The impact of gender in research
Does the gender of the interviewer effect the response rate, interview and the results obtained?
Many researchers have investigated the effect that interviewers’ gender has on research, this paper seeks to examine if there is an effect and how relevant that effect is. It will draw on academic papers, business examples and a case study.
Hyman et al (1954) were one of the first to examine the effect gender-of-the interviewer has on the interview process; they found respondents replied differently to male and female researchers. Since then there has been an abundance of research in the field, much of the research is based on and argues that gender-of-interviewer effects are evident on, topics related specifically to feminism, politics and other sensitive issues. This paper will analyse the research and apply it to a case study. With the main aim to determine if the gender-of-interview affects all interview areas or whether it is topic specific.
It is important to understand what is meant by interviewer effects; interviewer errors are expected to occur differently in every interview whereas interviewer effects refers to a specific interviewer characteristic i.e. gender (Dijkstra 1983). Interviewer effects look at how interviewer variance can bias the results of research.
The case study being looked at took place in Oldham, it was a consultancy project investigating participation rates of food waste recycling, with the main objective to increase participation rates. The methodology used was semi-structured interviews and drop off questionnaires. The interviewers were made up of two women and three men. Researchers went out in pairs, for safety reasons, the researchers went out in four pairings, three pairs made up by a women and male interviewer and the final pairing was comprised of two male researchers. Due to time constraints and the nature of the work this case study was unable to talk to residents about their impression of interviewer-gender-effects. However all interviewers were interviewed extensively by the author of this paper, to grasp their view of the effect gender had. The main topics covered were response rates, lengths of interviews and results obtained.
Does the gender-of-interviewer affect response rate?
Gender can affect the response rate; Smith (1972) suggests that women are less likely to invite men interviewers into their home explaining that it is due to the perceived ‘danger’, this argues Smith can be an issue for male interviewers conducting research. When looking at the Oldham Case study this was evident, the response rate for the mixed paring had a higher response rate to the male only pairing. The mixed pairings had a response rate average of ? and the male only pairing had a response rate average of ?. This is backed up by Dommeyer (2008) whose study examined how using a photo in the cover letter of a drop off questionnaire effects responses. Female interviewers prove to obtain a higher response rate. His research found that rivalling the gender was only productive if they were female. Bean and Medewitz (1988) had similar results when sending out cover letters with a female signature; a higher response rate was produced than when a male signature was used 35% and 26% response rate respectively. Moreover, Catonia et al (1996) experiments found that on a phone interview when respondents were given the opportunity to request a gender 82%, of women and 72% of male respondents did request with the majority selecting a female, suggesting that respondents prefer female interviewers. When interviewed, ‘Keith’ from the Oldham case study expressed: “Sandra seemed to get a higher response rate then I did, we quickly realised this and Sandra became the interviewer and I the scriber”. Johnson and Delamater (1976) discuss, whilst looking at response rates in sex surveys, the effect gender has on respondents agreeing to be interviewed. They argue that the gender of the interviewer can have a substantial effect on response levels; attributing it to the type of survey suggesting that if they are embarrassed about the topic being discussed they may be less likely to opt to participate, especially with someone from the opposite gender. In the Oldham case study this can be compared to respondents who do not participate in recycling and are therefore are less likely to agree to be interviewed, however the gender of the interviewer is unlikely to make a difference there. Benny et al (1956) notes that male interviewers gain fewer responses to female interviewers and most of these are from female respondents.
Rourke and Lakner (1989) discuss the gender bias that exists within the data collection…..
How does gender affect the results obtained?
Huddy et al (1997) looked at the effect that the gender-of-interviewer had in two surveys, where male and female interviewers were randomly assigned to interview male and female respondents. With the first survey; gender of interviewer had more of an effect on less educated and younger respondents. However these results were not replicated in the second survey. Yang and Yu (2008) argues that well educated people are not affected by gender as much because they are more use to inter-gender relations. Huddy et al (1997) also attribute it to the fact that people that are more educated are more confident in the company of the other gender. WHAT IS EDUCATION IN OLDHAM?
Many researchers have suggested that the gender of the interviewer only has an effect on certain topics. Huddy et al (1997) suggests that the gender-of-interviewer is more predominant when; politics and views on feminism are discussed. Bellou and Del Boca (1980) found stronger gender-of-interviewer effects among women respondents on questions about the existence of gender inequality whereas men tend to be more affected when questions about women movements arose. Flores-Macias and Lawson (2006) claim that in the past research has shown effect on social and political issues when interviewed by different gender. Using a survey on households in Mexico the research tries to add to the field, they found gender effects were confined to ‘sensitive’ questions, they concluded gender-interviewer effects are limited to gender topics. TOPIC OF RECYCLING
In contrast Kane and Macauley (1983) note the opposite; Women were most effected by questions on women’s movements for example collective action and women shared interest, while men were more effected by the interviewers gender when answering work-related gender equalities. Gender bias was highest with questions related to controversial politics and women’s movements. Kane and Macaulay (1993) research looks at the effects of interviewer gender on responses in particular on gender-related survey questions. They look at if gender effects are present and how it differs from male and female. Huddy et al (1997) argue respondents are susceptible to gender-of-interviewer effects across a broad spectrum of gender-linked items.
Groves and Fultz (1983) found that economic indicators receive more optimistic responses when interviewed by a male interviewer rather than a female interviewer. Landis et al (1973) reported that when women were interviewed by male interviewers they gave more feminist responses on women’s roles.
Kane and Macaulay (1993) summarise stating after analysing gender-attitude it is clear that interviewer-gender bias is present, it tends to include respondents giving a critical response to female interviewers than to male. On standard debate topics men are more likely to be effected by the gender of the interviewer and are less likely on less familiar topics. Women’s responses vary dramatically in their responses to male and female interviewers on various issues. They did not see a vast difference in interviewer gender effects for example respondents to male respondents. It is therefore questionable on it social power effects conversational power in the interview process. Both male and female respondents are at times affected by interviewer gender.
In contrast Herod (1993), who is a geographer, believes that the gender-of-interviewer can effect responses on any topic discussed. He argues that gender relations are an important aspect that can shape the interview process, gender can shape the type of data collected especially when carrying out interviews. Backing this up is McDowell (1992) who explains how interviews raise the issue of gender, Schaenberger (1992) agrees “gender makes a difference” (p.217). In the Oldham case study gender….. look at results!
Herod (1993) “gender can shape the use of interviews as a research tool” (p.306). Even when all respondents are of the same gender, gender bias still shapes the interactions between interviewer and interviewee. Feminists’ support this statement gender is significant in society it is always going to effect gender relations in the research process (Keller 1985). Herod’s paper looks at work conducted across different disciplines on gender and its effect on interviewing.
Turner and Martin (1984) in the classic work discuss how the gender-of-interviewer and the respondent has significant effect, the different options (male interviewer, female respondent/ male interviewer male respondent etc) have different effects in influencing opinions and feelings.
Eagly and Carli (1981) showed a statistical relationship between the gender of the interviewer and outcome of the interviews showing both; respondents giving different answers to male and female researchers, but also researchers interpreting it differently. Thus this demonstrates that an interviewer’s gender can affect respondents’ answers. Looking at the Oldham case this can be demonstrated through interpretation of answers, when looking at how the answers were interpreted it is completely a subjective task, when a male was interrupting it, the results appeard whereas when a female was interrupting it is seemed………….
Even when it is the same gender there is an effect, Aries (1976) suggested that men tend to be more aggressive and ‘macho’ to a male interviewer. Herod (1993) explains whereas a male interviewer may display one type of behaviour with female researcher he may show a different one to a male therefore obtaining different results. Oldham case study…..
Piliavan and Martin (1978) found that in a group setting men and women acted differently, in the Oldham council case study, researchers went out in pairs often a man and women researcher, Aries (1976) explains that she observed women were less likely to interact than men in a mixed group. This could have affected female respondents’ answers when being interviewed by a male and female researcher in the Oldham case study.
Moreover Herod (1993) notes the difference in the way male and female genders interpret information, on an interview he refers to the type of language used. Carli (1991) claims that women and men use language differently. Lakoff (1975) argues that women have to be socialised to use language that is less assertive then men and women’s expressions are tentative. Herod (1993) puts forward that these stereotypes have implications when men and women interpret language. Sociolinguist, Deborah Tanen (1990) claims that men and women have different beliefs of how conversations are meant to work, as well as different views on the role of conversational interaction and building relationships (Wolfram & Schilling-Estes, 1998).
Herod (1993) explains that there is a danger in trying to generalise about gender relations in such broad categories. Arguing that they themselves did not look at race or class and how these shape gender. There is a need to look at how gender can mean different things in different context. Interviewing is about how interviewers generate meanings and understandings.
Are men or women more affected by gender-of-interviewer?
Whelcher (1987) and Ballou and Del Boca (1980) both argue that male respondents show more effect from gender-of-interviewer. Whelcher (1987) documents that men give more democratic responses to male interviewers as they try to give the response that they feel the male interviewer wants to hear. Ballou and Del Boca (1980) states how men give more democratic responses to female interviewers.
Landis et al (1973) explain that it is expected that during an interview relationships are formed between respondent and interviewer and thus the behaviour of the respondent is influenced by their perception of the circumstances. Warren (1988) argues that women are better at building a rapport when interviewing. Landis et al (1973) found from their results that the women interviewed gave a more “feminist” response to the male interviewer, noting that the statistics were significant. They summarise that the gender of the interviewer does have a marked effect on response to women.
Hyman et al (1954) found that gender of interviewer effected female respondents’ results when a male asked the question; 61% agreed with the statement, but when they were interviewed by women only 49% agreed. Hyman concluded women felt more obliged to give conventional opinions to a male interviewer. Benny et al (1956) note that both men and women act differently in the company of the other gender; acting more formally and expressing less. Benny et al (1956) and Hyman (1954) both argue that women in the presence of men talk more traditionally are more formal and tend to give the expected answer. Oldham Case study then this >It could be argued that as both these papers are dated, the findings may be less relevant.
Macaulay (1993) claim the more egalitarian answers are more persuasive among male respondents when interviewed by a women. Flores-Macias and Lawson (2006) found that men are more likely to be effected rather than women. Furthermore that the social context has an effect, culture for example in Mexico City men were more susceptible to gender bias backing this up. Approximately 30% of men interviewed by men felt women rights were urgent however 40% interviewed by women felt it an issue. They also found women were more progressive when interviewed by men.
Holbrook et al (2003) argue that effect from respondents believe they try to answer what interviewers want. Oldham Case study
Davis et al (2010) argue that interviewer effects can impact the data obtained. They looked at measuring and controlling interviewer effects. Effects can occur from interviewer related issues such as the way questions are read, probes are used, instructions to survey etc (OLDHAM CASE study difference between male and female). Davis et al (2010) discuss how gender is the most noticeable characteristic of an interview and therefore is most susceptible to having an effect.
In the past women were considered better interviewers because they are seen as less threatening and therefore there has not been as much research on this topic. But since telephone interviews gender has become more of an issue as no longer can the respondent see “socioeconomic status, physical attractiveness, personal demeanour” (Huddy et al 1997, p.197). Huddy et al (1997) note that there is growing evidence that respondents are more likely to give a feminist view to a female interviewer as the respondent seeks to give the answer they think the interviewer wants to hear. OLDHAM CASE
Huddy et al (1997) had two goals from their study to test for the existence of gender-of-interviewer effects across a range of gender related questions. And to explore the characteristics of respondents most liable to gender-of-interviewer.
Huddy et al (1997) wanted to test to see if the existence of gender-of-interviewer effected a range of topics or just gender related questions to achieve this they used two surveys both containing questions that dealt with women’s issues and women’s movements. Their results showed respondents were more likely to give a feminist view to a female interviewer on 11/13 gender related topics. However the difference obtained by male and female interviewers was small and consistent and was only significant for a minority of questions. The topics that showed gender-of-interviewer effects differed from the two surveys in the first, carried out in 1991 the largest gender-of-interview effects occurred on questions relating to feminist identity whereas in the second, obtained in 1993 they were on topics on abortion and anti-sexual harassment legislation. In both surveys a female interviewer collected more feminist views then a man. They did find, however that gender bias occurred most on both surveys when topics on controversial politics were broached.
With Huddy et al (1997) second goal they discovered significant interaction between interviewer gender and education – less-well educated respondents were more influenced then well educated by the interviewers gender. Despite this the same results were not emulated in the second survey but they argue this could be because the gender bias was not as affluent either. To reinforce this Huddy et al (1997) assessed the statics of education and effect of gender on respondents with 12 and 17 years of education. They found that gender-of-interviewer effects were more prevalent with less educated respondents. Overall respondents with less formal education were more likely to be effected by the gender-of-interviewer and on gender related questions.
Huddy et al (1997) note that gender bias decreases with age and income was the only demographic characteristic that did not increase.
But how important is it, does it have a large enough effect to matter? Huddy et al (1997) found that small differences in their study. They conclude that it depends on the survey being administered. If it will effect then it is crucial that an equal number of men and women are randomly assigned to respondents.
Finally Huddy et al (1997) believe that from their results it can be seen that gender could effect any type of survey and use the beginning of their first survey to demonstrate this point; where gender bias is present and the topic of the survey has not been disclosed.
Huddy et al (1997) argue the view gender-of-interview effects questions is premature because; rarely have researcher controlled the individualism of interviews when examining gender-of-interviewer effects, few studies have measured the size of effects across a broad spectrum of questions to see if the effect is on feminist questions or all topics. Bellou and Del Boca (1980) did look into this in their 1980 study. Huddy et al (1997) continue explaining that effects are not standard even for questions on the same topic and few studies have tested the statistical significance across several variables with the same respondent.
Huddy et al (1997) suggest that not enough research on which gender is most prone to gender-interviewer-bias as there is a contrast in theories. Lueptow, Moser and Pendleton (1990) argue women are more likely to give feminist views to a female interviewer, which they proved through telephone interviews. In contrast Ballou and Del Boca (1980) contrasted stating male respondents are more vulnerable to female interviews and appear more feminine.
McDowell (1988) disagrees arguing that there is no gender split in research methods but rather a stereotyping in gender characteristics.
Little work has been done on the effect of the interviewers’ gender; as traditionally interviewing was a female occupation
Williams (1964) in his classic paper hypothesised that the greater the amount of social difference between interviewer and respondent the more likely of gender bias.
Demonstrates the importance of the gender of interviewer and that the subject can have an effect.
