Dealing with nonresponses

Emawk

New Member
#1
What can be done to compensate nonresponses in the sample (Aside from eliminating them, that is.)? Should they always be excluded from the sample size?
 
Last edited:

Emawk

New Member
#3
JohnM said:
Are you referring to complete nonresponses or a situation where some questions were not answered?
Like when a random sample of 249 lincese plates are drawn to determine whether they're insured or not, and the categories and frequency you get are say:

1. Insured (226)
2. Not insured (7)
3. Missing (16)

Should the missing data be excluded all together if you were estimating the proportion of the population that are not insured?
 

Emawk

New Member
#5
JohnM said:
Sorry to keep asking questions, but what exactly is missing - insurance info or license plate info?
The "Missing" category consists of lincence plates that were drawn for the sample but where investigators were unable to find the car or its owner.
 

Emawk

New Member
#7
I'm having a problem grouping the frequencies for the categories from the data set on excel? I don't know where to go. Would you be able to help me?
 

Emawk

New Member
#9
Oh thanks.


I have another question. What are two ways to deal with non responses? Are they:

1. To ingnore and include them in computation
2. Eliminate them all together

Or is the first one not one of them? The book I'm using just gives me one, but the problem I'm doing is asking for two. It wants me do caculations based on both. I can't say they should go and gather more samples, because I won't have the data to include in the calculation.

The question asks: Provide two methods of dealing with the missing Data. From each method, determine the upper and lower limits for the estimated number of for the thing I'm looking for.
 
Last edited: