152 AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH 2001 VOL. 25 NO. 2

In the literature on focus group methods,

there is ongoing discussion about the

relative merits of constructed and natu-

ral groups. Constructed groups (where par-

ticipants have not met before) are held to be

useful when the researcher wishes to mini-

mise the potential for group conformity

brought about by pre-established group

norms and patterns of leadership.1-4 Partici-

pants who are unlikely to meet again face

less personal cost if they express divergent

views and are more likely to be honest.2,5 In

addition, newly formed groups that lack pre-

established preoccupations are also less

likely to attempt to set the discussion agenda

– something that can be problematical when

the intention is a focused discussion on a

particular topic.

Natural groups must suffice when they

are necessitated by pragmatic constraints

or when the goals of the research require,

or are enhanced by, the presence of a pre-

existing group dynamic. They are typically

used when studying natural interactions,

such as family, peer group or organisational

dynamics and in very narrow target

populations, where recruiting groups of

strangers is impossible. They are also used

when the research goal is to elicit conversa-

tions and interactions, which might occur in

the normal environment where attitudes are

negotiated and formed.6 Natural groups have

been used by researchers interested in ex-

ploring interactions between participants as

a phenomenon in itself.6 Finally, natural

groups might be preferred when exploring

sensitive topics where a group, who are

brought together on the basis of factors

considered to be socially divergent, might

be reluctant to talk with strangers.

In an ongoing qualitative research project

aimed at developing communication strate-

gies for parents who were ambivalent about

childhood immunisation, a comparison be-

tween natural and constructed groups has

further illuminated circumstances in which

constructed groups are more appropriate.

MethodsTo explore parental reception of pro- and

anti-immunisation messages, four initial

focus group discussions were held with par-

ents recruited via Early Childhood Health

Centres (ECHC) in the northern suburbs of

Sydney in 1999. Two groups comprised pre-

existing first-time mothers’ groups (natural)

and two comprised mothers who had never

met before who were recruited individually

via the waiting room (constructed). Three of

the groups were recruited via the same

ECHC and the other, a constructed group,

was recruited via another centre.

During the focus groups, the same struc-

ture was followed, beginning with an open

discussion about childhood immunisation,

after which two video prompts were shown.

One was an excerpt from a controversial tel-

evision documentary. It was chosen as re-

flecting many of the anti-vaccination

discourses identified in a previous paper and

included footage of children allegedly

harmed by vaccines.7 The other prompt was

an excerpt from a tabloid current affairs pro-

gram about a pertussis outbreak and was gen-

erally pro-vaccination, featuring footage of

babies hospitalised with the illness.

Methodology

Focus group composition: a comparison between

natural and constructed groups

Correspondence to:Ms Julie Leask, Department of Public Health & Community Medicine, Edward Ford BuildingA27, University of Sydney, NSW 2006. Fax: (02) 9351 7420; e-mail: JulieL@health.usyd.edu.au

Julie Leask, Penelope Hawe and Simon ChapmanDepartment of Public Health & Community Medicine, University of Sydney,New South Wales

Abstract

Objective: To provide insight into the

effects of focus group composition.

Method: In an early phase of an ongoing

study of parental reception to messages

about childhood immunisation, we

conducted four focus groups; two with

participants who had never met before

(constructed groups) and two with

participants who were part of a pre-

established first-time mothers’ group

(natural groups).

Results: Marked differences were noted in

the group dynamics, depth of interaction

and diversity between groups. Discussions

with constructed groups were animated,

enthusiastic, expressed more divergent

views and articulated greater complexities

of the topic. Discussions with natural

groups were generally flatter and less

enthusiastic, displaying a higher level of

apparent conformity to conventional

wisdom. The need to protect other

participants from potentially disturbing

information about vaccination was

expressed across groups but acted to

censor natural groups, where participants

knew more of each others’ sensitivities.

Implications: Insight into the factors

contributing to such differences may

enhance understanding of the contexts in

which constructed groups are more

appropriate. The processes of social

censorship may be of primary interest to

the researcher. However, where it is

paramount to elicit a range of opinions

about a potentially controversial topic, we

suggest that natural groups in the delicate

stage of norming be avoided. The

peculiarities of each individual research

circumstance are best explored in pilot

studies.

(Aust N Z J Public Health 2001; 25: 152-4)

Submitted: February 2000

Revision requested: December 2000

Accepted: March 2001:

2001 VOL. 25 NO. 2 AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH 153

A day after the groups, all participants were telephoned by the

interviewer and asked for any further insights into the issues dis-

cussed and any other reactions to the television prompts. This

was also seen by the researchers as an opportunity to hear from

participants who were less vocal in groups, to debrief them about

the video prompts and allow them to take on a more consultative

role.

The recruitment, group moderation and post-group interviews

were all conducted by the same person (JL). In three of the groups,

another member of the research team was present as an observer.

All group discussions were taped and transcribed. At an early

meeting to review our methodological approach, investigators

unanimously agreed to abandon the use of natural groups. To in-

dependently verify our decision another investigator, not present

during the groups and blinded to our initial identification, read

transcripts from each category and correctly identified the ‘cold’

group.

ResultsParticipants in all four groups were from middle-class areas of

Sydney and did not differ markedly in age, education, ethnicity

(predominantly Caucasian) and socio-economic background. In

the natural groups, all women were primiparous. In the constructed

groups, there was more diversity in parity, ranging from mothers

of one very young baby to mothers of three children aged up to

16 years.

Strong differences were noted between the constructed and

natural groups. In the former, participants were more willing to

contribute, more animated about the topic and tended to interrupt

each other. The groups were more difficult to conclude and took

longer. There was also more diversity in opinion about vaccina-

tion and the reactions to the television footage appeared less

guarded. In contrast, the natural group discussions were ‘cold’ –

a phenomenon where “participants are quiet and seemingly re-

luctant to participate”.2 There was less enthusiasm and interest

demonstrated in the topic and the moderator needed to provide

many more prompts to generate discussion. Participants talked

less, tended to agree with each other more and the groups con-

cluded sooner. In addition, there was a higher level of conformity

to positive ‘conventional wisdom’ about vaccines that had emerged

at the beginning as being dominant. Differences in non-verbal

communication supported these findings, with participants from

the natural groups being more flat in their affect. Post-group

interviews supported our observations. Participants from the con-

structed groups described enjoying the opportunity to discuss the

issue and meet other mothers, whereas participants from the natural

groups described the session as informative, even though it was

not intended, nor perceived by the researchers, to be.

In all groups, support for immunisation dominated. Non-

immunisation tended to be talked about in terms of deviance and

something which put other children at risk. This dynamic made

the further discussion of doubts about immunisation a taboo and

it was necessary for the moderator to tease out such concerns,

which were more openly discussed in the constructed groups where

participants were more likely to voice disagreement. For example:

The risk of my child having a one in 10,000 chance of having brain

damage or something, is far less than him having polio or some-

thing ...

I don’t think it’s all that conclusive ...

During one post-group telephone interview, a participant from

a focus group held later in this study (not included in this analy-

sis) shed further light on the group conformity dynamic. As well

as participating in a constructed group for our study, she was part

of a new mothers’ group that she said had brought together women

from very different backgrounds who happened to attend the same

ECHC. In their attempts to establish group rapport and bridge

such differences, she suggested that members were seeking areas

of common ground and would do this by finding issues such as

immunisation to agree upon.

In addition to group conformity was a ‘protection dynamic’

where participants expressed a need to guard others in the group

from divergent or disturbing stories about vaccination generated

through discussion or via the videos. This was found across natu-

ral and constructed groups, but led to differing effects. In the con-

structed groups it was made explicit during the discussion. For

example, one participant with three children openly voiced her

concern about the effect of the anti-vaccination video on two first-

time mothers in the group:

I feel in a way disappointed, that especially (name) and (name)

saw that, because I think it could sow a seed of doubt ...

These women then had the opportunity to reflect on their own

responses to the videos and respond to this participant’s concerns.

Conversely, in the natural groups, the protection dynamic acted

as more of a censor. For example, in one post-group interview a

participant from a new mothers’ group who was also a health

professional admitted to feeling reluctant to discuss her own

experience of caring for hospitalised children she claimed were

vaccine-damaged and subsequent concerns about vaccinating her

own child because this might upset one group member whom she

described as being particularly sensitive.

Participation in the constructed groups was also more ‘fresh’.

The women who did not already know each other were more ea-

ger to share their own personal stories with the group. One new

mother, for example, asked the group “Can I tell you about my

experience?” and continued with a powerful account of her infant

being subject to a whooping cough scare in the maternity unit.

Had she been part of an existing group, it is likely that this story

would have already been told.

Discussion and implicationsIn this preliminary investigation, we found pressures of group

conformity and a protection dynamic were accentuated in natural

groups. The taboo of voicing concerns about vaccination appeared

to be mediated by group composition. Participants from the natu-

ral groups were at the delicate stage of establishing group norms

and wanting to fit in. This group consensus dynamic may apply

Methodology Focus group composition

154 AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH 2001 VOL. 25 NO. 2

with other controversial topics, where participants who will meet

again are less likely to express divergence of views or where the

level of homogeneity that comes with natural groups acts to re-

duce diversity in findings. In addition, a desire to be protective of

others may be intensified when participants feel they are familiar

with the personal sensitivities and vulnerabilities of others. For a

pre-established group, the job of being part of a research process

is usually novel because the group exists for entirely different

reasons. While successful in some circumstances, it may be awk-

ward for members to adapt to the new function of the group that

includes the presence of an outsider as leader. We investigated

the possibility that alternative explanations might account for our

observations, particularly given the fact that women in the natu-

ral groups were first-time mothers who were more eager to ap-

pear to be doing ‘the right thing’ by voicing support for

immunisation. Although this probably occurred to some extent,

our conclusion that group composition was the prime determi-

nant was supported by comments made by participants in the in-

dividual, telephone-based, post-group interviews.

The tendency of the natural groups to incline quickly towards

group consensus when discussing vaccination issues could be

treated as a ‘finding’ (rather than a design flaw) and, indeed, some-

thing that a researcher might seek to explore, should the central

aim of the research be to understand more about how social proc-

esses affect the way mothers talk about vaccination. In our case,

however, we felt that the need to determine the full range of par-

ticipants’ responses was greater and, hence, we are avoiding natural

groups, while continuing on the principle of placing ‘like with

like’ – for example, placing people with similar socio-economic

status together. We recommend that other researchers also en-

gage in pilot procedures such as this to determine which contex-

tual influences are most salient in relation to their particular

research topic. While general guidelines exist,1-4 each situation

brings unique considerations and tradeoffs. We were disinclined

to abandon focus group interviews for individual interviews, as

the benefit of having research participants identify with and spark-

off each others’ comments was seen to be essential to understand-

ing our underlying phenomenon of interest (how ‘pro’ and ‘anti’

immunisation messages appeal to parents). The testing process

also underscored our ethical responsibilities to have both skilled

facilitation and individual follow-up procedures in place to at-

tempt to ensure that any doubts about immunisation raised by the

research process did not cause harm.

AcknowledgementsThis research is being funded by a National Health and Medi-

cal Research Council project grant. The authors would like to

acknowledge the contribution and support of Margaret Burgess

from the National Centre for Immunisation Research and Sur-

veillance of Vaccine Preventable Diseases, Margaret Gibbons from

Western Sydney Area Health Service, Julie Rogers from North-

ern Sydney Area Health Service and the Early Childhood Nurses

from the centres used in this aspect of the study. The generosity

and contribution of the women who took part in the focus groups

is also gratefully acknowledged.

References1. Murphy B, Cockburn J, Murphy M. Focus groups in health research. Health

Prom J Aust 1992;2(2): 37-40.2. Kreuger RA. Focus groups: a practical guide for applied research. 2nd ed.

Thousand Oaks: Sage, 1994: 17.3. Yelland J, Gifford SM. Problems of focus group methods in cross-cultural

research: a case study of beliefs about sudden infant death syndrome. Aust JPublic Health 1995;19(3): 257-63.

4. Morgan DL, Krueger RA. When to use focus groups and why. In: MorganDL, editor. Successful focus groups. Advancing the state of the art. NewburyPark: Sage, 1993: 3-19.

5. Siegel M, Doner L. Marketing Public Health. Gaithersburg, Maryland: As-pen, 1998.

6. Kitzinger J. Qualitative research. Introducing focus groups. Br Med J1995;311: 299-302.

7. Leask J-A, Chapman S. ‘An attempt to swindle nature’: Press anti-immunisa-tion reportage, 1993-1997. Aust N Z J Public Health 1998;22: 17-26.

Leask, Hawe and Chapman Practice Note