A homophone is a word whose pronunciation is the same as that of another word having a different meaning and spelling (e.g., nun and none, doe and dough, etc.). Brain and Language (April 1995) reported on a study of homophone spelling in patients with Alzheimer's disease. Twenty Alzheimer's patients were asked to spell 24 homophone pairs given in random order, then the number of homophone confusions (e.g., spelling 'doe' given the context, 'bake bread') was recorded for each patient. One year later, the same test was given to the same patients.

1. Are the Paired-Samples T-test assumptions satisfied with this data set? (Are variables related, no outliers, are samples normally distributed?)

2. May we reject the null hypothesis? Do Alzheimer's patients show a statistically significant increase in mean homophone confusion errors over time?

T-Test Paired Samples Statistics Std. Error Mean N Std. Deviation Mean Pair 1 TIME1 4.15 20 3.498 782 TIME2 5.80 20 4.213 942 Paired Samples Correlations Significance N Correlation One-Sided p Two-Sided p Pair 1 TIME1 & TIME2 20 670 <.001 paired samples test differences significance confidence interval of the std. error difference mean deviation lower upper df one-sided p two-sided pair time1 time2 effect sizes point standardizer estimate cohen d .516 .977 .042 hedges correction .041 a. denominator used in estimating sizes. uses sample standard difference. plus a factor>