Computing confidence intervals for a proportion with summary data

This tutorial covers the steps for calculating confidence intervals for a single proportion in StatCrunch.
While this tutorial uses summary data, see Computing confidence intervals for a proportion with raw data to compute one-sample proportion results with raw data.

Calculating a confidence interval for the proportion

If the coin used is "fair", the proportion of of heads the coin will produce over a very long run of flips should be 0.5. A coin was flipped 50 times, resulting in 31 heads and 19 tails. StatCrunch can create a confidence interval for the proportion of interest. Choose the Stat > Proportion Stats > One Sample > With Summary menu option. In StatCrunch, a success is used to define the outcome of interest. In this case, consider a head result to be a success. Set # of successes to be 31, and set # of observations to be 50.Under Perform, choose Confidence interval for p. By default StatCrunch has a value of 0.95 for the Level input which will produce a 95% confidence level for the long run proportion, p. Changing this value to 0.99 would produce a 99% confidence interval. Leave the Level at the default 0.95 and click Compute!. The results below show a 95% confidence interval for the long run proportion of Heads with "L. Limit" representing the lower limit of this confidence interval and "U. Limit" representing the upper limit of this confidence interval.

Changing the confidence interval method

By default the Standard-Wald normal approximation is used for calculating the above confidence interval. To instead use the alternative Agresti-Coull method, choose Options > Edit to reopen the one-sample proportion dialog window, and change Method to Agresti-Coull. The results below show a new confidence interval. The "L. Limit" and "U. Limit" values have changed because of the change in method of calculation.

Always Learning
Pearson