Prejudiced attitudes vs prejudiced behaviour

Andrew Leigh has released interesting research he and co-authors have been doing on ethnic prejudice and discrimination, using several tests: sending employers CVs that are identical except for the ethnicity of names and seeing how many applicants get a call back, seeing whether there are differences in how many mis-addressed letters are returned to sender depending on the ethnicity of the sender’s name, and the implicit association test which I encouraged people to take last year.

Compared to an Anglo control group, Chinese and Middle Eastern names particularly were received less favourably in all three tests. Italian and Indigenous names generally did better. Here is the The Age‘s write-up of the research.

While I think this research is valuable, I differ slightly with Andrew L in how I look at the issue. For example, in discussing the various methods of examining ‘racism and discrimination’ Andrew and his co-authors Alison Booth and Elena Varganova say:

… social researchers have often used surveys to measure the degree of racism in a society. But if respondents know the socially correct response, then this approach will also provide a biased estimate of true attitudes towards racial groups.

I’m not sure that this is as big an issue as it might seem, since survey after survey has found that plenty of people will admit to some prejudice. Indeed, if we are simply trying to measure ethnic attitudes just asking people has the advantage that it is not influenced by the other factors that come into play when behavioural tests are used. For example, employers may take into account the actual or presumed views of existing employees and customers as well as their own prejudices, and in face-to-face situations social norms on politeness will come into play (none of Andrew L’s methods involve face-to-face situations).

On the other hand, the behavioural tests provide more important information than the prejudice questions. The smooth and fair running of society depends far more on how people behave than what they think, and from a policy perspective behaviour is much easier to control than opinions. The people bashing and robbing Indian students should be threatened with jail, not lessons in celebrating diversity.

24 thoughts on “Prejudiced attitudes vs prejudiced behaviour

  1. Andrew, thanks for your post (and for your verbal feedback in the Melbourne Uni seminar recently). I suppose what I’m reflecting is the standard economists’ prejudice for revealed preference over stated preference.

    Perhaps we could settle the issue by surveying respondents on racism, and then covertly seeing how they respond to interactions with actors of various races. Might be hard to get this one through the human ethics committee, though.


  2. Although The Age reports that the applications made it clear that the applicants had undertaken secondary schooling in Australia, I think there could still be a quite reasonable concern about accent/diction. I have a Vietnamese friend and a Singaporean friend who both came over at about 13 and I still find them difficult to understand at times. It’s a pity the CVs didn’t just state that the applicant was born in Australia. How else could you explain less discrimination against Aborigines than against Chinese?


  3. “How else could you explain less discrimination against Aborigines than against Chinese?”

    Maybe some people just don’t like Chinese.


  4. As I have said to Andrew L, I’d also like to see the results by how difficult the names are to say. In my experience, when you advertise for lower skill jobs you usually get quite a few applicants who could on paper do the job. So pretty minor factors will often dictate who gets an interview. It’s embarrassing ringing people and saying their name incorrectly, so this may be a factor at the margins. That many Asian people adopt Anglo first names suggests that they see advantage in having a more familiar name when living in an English-speaking country.

    That said, the self-reported prejudice surveys show roughly the same results. For example, in the 2007 survey of social attitudes the social distance questions showed that English and Indigenous people were more popular than Lebanese or Vietnamese people. So I don’t doubt that there is discrimination, even if as Rajat suggests it may not be intrinsic to the ethnic background but rather an assumption that on average people from certain backgrounds are more likely to have language difficulties or other attributes not favoured by employers.


  5. Does this not overlook a common practice. The job is advertised but the successful candidate is already chosen.
    So it is easier to get some Anglo interviewees to tell them to piss off. If you call in foreigners they may go all sensitive when they find the job is not theirs.


  6. Son of the Rat, I think there can be an element of “White Man Guilt” associated with being more likely to offer a chance to Aborigines. We feel like they have been mistreated by our Ancestors in the past, whereas other races we don’t have that sense of guilt towards.

    Andrew, I would be interested to know if a similar style of test has been done by testing the interview offer rate to females of breeding age. There is a likelihood that women in their mid 20s-mid 30s (especially married women) are less likely to get hired because of maternity absences. This would be industry specific and would probably only show up in career-path full-time jobs.


  7. M – I don’t know of one, though it would be fairly easy to do – just CVs identical other than gender. Most CVs don’t have marital status on them these days, but it could be added for the purposes of the experiment.

    This is a tricky one, since while discrimination on this ground is illegal, it is not ‘prejudice’ – staff turnover is a significant business cost even without the added expense of maternity leave, and so in an all-other-things-being-equal contest of CVs the rational employer would choose the male as a lower risk.


  8. Does it deal at all with how Chinese and Middle Eastern people feel about ‘Anglo’ people, or are the tests all one way?


  9. “I’m not sure that this is as big an issue as it might seem”
    That’s because you’re white. Personally, if you are going to brutally discriminate against certain groups, then you shouldn’t be too surprised when they end up not particularly caring about your laws or social customs. That might not be too bad with the Chinese, since they have the habit of becoming rich (obviously against the odds), and their social customs arn’t too different to Australian norms (although you can still find things to complain about if you want, my favorite anti-intellectual one is trying too hard at school!), but it’s easier to see the effect with other groups who have remained poorer.


  10. Conrad – It was a methodological point about polling, which I don’t think has anything to do with my skin colour. Andrew L was concerned that people would not reveal their true prejudices in a poll because of social norms against prejudice. While I agree that prima facie this is a reasonable concern, the fact that so many people do confess to it suggests that compared to other methodologies which have other bias issues it may be an ok method. I’m confident that even if the self-report polling does not give us a 100% reliable measure of the prevalence of prejudice, it does give us reliable measures of how different ethnic groups rank. For example, the social distance survey came to the same results as Andrew L’s research: Anglos and Italians get more positive responses than Asian or Middle Eastern people.


  11. Andrew N:

    I take it that Andrew L sent out what amounted to phony resumes. I don’t wish to overstate this however if the attendant survey methods were dishonest how can people verify if the results weren’t as well?

    What happened when people called and asked the applicant if s/he wanted to come in for an interview? Did the researchers fess up or lead them on?


  12. I just read the description in The Age, Andrew and can talk from personal experience and why at times academic research can be far removed from reality.

    It says:

    Australian National University researchers Alison Booth, Andrew Leigh and Elena Vargonova sent out 4000 fake job applications to employers advertising on the internet for entry-level hospitality, data entry, customer service and sales jobs, changing only the racial origin of the supposed applicants’ names.

    My family owns a business that at times, over the past few years required I help them select job applicants for low level entry jobs.

    They ran ads through The Age and later I ran ads for them through an internet service.

    I don’t have a racist bone in my body. However I found that it was extremely difficult being able to carry a conversation in regular English with a number of the Asian candidates. Most didn’t have an Australian driving license which is a requirement for some of the jobs.

    Naturally I gravitated to people that could converse in reasonable English and that meant locals. It was sometimes very difficult finding out if these people had Australian driving licenses before they ever came in as they at first said they did and later when they attended the interview the opposite came out. In fact I grew tired of this game and began screening people.

    An Indian was hired last year who managed to get through my screening simply because he had good English skills and promised me he had an Australian driving license after I asked him about 10 times over the telephone conversation 🙂

    1. numerous low level job applicants don’t have good English skills and it was really difficult to communicate over the phone.

    2. a large number are students that are unable to work every day. Again this came out later in the interview.

    3. May not have a work permit and therefore looking for a cash job.
    (most asked it it was a cash job and it wasn’t).

    4. Don’t have an Australian driving license which is required for insurance purposes.

    If AndrewL hasn’t allowed for some of these points, I would say his research is not only incomplete it would be downright dangerous to rely on.

    My own bias is as follows. Anglo’s at low level jobs are very unreliable in terms of even showing up to an interview. Italians were just as bad. Asians, when you find them can be really quite reliable and great employees if they have passable English.


  13. JC – As my original post says, I agree the employer tests have problems if we are just trying to measure raw prejudice, though the broad consistency with straight questions about prejudice suggests that this is a factor. On the other hand, they do measure something – even if it is just the heuristics employers use, based on experience, to quickly sort through a pile of CVs. These are likely to be ‘unfair’ to many applicants who never get a chance for various reasons (as a CV reader, I am particularly prejudiced against spelling errors, even when the job does not require much writing; and I am rarely keen on applicants I deem ‘over-qualified’, much like it seems the employers in this survey, because I don’t think they will stay long).


  14. Andrew:
    I can tell you that the number of Asian sounding names I simply ignored were numerous and this was done for the simple reason that time was a constraint and I didn’t want to deal with people that couldn’t speak English well enough to understand other people whose English was a second language or they didn’t have a driving license. Racism didn’t factor into it one bit for my part.

    Now the problem is that the Age went with the story and employers get a bad name for no other reason that they are time constrained like I was. People cut corners and what may appear to be the case isn’t at all.

    I agree the employer tests have problems if we are just trying to measure raw prejudice, though the broad consistency with straight questions about prejudice suggests that this is a factor.

    From my experience it is a factor only in so far as to screen potential candidates so you do waste your time. For one job, I had over 150 applicants. Frankly I had no intention to respond to all of them, which obviously means that people screen. However I can tell you that there was never any sense of being racially prejudiced in doing so. It was simply a case of time vs likelihood, that’s all.

    On the other hand, they do measure something – even if it is just the heuristics employers use, based on experience, to quickly sort through a pile of CVs.

    Well yes, they probably measure the methods employers use to screen people. Let me tell you anecdotally that I was actually trying to screen away from European candidates last year as for the most part they were unreliable, such as not turning up for work or not showing up at the interview. Eventually I found that more than a few Europeans were using the call or the job interview to use as evidence they were actively seeking a job with the Centre link office.

    Even entry level hospitality requires reasonably good English skills.

    Not blaming Andrew L, however The Age ploughs along and cries ” racism”. This shows prejudice of ignorance, however I expect no less from The Age.


  15. JC, assuming that people with Asian sounding names have inadequate English skills sounds like racial profiling and prejudice to me.


  16. “JC, assuming that people with Asian sounding names have inadequate English skills sounds like racial profiling and prejudice to me.”
    I agree. Chinese have been in Australia almost as long as whites. If you want to see the effect of long term discrimination, feel free to go to France and the UK. Personally, given the amount of discrimination Chinese have had over the last 100 years, Australians should consider themselves lucky in terms of the outcomes of that group.


  17. No matt, it’s profiling, not racial prejudice as you infer. Lots of people don’t have time to go through countless resumes when its likely to be the same result.

    Have you thought that perhaps reasonable English skills may be a required or is that racial prejudice too?


    Stop kidding yourself, the numbers around before the termination of the white Australia policy were too small to even mention. That’s a mighty thin straw you’re clutching.

    You two guys don’t really get it do you? If has nothing to do with hiring people as a result of racial background.


  18. JC, I really don’t see how profiling based on race can be devoid of prejudice. Reasonable English skills is definitely a must in an English speaking country – that was never in question. This is completely irrelevant to the point that assuming a person with an Asian sounding name will have inadequate English skills is prejudice.

    Perhaps you are missing the point that regardless of your motivation, when you decide to not consider someone for a job based on an ‘Asian sounding name’, the result is discrimination based on a person’s ethnic background.

    It may not be your intention to discriminate on the basis of ethnicity, but it is an inevitable outcome of your actions.


  19. Matt:

    And the choice according to you is that one should go through 150 resumes in the hope of finding someone that has adequate English speaking skills and an Australian driving license in order to prove to the outside world that you’re definitely not being a racist? Is that it?

    ……..thanks, but I much rather save the time.


  20. We all use heuristics all the time, to deliver fast, usually good enough, judgments. Sometimes these heuristics use sensitive characteristics (eg Chinese names) to provide short-cuts to non-sensitive characteristics (eg English language proficiency). The problem is that this is not just prejudice, since there are lot of people with Chinese names for whom English is a second language. The more realistic way to deal with this is for people with Chinese names but high English proficiency to make this clear on their CVs and/or follow up with a phone call.


  21. True enough, Andrew.

    However I always opt for the lazier route if it produces the same result 🙂

    Why over exert when it’s totally unnecessary? Lol


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s