Skip to main content

Table 1 Performance of approximations of precision of approximations to \( SE\left(\hat{\beta}\right) \) in 51 cities

From: Sample size issues in time series regressions of counts on environmental exposures

Estimator (text expression number)

Distribution of % error in estimators of \( SE\left(\hat{\beta}\right) \) over the 51 cities

Mean (bias)

Mean absolute

Lowest

Highest

A: Using known exposure distribution

 Poisson crude (1)

− 39.3

39.3

− 58.0

−19.7

 Poisson (2)

−3.8

3.9

−11.5

2.6

 Q-Poisson (3)

−2.3

2.5

−5.0

4.0

B: Using exposure distribution from year 1

 Poisson_crude (1)

−37.1

37.1

−58.6

−18.3

 Poisson (2)

7.1

15.7

−28.9

44.5

 Q-Poisson (3)

9.1

17.1

− 32.4

46.7

  1. The table summarises the distribution of errors in three approximations to the standard error of the coefficient of heat (\( SE\left(\hat{\beta}\right) \)) in the 51 Spanish provincial capitals
  2. % error = 100*[approximation-(true value)]/(true value)]