Deck 13: Collecting the Data: Field Procedures and Nonsampling Error

Question 1

Which of the following is FALSE?

A) Better sampling designs and more sophisticated samples may well be instrumental in eliminating nonsampling errors in research. The more efficient the sample design, the less important are nonsampling errors.
B) The researcher should experience little difficulty in constraining sampling error to whatever bounds he or she desires.
C) It is somewhat paradoxical but true that the more statistically efficient the sample design, the more important nonsampling errors are likely to be.
D) Nonsampling errors arise from weaknesses in research procedure. They are likely to increase with increases in the sample size.
E) Nonsampling errors contribute more than sampling errors to total error in most surveys.

Answer 1

Better sampling designs and more sophisticated samples may well be instrumental in eliminating nonsampling errors in research. The more efficient the sample design, the less important are nonsampling errors.

Question 2

What type of error is usually treated in the usual standard error formulas?

A) nonsampling error
B) sampling error
C) noncoverage error
D) overcoverage error
E) nonresponse error

Answer 2

sampling error

Question 3

Nonsampling errors include

A) method errors.
B) observation errors.
C) Non-observation errors.
D) a and b.
E) b and c.

Answer 3

b and c.

Question 4

An observation error is the

A) difference between the reported value and the "true" value.
B) failure to obtain information from some elements of the population which were selected from the sample.
C) duplication in the list of sample units.
D) failure to include some units of the defined survey population in the actual sampling frame.
E) difference between the observed values of a variable and the long-run average of the observed values in repetitions of the measurement.

Question 5

Methods to exclude population elements from a sampling frame due to noncoverage biases include

A) research objectives.
B) eliminating duplicate entries.
C) explicit statements identifying exclusion.
D) b and c.
E) a, b, and c.

Question 6

Noncoverage errors can be eliminated in quota samples by

A) increasing the number of characteristics for selection.
B) increasing the interviewer's flexibility in choosing respondents.
C) sending some interviewers to strictly high-income neighborhoods.
D) thoroughly reviewing completed interview forms.
E) none of the above.

Question 7

The best way of handling duplicates when sampling from lists is to

A) skip the duplicate and choose the next name on the list.
B) get an updated list.
C) use the standard error formula to estimate the effect.
D) weight the results by the inverse of the probability of selection.
E) do all of the above.

Question 8

Noncoverage biases

A) do not affect standard error formulas.
B) can be eliminated by increasing the sample size.
C) are small in comparison to sampling errors.
D) can be eliminated by a number of methods.
E) all of the above

Question 9

The main source(s) of nonresponse bias is (are)

A) not-at-homes.
B) nonreceipt of survey instrument.
C) death of the respondent.
D) interviewer training.
E) a and d.

Question 10

Let
1=no answer
2=not at home when contacted
3=disconnected
4=refused to participate in the study
Which of them contribute to the nonresponse bias in a telephone survey?

A) 1,2
B) 1,2,3
C) 3,4
D) 2,4
E) 1,2,3,4

Question 11

Which of the following is the most effective method to decrease respondent refusals in a mail study according to the empirical evidence?

A) provide return postage
B) specify a deadline
C) guarantee anonymity
D) offer an incentive
E) offer to provide survey results

Question 12

Common strategy(ies) to correct for nonresponse bias is(are)

A) extrapolation of information contained elsewhere in the instrument.
B) use of follow-up methods.
C) increasing the initial response rate of the instrument.
D) a and c.
E) a, b, and c.

Question 13

Given $\overline{a}$ = $p_{r}\overline{a}_{r}$ = $p_{nr}\overline{a}_{nr}$ , where a is the overall mean and "r" refers to those responding and "nr" to those not responding, which of the following statements is most correct?

A) Both $\overline{a}_{r}$ and $\overline{a}_{nr}$ can be calculated by splitting up a (which is known or can be estimated) into its component parts.
B) The nonresponse component can be ignored as it is reasonable that $\overline{a}_{r}$ equals $\overline{a}_{nr}$ .
C) $\overline{a}$ cannot be calculated because of the difficulty in estimating $\overline{a}_{r}$ and $\overline{a}_{nr}$ .
D) A rough estimate of $\overline{a}$ can be made by adjusting the initial results after $\overline{a}$ is estimated by follow-up.
E) A rough estimate of $\overline{a}$ can be made by adjusting the initial results after $\overline{a}_{nr}$ is estimated by follow-up.

Question 14

A researcher conducting a mail survey develops a number of subgroups by noting the order of those responding: e.g., group 1, those responding after initial contact; group 2, those responding after first follow-up, etc. Statistics are then computed to see if there are any significant differences between the groups. If there is no significant trend the mean for the nonrespondents and respondents is ___. If there is a significant difference the researcher ____.

A) biased; scraps the project and starts all over again
B) assumed the same; extrapolates to allow for the respondents
C) assumed equal; assumes the nonrespondents are like the respondents
D) assumed different; extrapolates to allow for the nonrespondents
E) assumed equal; extrapolates to allow for the nonrespondents

Question 15

Studies which have been specifically designed to estimate the direction and magnitude of data collection errors have typically

A) relied on the actual availability of bias-free information which then served as a validity check.
B) used split-run comparisons.
C) measured the consistency of replies over time from the same respondent.
D) been applied in substantive as opposed to method studies.
E) allowed the information gained to be taken at face value.

Question 16

Item nonresponse can be treated by

A) inferring the information for the missing item from other information in the questionnaire.
B) using regression analysis.
C) substituting the average response for the item of those who did respond.
D) all of the above.
E) none of the above.

Question 17

When only a few items are left unanswered, replies can be made useful by all of the following techniques, EXCEPT

A) inferring the response from other answers on the form.
B) using the average response for the item of interest.
C) combining the "don't know" and "no response" categories.
D) using regression analysis to estimate the response.
E) All of the above are appropriate techniques.

Question 18

Which of the following can NOT be used to correct for the nonresponse bias introduced by designated respondents refusing their participation?

A) The initial response rate can be increased.
B) The impact of refusals can be reduced through follow-up.
C) The obtained information can be extrapolated to allow for nonresponse.
D) The sample size can be increased.
E) All of the above can be used.

Question 19

The most prevalent type of observation errors are

A) sampling errors.
B) field errors.
C) nonresponse errors.
D) noncoverage errors.
E) office errors.

Question 20

Which of the following does NOT need to occur for respondents to answer questions put to them?

A) understand what is being asked
B) process the information cognitively
C) retrieve the pertinent facts
D) evaluate the response in terms of its accuracy
E) all of the above need to occur

Question 21

In order to ensure interviewers understand the instructions of a data collection procedure, the following technique(s) can be implemented:

A) have the interviewer complete the questionnaire
B) observe the interviewer in action
C) test the interviewer as to the study objective and procedures
D) b and c.
E) a, b, and c.

Question 22

Which type of cheating is most detectable with normal validation procedures?

A) The interviewer asks only the key questions and fills in the remaining information later.
B) The interviewer interviews respondents in groups rather than separately as instructed.
C) The interviewer fabricates the whole interview.
D) The interviewer interviews the "wrong" respondent because the designated respondent is unavailable.
E) The interviewer employs one contact for information for two separate studies and thereby introduces contamination through respondent fatigue.