This is a mixed methods study aimed at assessing whether there is a variance in the perceptions of the size of the gender wage gap, and if there is a variance in the way different ethnicities perceive the causes of the gender wage gap, focusing specifically on the employer discrimination variable. Pre-existing research on gender wage gap perceptions specifically of African or Caribbean Black was very limited, so the interest in addressing a specific demography was sparked. However, pre-existing research notes that the causes of the gender wage gap differ due to a multiplicity of variables. This study consists of 107 respondents recruited in a convenience sampling framework, and 12 interviews were pooled for qualitative analysis. A bivariate and multivariate regression were carried out on independent variable ‘ethnicity’ and dependent variables ‘perceived pay gap’ ‘employer discrimination’ whilst controlling for gender, age and educational qualifications. The quantitative data did not return a statistically significant between the pairs of variables, and the controls. The qualitative data semi-structured interviews allow for a wider breadth of understanding of the perceptions of the gender wage gap.
INTRODUCTION
The gender wage gap is, if anything, a persistent issue for the UK society. Bøler et al (2015:1) argue that even in ‘relatively equal societies such as the Northern European countries’, the gender wage gap is an often ignored dimension of inequality. Khoreva (2011:234) argues that the ‘existence of the gender pay gap almost all over the world has generated the need to better understand various factors contributing to it’ and she explains that there are both micro and macro explanations for its occurrence- the macro level being focused on economic theories and exploring the differences in ‘education, work experience’ and ‘different types of discrimination’, whilst micro level explanations examine personal factors such as ‘individual preferences’. The 2011 census showed us that in the ten years since 2001, Britain became more ethnically diverse: there are now 1,865,000 (15% of the population) ‘Black, African, Caribbean, Black British’ people in England and Wales (ONS, 2015). Considering this, a more in-depth analysis of how different ethnicities, especially this large a percentage of the population, perceive and understand the gender wage gap would be helpful in policymaking and awareness in order to remedy or tackle the issue. As the gender wage gap perceptions of African Black and Caribbean Black people in the UK seem to be relatively unexplored, which I will address in a review of literature related to this issue, the purpose of this study is to attempt to highlight the perceptions of the gender wage gap from people whom are members of these groups.
The range of literature addressing this same area of concern is seemingly significantly limited, and this sparked my interest in the topic area. Examining the literature, I discovered that although many studies explored the gender wage gap, or the differences in ethnic minority employment levels and wage levels, there was a lack of attention given to perceptions of the gap. Blackaby et al (2005), using the Office of National Statistics Labour Force Survey (data from around 60,000 UK households) to examine how native ethnic minorities are fairing in the British labour market, found large differences in employment levels by ethnicity. Using data pooled from 1993 to 2000, of 1463 British born ethnic minority people, they found that there is a 18.8% difference in employment decompositions between whites and British born ethnic minorities, and they describe this employment and earnings disadvantage as ‘particularly severe’. This research highlights that these examinable ethnic inequalities in pay and employment exist, and includes analysis of multiple ethnicities to build a more complete picture, however it doesn’t measure or analyse perceptions as it uses a large scale sample and analysing perceptions was not the intention of the research. Green and Ferber’s (2005) study of 2010 men and 1780 women also draws light to the existence of a difference in wages for minority ethnic communities, although unlike Blackaby et al, this study was conducted in the U.S., which makes the findings of it less generalizable to the population I am concentrating this study on. Green and Ferber also offer explanations for the ethnic wage gap, concluding that although discrimination plays a significant role in the differentials, other factors such as occupation and education contribute to the discrepancies. As useful as this research is for understanding the existence and scale of wage differentials, it doesn’t focus on perceptions of the gap. Following on from this, Mandel and Semyonov (2016) used data from five decennial years between 1970 and 2010 to investigate gender differences in trends and sources of the racial pay gap. Although the aim was not to identify a gender wage gap, they found that the data suggested earning inequality is more gendered than racialized and women of both races share considerable earnings disadvantage relative to men. This study identifies that women of colour tend to occupy jobs considered as being the ‘lowest positions’ in the labour market. This finding was interesting for me, as it paves the way for further questioning- how does this concentration of labour into a particular earnings bracket affect a person’s perceptions of the gap in gendered wages?
Although the three pieces of literature I reviewed gave great detail and attention to identifying the gap, and at times considering the drivers of this, I felt that my study needed to focus on the perceptions of the gap of black African and Caribbean people, as it had not seemed to have been attempted before. Hence, the focus of this research will be to understand gender wage gap perceptions using a mixed methods approach, attempting to address two research questions:
a) How do people of African and Caribbean descent (Black) perceive the size gender wage gap relative to another ethnicity (White and other ethnic categories).
b) Does ethnicity play a role in the perception of the causes of the gender wage gap?
METHODS
This study set out to use a mixed methods approach to examining perceptions of the gender wage gap, to aid depth and breadth of understanding and corroboration (Bryman 2006:105) in the findings produced. By using a process of triangulation, defined as the combination of different methodologies when studying or examining the same phenomenon (Jick, 1979), it is hoped that this study will address perceptions of the gender wage gap with a more holistic and complete approach. It is hoped that by using more than one method, collated results can be used to explain each other and uncover potential unexplored aspects of the phenomenon (Bryman, 2015)- the quantitative results generate a picture of the relationship between the respondent’s ethnicity and their perception of the size and experience of the gender wage gap, whilst the qualitative results offer an explanation, or uncovers the drivers behind the data examined in the quantitative analysis. From this thinking, I developed research questions and hypotheses aimed at investigating the relationship of ethnicity and perception of gender wage gap utilising the capabilities of investigation the mixed methods approach allows for.
Research Question 1
How do people of African and Caribbean descent (Black) perceive the size gender wage gap relative to another ethnicity (White and other ethnic categories).
H0: There is no association between perceptions of the size of the gender wage gap and ethnicity (African/Caribbean and other ethnicities)
Ha: There is an association between p
erceptions of the size of the gender wage
gap and ethnicity (African/Caribbean and other ethnicities)
Research Question 2
Does ethnicity play a role in a person’s perception of the causes of the gender wage gap? (Using the variable of employer discrimination)
H0: There is no association between ethnicity and perception of the cause of the gender wage gap
Ha: There is an association between ethnicity and perception of the cause of the gender wage gap
Before proceeding further with the study, it was important to complete an Ethics checklist to ensure that the study did not potentially include unethical practice.
Quantitative Methodology
For the quantitative element of this study, data was collected from a purposive sample of 107 respondents who were chosen based on their availability and accessibility. Due to both the time and location constraints, the data collected was a result of non-probability sampling, which means we must be careful when generalising results from the dataset and interviews. Convenience sampling was used to select respondents, this is when a sample is picked from the target population just because they are willing and available to be interviewed at the time. It is easy to do and requires little time and money in comparison to other sampling techniques that could have been used, such as a probability method like stratified sampling. However, the issue with this type of sampling is that it is highly susceptible to selection bias and interviewer effects are likely to be heightened when the interviewer is familiar with the interviewee. Thus, the credibility of my research is reduced. In addition to this, there is a reduction in external validity due to the size of the sample further which negatively impacts the generalisability of the results to the wider population, as it is not a representative sample. Despite being unable to generalise the results to target population due to external constraints, this research is still useful in exploring perceptions of the gender wage gap, and conclusions can still be drawn despite being unable to generalise. To improve on this, one could allocate a greater time and monetary investment in accessing a greater pool of respondents, to improve generalisability.
Each respondent in the dataset was given an identical quantitative questionnaire prior to the qualitative interview that was conducted. This questionnaire asked seven demographic questions, five questions about gender role orientations and eleven questions about perceptions of the gender wage gap. From this dataset, the responses were converted into variables and the following were selected for the quantitative and recoded when necessary
Independent Variable
Ethnicity
• This is a categorical variable, with 7 ethnic categories to choose from: White, Mixed, Asian (inc Asian British and Asian Scottish), African, Caribbean or Black, Arab, and an option for ‘other ethnic group’. As my analysis focuses on perceptions of African Black and Caribbean Black respondents, I decided to recode the variable from a categorical to a dichotomous one, to allow for comparison. I put the ‘African’ and ‘Caribbean or Black’ values together and the other values together as the comparative data. This variable consists of 106 values.
Dependent Variables
Perceived Gender Wage Gap
• This variable is a continuous variable recorded the perceptions of the size of the gender wage gap from respondents. Its values ranging from .2 (20p) to 1 (£1). This variable consists of 105 values (2 missing values recoded).
Employer Discrimination
• This variable was recoded as a dichotomous variable (originally continuous values obtained from a likert scale, I thought it would be easier to treat it as a dichotomous variable for analysis). I grouped the three responses associated with employer discrimination being unlikely to explain the gender wage gap (‘Very Unlikely’, ‘Unlikely’ and ‘As likely as not’) and grouped together the responses that imply employer discrimination is a likely cause of the gender wage gap (‘Likely’ and ‘Very likely’).
Control Variables
Age
• This is a categorical variable, which was recoded into three different categories because I wanted to group similar possible experiences and perceptions together to analyse a difference between them. I made variables of age 10-30, 31-50 and 51+. It is important to control for age because different ages represent different life experiences and therefore different ideological perceptions, which can impact one’s perception of the size of the gap.
Gender
• This is a dichotomous variable as there were only two options on the questionnaire (‘Male’ and ‘Female’). As we are measuring the gender wage gap, this variable is key in accounting for differences in perceptions.
Education
• This is a categorical variable, which I recoded and reduced into three categories (Degree Holders, A Level or equivalent and GCSE or equivalent) to examine the differences in perceptions in the regression analysis. I chose to group together the degree holders as having a degree means one will have a likely access to more information about social issues such as the gender wage gap, and in addition to this the degree influences the earnings one will go on to earn, and therefore affects their experiences of the gender wage gap. I chose to group together those who hold A Levels or equivalent values and a separate group for GCSE or equivalent values for the ease of analysis, and because they are different types of education- degrees are not compulsory, whereas education or apprenticeship training is compulsory until the age of 18 in England (GOV.UK, 2017).
To analyse the data, the variables were used to produce a table of descriptive statistics followed by a multivariate regression to examine the relationship between the independent and dependent variables.
Qualitative Methodology
In a team of four, my group and I decided to use semi-structured interviews to obtain a deeper depth and breadth of understanding of the gender wage gap. To conduct the interviews, each of us contacted and arranged to meet with four people each. This was a purposive sample, and convenience sampling was used. When engaging with respondents, we obtained informed consent to participate in the study. We aimed to interview an equal mix of males and females, of Black African and Black Caribbean ethnicities. Much like the quantitative data collected on a larger scale, the drawback to only conducting four interviews (a total of twelve pooled interview resources) is that it renders the results unable to be generalised to the target population. In addition to this, the ability to generalise is reduced further by the presence of sampling bias. It is not possible to generalise the responses of twelve people to an entire ethnic identity; however, the results still reveal interesting perspectives that only strengthen the case for further research. To improve the quality of said future research, using a much larger sample and a different sampling frame to randomise participants and eliminate the potential occurrence of interviewer bias.
We firstly created a topic guide, which identified key areas of focus that we would base the questions on. The first four questions focused on finding out the participant’s awareness of the gender wage gap, concentrating on what they know about it, and if they perceive it to be an issue. The following two questions were intending to find out the respondent’s perception of the gender wage gap from their personal
experiences. It concentrated on finding out whether the respondent had experience
d the wage gap personally, and explored their occupational background and opinions.
The next set of questions focused on the participant’s perceptions on the causes of the gender wage gap, and aimed to identify the biggest perceived factor for driving the gap. The aim of these questions was to address the pre-existing literature (Mandel and Semyonov 2016; Green and Ferber, 2005) and further examine the links between employer discrimination and wage differentials, from the perspective of the ethnicity we chose to focus our research on and in the context of explaining the gendered, not ethnic (Blackaby et al., 2005), wage gap. By including direct questions about what the respondent thinks the drivers of the wage gap are, it is hoped that the qualitative findings will add another level of analysis to the quantitative findings. Throughout the interview, we addressed questions relating to ethnicity and culture, in attempt to understand if their views on ethnicity influence their perceptions of the gender wage gap.
Although the interview had a set of questions to ask the respondent, the semi-structured nature of the interview meant that we could elaborate on responses to gain more depth of understanding and data, whilst following a guide of questions to still obtain relevant responses. After the interviews were conducted, the notes and recordings were summarised and transcribed to allow for analysis and the ease of pooling the data. There was an emergence of key themes in the interviews, which will be addressed alongside the quantitative results.
RESULTS
Quantitative Findings
The following table illustrates the demography of the quantitative sample over the variable of ethnicity. It details the sample in each of the groups, Black African and Caribbean (this group includes 26 respondents accounting for 24.5% of total sample) and ‘Other Ethnicities’ (this group includes the remaining, non-black, ethnic categories from the sample, and includes 80 respondents accounting for 75.5% of total sample).
Table 1. Table of descriptive statistics displaying the number of observations, percentages of chosen dependent and control variables, means and standard deviations over ethnicity.
Black African/ Caribbean
(n=26)
Other Ethnicitya
(n=80)
Total
(n=106)
%
M
SD
%
M
SD
%
M
SD
Dependent Variables
Perceptions of Wage Gap (£)
0.70
0.18
0.79
0.14
0.77
0.15
Employer Discrimination
4.00
1.17
3.86
0.90
3.90
0.73
Control Variables
Gender
Male
32.14
67.86
27.36
Female
22.08
77.92
72.64
Education
Degree Holders
21.21
78.79
63.21
A Level, Equivalent or Similar
20.00
80.00
28.30
GCSE, Equivalent or Other
66.67
33.33
8.49
Age
10-30
32.69
67.31
48.60
31-50
8.57
91.43
33.64
51 and over
31.58
68.42
17.76
Source: LSE SA201 Student Survey November/January, 2016-17
a Other Ethnicity includes White, Mixed, Asian, Asian British and Asian Scottish, Arab and Other Ethnicities.
Participants were asked to give a figure in pence for how much they thought for every £1 a man makes a woman with the same hours and qualifications would make. From observing the descriptive statistics, it can be noted that the mean of the responses for Black African and Caribbean respondents is 0.70, or 70p, whereas for other ethnicities the mean is 0.79, or 79. It is reported that, as a national average, for every £1 a man makes, a woman makes just 81p (Women’s Equality Party, 2017), so these results show that Black African and Caribbean respondents perceive the wage gap to be slightly larger than the national reported average gap. Other ethnicities (67.5% White, 4.5% Mixed, 23% Asian, 5% Other) on average also perceived the size of the gap to be bigger than it is, however only slightly (0.02 or 2p). When looking at the means for the perceptions of employer discrimination being a likely reason for the gender wage gap, we find that the mean for African and Caribbean Black respondents was 4, meaning that respondents in this group think it is ‘Likely’ that employer discrimination is a cause for the gender wage gap. Interestingly, other ethnicities perceive employer discrimination to be a slightly smaller driver of the gender wage gap, with a mean of 3.86. This means that they perceive it to be ‘as likely as not’ that the gender wage gap occurs because of employer discrimination. To further examine the relationship between perception and causality, I will ran a multivariate regression for both dependent variables (perceived gap and employer discrimination) to test whether a relationship between the variables exists to any statistical significance whilst taking the effect of other variables into account.
Table 2: Bivariate and Multivariate regression estimates of the relationship between ethnicity and perceptions of the size of the gender wage gap, and the relationship between ethnicity and perceptions of a causal factor of the wage gap (employer discrimination)
(N=106)
Model 1
Perceived Gap
Model 2
Perceived Gap
Model 3
Employer Discrimination
Model 4
Employer Discrimination
Respondent Ethnicity
(African/Caribbean Black)
0.0859** (0.0339)
0.0843**
(0.0385)
-0.1375
(0.2186)
-0.2959
(0.2386)
Gender
(Female)
-0.0052
(0.0355)
0.4974**
(0.2182)
Age Group
31-50
0.0049
(0.0365)
0.1714
(0.2260)
51 and over
0.0393
(0.0446)
0.2349
(0.2721)
Education
A Level or Equivalent
0.0090
(0.0358)
0.0641
(0.2220)
GCSE or Equivalent
-0.0134
(0.0589)
-0.4080
(0.3655)
Constant
0.7027***
(0.0294)
0.6975***
(0.0419)
4.00***
(0.1899)
3.6801***
(0.2582)
R2
0.0592
0.0669
0.0038
0.0780
Note: Standard errors in parentheses.
***p<0.01; **p<0.05; *p<0.1. Reference group for Age is the 10-30 age group. Reference group for Education is degree holders.
Data Source: LSE SA201 Student Survey 2016-17.
In this analysis, it was investigated whether there is a statistically significant association between ethnicity and the two dependent variables ‘perceived gap’ and ‘employer discrimination’. Model 1 illustrates the results of a bivariate regression of the perceived wage gap variable and ethnicity. The results show that African and Caribbean Black perceptions of the size of the gender wage gap are expected to be 8.6% bigger than the perceptions of the size of the gender wage gap of other ethnicities (t=2.53, p<0.05). We can also interpret the R2 statistic, which shows the percentage of the dependent variable predicted by the variables in the model. Therefore, we see that ethnicity explains around 6% of the variance of perception of the size of the pay gap. Following this, I decided to test whether the somewhat small differences in perceptions across the two ethnic categories would hold when controlling for the effects on perceptions of different age groups, different education levels, and different genders in a multivariate regression. In this model, the R2 value increased to 0.0669, meaning that around 6.7% of the variance of the perception of the size of the wage gap can be explained by the variables included in the model. This regression shows that when controlling for variables, it is expected that African and Caribbean Black perceptions of the size of the wage gap will be 8.4% bigger than perceptions of other ethnicities (t=2.18, p<0.05). However, this model is not significantly associated with perceptions of the size of the gender wage gap because the F-statistic is not significant [F(6, 95) =1.13, p=0.348]. Therefore, our results show that there is not a significant relationship between perceived sizes of the gender wage gap and ethnicity.
Following on from this, Model 3 illustrates the results of a bivariate regression of employer discrimination as a driver of the wage gap. It shows that for the employer discrimination dependent variable, African and Caribbean Black respondents perceive it as a driver of the gender wage gap 13% less than other ethnicities in the sample (t=-0.63, p=0.531). The statistic for R2 is 0.0038, telling us that ethnicity explains 0.38% of the variance in perception of employer discrimination as a driver of the wage gap. I then decided to further test this relationship using multivariate regression whilst controlling for the same variables in the previous regression. In Model 4, the R2 value increased to 0.0780 tells us that around 7.8% of variance, when controlling for other variables, in perceptions of employer discrimination as a driver of the wage gap can be explained by ethnicity. As Model 4 illustrates, African and Caribbean Black respondents perceive employer discrimination as a driver of the wage gap 25.9% less than other ethnicities in the dataset (t=-1.24, p=0.218). As with the previous model of regression, the F-statistic is not significant [F(6, 97)= 1.37, p=0.235] meaning that we cannot claim a significant association between ethnicity and perception of discrimination as a driver of the wage gap.
As we have examined, although regression and descriptive statistics offered interesting points of analysis, due to the lack of statistical significance between the independent (ethnicity) and dependent variables (pay gap and employer discrimination), the findings mean we fail to reject the null hypotheses.
Qualitative Findings
When analysing the data obtained from the interviews, key themes emerged. The qualitative interviews add another level of understanding to the key data noted in the quantitative results. The interviews provide a background, or a voice, to the data points that have been analysed thus far.
Gendered Social Roles
One of the re-occurring themes that emerged from the interviews is the acknowledgement of gendered social roles within both the familial and societal structures. Throughout the interviews the respondents generally seemed to note the differences in traditional gendered employment. The respondents noted the influence of parents and culture in their perceptions of gendered social roles
“Growing up my mum was always the one at home with us, whilst my dad was the one who went out to work; mum always took care of the cleaning and cooking, whilst dad earned the money to pay for it”
(IV3)
When discussing the expectations of gendered social roles, respondents highlighted the occupational segregation that gendered social roles further promote:
“… women would go into low paid jobs just to get their identities back, and it meant that the men were still seen as the breadwinners, as they were still the ones doing the ‘hard’ jobs”
(IV2)
Furthermore, included in the interview questions was one that focused on perceptions on employer discrimination, which allows us to establish more depth in our understanding of the quantitative data findings.
Discrimination
In several responses, interviewees were critical of the element of discrimination in the gender pay gap and in employer discrimination (both ethnic and gender). Respondents were asked whether they thought someone’s ethnicity might make them more or less aware of the gender wage gap in order to try and understand if their views on ethnicity influence their perceptions of the gender wage gap:
“Men are more likely to hire men and give promotions to men, where does that leave women?”
(IV4)
Furthermore, the interviews highlighted the racial discrimination faced by one of the respondents in their employment histories;
“I was treated like I was alien because I worked with a group of foreign ladies. People assumed I was also foreign because I cleaned. The English girls earned £8 per hour, whereas we only earned £7 per hour, they tried to say it was an accounting error”
(IV2)
DISCUSSION
This study aimed at investigating the perceptions of both the size and causes of the gender wage gap, from the perspective of African and Caribbean Black respondents comparatively analysed with the perspectives of other ethnicities in the sample, using a mixed methods approach. We have observed that the data failed to show statistical significance b
etween ethnicity and perceptions despite controlling for gender, age and education, meaning we failed to reject the null hypotheses for both research questions. As the study used mixed metho
ds, despite the lack of significance in quantitative findings, the qualitative analysis allowed for opportunity to gain a better understanding of the reasons and perceptions behind the variables, which larger scale literature is not as able to do. For example, Blackaby et al., (2005) used a large sample to investigate the extent of possible employer discrimination by examining employment decompositions. The data from Blackaby et al is useful because it addresses the size of the inequality, however findings from the qualitative section of my study could possibly be used to attach meaning or reason to statistics. Green and Ferber’s (2005) study suggested that earnings differentials cannot be explained solely by discrimination, and my study draws on themes of gendered social roles which could offer as a focal point for further research.
Strengths and Limitations
A key strength of my study is that it used a comprehensive mixed methods approach to examine the research questions. By obtaining both quantitative and qualitative data, I can use one to support the other, or allow for ease of inference. In addition to this, a strength of this study is that it focused on the perceptions of the pay gap of African and Caribbean Black people specifically, and this is something that doesn’t appear to have been attempted before.
There are multiple limitations in my study which could be remedied or rectified if repeated in the future. The study in its entirety is not generalizable, from the sample of 107 and 12 to the insignificant results. If this study were to be conducted again with a much larger sample size (both for quantitative data and qualitative data- more interviews) if we wish to be able to generalize the findings to the wider population. Although the aim was to focus on the perceptions of African/Caribbean people, this study could have made comparisons across multiple ethnic groups rather than grouping them together. Another issue with this study is that mixed method designs can be extremely time consuming and expensive.
Conclusions and Policy Implications
To conclude, although these results didn’t arrive at a significant relationship between ethnicity and the perceptions of the size of the gender wage gap and its causes, they are not entirely redundant. The qualitative interviews provided us with real voices and opinions, which added another level to the numerical data already obtained, meaning that they can be utilized in the policy decision making process. Jeremy Corbyn of The Labour Party announced in their 2017 that they will, if elected, eradicate the gender pay gap. Based on the respondent’s answers to the qualitative questions, it could be suggested that to take steps to reducing or eradicating the gender pay gap, means that steps should be taken to ensure a growing trend of transparency and awareness about pay.