Module Components

• Learning Activities

Topics

• Results Overview
• Statistical Conclusion Validity
• Descriptive Statistics
• Inferential Statistics
• Module 4 Self-Review

NRSG 790-Methods for Research and Evidence-Based Practice

Module 4: Understanding Results

Statistical Conclusion Validity

As Trochim notes (p. 205), conclusion validity is one of the four types of validity examined in a study (the others are internal validity, construct validity, and external validity).  While Trochim drops the word “statistical” and just calls it “conclusion validity”, much of the appraisal of conclusion validity has to do with the statistics used.  Inferential statistics are used to test research hypothesis.

Trochim (2005) reminds us that there are many reasons why it is difficult to reach conclusions such as measures that have low reliability, relationships of interest may be weak or not enough information was collected. He uses the terms “signal” to represent what you are trying to see and “noise” to represent things that stop you from seeing it.  This is a good analogy because many of our statistical tests represent this relationship (e.g., F test is the ratio of explained variance to unexplained).

When assessing studies for statistical conclusion validity, review the study methods, the  data preparation, the descriptive and inferential statistics and the text where conclusions are stated in text.  Review the difference between descriptive statistics and inferential statistics.

Threats to conclusion validity can also be considered through errors of inference.  In Type 1 errors, relationships are claimed when there is not one.  In Type 2 errors, no relationships are claimed when there is one. Review these concepts.

Investigators can improve the statistical conclusion validity by estimating power, using measures  that are reliable and good implementation. Review ways to improve statistical conclusion validity.