Chapter 11 - Paired Samples t-test

paired samples t-test

• comparing 2 means from the same people

• the dependent variables must be interval/ratio

• each person must be in both groups

• within-subjects research design or repeated measures research design

• working with difference scores — difference in means is compared within each person

Within-Subjects Research Design

• main advantage: the exact same people are in both groups

  • no risk that people in group 1 are substantially different than group 2

  • we always want to reduce “noise” to find our true “effect”

  • this design removes some of that extra “noise)

Hypotheses

Null Hypothesis: H₀ : μ_D = 0

Alternative Hypothesis: H₁ : μ_D ≠ 0

Large Effect

0.80 and above

Medium Effect

0.50 and above

Small Effect

0.20 and above

Results Statement Example

• Participants’ self-esteem scores were significantly higher before performing a comedy routine compared to after the routine, t(8) = 2.99, p < .05, d = 1.00

  • significance

  • direction

  • t statement: t(df) = t-statistic, alpha level, d = effect size

Assumptions of independent t-tests

• scores are independent within each sample

  • e.g., one person’s score is not dependent on another person’s score

• normality

Independent vs Paired Samples

• generally, more pros to within-subjects (paired) design

  • uses participants more efficiently (often require lower n because 1 person is used for 2 measurements)

  • can study change within individuals

  • reduces problems with individual diffenrences → increases the chance of finding a significant effect (removing noise)

Cons to within-subjects designs

• time may have an effect on the score

  • adding more noise by allowing time to pass

  • ex; 1 week goes by and now it’s exam season so people are stressed

• order effects

  • could response on one measure affect the other

  • ex; respond differently when looking at child first or adult first

• practice effects

  • repeating the same test means scores could be improving because you remember the test and are getting better at it not because of manipulation