Question

"Environmental Mercury, Autism Linked by New Research" reported in Environmental News Service in March 2005 that Texas researchers had found a possible link between autism and mercury in the air and water. "The study compared mercury totals reported for 2001 in the 254 Texas counties to the rate of autism and special education services in nearly 1,200 Texas school districts. The districts, which range from urban to small metro to rural, enroll four million Texas children. 'The main finding is that for every 1,000 pounds of environmentally released mercury, we saw a 17 percent increase in autism rates,' said lead author Raymond Palmer, Ph.D., associate professor in the Health Science Center's Department of Family and Community medicine."

(a) How do we best describe the individuals studied?

Texas school districts

Texas children    

Texas counties

(b) Which of these is the response variable?

pounds of environmentally released mercury in 2001

rate of autism     

(c) Which of these is the slope of the regression line?

1,000/0.17

0.17/1,000     

(d) How should they denote the slope for their findings?

b₀

β₁     

β₀

b₁

(e) Since the study produced evidence of a relationship, can we conclude that a 95% confidence interval for β₁ contains zero?

Yes

No     

(f) Considering that the sample size was quite large, was the P-value probably much greater than 0.05, approximately 0.05, or much less than 0.05?

approximately 0.05

much less than 0.05     

much greater than 0.05

(g) Which one of these possible confounding variables would have been most important for the researchers to take into account?

time of year

age of children     

gender

setting as urban, small metro, or rural

Answer 1

(a) The individuals studied are,

Texas children    

(b) The response variable is,

rate of autism     

(c) The slope of the regression line is,

17% per 1000 pounds = 0.17/1,000     

(d) The slope for their findings is a statistic and is is denoted by,

b1     

(e) Since the study produced evidence of a relationship, we cannot conclude that a 95% confidence interval for β1 contains zero.

No     

(f) As the sample size was quite large, the model would be significant and the P-value probably much less than 0.05.

much less than 0.05     

(g) The researchers to take into account below confounding variable.

setting as urban, small metro, or rural

The school districts location may affect the mercury levels in air and water and thus can be a confounding variable.