Confidence Intervals

You will be using the spreadsheet that I have prepared to do confidence intervals. The link is located on the Web site. For each variable in your study you will calculate one confidence interval. You will have to decide whether to calculate the confidence interval as a percentage or as a mean. First, I will describe how to do a confidence interval for a percentage. In the output below, it makes sense to use a percentage since it is a yes/no question. For each variable only choose one value to do a confidence interval. In this case I am going to choose the percentage that said “yes.” This will be 84.1 percent or .841. The only other value I need from the output is the number of valid cases. In this case it is 170.

I then need to put the results in the spreadsheet. The spreadsheet looks like the one below.

I have entered the information on the last row. Of course I would want to delete the dummy data in the previous rows too. The spreadsheet will calculate the standard error and the upper and lower limits of the confidence interval. If the low limit is less than zero then make it zero and if the upper limit is over one, make it one. In our case above the 95% confidence interval is .786 £ p £ .896.

CONFIDENCE INTERVALS: PROPORTIONS (adjust if necessary -- .00 to 1.00)
Variable / Values / Proportion / Samp. Size / Std Error / Lower Lim. / Upper Lim.
V1 / 4,5 / 0.746 / 205 / 0.030 / 0.686 / 0.806
V2 / GT 5 / 0.200 / 215 / 0.027 / 0.147 / 0.253
V51 / Yes / 0.567 / 209 / 0.034 / 0.500 / 0.634
V97 / 1,2,3 / 0.876 / 200 / 0.023 / 0.830 / 0.922
V25 / Yes / 0.841 / 170 / 0.028 / 0.786 / 0.896

Now I will demonstrate how to do a confidence interval for a mean. In the output below it makes sense to use the mean since we have a scale for which the mean has meaning. In this case the mean is 2.595 and we also need the standard error. It is .084. Now we enter these in the spreadsheet. However we need the part of the spreadsheet for means.

The spreadsheet looks like the one below. As I did for the example above, I will enter the values in the last row.

CONFIDENCE INTERVALS: MEANS
Variable / Mean / Std Error / Lower Lim. / Upper Lim.
V25 / 3.561 / 0.078 / 3.408 / 3.714
V36 / 2.540 / 0.100 / 2.344 / 2.736
V43 / 2.745 / 0.035 / 2.676 / 2.814
V72 / 4.123 / 0.321 / 3.494 / 4.752
V26 / 2.595 / 0.084 / 2.430 / 2.760

Again, the spreadsheet computes the upper and lower level of the confidence interval. In this case the 95% confidence interval is 2.430 £ m £ 2.760.

Again, you only need to pick either one proportion or a mean for each variable in your study. Usually, most of the time it will be a proportion rather than a mean since you have used a percentage in your final report (and in your frequencies book) in reporting your findings. So in your confidence intervals you should only have one row for each of your variables. Remember if you copy rows and thus the formulas make sure you copy all the cells with values. When you are done just print the spreadsheet and put it in the appendix of your final report. Confidence intervals are relevant for all reports. In the rare case that your sample is more than 10 percent of the population, please see me because we will have to make an adjustment in the spreadsheet calculations.