Researchers cannot research without knowing the ethical rules, but, even then, some of the rules seem to contradict each other.
Before an interview, members
of the Market Research Society must ensure that the name of the interviewer is given; there is
an assurance that the interview will be carried out according
to the MRS Code of Conduct.
Respondents will also be told the general subject of the interview, the purpose of the interview, and a few other details. This obligation goes beyond the interview itself because a follow-up interview cannot be carried out if the respondent’s permission was not obtained at the previous interview.
The MRS has a rule to say that respondents can be recontacted for quality control purposes; this means we can go back to respondents. As a matter of course, many respondents are recontacted
to ensure the interview was actually carried out. Some agencies recontact 5 per cent; some will contact as many as 10 per cent. The ‘back-checking’ levels of 5–10 per cent appear as part of national and international quality standards.
However, there are some research techniques that require researchers not to introduce themselves. One of these approaches is mystery shopping. If the research is carried out for
a client, but in a store belonging to one of their competitors, the MRS rules say the opposite:
‘Members must ensure that their identities are not revealed’. But there is some comfort for employees who have not been warned that a mystery shopper may appear: that small comfort
is that they cannot be recorded—in other words, photographic or sound recording equipment, hidden or not, must not be used.
Compiled by Nigel Bradley 2012.
Sources:
ESOMAR’s Mystery Shopping Guidelines, http://www.esomar.org. Reproduced with kind permission.
MRS Code, rules B11, B12, B21 and B43–B46, http://www.mrs.org.uk/standards/codeconduct.htm
Questions
1 List the things that an interviewer must tell a respondent before the interview.
2 What are the benefits of ‘back-checking’ respondents who have already been interviewed?
3 When can a researcher use a camera in mystery shopping?
© istockphoto.com/narvikk
Part 2 Data collection
132
runs the risk of breaking national laws if the respondent is not told that the call is being
recorded. CAPI and CATI interviews, which use computers for interviewing, may be equipped
with voice capture devices making a digital recording. Automated speech recognition (ASR)
software can then be used to transfer the content into written words. There are other devices
that can be used in the analysis of voices.
The basic idea with ‘voice devices’ is that changes in voice pitch may indicate emotion. In
contrast, ‘sensory devices’ examine the skin. Decreases in electrical resistance of the skin come
from increased perspiration, which may in turn be due to excitement or interest in a stimulus;
this has been used for new product tests. One device is known as the ‘psycho-galvanometer’.
Another device is the ‘electroencephalograph’, whereby electrical frequencies of the brain are
measured. Proponents of the method suggest that advertising recall and levels of attention
should be detected ‘at source’. Hidden observation or listening devices can be used in public
places, but from an ethical viewpoint, the MRS says that signs must advise that monitoring is
taking place, the purpose of the monitoring, and contact details.
‘Scanning devices’ read bar codes, which are a series of vertical or horizontal parallel
lines. The code is read using a bar code scanner. Packaging codes identify items, prices,
and manufacturer details. Scanners on electronic cash registers allow prices to be totalled
and update inventory fi gures. When linked with loyalty cards, the devices are an extremely
powerful way of mining data and identifying clusters of purchases. They can be used to
understand which merchandise to stock in particular localities and, specifi cally, how to
position products in stores. For researchers, bar code scanners can be used in three ways:
1. At the checkout, to facilitate sales and record the time, price, and quantity for analysis later
2. By auditors to analyse stock in the store
3. By researchers (or respondents themselves) at home.
The TNS Superpanel, the UK’s leading continuous consumer panel, consists of 15,000
households. Data is collected twice weekly via electronic terminals in the home, with
purchases being recorded via home-scanning technology (see http://superpanel.tns-global.
com/superpanel/).
An ‘audimeter’ is a recording device attached to a television to monitor the time and
channel. Diff erent versions exist, whereby respondents are obliged to register their presence
in the room. The term ‘audimeter’ is derived from early use with radio. The ‘instantaneous
audimeter’ was developed to allow telephone lines to relay the information to the researcher.
These are sometimes known as ‘peoplemeters’.
Factors that affect
cooperation
Response quality
Whether we are asking questions or making observations, there will be problems with
measuring human nature. Questioning techniques are explained fully in Chapter 6 and
Chapter 4 Primary data
133
these off er the researcher ways
to identify and to measure aspects of the human situation.
However, all instruments employed share the potential problem of instrumentation effect. Any instability of the instrument or the mere fact that it is being used may result in an inaccurate reading.
This means that, however well considered, any research study will have some defect. The good researcher tries to anticipate the best ways of minimising any problems that may aff ect results.
If we test the knowledge of a respondent and then repeat the test, there will be a diff erence in outcome. This is due to learning from the fi rst round and this
is a good example of ‘instrumentation eff ect’. Similarly, there is a concept called mere measurement effect, which suggests that ‘the mere measurement’ of attitudes creates them, and where they exist, may change them (see Dholakia and Morwitz 2002 for more on this). Researchers must accept that there are respondents with poor literacy skills and we also know that people do not always tell the truth. Being untruthful can be due to various things: the situation of the interview, a like or dislike for the interviewer, lack of knowledge, even simply forgetting.
The relationship with the interviewer can aff ect response. Acquiescence response bias is a form of sympathy where a respondent agrees with a proposition, rather than disagreeing. This
is particularly a problem with children and the elderly, who, in an attempt to be cooperative and polite with strangers, will tend to agree before considering the full implications of a question.
Therefore, the choice of interviewer is important. A ‘same sex’ interviewer may be best for some topics: for example, a female for a study of breast cancer and male for a study of testicular cancer. A ‘same race’ interviewer may be appropriate if cultural diff erences aff ect cooperation; a ‘same age’ interviewer may be appropriate for some topics. Interviewers may work in pairs for such reasons and for reasons of safety.
Satisficing is relevant here. Krosnick, Narayan, and Smith (1996, p. 29) point out the relevance to the researcher when they say that informants ‘shortcut the cognitive processes
necessary for generating optimal answers’. To carry out the task of answering questions
effi ciently, the respondent: must be fully exposed to the question, either by listening carefully or reading carefully; must understand the meaning, or at least interpret it; must recall applicable information; must process these data into something meaningful and then must be able to articulate this in a spoken or written manner, or in some cases, by non- verbal communication. Satisfi cing is said to occur if any of these is not carried out to the respondent’s best ability or if his/her ability is not suffi cient for any one part to be completed.
Part 2 Data collection
134
Satisfi cing can mean that a ‘suffi cient’ answer is given when a better one might have been
available.
When designing research, we must be mindful that one approach may only give one
viewpoint of the issue under inspection. For that reason, researchers are advised to consider
triangulating their studies. Triangulation means that we examine a problem from three or
more viewpoints. The benefi t is that there is more chance that the other approaches may
identify something that was not apparent in one. This is not a case of repeating the study, but
it is a case of carrying out three studies simultaneously, in a complementary way. Triangulation
has been divided into various types: we can add to the data, add another investigator, add
a theory, and add a diff erent method. These interests may have arisen from concerns about
response quality and response rates, which we will now examine.
Response quantity
The ‘response rate’ refers to the number of complete responses that have been obtained by
a researcher compared with the number of eligible individuals. The response rate is always
expressed as a percentage. Because the percentage will be between 0 and 100, response rates
are often seen as a ‘barometer’ to suggest that a survey has been ‘good’ or ‘bad’, depending
on how high the percentage is. This can be quite misleading, particularly when non-random
sampling methods have been adopted.
Response rates vary for many reasons. Let us divide these into intrinsic factors and extrinsic
factors. ‘Intrinsic factors’ are those that are a part of the project itself, ones that the researcher is
more able to change, and extrinsic factors are exterior ones, ones that will aff ect response, but
about which the researcher can do little (see Table 4.12).
Intrinsic factors include the interviewer. Even the most experienced and trained interviewers
can introduce bias simply by interacting with people where there is a certain level of comfort.
Another intrinsic factor is the sample quality: a good list of appropriate respondents is
clearly superior to a poorly formatted one that contains duplicates, misspellings, and ‘dead’
respondents. The choice of words for use in the introduction can make or break cooperation.
Intrinsic factors and extrinsic factors that aff ect response
Intrinsic factors Extrinsic factors
Interviewer Weather
Sample quality Temporal factors (time of year, day)
Type of introduction Interest in topic
Instrument design Respondent profi le (age, experience with surveys, personality,
mood) Motivating factors Competing events (holidays, national events, other surveys)
Incentives Suggers, fruggers, and duggers
Table
4.12
Chapter 4 Primary data
135
The professionalism of the questionnaire or other research instrument can also aff ect response. Similarly, respondents can be motivated by various things: they may cooperate for altruistic reasons, or for incentives.
Conversely, ‘extrinsic factors’ include: the weather; the time of year or day; interest in the topic; the respondent’s age, experience with surveys, personality, and mood. There may be events that compete for the respondent’s attention such as holidays, national events, or other surveys. In the 1990s, interviewers were banned from the city-centre streets of Sheffi eld, which had extreme eff ects on interviewing. In 1998, this was relaxed somewhat and interviews could take place if a
‘Sheffi eld City Badge’ was worn, each of which was numbered and kept on a register.
Here we can introduce three terms specifi c to the market research world: the work of researchers is compromised by ‘suggers’, ‘fruggers’, and ‘duggers’.
Suggers are people or fi rms who ‘sell under the guise of research’; fruggers are people who
‘fund-raise under the guise of research’, and duggers are people who ‘gather data under the guise
of research’. These activities may be face-to-face, on the telephone, or even by post. For example,
on the street, a member of the public may be approached and asked a few simple questions, but the questions will be there simply to decide if the person is a prospective customer or charity donor. Who can refuse to give money when asked: ‘Do you agree that children should be given
food?’. Forms that ask lifestyle questions arrive by post or are inserted into magazines; sometimes,
they come with a free pen, a prize draw, or some other incentive. If they request a donation, they are frugging; if they do not, they are database-building or dugging.
The spiral of silence
Elisabeth Noelle-Neumann developed the ‘spiral of silence’ concept in 1974. Her main points are these: society threatens deviant individuals with isolation; individuals experience fear of isolation continuously; this fear means that individuals assess the climate of opinion continuously; this assessment aff ects whether they express opinions openly or conceal them. Let us apply this to political voting. In order to avoid isolation on favouring unpopular policies, people will look to their environment for clues about the dominant opinion; they will identify which views are gaining strength and which are in decline. If they feel that their personal views are among those in decline, they are less likely to express them openly.
As a result, the views perceived to be dominant seem to gain even more ground and other possibilities decline further. These are Noelle-Neumann’s own words:
The more individuals perceive these tendencies and adapt their views accordingly, the more one faction appears to dominate and others to be on the downgrade. Thus the tendency of the one
to speak up and the other to be silent starts off a spiralling process which increasingly establishes one’s opinion as the prevailing one.
The spiral of silence concept therefore suggests that support for one or other political party
is not recorded in opinion polls because people prefer not to admit their allegiance to an interviewer or researcher. From this, we have learnt that two important areas must be probed
in addition to a question on voting intention:
‘ ’
Part 2 Data collection
136
Common mistakes