Question

One poll has Trump leading in Florida by 45% to 44% while another poll has Hillary leading by 48% to 44%. Sample size was n=900 for both pools. {Use the average value of Hillary in computing the standard error, i.e., 46%.}

On the basis of the Hillary results, what is the standard error of the difference between proportions (Poll 2 - Poll 1 or .48 - .44)?

The critical value of the difference in proportions is, Ho: Poll2 - Poll1<0, 95% confidence?

The calculated test value of Z is?

T/F These results tend to support the hypothesis that the second poll has a significantly higher proportion than the first pool, using 95% confidence, upper-tail test.

Answer 1

(a) Standard error of proportions

p = 0.46 and 1 - p = 0.54

SE = SQRT[ p ( - p) / n] = Sqrt ( 0.46 * 0.54 / 900) = 0.0166

(b) For the upper tailed test, the Z critical at alpha = 0.05 is 1.645

(c) The Calculated Value of the Z statistic = Difference / SE = (0.48 - 0.44) / 0.0166 = 2.41

(d) Since t observed (2.41) is > t critical (1.645), the statement is TRUE that the results tend to support the hypothesis that the second poll has a fignificantly higher proportion than the first poll, using 95% confidence, upper tail test.