Gender framing effects in victim surveys
| Author | Petri Danielsson,Janne Kivivuori,Reino Sirén |
| Published date | 01 March 2012 |
| Date | 01 March 2012 |
| DOI | 10.1177/1477370811424383 |
| Subject Matter | Articles |
Article
European Journal of Criminology
9(2) 142 –158
Gender framing effects
© The Author(s) 2012
Reprints and permission: sagepub.
in victim surveys
co.uk/journalsPermissions.nav
DOI: 10.1177/1477370811424383
euc.sagepub.com
Janne Kivivuori, Reino Sirén and
Petri Danielsson
National Research Institute of Legal Policy, Finland
Abstract
Crime victimization surveys are important sources of trend information and provide data for basic
criminological research. In recent years, victim surveys have proliferated and their strengths are
well known. The aim of this study is to increase the methodological literature on victim surveys
by analysing framing effects, defined as the way the survey instrument communicates its topic and
aim, especially in terms of the gendered nature of violence. Three experimental frames were
applied to independent, random samples of the adult Finnish population: male-to-male violence
frame, female-to-male violence frame, and male-to-female violence frame. The impacts of these
frames were analysed in relation to two outcome variables: self-assessed propensity to report
hypothetical borderline incidents in a victim survey and reporting of prior personal violent
victimization. Thus, we utilized measures of both intended survey reporting behaviour and real
survey reporting behaviour. The findings indicate that the male-to-female violence frame increases
the willingness of the respondents to report borderline cases to survey researchers, regardless
of other factors. It also increases the prevalence of reported prior victimizations. The female-to-
male frame has a similar but weaker framing effect. The findings are discussed from the point of
view of the ‘conversation’ paradigm of survey methods research.
Keywords
experimental design, framing effect, gender, methodology, victim survey
Crime victimization surveys are standard tools used in modern criminological research.
They are used in most developed nations to monitor crime trends. In Finland, national
Corresponding author:
Professor Janne Kivivuori, Director of Criminological Research Unit, The National Research Institute
of Legal Policy, PO Box 444, FI-00531 Helsinki, Finland
Email: janne.kivivuori@om.fi
Kivivuori et.al
143
crime victim surveys were launched in 1980. Since then, and especially beginning in the
1990s, there has been a considerable proliferation of crime victim surveys targeted at
particular populations, such as women, young people and children (Haggerty, 2001).
However, the growing number of victim surveys has not been accompanied by a corre-
sponding interest in methodological research, especially with regard to how the targeting
of specific sub-populations affects the results. In this article, we examine the methodological
foundations of crime victim surveys by probing framing effects in a survey experiment.
Prior research
In survey research, a large number of factors influence the respondents’ capability and
willingness to reply to questions. Some of these are external to the research instrument,
such as the perceived climate of opinion during the research process or even the weather
at the time of research contact. Others are internal to or embedded in the research instru-
ment. These range from the method of data collection (phone interview, mail survey,
computer-assisted survey, etc.) to the wording and sequence of questions and topics within
the instrument. Factors such as the sponsor and the name of the study are also likely to
affect the responses (Bowling, 2005; Tourangeau and McNeeley, 2003). In this article,
we use the concept framing effect1 to denote effects that emerge from how the survey
instrument communicates (explicitly and tacitly) the topical focus of the survey. An important
aspect of framing effects is that they can be largely controlled by the researchers – unlike
other context effects, such as changing climates of opinion or the presence of other people
when people respond to questions. Our concepts ‘gender framing’ and ‘gendered violence
framing’ thus refer to how and whether the questionnaire communicates to the respondent
that the research is about violence by one gender against the other.
For the respondent, answering a question is a highly contextual affair. People engage
with survey questions as if participating in a natural conversation (Tourangeau and
McNeeley, 2003: 25). Respondents draw upon subtle cues to make inferences about the
intent of the questions and how they are ‘expected’ to answer (Tourangeau and McNeeley,
2003: 27). The apparent purpose of the survey as perceived by the respondent thus poten-
tially influences the response. People who respond to a survey tend to be sympathetic to
the goals of the survey. They have made the decision to engage in altruistic behaviour by
responding, and they wish to cooperate with the demands of the study (Tourangeau and
McNeeley, 2003: 28–9, 30). For example, Tourangeau and McNeeley (2003: 29) have
suggested that some US rape surveys are based on the assumption that rape is generally
underreported. In their view, it is likely that such surveys ‘cast a wide net’ by triggering
the perception that the intent is to apply a broad concept of victimization. Motivated by
the wish to cooperate with the researcher, respondents interpret the questions so that they
allow a broad inclusion of incidents. Additionally, people who already have a broad concept
of violence may be more likely to respond.
Previous research has identified at least three mechanisms through which framing
effects operate. First, earlier questions can provide the respondents with an interpretive
framework that is applied as they decipher the meaning of the subsequent questions.
Second, the preceding questions can produce a priming effect, which activates the retrieval
of topical memories and facilitates responding to subsequent questions. Third, prior
144
European Journal of Criminology 9(2)European Journal of Criminology
questions can produce judgemental contrasts so that subsequent questions are interpreted
in relation to questions presented earlier in the survey (Tourangeau et al., 2003: 491)
One potential source of framing effects is the title of the study. In a study conducted
in Norway, the researchers investigated the effect of the title and the length of the study
on the response rate in a mail survey that dealt with women’s health (Lund and Gram,
1998). The study was based on five independent random samples of middle-aged women
using five experimental conditions, each questionnaire having a different length and title.
It appeared that emotionally loaded concepts used in the title, such as the words ‘cancer’
and ‘woman’, had a positive effect on the response rate but did not influence the distribu-
tion of the risk factors. However, in studies targeting specific populations that are known
to suffer from a disease, the introduction of the questionnaire may have an impact on the
variable associations (Smith et al., 2006). In the field of youth crime surveys, a small-scale
pilot study conducted in Finland suggested that the sequence of topics matters when the
respondents are asked about attitudes towards crime. The respondents tended to be less
tolerant towards crime if the questions about attitudes were preceded (rather than followed)
by the questions about crime victimization (Kivivuori et al., 2001: 37–8). Conceivably,
attitudes related to crime are more punitive if people are first asked to report their own
victimization experiences.
Gender is highly relevant in the conversation that takes place between the respondent
and the questions formulated by researchers, especially if the research is about gendered
phenomena. There is a scarcity of research on this topic as well,2 although an important
related study was conducted by Galesic and Tourangeau (2007). Their study was a rand-
omized controlled trial examining the impact of framing on responses about sexual harass-
ment in the workplace. The respondents were randomly assigned to one of two conditions:
one questionnaire stressed the researcher’s connection with a feminist organization, while
the other described the researcher as neutral on the issue of sexual harassment. The titles
of these questionnaire versions were ‘Sexual Harassment Survey’ and ‘Work Atmosphere
Survey’, respectively. Respondents in the feminist frame reported more experiences with
sexual harassment and fewer experiences with non-sexual problems. Similar differences
emerged in the questions asking respondents to comment on how bothersome the behaviours
were. Respondents in the feminist frame condition were more likely to define all incident
descriptions as sexual harassment. Clearly, the framing of the study influenced how the
respondents replied to the questions.
In telephone surveys, the gender of the interviewer is a potential source of a framing
effect. In a study conducted in the United States, the researchers found that the respondents
were influenced by the gender of the interviewer when responding to questions about
gender roles. Both female and male respondents were more supportive of gender equality
if they were interviewed by a woman (Kane and Macaulay, 1993; Tourangeau and Yan,
2007: 876). In mail surveys, however, the paper questionnaire is ‘mute’ but not necessarily
genderless. In the...
To continue reading
Request your trial