For the year 2003, the sample proportion of adults who believed the United States made the right decision is calculated as follows:
\[
\hat{p}_1 = \frac{x_1}{n_1} = \frac{1080}{1500} = 0.72
\]
For the year 2008, the sample proportion is:
\[
\hat{p}_2 = \frac{x_2}{n_2} = \frac{573}{1500} = 0.382
\]
The difference in sample proportions between the two years is:
\[
\hat{p}_1 - \hat{p}_2 = 0.72 - 0.382 = 0.338
\]
To construct the 90% confidence interval for the difference between the two population proportions, we use the formula:
\[
(\hat{p}_1 - \hat{p}_2) \pm z \sqrt{\frac{\hat{p}_1(1 - \hat{p}_1)}{n_1} + \frac{\hat{p}_2(1 - \hat{p}_2)}{n_2}}
\]
Where \( z \) is the z-score corresponding to a 90% confidence level, which is approximately \( 1.645 \).
Substituting the values:
\[
0.72 - 0.382 \pm 1.645 \cdot \sqrt{\frac{0.72(1 - 0.72)}{1500} + \frac{0.382(1 - 0.382)}{1500}}
\]
Calculating the standard error:
\[
\sqrt{\frac{0.72(0.28)}{1500} + \frac{0.382(0.618)}{1500}} \approx \sqrt{0.0001344 + 0.0001576} \approx \sqrt{0.000292} \approx 0.0171
\]
Now, substituting back into the confidence interval formula:
\[
0.338 \pm 1.645 \cdot 0.0171
\]
Calculating the margin of error:
\[
1.645 \cdot 0.0171 \approx 0.0281
\]
Thus, the confidence interval is:
\[
(0.338 - 0.0281, 0.338 + 0.0281) = (0.3099, 0.3661)
\]
The 90% confidence interval for the difference between the two population proportions is:
\[
\boxed{(0.310, 0.366)}
\]