PS296 Lecture Notes - Lecture 13: Standard Deviation, Statistic, Statistical Hypothesis Testing
Effect Size: All steps in 1 question
Step 1 State hypothesis:
H0 = Spring break does not effect wellness ratings. Mu = 22.5
HA: Spring break improves (one tailed) wellness ratings. Mu > 22.5
Step 2 Assess results based on SPSS output:
Based on the output, t(29) = 2.35, p = .026
We conducted one tailed, therefore, t(29) = 2.35, p = .013
The p value is less than alpha = .05, therefore we reject null
Step 3 Calculate the 95% CI;
t.05(29) = 2.045, two tailed (get this from table E.6)
CI.95 = (22.698, 25.368)
Step 4 Calculate the effect size:
dhat = xbar – mu / s
=.429
Step 5 Write a conclusion:
Based on a one sample t-test at alpha = .05 (one tailed), we can conclude that spring break improves
uni students' wellness ratings (M = 24.03, SD = 3.58) compared to the average (M = 22.50), t(29) =
2.35, p = .013.
The effect size (dhat = .43) is small - moderate, and shows that wellness ratings increased .43 SD units
more than we would have expected by chance. With repeated sampling, the true population wellness
score in university students would fall between 22.70 and 25.37 points 95% of the time.
Hypothesis Testing: Two Sample Means
-lots of names for related samples t test (related samples, repeated measures, matched samples)
Related samples:
-the same participants provide two sets of data
-in subjects design
Repeated measures:
-an experimental design involving the same participants observed over different treatments
-same participants with different scores provides
Matched samples:
-an experimental design in which two participants; individual scores produce one par of scores
-often used in twin studies, or a husband and wife's relationship (each gives one score, combined to
make one paired score)
-often we are using tests where we compare the means on two different tests
-ex participant gives anxiety score, does something anxiety producing, and then gives another score
after
Advantages of related samples design:
-no variability among participants (observed differences not due to pre-existing differences b/w groups)
-control over extraneous variables (have high power; ability to detect effect of IV on the DV)
-requires fewer participants (the same participants or matched participants experience each condition)
Document Summary
H0 = spring break does not effect wellness ratings. Ha: spring break improves (one tailed) wellness ratings. Step 2 assess results based on spss output: Based on the output, t(29) = 2. 35, p = . 026. We conducted one tailed, therefore, t(29) = 2. 35, p = . 013. The p value is less than alpha = . 05, therefore we reject null. Step 3 calculate the 95% ci; t. 05(29) = 2. 045, two tailed (get this from table e. 6) Step 4 calculate the effect size: dhat = xbar mu / s. Based on a one sample t-test at alpha = . 05 (one tailed), we can conclude that spring break improves uni students" wellness ratings (m = 24. 03, sd = 3. 58) compared to the average (m = 22. 50), t(29) = The effect size (dhat = . 43) is small - moderate, and shows that wellness ratings increased . 43 sd units more than we would have expected by chance.
