November 11, 2003

Hypothesis Testing


Testing the null hypothesis that p = 0.50 for 10 up/down forecasts of the stock market provides insights into the nature of hypothesis testing.


Increasing the sample size to 10,000 improves our ability to distinguish between p = 0.50 and probabilities just a few percentage points higher.


The upper 5% tail of the standard normal distribution is above 1.645. The critical value for the sample proportion is 0.50 + 1.645 (0.005) = 0.5062.


Posted by bparke at 03:24 PM