of
  the expected frequencies
• Each score must be independent of every other
  score

6
Sample Research Question

• What is the effect of gender on seating
  preference among college students?
• In this question, gender is dichotomous, seating
  preference was a 5-item Likert question (front
  left side right side middle back).
• Chi-square table would be a 2 x 5.

7
Sample Research Question

• What is the effect of theoretical orientation on
  instructional units taught among classroom
  teachers?
• In this question, theoretical orientation was
  categorical and instructional units taught was
  categorical.