Survey Methodology by unknow

Author:unknow
Language: eng
Format: epub, pdf
Publisher: John Wiley & Sons, Inc.
Published: 2016-03-22T04:00:00+00:00

where a is the proportion of events that are reported (reflecting concerns about sensitivity or social desirability), b is the rate of decline in reporting as a function of time, and εi, is a deviation from the model for the ith respondent. The e is just Euler’s number, the base of the natural logarithms. Thus, the model specifies that the proportion of events correctly reported exponentially declines (rapid decline in the time segments immediately prior to the interview and diminishing declines far in the past). The literature implies that for events that are distinctive, near an easily recalled temporal boundary, and important in the life of the respondent, a is close to 1.0 and b is close to 0.0. For nonsensitive events that are easily forgotten, a may be close to 1.0, but b is large. As seen in Figure 7.2, the exponential decay model fits some empirical data better than others.

A final form of memory failure results from our efforts to reconstruct, or fill in, the missing pieces of incomplete memories. Such reconstructions are often based on what usually happens or what is happening right now. For instance, Smith’s studies of dietary recall suggested that respondents filled in gaps in their memories for what they actually ate with guesses based on what they usually eat. In a classic study, Bem and McConnell (1974) demonstrated a different strategy. Respondents in that study inferred their past views from what they now thought about the issue. This “retrospective” bias has been replicated several times (e.g., Smith, 1984). Our current state influences our recollection of the past with other types of memory as well, such as our recall of pain, past use of illicit substances, or income in the prior year (see Pearson, Ross, and Dawes, 1992 for further examples). We seem to reconstruct the past by examining the present and projecting it backwards, implicitly assuming that the characteristic or behavior in question is stable. On the other hand, when we remember that there has been a change, we may exaggerate the amount of change.

Table 7.1. Summary of Factors Affecting Recall

Neter and Waksberg (1964) on Response Errors

In 1964, Neter and Waksberg published a study comparing different designs for reporting past events.

Study design: Two design features were systematically varied: whether the interview was bounded (i.e., the respondents were reminded about their reports from the prior interview) and the length of the recall period (i.e., 1, 3, or 6 months). The context was a survey of the number of residential repair and renovation jobs and the expenditures associated with them, using household reporters.

Findings: With unbounded interviews, there were much higher reports of expenditures than with bounded interviews (a 55% increase). The increase in reports was larger for large jobs. The authors conclude that respondents were including reports of events that occurred before the reference period (this was labeled “forward telescoping”), and that rare events were subject to greater telescoping. Asking people to report events 6 months earlier versus 1 month earlier led to lower reports per month, with smaller jobs being disproportionately dropped from the longer reference periods.

Download

Copyright Disclaimer:
This site does not store any files on its server. We only index and link to content provided by other sites. Please contact the content providers to delete copyright contents if any and email us, we'll remove relevant links or contents immediately.