NC SCHS: Interpreting Confidence Intervals in CHAMP

North Carolina State Center for Health Statistics

Vital Records

CHAMP Home Page

Questionnaires

Interpreting Confidence Intervals in CHAMP

The prevalence rates reported from NC CHAMP estimate the percent of the North Carolina population of children who exhibit certain health behaviors or characteristics. Since the estimates are based on a sample and not the entire population, the percentage of children in the survey with a given health characteristic may differ from the “true” prevalence of that characteristic in the North Carolina population simply by chance. Confidence intervals take this error into account and present a range in which the “true value” is likely to fall. For example, based on the 2007 CHAMP survey between 14.0% and 17.5% of children in North Carolina have ever had asthma, with a point estimate of 15.7%.

2007 North Carolina Statewide CHAMP Survey Results

Asthma

Has a doctor ever told you that (CHILD) has asthma?***

Select a different topic:

	Total Respond.^	Yes			No
	Total Respond.^	N	%	C.I.(95%)	N	%	C.I.(95%)
TOTAL	2,573	414	15.7	14.0-17.5	2,159	84.3	82.5-86.0
GENDER
Male	1,274	237	17.5	15.1-20.2	1,037	82.5	79.8-84.9
Female	1,299	177	13.7	11.5-16.3	1,122	86.3	83.7-88.5
RACE
White	1,884	293	14.5	12.6-16.6	1,591	85.5	83.4-87.4
African American	378	88	23.2	18.6-28.5	290	76.8	71.5-81.4
Other Minorities	311	33	8.4	5.5-12.6	278	91.6	87.4-94.5
HISPANIC
Yes	231	14	7.1	3.8-12.7	217	92.9	87.3-96.2
No	2,338	399	16.7	14.9-18.7	1,939	83.3	81.3-85.1

The width of the confidence interval is influenced both by the degree of certainty sought (e.g., 95 percent versus 99 percent certainty) and the standard error. A high degree of certainty (e.g. 99 percent) increases the width of the confidence interval. Sample size will also influence the width of the confidence interval, with small samples increasing the width of the confidence interval. The wider the confidence interval is, the less reliable the point estimate is indicating that the point estimate should be interpreted with caution.

The confidence interval also tells you about the stability of the point estimate. A stable estimate is one that would be close to the same value if the survey were repeated. An unstable estimate is one that would vary from one sample to another. A wider confidence interval around the estimate indicates greater instability, and thus less reliability. For example, in the 2007 CHAMP survey, 7.1 percent of Hispanics reported that their child ever had asthma, but the margin of error is 3.8 to 12.7, indicating that the point estimate is fairly unstable. In one sample of NC residents, you might have 3.8 percent of Hispanics say their child has ever had asthma, and in the next sample, as many as 12.7 percent might say their child has ever had asthma. This is more than three times as many children with asthma, but both values are still within the margin of error for the survey sample.

On the other hand, narrow confidence intervals in relation to the survey estimate tell you that the estimated value is relatively stable, i.e., repeated CHAMP surveys would give approximately the same results. In 2007, the asthma rate for white children was 14.5 percent, with the range of the confidence interval from 12.6 to 16.6 percent. The upper and lower range of the confidence interval are fairly close, therefore this is a somewhat reliable point estimate.

When making comparisons between estimates, how do I determine if the differences are statistically significant?

Confidence intervals are similar to margins of error. When the confidence intervals of two estimates of the same indicator from different groups do not overlap, they may be said to be statistically significantly different, i.e., these differences are unlikely related to chance and are considered true differences. For example, in 2007, the asthma rate for African American children was 23.2%, with a confidence interval of 18.6-28.5%. The asthma rate for white children was 14.5% (C.I. 12.6-16.6%). The confidence intervals do not overlap, so the difference between asthma rates for African American and white children is statistically significant.

In contrast, if there is any value that is included in both intervals, then the two estimates are not statistically significantly different from one another. For example, in 2007, 17.5% of males had ever had asthma, with a confidence interval of 15.1-20.2%. The asthma rate for females was 13.7% (C.I. 11.5-16.3%). The confidence intervals for male and female estimates overlap, so the difference between the male asthma rate (17.5%) and the female estimate (13.7%) is not statistically significant.

Page Last Updated March 09, 2012