Word count: 3133

model <- lm(persinc2 ~ rsex + religcat + orient + uninatid + tunionsa + rsuper + rage, data = nilt)

Introduction

In determining my research question and focus for this interpratative quantitative findings report, I chose to narrow down to the aspect of trade union membership and it’s relationship with annual income(GDP).

Subsequently my research question is as follows; What are the effects of membership of trade union’s on annual income?

This focus of research will allow me to delve deeper into the pay divide and the impact of trade union’s in specifically Northern Ireland and help to determine a wider trend for the UK. Trade unionism has been long synonymous with the working class core in the UK emerging from the period of industrialization in Europe, it required a form of regulation and representation by providing a channel through which one may formally reach out to the employer (Lovell p74 1973). However, whilst trade union’s have been traditionally strong, this focus of research will shed light on their impact on annual income in the data analysis.

Following on from this observation, when formulating my hypothesis to this research report, I came to the conclusion that ultimately membership of a trade union will have a positive effect on annual income compared to non-membership. Thus, individuals who exist within a kind of trade union will earn more on an annual basis.

This hypothesis is drawn from existing literature that purports the importance of trade unions in both the public and private sectors. Central to their influence is the idea that they act as political and economic actors, in the aspect of interest groups as well as labor market monopolists. Collective bargaining with employers enhances workers’ terms and conditions. A critical component of bargaining strength revolves around the percentage of all the workers they represent, which leads to a higher union wage premium. Direct effects on wages stems from several aspects, including the ‘threat effect’ whereby non union members raise their wages to avoid the threat of unionization. To extend this further, unions play an indirect role, their voice face assists in lengthening job tenure, correlating to higher wages. This also alters the incentives employers and workers face when investing in their human capital (Blancflower and Bryson, p100 2009). In diverging from this aspect, there has also been a noticeable shift in trade union activity in the twenty-first century, despite occurring more recently in the past few months strike action from school teachers to train staff demonstrate the collective power of unions to shift the wage bracket. Therefore, one should expect a positive correlation in the focus of this report.

In reaching this hypothesis from literature and data extrapolation, this research report will seek to evaluate the NILT data set against the model sample, drawing on it’s multidimensional characteristics, whilst providing analysis on data visualizations to underpin the report. All of which will contribute in testing this hypothesis.

Data and method

In tackling the data set, it is important to begin by understanding the nature and sources of the results. In the process of research design one must consider the background of a data set in order to fully comprehend its credibility and meaning. The data set provided below originates from the Access Research Knowledge Hub (ARK), dating back to 1998 they have conducted the Northern Ireland Life and Times Survey. This survey has been responsible for recording the attitudes, beliefs and values of local adults to a wide range of social issues(NILT, 2012).It is of important note to recognise this as primary data research, in order to acknowledge its full depth and reliability as the data is derived from first hand data sets. Although, the primary data collection has evolved from the era of conducting interviews and focus groups, because of the vast network of data readily available social scientists can produce forms more efficiently. This particular form of research report is constructed on a secondary basis from a pre defined dataset, manipulated for a set purpose (Fogarty ch1, 2019). In reviewing the NILT data set it is important to note that is carried out on an annual basis, the survey therefore differs from similar processes such as a census, this enables one to gain a sense of the change in attitudes and behaviors over a prolonged period of time.

When viewing the data set below it is clear that are ranging variables from sexual orientation to religious beliefs, these are measured against Annual Personal Income(GBP). Subsequently for this research report one is measuring Annual Personal Income(GBP) against the membership of trade unions. In referring back to the results of the data, there is a necessary contention to make behind the collection process. The sample size for the 2012 survey included the drawing up of 2350 total addresses using a Postcode Address File, with the total respondents equaling 1204 face to face interviews with eligible candidates (NILT). Subsequently, a randomized system of sampling is on display here, such a system means the research design was inclusive of the whole population size, as the whole had a equal chance at being picked. Moreover, it is free from bias and allows for greater statistical inference in this case against the wider population (Taherdoost, p19 2016). Whilst on a general level the sampling process and overall data set account for the percentage of trade union members against their annual income, on a more focused framework it fails to account for additional factors such as general opinions towards organizations such as trade unions, limiting its scope. ### Categorical Analysis

figure 1

ggplot(model, aes(x= tunionsa, y=persinc2)) +
  geom_boxplot() +
  labs(title= 'Annual Income by Trade Union Membership', 
       x='tunionsa', y='Annual Income (GDP)')

From the plot above which is laid out as a box plot, there is a general overview of the relationship between annual income and trade union membership. Before this is broken down into smaller components, with additional variables considered it is clear that the population whom exist in a trade union earn more on average. When examining the plot one can see that the distribution for membership of trade union covers a wider salary scale starting at a higher base salary compared to non members who start on a lower level. The upper quartile range on the graph further demonstrates that the top earners exist within a union reaching up to £50,000. Furthermore, the median is represented by the middle point within the box plot, with trade union members equalling to 15,000, as a result 50% of respondents are on a higher/lower income, whereas the median amount of non union members equates to roughly 10,000. Though useful, the results are not conclusive without consideration of the internal makeup of variables, yet it

install.packages('ggplot')
library(ggplot2)

figure 1.1

ggplot(model, aes(x = rage, fill = tunionsa)) +
  geom_histogram() +
  labs(title = 'Trade Union Membership by Age', x = 'Age')

When breaking down the variables that are included in the dataset, it is of particular importance to view the distribution of age in order to establish a relationship with the research question. The X axis (horizontal) on the histogram represents the age of respondents, whilst the Y frequency. The lowest age seems to be around 20 years old, with upwards of 15 respondents and the oldest going up to 100. The most frequent age consists of around 40-50 years old, with upwards of 50 respondents being 40 years old, this is exemplified through the median which lies at 48. Such a figure highlights that that means that 50% (or half) of the respondents are equal or younger than this, and the other 50% is equal or older. As will be discussed, there seems to be a stronger relationship with the older workforce and their commitment to trade unions.

install.packages('vtable') 
library(vtable)

figure 1.2

sumtable(model[["model"]], vars  =  c('rsex', 'persinc2', 'rage', 'rsuper', 'religcat'), group = 'tunionsa')
Summary Statistics
tunionsa
Yes
No
Variable N Mean SD N Mean SD
rsex 301 374
… Male 135 44.9% 149 39.8%
… Female 166 55.1% 225 60.2%
persinc2 301 20159.668 13833.279 374 14262.299 12542.526
rage 301 49.953 15.587 374 44.195 17.866
rsuper 301 374
… Yes 113 37.5% 98 26.2%
… No 188 62.5% 276 73.8%
religcat 301 374
… Catholic 112 37.2% 165 44.1%
… Protestant 136 45.2% 147 39.3%
… No religion 53 17.6% 62 16.6%

figure 1.3

sumtable(nilt, vars  =  c('rsex', 'persinc2', 'rage', 'rsuper', 'religcat'), group = 'tunionsa')
Summary Statistics
tunionsa
Yes
No
Variable N Mean SD N Mean SD
rsex 440 739
… Male 209 47.5% 318 43%
… Female 231 52.5% 421 57%
persinc2 348 20013.391 13884.507 543 14142.541 12723.902
rage 438 52.678 15.938 738 47.977 19.714
rsuper 383 486
… Yes 139 36.3% 125 25.7%
… No 244 63.7% 361 74.3%
religcat 431 717
… Catholic 160 37.1% 323 45%
… Protestant 201 46.6% 290 40.4%
… No religion 70 16.2% 104 14.5%

Above is a contingency table visualizing the statistics within the model compared to the full NILT dataset below. There are 675 respondents in this sample size which is a significantly smaller set in comparison to the overall dataset 1204. The top column represents trade union membership whilst the left hand column illustrates the variables; sex, type of role and age. Initially when examining the demographics contained in the table, there is an even number of male and female respondents with slightly more females in trade unions at 55.1%. Although these variables are significant they do not offer any great deal of subjectivity in the line of research. Instead there should be a distinct focus on the variables of the type of role in employment. Noticeably, there is a large differential in the type of role, with only 37.5% of supervisors existing within a trade union. This compares drastically with 62.5% of regular employees, such a large gradient suggests there is an uneven concentration in the workplace whom are members of unions. Furthermore, when examining the statistics surrounding religious orientation, there is an even distribution between Catholics and Protestants in relation to union membership suggesting religion is not a key determinant in correlation with annual income.

Results and discussion

Before undertaking analysis on the data model and it’s relationship with different variables, there must be a focus on the overall type and this significance. Evidently, the model is a measure of annual income(GBP) against set variables. For the purpose of this model the dependent variable, is related to the the factor (annual income) that varies when there is a change in the other independent variable (Trade Union Membership). The dependent variable is thus quantitative in this instance as it can be applied with a numerical value, it is ‘unbounded’ meaning there are no constraints on the variability of the outcome. This stands in stark opposition to a dichotomous variable which is categorical with only two set outcomes (Field p229, 2012). Subsequently, in explaining the model one can see that there exists some negative and positive relationships between the dependent and independent variable. For example, the model explicitly states in terms of statistical figures that trade union members earn more, non-members typically earn £5,277 less. Such a representation means the null hypothesis is rejected; there is no relationship between annual income(GDP) and trade union membership. Closely linked to the results produced by trade union membership was the age of respondents, gender and more importantly the type of role in employment, as regular employees earn a considerable lower amount at £8,648 less. However, a discrepancy appears, as in the table provided above only 37.5% reside in trade union, from this one can assume that within the dataset respondents of employee level dominate membership. Therefore, these two variables ultimately have a positive relationship on annual income. In comparison, the analysis of the other variables in the model,religious orientation, sexual orientation and constitutional view provide little significance in contributing to the relationship on annual income, as a result they will be excluded from further analysis.

figure 1.4

ggplot(model, aes(x=rage, persinc2, color = tunionsa)) +
  geom_point(position = 'jitter') +
   moderndive::geom_parallel_slopes(se = FALSE) +
  labs(title = 'Annual income vs Trade Union Membership by Age',
       x= 'Age', y= 'Personal Income(£ a year)',
       colour= 'Union Membership')

m4 <- lm(persinc2 ~ rage + tunionsa, model[["model"]])
summary(m4)
## 
## Call:
## lm(formula = persinc2 ~ rage + tunionsa, data = model[["model"]])
## 
## Residuals:
##    Min     1Q Median     3Q    Max 
## -19521  -8369  -3667   5216  62039 
## 
## Coefficients:
##             Estimate Std. Error t value Pr(>|t|)    
## (Intercept) 23284.81    1673.72  13.912   <2e-16 ***
## rage          -62.56      29.90  -2.092   0.0368 *  
## tunionsaNo  -6257.61    1028.95  -6.082    2e-09 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 13100 on 672 degrees of freedom
## Multiple R-squared:  0.05375,    Adjusted R-squared:  0.05094 
## F-statistic: 19.09 on 2 and 672 DF,  p-value: 8.665e-09

As demonstrated in the graph above there is clear evidence that age and trade union membership are core factors in affecting respondents’ annual income. The graph is separated in colour by trade union membership, thus it is evident that the line of best fit which indicates the averages of the model is significantly higher in the case of trade union members. The orange line as represented by trade union members, intercepts at the figure of 23000 dropping to 19000. Despite this reinforcing my original hypothesis, age is an important determinant within trade union membership. The scatter plot highlights that between the ages of 20-40, respondents earn a higher amount. In reviewing the literature outside of the NILT dataset there is recurrent theme that appears consisting of the contention that the income of young adults tends to fall behind the older population in the case of Europe. Between 2008-2017 a research suggested that equivalised household income of young adults with young children stagnated or decreased with the financial crisis. Bernhard Hammer places this down to employment trends, contending that the disparities of the labour market produce uneven results. The older generation find themselves as permanent employees with greater protection and stability, whilst the younger generation find themselves increasingly entrapped in insecure and temporary work (Hammer, et al.526, 2022). The mean age in the sample is 49, with the core spread of respondents spanning the 30-50 years old bracket, with the highest earners also existing in this bracket. This to an extent undermines the hypothesis that membership of a trade union leads to greater annual income, as age is an additional sub component, with this sample exhibiting that younger respondents earn a greater amount. However this may be due to the fact that there is a higher concentration of them in trade unions.

In continuing to focus on affecting variables, there must be a focus on the type of role(rsuper) as it reveals that trade unions are dominated by lower level employees. This in turn is has a positive relationship with annual income with supervisors earning more. Whilst this does not specifically coincide with the stated hypothesis, it reveals certain inconsistencies. In explaining this disparity, one must look to factors that influence member participation in trade unions. Class consciousness is a prevalent theme in this context due to its intrinsic link with employee level. This refers to Marxist theory, comprising of the idea that one is aware of their social and economic class relative to others, as well as understanding of economic rank in the context of the larger society. Additionally, Marx saw trade unions as not only a method of redressing workers’ grievances, but also a setting for a more general process of socialisation in the class-consciousness (Patel 2368, 1994). In a contemporary setting, this translates into the work place as a lower level employee is in a position of increased likelihood of joining a union. Specific mental attitudes means the individual identifies with the idealistic union norms, further relating to the idea that they see them self as a worker tending to perceive industrial disputes from the vantage point of workers rather than the seniority of management (Perline and Lorenz, 426, 1970). Thus this is critical for explaining the large differential in union membership in relation to type of work role, whereby a divide occurs in the hierarchies. One can attribute this to the collective sense of false consciousness within the management, as they are more individualistic in their outlook producing an overtly competitive view with others of one’s social and economic standing, instead of a unified outlook with particular groups to further interests.

Together these approaches shed light on the internal makeup of trade unions and their demographic, whilst ultimately revealing that members of trade unions typically earn more on average. In directing focus to the reason as to why members of trade unions earn more is owing to several aspects including the point that they alter the relationship between the employee and employer. As individuals are often part of a collective union, a process of collective bargaining is at the forefront of the labor market. Individuals are better protected, as they are represented in meetings with management over a range of issues revolving around conditions of employment. Unions thus help in reducing the labor turnover rate which helps both the employee and employer, as stability in the role is enhanced whilst the high cost of labor turnover is reduced (Hassel and Addison, 6 2003).

Regression results
Annual Personal Income (GBP)
Sex: Female (ref.: Male) -5,068.737***
(994.748)
Religion: Protestant (ref.: Catholic) 465.188
(1,458.367)
Religion: No religion 895.169
(1,533.323)
Sexual Orientation: Homosexual (ref.: Heterosexual) -6,247.777
(3,437.048)
Sexual Orientation: bi-sexual -2,826.980
(8,698.806)
Sexual Orientation: Other 1,323.336
(8,737.282)
Constitutional View: Nationalist (ref.: Unionist) 1,788.873
(1,898.294)
Constitutional view: Neither 1,438.036
(1,350.423)
Trade union membership: No (ref.: Yes) -5,277.978***
(977.008)
Supervisor: No (ref.: Yes) -8,648.320***
(1,037.559)
Age -84.369**
(29.430)
Constant 31,343.540***
(2,488.067)
Observations 675
R2 .183
Adjusted R2 .170
Residual Std. Error 12,252.840 (df = 663)
F Statistic 13.533*** (df = 11; 663)
Notes: *P < .05
**P < .01
***P < .001

Conclusion

In conclusion, this research report was guided by the relationship with the effect that trade union membership had on annual income. Before analysing the data at hand, there was a preconceived idea that trade unions would have a positive effect on personal income due to their nature and prominence in society today as a result, my hypothesis reinforced this. By focusing on quite a prominent independent variable I was able to examine various aspects in connection to the makeup and demographics within unions, but also reviewing their relationship with annual income. In first drawing the necessary statistics for my research, I had to separate the new sample from the original dataset ensuring ‘model’ was created as an object for further insight. Consequently, this reduced both the respondent size from 1204 down to 675 as well as the variable set to a list of 14. In performing this, I was able to produce visualisations of the data set assisting in focusing on the critical statistics for wider comparison between variables.

A particular vital component on the research design which enabled me to draw overall conclusions, was figure 1 and figure 1.2 as I was able to determine the differences in personal income. Overall, in the sample the difference in personal income over the membership of trade union was £5,897. This subsequently reinforced my hypothesis and was enhanced further by the existing literature which purported similar views. A widespread view emerged that trade unions possess considerable leverage and power over the labour market, particularly in shaping firm directives. As the collective bargaining power of a firm’s employees force companies to negotiate with unions into altering terms and conditions of employment, along with other several indirect effects. In reaching this conclusion, a limited number of variables transpired which contradicted popular theories concerning income and union membership. The results produced by the dataset challenged the notion that the older population typically earn more, as represented by figure 1.4, the lowest earners in trade unions were scattered between the ages of 50-75, ultimately opposing the existing literature in the belief that the older population earned more on average. Despite this, additional variables seemed to display discrepancies in the demographic of trade union membership, with supervisory type roles in employment having a sparse existence in unions.

Overall in spite of facilitating the requirements for an extensive research report, the dataset narrowed much of the research to a salary based approach, with a wider scope of data I would have explored differing approaches to conceptions of attitudes towards trade unions among the population and their significance.

REFERENCE

Lovell, J. (1977). British Trade Unions 1875–1933. In: Clarkson, L.A. (eds) British Trade Union and Labour History A Compendium. Studies in Economic and Social History. Palgrave, London.

Blanchflower, D.G. and Bryson, A. (2010), “The Wage Impact of Trade Unions in the UK Public and Private Sectors. Economica”, 77: 92-109.

Queens University Belfast, https://www.ark.ac.uk/teaching/NILT2012TeachingResources.pdf last accessed: 09/01/23

Fogarty, G.B (2019) Quantitative social science data with R: An Introduction. London

Taherdoost, H. (2016). “Sampling Methods in Research Methodology; How to Choose a Sampling Technique for Research”. International Journal of Academic Research in Management. 5. 18-27.

Field, A.P (2012) Discovering Statistics Using R. London

Hammer, B., Spitzer, S. & Prskawetz, A. “Age-Specific Income Trends in Europe: The Role of Employment, Wages, and Social Transfers”. Soc Indic Res 162, 525–547 (2022).

Patel, P.J. “Trade Union Participation and Development of Class-Consciousness.” Economic and Political Weekly, vol. 29, no. 36, 1994, pp. 2368–77.

Perline, Martin M., and V. R. Lorenz. “Factors Influencing Member Participation in Trade Union Activities.” The American Journal of Economics and Sociology, vol. 29, no. 4, 1970, pp. 425–38.