Question

6) A researcher wants to find out the association between hospitalizations and infections while in the hospital. He found that people who had longer stay in the hospital tend to have more infections at a rate of 3.0 per 35 days of hospitalization. People who did who discharged relatively quickly had infections at 1.2 per 35 days. Calculate the rate difference associated with hospital stay. Briefly explain your results.

7) In your own words explain the difference between random error and non-random error and provide an example for each.

8) Explain why adjustments are needed in determining risk and rates.

Answer 1

(6). The rate difference is (3 - 1.2) per 35 days i.e 1.8 per 35 days. This means that if a patient has a relatively quick discharge form hospital then his/her rate of infection is likely to be reduced by 1.8 per 35 days.

(7). Random errors are statistical fluctuations that can occurs in either direction of the true value due to limitations of precision in the experiment. For e.g. an experimenter repeating the same experiment several times may not be able to take the measurements in exactly same way every time to reach to the same value and thus a error creeps in. These errors can be studied through statistical analysis and can be reduced by averaging out over large no. of observations.

Non-Random errors on the other hand are the errors which are consistent throughout the experiment and are in any one direction of the true value. For e.g. there is some error in the instrument like zero setting error in a weighing machine, the calibrations on the instrument are wrong etc. Diagnosing these kind of errors is very difficult as they are in the same direction.