Thesis Student Skills Sessions - Postgraduate Studies - University of Canterbury

Thesis Student Skills Sessions

Semester Two 2015

A number of sessions are available for thesis students as detailed below. You are encouraged to discuss these sessions with your supervisors and to attend sessions of interest to you. As well as learning valuable skills these sessions also provide you with an opportunity to meet other thesis students and staff from across the university (and in some cases you get free lunch!).

Surviving your Thesis Sessions

Three day-long sessions are planned for August/September 2015. On each day we have grouped together sessions that are common in theme and/or are suitable for students at certain stages of their candidature (early/mid/late). It is hoped that students will attend for a full- or half-day but please do feel welcome to come along for just specific sessions.

Please register your attendance through the links below.

(a) Communicating Research – Monday 24th August; Room 252, Sociology/Geography Building





Giving an oral presentation

Lucy Johnston


Morning Tea


Conference attendance: Getting the most out of going

Tom Wilson


Thesis in Three and lunch



Communicating with the media

Tara Ross


Afternoon Tea


Publishing Research and co-authorship

Philip Schluter and Paul Gardner

Register for the Communicating Research Session

(b) Getting started – Monday August 31st; KG06 Kirkwood Village



Presenter(s) – additional to be confirmed


The Thesis Journey

Lucy Johnston


Career Objectives and Planning

Dave Petrie


Morning Tea


The student-supervisor relationship: Setting up and managing expectations

Janet Carter


Networking and lunch

Heidi Quinn; Lynn Clark


How to receive and respond to feedback

Lucy Johnston


Afternoon Tea


Dealing with problems in supervision

Jeanette King; Catherine Moran


Overcoming obstacles to completion

Kate Pedley

Register for the Getting Started Session

(c) The Broader Context – Monday 7th September; KG06 Kirkwood Village





Cultural Awareness

Dr Mary Boyce (Director of Māori Teaching and Learning)



Morning tea



Ethics and professional practice: Doing research properly






Preparing for the oral examination

Dr Lucy Johnston (Dean of Postgraduate Research)

Register for the Broader Context Session


Data Analysis and Statistics Workshops Semester 2

A series of 4 workshops are being offered in Semester 1 as detailed below. The sessions are to be run by Dr Elena Moltchanova and Dr Daniel Gerhard from the Statistical Consulting Unit (School of Mathematics and Statistics).
Please ensure that you register for each session that you wish to attend by following the link after each abstract. There is no limit on attendance at the seminars but there are for the lab. Streams are limited to 20 people per stream – if you sign up and are subsequently unable to attend please ensure that you let me know ASAP so that I can open the place up to another student.
You can access materials from the past seminars, slides and lab worksheets on the Statistical Consulting Unit page on Learn which is open to everybody with a UC password.







Classification and Cluster Analysis

Seminar (1hr) + lab (1 hr)

Thursday August 13th; 2-4pm
Seminar – Room 111, Erskine Building
Labs – Rooms 010 and 248 Erskine Building

Multiple Hypotheses Testing

Seminar (1hr) + lab (1 hr)

Thursday September 17th; 2-4pm
Seminar – Room 111, Erskine Building
Labs – Rooms 010 and 248 Erskine Building

Principles of regression analysis and ANOVA (EM, 1h + 1h lab in SPSS/R)

Seminar (1hr) + lab (1 hr)

Thursday October 15th; 2-4pm
Seminar – Room 111, Erskine Building
Labs – Rooms 010 and 248 Erskine Building

Mixed-Effects Modelling (EM, 1h + 1h lab in SPSS/R)

Seminar (1hr) + lab (1 hr)

Thursday November 19th; 2-4pm
Seminar – Room 111, Erskine Building
Labs – Rooms 010 and 248 Erskine Building

Classification and Cluster Analysis (Daniel Gerhard, 1h + 1h SPSS/R Lab)
With classification methods we assign new observations into one of several predefined categories, e.g. predicting the status of a patient as healthy/illness based on medical test results. Whereas classification methods 'learn' from a training data with complete knowledge of the correct classification, cluster analysis is used to allocate observations into a number of groups only based on the data at hand.
In this seminar we take a look at several methods for classification (logistic regression, random forest) and cluster analysis (hierarchical and model-based clustering).

Register here to attend one of the lab streams

Multiple Hypotheses Testing (Daniel Gerhard, 1h + 1h SPSS/R Lab)
When testing several hypotheses simultaneously, e.g. pairwise comparisons of multiple treatments or using hypotheses tests for variable selection in gene expression studies, the risk to reject at least one of these null-hypothesis just by chance exceeds the type-I-error rate of α=0.05 for each single test. Neglecting this multiple testing problem might lead to an inflated number of reported statistically significant effects and consequently the reproducibility and reliability of research results cannot be ensured.
In this seminar we will present various approaches to control different error rates (Family-Wise Error Rate (FWER), False Discovery Rate (FDR)), and discuss when to use these multiple testing adjustments in practice.

Register here to attend one of the lab streams

Principles of Regression Analysis and ANOVA (Elena Moltchanova, 1h + 1h lab in SPSS/R)
Multiple regression and ANOVA are among the most frequently applied statistical methods. In this seminar we will go through the four steps of the regression modelling: (i) exploratory data analysis, (ii) model fitting, (iii) diagnostics, and (iv) interpretation of results. Model selection within ANOVA framework will also be considered.
Register here to attend one of the lab streams

Mixed-Effects Modelling (Elena Moltchanova, 1h + 1h lab in SPSS/R)
The assumption of (generalized) linear regression and ANOVA requiring independent identically distributed observations often does not hold, because there are repeated measurements in the experiment or because there is some structure in the data set: classes and schools, plots and sites, families, different personnel performing experiments on different days. All this can be taken into account via mixed effects regression models, which are the topic of this seminar.
Register here to attend one of the lab streams