Week 3 Theory and Logic of Hypothesis Testing Flashcards
What problem does randomization create?
To determine whether the differences between treatment conditions is entirely or partly due to chance. Will not be able to answer unless determine whether difference is significant or not. (CHANCE FACTORS)
What is the goal behind hypothesis testing?
To make inferences from a sample to a population
What is the statistical hypotheses used in psychology?
The null hypothesis (Ho) = there are no differences between means. Means equivalent
The alternative hypothesis (Ha) = means not equivalent
The decision to reject the Ho means what?
Experiment supports the original hypothesis
The decision to accept the Ho means what?
Experiment fails to support the original hypothesis
What is experimental error?
chance factors - individual differences, measurement errors etc.
Unsystematic variance
What are treatment effects?
The differences between means due to the treatment effects
Systematic variance
When Ho is true, what does this mean?
When Ho is false, what does this mean?
- Unsystematic variance alone (population means are equal)
- Systematic variance and unsystematic variance (population means are not equal)
When Ho is true, what are the two estimates of experimental error in the ratio? (numerator/denominator) What would the ratio be approx equal to?
- differences among treatment means
- differences among subjects treated alike
Ratio would equal approx. 1.0
When the Ho is false, what other component enters the ratio? What would the ratio be approx?
- Treatment effects
Ratio would equal greater than 1.0
What is the error when Ho is true, but rejected?
Type 1 error (1 - a)
What is the error when Ho is false, but accepted?
Type 2 error (1 - B)
What is true when the type 2 error is small?
Power to make correct decision increases
How can you directly control the type 1 error?
the selection of significance level
