Chapter+10

__**Chapter 10: Data Collection and Basic Descriptive Statistics (pp 323-329)**__
=**1.1 ** Characterize types of errors commonly made during data collection. = There a number of different types of **nonsampling errors** (errors in the research process involving anything except the sample size) that can be made during data collection.

If a researcher uses **fieldworkers** (Individuals hired to administer the survey to respondents) to collect data then the results may be subject to **intentional fieldworker errors.** In this type of error fieldworkers will deliberately falsify information and may submit fraudulent results.

Fieldworker errors may be minimized by hiring a reputable data collection company that utilizes **training, supervision** and **built-in validation techniques** (there is a move in the UK to create interviewer certification).
 * Unintentional fieldworker ** errors are another risk associated with using fieldworkers to collect information. These mistakes are often caused by fatigue or a lack of understanding on how to conduct the survey.


 * **Respondent errors ** are another type of error that may be made during data collection.
 * **Intentional respondent errors ** arise when the respondent knowingly provides false answers or fails to give an answer.
 * **Incentives, anonymity assurances, confidentiality, ** and **follow-up validation** may be employed to reduce the frequency of intentional respondent errors.
 * **Unintentional respondent errors ** occur when the respondent is confused, distracted or inattentive in any way.
 * **Good survey/questionnaire design, adequate pretesting of the questionnaire, “no opinion” or “no response” options, negatively worded items and prompters ** (Example: Asking respondents “Do you have any other things come to mind?”) are effective ways to reduce unintentional respondent errors.

=**<span style="font-family: Calibri,sans-serif; font-size: 11.5pt;">1.3 ** Describe the various types of nonresponse that a researcher may encounter with a survey. =
 * <span style="font-family: Calibri,sans-serif; font-size: 11.5pt;">For Online Surveys there are 3 unique data collection errors. **
 * **<span style="font-family: Calibri,sans-serif; font-size: 11.5pt;">Multiple Submissions by same respondent: ** <span style="font-family: Calibri,sans-serif; font-size: 11.5pt;">Unless an online survey is controlled then it is possible for respondents to submit their completed surveys multiple times which may result in an overrepresentation of the individual’s view. Multiple submissions may be reduced by asking for an email address (although this does not stop people from making false/duplicate email addresses).
 * **<span style="font-family: Calibri,sans-serif; font-size: 11.5pt;">Bogus respondents and/or responses: ** <span style="font-family: Calibri,sans-serif; font-size: 11.5pt;">Due to the anonymity of the internet many individuals will provide false responses to the questions. Coupled with the multiple submissions error, bogus responses have the ability to wreak havoc on a survey’s results.
 * **<span style="font-family: Calibri,sans-serif; font-size: 11.5pt;">Misrepresentation of the population: **<span style="font-family: Calibri,sans-serif; font-size: 11.5pt;">Not all consumers are equally computer literate and thus online surveys will not reach all segments of the population equally. Some segments of the population that have less access to internet surveys include the elderly, low-income families, technophobic people and people living in remote areas.
 * **<span style="font-family: Calibri,sans-serif; font-size: 11.5pt;">Nonresponse Errors: **<span style="font-family: Calibri,sans-serif; font-size: 11.5pt;">Nonresponse errors are failure on the part of the prospective respondents to participate in the survey or answer a specific question.

**<span style="font-family: Calibri,sans-serif; font-size: 11.5pt;">3 Types: **

 * <span style="font-family: Calibri,sans-serif; font-size: 11.5pt;">Refusal: **<span style="font-family: Calibri,sans-serif; font-size: 11.5pt;">Prospective respondent declines to participate in the survey.


 * <span style="font-family: Calibri,sans-serif; font-size: 11.5pt;">Telephone surveys **<span style="font-family: Calibri,sans-serif; font-size: 11.5pt;">have to refusal rates estimated to be as high as 50%. Offering to call when it is more convenient, identifying the name of the research company, making the survey as short as possible and emphasizing you are not selling anything are all ways you can reduce the refusal rate on telephone surveys.

<span style="font-family: Calibri,sans-serif; font-size: 11.5pt;">A **Break-Off** is when a respondent decides to stop answering questions for the survey for any reason. Potential reasons for a break-off to occur include the survey taking too long; topic or questions being too distasteful or too personal; instructions becoming too confusing; instructions may be too confusing; a sudden interruption may occur; or the respondent may choose to take an incoming call rather than finishing the interview.

<span style="font-family: Calibri,sans-serif; font-size: 11.5pt;">Successful research companies today have realized that including interviewers in the marketing process is the key to their success in gathering market information.
 * <span style="font-family: Calibri,sans-serif; font-size: 11.5pt;">Item Omission/Refusal to Answer Specific Questions: **<span style="font-family: Calibri,sans-serif; font-size: 11.5pt;">If a marketing researcher suspects ahead of time that a particular question, for example income, will have some degree of refusals it is appropriate to include “refusal” as a choice on the survey.


 * <span style="font-family: Calibri,sans-serif; font-size: 11.5pt;">Completed Interview: **<span style="font-family: Calibri,sans-serif; font-size: 11.5pt;">Because researchers will often experience break-offs and item omissions the researcher will have to judge which surveys may be considered “completed”. If sufficient data has been gathered for the results to be analyzed then the results may move on to the analysis stage. Some surveys, however, require all questions to be answered in order to move on to the analysis stage. Some surveys will include secondary details at the end of the survey so even if a respondent does not finish responding to the survey they may be able to submit the results to the next stage (“if respondent answers past question 18, then the survey is considered complete, demographics begin on question 19”).