Introduction to Structural Equation Modelling using Mplus

Details: 

This course is being led by Nick Shryane from The University of Manchester.

This two-day course introduces the family of statistical methods know as structural equation modelling (SEM). The course also introduces Mplus, a powerful statistical modelling and simulation software package for carrying out SEM.

SEM allows researchers to evaluate the quantitative consequences of their theoretical models. It is therefore most useful for testing hypotheses rather than data exploration. SEM is appropriate for modelling all kinds of multivariate data from many different research designs. The examples used in the course will be drawn mainly from the social, psychological and clinical realms, using publically available population survey data.

The course is aimed at people with experience of the empirical research process and statistical modelling such as regression analysis, who want to an introduction to the much broader and more flexible suite of analytical tools offered by SEM.

Day 1
Day one of the course will start by giving a brief introduction to Mplus by using it to run linear regression models. We will then see how regression models can be extended into path models, which can have more than one outcome variable and can therefore model putative causal chains. The classic example of a path model, mediation analysis, which aims to evaluate the mechanisms of action by which causes bring about effects, will form the core of this part.

In the afternoon we will look at how to model theoretical constructs that are not part of the observed dataset, using confirmatory factor analysis (CFA). CFA is a further extension of regression, integrating factor modelling into the broader SEM framework. Factor models also allow for the effects of random measurement error to be disentangled from variation in the target measurement construct.

Day 1 will give participants knowledge about:

• Mediation analysis; its relationship to regression, its additional assumptions, and the types of hypotheses it can be used to evaluate
• Confirmatory factor analysis; its relationship to regression, its additional assumptions, and the types of hypotheses it can be used to evaluate
• The application of mediation and CFA to real-world research problems
• Using Mplus to specify and estimate mediation and CFA models
• The output of Mplus, and using this output to evaluate research hypotheses

Day 2
On day two of the course we will start by consolidating the learning from day one, then looking at how SEM models can be built by combining elements from path and CFA. We will then look at how these models can be tested and evaluated. For the afternoon session on day two we will see how the ‘traditional’ SEM methods that we have used thus far, which rely upon strong assumptions of normality, can be extended into Generalized SEM, where the observed variables can be binary and ordinal as well as continuous.

Day 2 will give participants knowledge about:

• How to combine mediation and CFA models to create models designed to evaluate specific research hypotheses
• Methods of evaluating relative and absolute model fit, and how to use these to guide selection of the ‘best’ model
• How to extend SEM to include binary observed outcome variables, and how to interpret the output of such models

Pre-requisites
The course does not assume any experience with SEM or Mplus.
Basic understanding of linear and logistic regression modelling (e.g. what the regression coefficients represent, what R-squared is) is assumed.
Competence in using a PC and Microsoft Windows is assumed.

Application process:
There are 20 places available on the course and places will be allocated following a process of application.  Priority will be given to doctoral students enrolled in a social science degree programme at any Scottish University or to doctoral students funded by an ESRC-funded UK Doctoral Training Centre.  Remaining places will be offered for a fee to other doctoral students, academic staff, researchers and non-academics who have submitted an application.

To apply for this event, a completed on-line application form should be submitted by 12 noon on 30th November 2016.  Successful applicants will be notified by 2nd December 2016.

Course costs:
This training course is offered free of charge to doctoral students registered on a social science degree programme at a Scottish University or to doctoral students funded by an ESRC-funded DTC (see above).

Remaining places will be open to others for a fee:

Masters students £60
Doctoral students not funded by an ESRC UK DTC £60
Academic staff, ESRC funded researchers and UK registered charitable organisations £120
Others £500

Please note that AQMeN can only accept payment for training via the University of Edinburgh ePay system.  We are unable to issue invoices or accept cheques for payment.  All payments must be received within 14 days of a place being offered otherwise the place will be released.  Payment should not be made until you have received confirmation that you have been allocated a place on the course.

Travel and accommodation bursaries:
Doctoral students registered on a social science degree programme at a Scottish university or at an ESRC-funded DTC may be eligible to claim travel and/or accommodation costs to attend.

In order to be eligible for a bursary you must reside outside Edinburgh and attend a university outside Edinburgh.  Reimbursement can only take place if you follow the reimbursement process detailed below and bursaries for eligible students will be capped at the following rates:

Travel time (by rail) from Glasgow Maximum Travel Contribution Accommodation Contribution
 Up to 1 hour (travelling each day) 2 return trips @£25/trip = £50 NIL
Up to 1 hour (with accommodation during event) 1 return trip @£25/trip = £25 up to 2 nights accommodation at maximum £60 per night
Up to 2 hours 1 return trip @£50/trip = £50 up to 3 nights accommodation at maximum £60 per night
Over 2 hours 1 return trip @£100/trip = £100 up to 3 nights accommodation at maximum £60 per night

Reimbursement process:
Students are responsible for arranging travel and/or accommodation (if applicable) themselves and will only be reimbursed upon presentation of original receipts (no photocopies or credit card receipts will be accepted) and completion of the relevant expense claim form which will be provided post-event.
AQMeN will not reimburse the following costs (unless agreed prior to the event):

  • Mileage
  • First class travel
  • Meals or room service
  • Inter-city travel (e.g. buses to and from event venue)
  • Taxi fares
  • Credit card fees
  • Sundries (e.g.wireless internet access or newspapers at accommodation)

All expense claims and receipts must be received by the AQMeN office no later than 2 weeks following the last day of the event in order to be eligible for reimbursement.

Cancellation and Non-Attendance
By applying for this course, you are expected to attend for the full duration.  Failure to attend all sessions (unless by prior agreement with the AQMeN core team) may result in a charge of £50 per day of the course missed, non-reimbursement of expenses and ineligibility to be considered for future AQMeN training courses and events.


Non-paying delegates
If after receiving an allocated place on the course you are no longer able to attend; you must notify AQMeN as soon as possible to allow the place to be offered to someone on the waiting list.  Failure to do so may result in a charge of £50 per day of the course, non-reimbursement of expenses and ineligibility to be considered for future AQMeN training courses and events.

Paying delegates
Course fees are non-refundable.  In exceptional circumstances a refund may by permitted at the discretion of the AQMeN Research and Development Manager.

Contact:
If you have any questions regarding the course, please feel free to contact events@aqmen.ac.uk or + 44 (0) 131 650 2105

Application Form

Date: 
Monday, December 5, 2016 - 09:00 to Tuesday, December 6, 2016 - 17:00
Organiser: 
AQMeN
Location: 
Glasgow
Venue: 
University of Glasgow

Research Strand: