Significance Testing and the
Hybrid Approach to Testing
The Component Parts of Significance Testing:
State a null hypothesis
Make a one-tailed versus two-tailed decision
Compute a calculated value
Determine the data-based p-value
Assessing the likelihood that Ho is true
Significance Testing Using "p-less-than" and "p-greater-than"
If the precise p-value can't be computed, the researcher might
say something like "p<.001" or "p > .15" (but the number
to the right of "<" or ">" isn't the a-level,
because there's no a in significance
Statements in which p is "bracketed" (e.g., ".01 < p <
.05" ) might also come from statistical testing
A Popular Hybrid: Significance Testing With an a-Level
The simple idea of the "hybrid" approach that combines parts hypothesis
testing and significance testing
"Marginal significance" and other terms for "near misses"
The notion of "highly significant"
Reporting fail-to-reject decisions with "p-less-than" statements
The "one-sided" alternative to the hybrid approach
The 7-, 8-, and 10-step versions of the hybrid approach