Professional Documents
Culture Documents
(2013-14)
Contents
Fo rewo rd . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
M e t h o d o l o g y. . . . . . . . . . . . . . . . . . . . . . 2
Data Collection and Survey Design.....2
Data Cleaning................................... 3
D e s c r i p t i v e S t a t i s t i c s . . . . . . . . 4
Model............................... 6
Rationale..........................................6
Regressions...................................... 8
Results............................................. 9
Conclusion.......................13
Limitations......................14
S L F S Te a m . . . . . . . . . . . . . . . . . . . . . . . . 1 5
Foreword
Dear Reader,
The research division of the LSE SU Economics Society is pleased to present you the
first report on Student Labour Force Survey (SLFS). This report is part of a research project
that studies the part time work and internships carried out by the students of LSE and marks
a shift from the Student Consumer Price Index (SCPI), another research undertaken by the
division previously.
The idea of this project came into being in a discussion between Dr. Judith Shapiro,
the undergraduate tutor in the department of Economics and William Matcham, an LSE postgraduate who then was in his second year of undergraduate studies. The project was initiated in 2013 and was divided into four key stages, namely Data Collection, Data Management,
Data Analysis and Reporting. Data Collection was done through an electronic survey that
asked questions such as the field(s) students are interested in working in, their department,
their positions in societies, their full and part time work experience etc. We received a total
of 615 responses which seems to be a good sample size and represents the LSE population
well. The incentive given was a lucky draw with a variety of monetary and non-monetary
prizes, including an iPad.
The econometric analysis of the collected data was done through running three multivariate OLS regressions of telephonic interviews, assessment centres and job oers on a
number of independent variables such as career events, position in societies, gender, work
experience etc. Our study shows that attending career events, networking with professionals and having work experience are significant factors in improving a candidates employability and prospects of getting the desired job. Another interesting and surprising result
is that gender is significant for the second year students, which indicates that female students may have an edge over male students in getting a call for the telephonic interview and
reaching the assessment centre.
The successful completion of each stage of this project was a result of the combined
effort by the SLFS team and I thank them all for their constant support and enthusiasm. The
project was backed and generously funded by the LSE SU Economics Society. I sincerely
thank Mr. Steve Bond of the Centre for Learning and Technology (CLT) of LSE for advertising
the survey on the LSEs Moodle page. I also offer my sincere thanks to Ms. Jenny Blakesley of
the LSEs Career Hub for publishing the survey on their webpage. I also take this opportunity
to express my profound gratitude to Dr. Judith Shaprio for her continuous assistance and
guidance throughout the course of this project. I hope that you will have as much pleasure
reading this report as we had writing it.
Warm regards,
Amit Singhi
(SLFS Team Leader)
Methodology
By Anshuman Bhargava
Methodology
by Arnaud Dyvre
Data Cleaning
The sample consisted of 615 respondents from whom answers were collected either
through a face-to-face administration of survey or in an electronic format. The survey was
open for four weeks from mid-January to mid-February.
The data can be accessed here:
https://www.dropbox.com/sh/3zqfxlqglpys013/AABUlWSI9BywkxyYlTuLcEIDa?dl=0
The data collected needed to be cleaned for 3 main reasons. Firstly, many responses
were missing for a number of categories. Although the students had taken the survey from
beginning to end, many skipped some questions or left some incomplete. Secondly, open
ended questions needed to be harmonised into a unied format. And finally, string values
had to be coded into numerical variables. The important task then was a translation from
the kind of data Survey Monkey provides to a dataset that can easily be analysed in statistical packages. Regarding the incomplete responses received, many values were missing
for variables like Examination Results (including year first, second and third) and hours
worked if engaged in Part Time Employment. It was also observed that many students
either did not have their scores or were reluctant to communicate them. Moreover, fewer respondents had part time work-experience. Had we thought of using only complete
responses in regression analysis the sample size would have reduced considerably, thus,
resulting in poor analysis. Therefore, due to limited availability of data, these variables were
not included in our models. Non-numerical variables like a respondents country of origin, language, gender, year of education, interests, societies, department etc. were transformed into categorical variables. An additional variable was created which is the ratio of
hours dedicated to study divided by the total time spent in extra-curricular activities. Since
these two variables are negatively correlated, including both of them simultaneously
would have given rise to multicollinearity issues.
D e s c r i p t i v e S tat i s t i c s
by Krzysztof Zaremba
Year of study
Count
Percentage
First year
114
19%
Second year
140
22%
Third year
103
14%
Master
243
17%
PhD
12
2%
Total
612
100%
Table 2: Sample composition by year of study
D e s c r i p t i v e S tat i s t i c s
It was also observed in our study that students studying Economics attend on average
1.9 events. Whereas, this number is 2.35 for students of Law, 1.65 for students of International Relations and 2.85 for students of Management.
Program Count Percentage
Accounting 18
3.6%
Anthropology 4
0.8%
Economic History 20
4.01%
Economics 84
16.8%
European Institute 3
0.6%
Finance 12
2.4%
Gender Institute 3
0.6%
Geography & Environment
28
5.6%
Government 49
9.8%
International Development 11
2.2%
International History 14
2.8%
International Relations 27
5.4%
Law 20
4.0%
Management 46
9.2%
Mathematics 25
5%
Media & Communications
14
2.8%
Methodology 1
0.2%
Philosophy, Logic and Scientific Method
17
3.4%
Public Affairs (IPA)
7
1.4%
Social Policy 34
6.8%
Social Psychology 4
0.8%
Sociology 16
3.2%
Statistics 32
6.4%
Accounting 6
1.2%
Managerial Economics and Strategy Group 1
0.2%
Philosophy 3
0.6%
Total 499
100%
Table 3: Sample composition by department
Model
by Xiaotong Hong
Rationale
This project is an attempt to understand the internship and part time work carried out
by the students of LSE to study the dynamics of LSEs labour force. Keeping in mind our aim,
the team contemplated on finding the significance of factors crucial in successfully passing
each stage of the recruitment process. A model was thus prepared to investigate how work
related experience, gap year, time spent in studying and socializing in a week, number of
career events attended, part-time jobs etc. would affect the employability of students.
We ran three baseline OLS regressions to explore and examine how specific explanatory variables are related with dependent variables. The three dependent variables
in these regressions were phone interview, assessment center and job offer. We understand that students in different year of studies have different interests and the opportunities available to them vary, too. Consider an example that career events regarding internships and job opportunities are usually recommended to students in year two or three
of their undergraduate studies. Whereas, students in year one usually attend events for
short spring internships. Therefore, our econometric analysis takes into account these differences and is done using observations in the same year group. This also helps in comparing students to their peers with similar amount of work experience and qualifications.
The explanatory (independent) variables used are as follows:
1. Gap year
This refers to the year taken out prior to commencement of university study. Students
who take a gap in starting their education in college are assumed to have an edge in job applications and in general the recruitment process. This is because they generally spend the year
doing some part time work or an internship or any research which adds to their work experience.
Recruiters and employers usually expect students to have a strong academic background
with high examination scores such as a first class or an upper second class. Hence, there was a
need to explore the effect of weekly study time on dependent variables. On the other hand, it is
generally believed that students who socialize more have higher chances of successfully passing
a telephonic interview and the assessment center thereafter. We have taken a ratio of the two
variables because both when used simultaneously would give rise to multicollinearity issues.
Model
3. Career Events
We were interested to check whether attendance in career events helps students in
recruitment process. Such events also provide an opportunity to build important networks
that can aid in the hiring process. Together with this, students get meaningful insight and
important information about any company they are considering applying from these events.
4. Gender
To assess any relative advantage of one gender group over another, it was decided to
examine the effect of gender on the dependent variables. Gender is a dummy variable and
we have represented it in binary form with number 1 being allotted to female students and
0 to male students.
5. Position
Having a formal position in a society is supposed to improve a students employability
by improving management, communication and team-work skills. Such skills are very valuable and may help students in demonstrating their abilities in an interview at assessment
center. We have taken Position as a dummy variable. Number 1 indicates that a student
has a position whereas 0 points to not holding any position.
Doing a part time job requires an ability to balance work and studies. Students with
a part time work may signal their time management and skills along with the willingness to
take challenges and therefore we are interested in finding whether having a part time job
help students to differentiate themselves from their competitors in the hiring process. This
again is a dummy variable with numbers 1 and 0 representing those with and without a
part time job, respectively.
Students with some experience in their field(s) of interest are said to have developed
skills that may signal employers that they are better prepared to take up the job and will
have less difficulty in understanding their role. It has also been seen that employers put more
value on work experience of an applicant as it helps them in selecting the right candidate for
the role.
Model
Regressions
1) The first stage of any formal application process involves the submission of an
application electronically along with a covering letter and a rsum.Thereafter, the shortlisted
applicants are invited for a telephonic interview. Here we are interested to find the significance
of factors that help students get through the initial screening and we do so by running the first
regression that includes the independent variables that best reflect the qualifications required
by employers to get a call for telephonic interview. Hence, all the independent variables are
used, namely,gap year,work experience, study to social time ratio, gender, position and parttime job. The hypothesis for this regression is that all the variables save gender are significant.
2) Subsequent to successfully passing the telephonic interview comes the stage of visiting the assessment centre where students are closely assessed by employers (interviewers) for the job they are applying. Independent variables are carefully selected for the second regression and our hypothesis is that having a position in a society or having work
experience will increase students chances of getting an invitation to assessment centre.
3) Following assessment centres, some companies will invite candidates to partner interviews as the last stage of the application process. In the final regression, we examine the
factors that help students get the job offer and to do this we have included all the explanatory variables along with a set of additional networking indicators. Networking is believed to have a huge role in helping an applicant skip the intermediary stages to reach
the final stage of the hiring process and eventually get the job offer. Our hypothesis is
that networking is crucial in getting the offer (or the networking variable is significant).
Model
Results
First regression: The number of telephone interviews
Model: telephone interviews First year Second year
Third year Full sample
Variables
Gap 0.856 -1.521 -0.719 -0.102
(1.163) (-1.78) (-1.03) (-0.34)
(4.28)***
(2.03)** (1.59) (2.12)**
Observations 44 81 63 340
Adjusted R2 0.38 0.14 0.20 0.02
t-statistics between parentheses. ***, ** and * significant at 1, 5 and 10 percent respectively
Interpretations:
The results indicate that our model explains 38% of the variation in the number of telephone interviews from its average value for the first year students. The adjusted R-squared
is, however, lower for other regressions such as 14% and 20% for the second and third year
students, respectively.
A careful look at the results table above also indicates that career events is a positive
and significant variable for almost all year groups, excluding year three which implies that
attending career events does help in getting through the first stage of the hiring process
successfully. Another surprising but interesting result is that the variable gender is significant
Model
for the first and second year students (with opposite signs) but insignificant for the complete
sample.
The absolute value of t-statistic presented inside the brackets is less than 1.96 for all
year groups for variables such as gap year, s-s ratio and part time job. This implies that all
these variables do not play an important role in passing the first stage of the recruitment
process.
The variable position, however, is significant for only third year students. Similarly,
getting some work experience can improve the chances of a successful telephonic interview
for second year students.
(0.74)
(-1.08 )
(-0.74)
(1.25)
(0.33)
(1.79)
(0.95)
(-2.94)**
(1.04)
(-0.81)
(-0.11)
(0.11)
(2.13)**
(1.51)
(-1.68)
(-0.50)
-0.654
Work experience
0.148
(3.17)**
-0.225
-0.150
-0.106
0.088
-0.042
-0.023
(0.84)
(-0.25)
(-0.33)
(-0.43)
(-0.43)
Observations 42 77 55 284
Adjusted R2 0.33 0.05 0.09 0.03
t-statistics between parentheses. ***, ** and * significant at 1, 5 and 10 percent respectively
Interpretations:
The independent variables jointly explain 33% of variation in the dependent variable
from its mean for the 1st year students. The explanation percentage, which can be interpreted from the Adjusted R-Squared values, decreases further down the table to 3%.
For the first year students, the coefficient for study to social time ratio is positive and
significant. This implies that there is a positive correlation of the s-s ratio with the chance of
receiving a call for visiting the assessment centre. Our study also indicates that work experience is highly significant as indicated by the high value of the t-statistic for this variable.
10
Model
For the second year students, gender has a negative coefficient but is significant at 5%
level. Whereas, other variables such as career events, work-experience, part time job and gap
year do not seem to be highly correlated with the dependent variable.
It is interesting to note that for the third year students, the variable position is positive
and significant which means that having one or more formal positions in societies can be
crucial in this stage of hiring process. On the other hand, the analysis does not highlight
significant variables for masters students. This suggest that there would be other factors
important for getting an invite to the assessment centre such as attending academic and
non-academic competitions, volunteering, teaching or taking up a research job.
Note: in the Network we ask about five different kinds of networks. (1): network + interview, (2) network
+ interview + formal selection and application, (3) personal network, (4) personal network + interview
11
Model
Interpretations:
The results from this regression suggest that attending career events is important in
getting the job offer by successfully passing the assessment centre. This can be inferred from
the positive coefficient of this variable in the Full sample row whose t-value is greater than
1.96. The s-s ratio, however, is significant for only third year students, which this time has
a negative coefficient. This means that the lower the weekly study to social time ratio, the
higher are the chances of getting a job offer.
A new feature of this regression is the inclusion of network indicators to examine how
different ways of networking (with and without formal application) help in getting a job offer.
The explained variation for the third year group is 27% which gradually decreases to 3% for
the complete sample. It is interesting to note that having a personal network or a connection
in a chosen field is influential in getting a job as shown in the last row of the results table for
complete sample.
12
Conclusion
by Arnaud Dyvre
Ultimately, this first edition of the Student Labour Force Survey contains some useful
advice for LSE students. There is strong evidence that the services provided by the LSE Career
Hub help students in catching the attention of recruiters and further, in obtaining jobs and
internships. We believe the Career services are efficient in 3 ways:
They connect students to employers. By doing so, students are able to present their skills
and experiences in person.
They help students in building serious CVs and cover letters. This may be especially true
for first year students whose experience in applying for jobs is limited.
They help students in identifying the sectors in which they are most interested. The
Career services at LSE are doing a great job of presentation of the various industries in which
LSE graduates may be interested to work. By tailoring and focusing their applications to specific sectors, students who attend more Career events are then more successful in landing a
job.
In addition to this, we found that holding a formal position in a student society has a
positive impact on job market outcomes, especially for undergraduates in year 3 and postgraduates. This turns out to be a valuable lesson for students with little or no work experience as taking up responsibilities in a society can, in many cases, be a substitute for work
experience.
Our final finding is that putting more weight on studying rather than engaging in social activities or striking a balance between studying time and social time seems to matter
very little. Although, large prestigious organisations and firms usually require excellent academic credentials for taking up jobs and internships, however, we did not make a distinction
in the quality of jobs in our study. All jobs were treated in a similar way without consideration
of prestige, thus diminishing the relevance of the ratio of weekly study time to weekly social
time, an independent variable used in the study. This is a possible avenue for future research.
13
L i m i tAt i o n s
Challenges
Although this study sheds light on some interesting facts, it does not inform us about
causal mechanisms. It would have been interesting to collect data at the beginning and at
the end of the academic year, or even better, over several years. Getting a hold on such
interesting data would go beyond the scope of this project. Given our time and resource
constraints, our results should be taken as suggestive rather than explicative.
A serious limitation of our results is due to problems of sample size. The explanatory
power of the model, as given by the adjusted R-squared, was high for the first and second
year students but gradually decreased for the third year and master students, eventually
reaching below 4 percent for the complete sample in each case. The sample size for a single
year was small (below 100) in a lot of regressions. This should be corrected in two ways for
subsequent editions of the SLFS: firstly by interviewing more students and secondly by making sure that all respondents indicate their year of study.
Finally, we believe that some limitations might have risen due to the timing of the survey. This survey was taken during the months of January and February when many students
still were in the process of either getting a call for the assessment centre or receiving a job
offer. Thus, we believe that the numbers in the fields of successful assessment centre(s) and
job offers might have been underreported leading to variations in the analysis of results.
Similarly, many students would not have engaged in serious studies during the time this
survey was taken as examination period was still a few months away. Therefore, the data collected from the field average weekly study time would have been underestimated and the
field average weekly socializing time would have been overestimated as this excluded the
time students devoted on average to studies and social activities during examination period.
Going further
Acknowledging these limitations, the Research division has taken steps to improve the
quality of its research in the forthcoming editions of the SLFS. We got in touch with other
student-led Economics Societies in renowned UK universities so as to conduct this research
with them in the future. Economics societies in Bath University, Kings College, the University
of Manchester and the University of Oxford have have accepted to run this project jointly
with the LSE SU Economics Society. Their involvement would enable us to have a larger and
more diverse sample of respondents.
Finally, conducting this project year after year would certainly unveil interesting patterns in job market outcomes for LSE graduates over time. Our study is more a snapshot.
14
SLFS Team
Acknowledgments
SLFS Project Head: Amit Singhi
BSc. Economics
MSc. Economics
Econometrics Team:
Xiaotong Hong (Head)
Han Loong Ng
Rumi Jahani
Reporting Team:
Krzysztof Zaremba (Head)
General Course
Anshuman Bhargava(Co-Head) BSc. Economics
Tina Uniyal
BSc. Economics
15
SLFS Team
Project Members
Back (From Left to Right): Nadim Karim Choudhury, Krzysztof Zaremba, Arnaud Dyvre, David Yong
Front: Han Loon Ng, Anshuman Bhargava, Amit Singhi, Ateeq Hazari, Xiaotong Hong, Sakshi Tomar
16
17