50>Hang Ups: Looking at Non-Response in Telephone Surveys

Hang-ups-Looking at Not-Response in Telephone Surveys

Nancy McGuckin
Travel Behavior Analyst/NHTS Team Member Nancy.McGuckin
moment-g.com

ABSTRACT

Since the mid-1980″due south, telephone surveys take go the standard practice for obtaining information on household travel in the U.S. (Stopher, 1996). But, for a variety of reasons including changes to the North American telephone numbering system, the availability of intercepting technologies, such as caller-ID, and the multiple contacts required to complete a two-stage survey, telephone-based travel surveys seem to be suffering from failing response rates. Recent regional telephone surveys of household activeness or travel surveys in the U.Due south. accept had household response rates ranging from 20 to xl percent (Zimowski, Tourangeau et al, 1997).

The information presented in this paper was obtained from the pretest of the National Household Travel Survey (formerly the NPTS/ATS). The 2000 pre-test included a number of method and content tests, but for this research nosotros examined the test of nine contact attempts versus xix contact attempts and the imbedded non-response follow-up survey. For the pre-test equally a whole, the household recruitment charge per unit (called the cooperation rate in this paper) was 44 percent, and the last response rate was 28 percent. This very depression response charge per unit prompted a difficult await at where we lost potential respondents in the survey procedure.

You are watching:
One of the biggest problems with telephone surveys is that

Non-response is made up of refusals and non-contacts. We have used the telephone call-disposition file (a tally of the results of each attempt to contact each household) and the non-response follow-up survey to try and empathize more about not-response. From the telephone call disposition file we have plant that non-contact is a much larger area of loss than direct refusal. For example, about 30 percent of call attempts reached an answering machine, and never contacted a person. Some other 25 pct of phone call attempts rang with no respond.

Popular:   Warframe Orokin Text on Screen

When response rates fall depression plenty, questions about the representativeness of the respondents are raised. This is probably the biggest challenge facing telephone-based and random-digit dialing (RDD) surveys. Equally contact rates and response rates fall, survey practitioners should consider a range of techniques to increase contact and participation.

1. Overview:

One of the greatest dilemmas faced by survey practitioners in the Usa today is weighing the benefits of CATI (computer assisted telephone interviewing) with an RDD (random digit dialing) sample frame compared to the seriously declining response rates for this method. The CATI/RDD method for big-calibration surveys gained popularity in the U.S. throughout the 1970″s and past the end of the 1980″s even some very traditional government agencies had switched from accost samples to RDD. The benefits of the telephone contact, using an RDD sample, compared to an address sample, include:

Lower cost per survey unit in a CATI/RDD survey allows for a much larger sample size. Complication of the questionnaire and bug arising from actual interviews are dealt with effectively since CATI technology allows for automated skip patterns and on-line edits. A centralized interviewer pool allows for more training and interviewer feedback The telephone as the contact instrument allows for more contact attempts and much greater variation in times contacts are attempted. In the U.s. potential respondents in some neighborhoods were getting more than and more wary of having interviewers prove upwards at their door.

However, in the U.S. lately we appear to be facing a serious pass up in response rates with this method. People have become wary of telephone solicitation and find it piece of cake to hang-upward the telephone. Families are busy and resent intrusions on their home fourth dimension. Perhaps the distrust of government has increased, as take concerns well-nigh privacy.

Popular:   Geeks for Geeks Company Wise

Some researchers are looking at the assumptions integral to the CASRO standard of computing response rates, such equally the supposition that not-contacts are proportionally similar to contacts in terms of eligible units. Some researchers go on to look for methodological approaches, such as pre-contact letters to legitimize the survey, very intensive contact attempts (one federal survey volition attempts each number upwards to 40 times), and multiple modes of contact.

In addition, the use of an RDD sample is further complicated by changes in the phone industry, such every bit the well-nigh six-fold increase in the number of potential phone numbers in the fourth dimension betwixt 1990 and 2000, greater availability of prison cell phones, and multiple lines in many residences.

In this research nosotros looked at non-contacts and refusals by using the call disposition file (which tallies the results of each call attempt to each sample number), and the non-response follow-upward survey imbedded in the pretest.

The telephone call disposition file is a rich source of information but cumbersome to analyze. Data on each call attempt allows us to go along track of contact patterns and telephone answering behavior, and potentially to link contact patterns to concluding outcomes. For case, some intriguing enquiry in political stance surveys (Couper 1997) institute that people who said they were not interested in participating, but later completed, had very dissimilar political attitudes then those who readily agreed to participate. Follow-up research linking the contact behavior to travel patterns might discern similar biases.

2. Changes in the Sampling Frame

Popular:   Forget Everything You Know About Mobile Games

There has been considerable discussion in the market inquiry field almost the changes in the U.South. telephone sample frame. The growth in telephone exchanges because of telephone company competition, the release of telephone numbers from geography (called portability), the proliferation of multi-line households, the projected surge in cellular phones as a main residential phone take complicated the film fifty-fifty more than.

Meet more:
The Coattail Effect Is Most Likely To Help Elect Which Kind Of Candidate?

In improver, the many barriers to contacting a person continues to business concern practitioners, as use of an answering motorcar grows in acceptance and more and more methods of screening calls are available to the telephone user. Call identifying technologies, such as caller-ID and privacy baby-sit, also equally multi-use telephone lines, increase the difficulty of identifying an eligible household from the RDD sample.

These concerns focus on 2 things: fishing for residential numbers in a more dispersed universe of working eligible samples, and barriers to contact a potential respondent via the phone line. The concerns about the sample frame abound from the proliferation of exchanges, which in turn lowers the household “hit” rate, and also brings into question some of the techniques used to increased sample efficiency (such as sampling only blocks where at to the lowest degree one number is listed). The proliferation of exchanges and the effect on the probability of connecting with a working residential phone are shown in Table 1.