Reaching beyond the white middle classes
BMJ 2004; 328 doi: https://doi.org/10.1136/bmj.328.7453.1433 (Published 10 June 2004) Cite this as: BMJ 2004;328:1433Data supplement
Problems with researcher assisted completion10
The problem of mutually exclusive response options
The depressing experience of a morale scale
Table A Common areas of misunderstanding or sensitivity in demographic questions
Who’s giving the answers?
Beryl and Jim are a couple in their 80s, living in sheltered accommodation. They are part of a housing study commissioned by the local council, being completed by researcher Caitlin. Jim has been selected to complete the study, but as soon as Caitlin sits down to interview Jim, Beryl appears with tea and proceeds to answer all the questions for him. Caitlin explains that it’s Jim’s views she wants, and suggests she’ll talk to Beryl later. Beryl leaves them to talk, muttering that "we both think the same thing anyway," but within five minutes is back again, answering for her husband. On several occasions Jim begins an answer one way, but after Beryl’s intervention, changes his mind. Caitlin prompts that it’s important he says what he feels, but he states "Beryl knows me better than anyone else, she always knows the right thing to say."
Problems with researcher assisted completion10 In the following example, the participant is required to answer the question: "Overall, would you say that life brings out the best in you, treats you like everyone else, or doesn’t give you a chance?"
Interviewer: Overall, would you say that life brings out the best in you ...
Participant: Yes
Interviewer: treats you like everyone else ...
Participant: Yes
Interviewer: or doesn’t give you a chance?
Participant: Eh?
The problem of mutually exclusive response options
Interviewer: Is the pain better, worse, or the same?Participant: Yes
Interviewer: No, I mean is it better OR worse OR the same?
Participant: Yes
[In a later interview to clarify the response]
Interviewer: Tell me how your pain is better, worse, and the same
Participant: [after much discussion through interpreter] Yes. Better than a really bad pain, worse when I don’t take the tablets, and the same as my wife’s pain.
The depressing experience of a morale scale
Bruce was working on a study of ulcers in older people. He included in his package of measures the Philadelphia Geriatric Morale Scale.w1 This includes questions on sadness, loneliness, and whether life is worth living. Bruce had piloted his questionnaire and felt that this measure was best placed nearer the end of the questions since it contained potentially upsetting questions. Nevertheless, he was unprepared for the distress he experienced on realising some participants were isolated and had contemplated suicide. He felt that having raised this issue, it was his duty as a researcher to comfort and offer support services to participants. However, his project had been funded on a tight budget, and he was unable to negotiate extra time in his schedule for supporting participants. Furthermore, the principal investigator (who was not directly involved in the fieldwork) discouraged Bruce from talking about his own emotions.
- Lawton MP. The Philadelphia geriatric morale scale: a revision. J Gerontol 1975;30:85-9.
- Bradby H. Describing ethnicity in health research. Ethn Health 2003;8:5-13.
- Aspinall PJ. Describing the "white" ethnic group and its composition in medical research. Soc Sci Med 1998;47:1797-808.
- Foster MW, Sharp RR. Race, ethnicity, and genomics: social classifications as proxies of biological heterogeneity. Genome Res 2002;12:844-50.
- Goodman AH. Why genes don’t count (for racial differences in health). Am J Public Health 2000;90:1699-702.
- Mays VM, Ponce NA, Washington DL, Cochran SD. Classification of race and ethnicity: implications for public health. Annu Rev Public Health 2003;24:83-110.
- Nanchahal K, Mangtani P, Alston M, dos SS, I. Development and validation of a computerized South Asian names and group recognition algorithm (SANGRA) for use in British health-related studies. J Public Health Med 2001;23:278-85.
- Bornstein K. My gender workbook: how to become a real man, a real woman, the real you, or something else entirely. London: Routledge, 1998.
- Carstairs V. Deprivation indices: their interpretation and use in relation to health. J Epidemiol Community Health 1995;49(suppl 2):S3-8.
Table A Common areas of misunderstanding or sensitivity in demographic questions
Item
Potential misunderstanding or sensitivity
Suggested solution
Age
Older participants may find this question rude or embarrassing, especially when put to them by a young researcher.
Consider asking for date of birth and written age rather than putting the direct question "what is your age?".
Ethnicity
Problems with fixed-response categories include the reproduction of racialised categorisations, overemphasis of homogeneity within groups and differences between them, and failure to offer terms with which people identify and which can express complex identities.w2 Traditional ethnic categories are too heterogeneous to be meaningful (eg in the UK, "white" encompasses British White, Irish, and Eastern European, and "Asian" includes many different Bangladeshi, Indian, and Pakistani groups).w3 Sociocultural and material differences generally greatly outweigh biological inheritance as determinants of health, hence white or Asian.w4 w5
Self-assigned ethnicity allows multiple affiliations to be described, but introduces the costs of processing free text.w2 An up to date and standardised range of fixed categories, with the option of other is sometimes preferable to a free text response.1 w6 Researcher assigned ethnicity can increase comparability and consistency but may be at odds with self identity.w2 Classification by surname has some value in British Asians.w7 As society becomes more interbred, social classifications are likely eventually to replace ones based on race or ancestral origin.w4
Gender
Although most people understand the terms male and female, researchers may wrongly assign gender based on observed social role, a practice facetiously referred to as counting shirts and blouses. Often the terms are incorrectly used interchangeable. In some research, transgender and intersex options may be needed.w8
Do not assume the gender of a participant. Gender refers to your social role--how you act "male" or "female" or how other people see you. Sex refers to your biological makeup (a male would be defined as having a penis and testicles and female by a clitoris, labia, and vagina)
Income
People are rightly suspicious of the researchers’ need to ask this question, which is generally considered private information. Those at extremes of income may give responses that regress to the mean. Those with illegal sources of income will not declare this.
Explain why the information is needed and what it will be used for. Give income brackets for participants to pick from rather than asking for a free text response.
Job description
Collecting job titles can be used to ascertain social status through standard occupational classification. However, these can be very costly and time consuming to recode, titles can be vague (eg nurse could mean any grade, or level of experience), and people may conceal or talk up a low status or stigmatised occupation.
If you wish to collect this information, build in time for additional re-coding.
Occupational status
Being unemployed is generally a source of low self esteem, and asking about it may exacerbate this. Full time carers or home workers may describe their occupation as none.
If possible, ask the participant what they actually do, and code accordingly. Use easily understood terms. ‘Job’ may be a clearer term than ‘occupation’. ‘Home manager’ may be more politically correct but ‘housewife/husband’ better understood.
Post code
Postcodes can be a valid indicator of social deprivationw9 and access to services, but be sure you keep it protected if you do record these details. Potentially allows participants to be identified, hence collection of full postcode arguably breaches the Data Protection Act.
You could use the first three letters of a postcode, but this may not reflect the socioeconomic diversity of the area. Be wary of using a health centre or surgery postcode as a cover all. Some participants may live in a socially deprived area but be registered with a GP in a nearby wealthier ward.
Relationship status
Widowed, divorced or single status may be perceived as stigmatising, and a bereavement or divorce may be distressing to recall. Most demographic menus exclude ‘going out with someone’. People may be in a relationship that isn’t marriage or living together, but have to choose ‘single’ as there is no other option.
Only ask this information if it informs your research. If relationship status is important, then ensure all types of relationship are represented.
Religion
May be perceived as an intrusive (and possible racially motivated) question. Religion itself is relatively uninformative unless religiosity (the extent to which the person practises their religion) is also explored. The devout in one religion may have more in common with the devout in other religions than with more liberal sects in their own.
Only ask about religion if it directly adds to your research. If it does, considering measuring aspects of religiosity – for instance, to what extent does the participant’s religion have an impact on their day-to-day lives, moral decisions, and so on?
Sexuality
A complex taxonomy of different sexual preferences and activities may be beyond the comprehension and tastes of many respondents. On the other hand, simple classification as ‘gay’ or ‘straight’ wrongly implies homogeneous categories. It will miss those who are not sure, and individuals who perform same-gender sex acts without identifying themselves as ‘gay’ or ‘bisexual’. Note the bias of recruiting gay participants from ‘gay’ venues: you will miss those who are not ‘on scene’.
Use judgement to decide how much information on sexuality your research needs. If appropriate, pose a number of questions about sexuality rather than offering a set of closed categories – for example, sexual preference, sexual behaviour (Do you have sex only with women? Only with men? Both equally? etc).
See more
- Return of the “firm” gets cautious welcomeBMJ December 07, 2016, 355 i6556; DOI: https://doi.org/10.1136/bmj.i6556
- Cancer drugs remain FDA approved despite lack of benefit, study findsBMJ December 06, 2016, 355 i6568; DOI: https://doi.org/10.1136/bmj.i6568
- Whole brain radiotherapy for brain metastasesBMJ December 05, 2016, 355 i6483; DOI: https://doi.org/10.1136/bmj.i6483
- Five thousand dead and counting: the Philippines’ bloody war on drugsBMJ November 28, 2016, 355 i6177; DOI: https://doi.org/10.1136/bmj.i6177
- Fair vaccine pricing please, not random acts of charityBMJ November 23, 2016, 355 i6173; DOI: https://doi.org/10.1136/bmj.i6173
Cited by...
- Exploring views and experiences of how infections are detected and managed in practice by nurses, care workers and managers in nursing homes in England and Sweden: a survey protocol
- Survey research in anesthesiology: a field guide to interpretation
- The use of internet-mediated cross-sectional studies in mental health research
- Rational use of medicines in the hospitals of Poland: role of the pharmacists
- Caring for those who die at home: the use and validation of 'Care Of the Dying Evaluation' (CODE) with bereaved relatives
- Patients' experiences and expectations of general practice: a questionnaire study of differences by ethnic group