Help pages for GEMDE: Grid Enabled ethnic Minority Data Environment

Quick links:

GEMDE is a resource for supporting quantitative data broadly related to the analysis of ethnicity for social science research (spanning topics such as ethnic itentity, nationality and national origins, religion, immigration, language). It boils down to a means of providing access to organised, easy to use statistical resources on the topic. We refer to these resources as either 'MUGs' or 'MIRs' (see our FAQs).

We are trying to make GEMDE easy to use, but there are lots of features to the service which are likely to be new to most people. This page gives detailed instructions including screenshots illustrating features of the GEMDE online 'portal' environment where the data resources are made available: the GEMDE portal (needs a login).

The portal iself does have some help notices and instructions for users within its pages, but you will probably benefit from using these webpages in combination to make sense of the portal. We are also preparing pdf instruction guides (to appear under our GEMDE references list).

The portal should work smoothly for most mainstream internet browsers, but we have only tested it extensively for Internet Explorer (v8 or above) and Firefox. Please let us know if you experience difficulties with the portal and we will address them if we can.

GEMDE PORTAL SERVICES / FAQS / MOST USEFUL THINGS / Contact us / References

External link: ACCESS THE GEMDE PORTAL

Workshop materials (28 January 2010) : including slides and handouts introducing GEMDE and the portal service for an 'expert workshop' held at Stirling on 28/1/2010

GEMDE PORTAL SERVICES

If you haven't already read them, it is probably helpful to first read our notes on the background to GEMDE and our higher-level description of using GEMDE

.

This note provides more extended descriptions of the following GEMDE services:

We're still developing GEMDE. Please contact us with feedback/comments on the service and its usability.

Guest login instructions

You can login to the GEMDE portal as a guest. This will allow you to search and browse across, and subsequently download data from, the portal resources. As a guest, however, you won't be able to access the functionality which covers uploading files to GEMDE or making other contributions to the service. For many users, 'guest' level access will be all that you need from the system.

To login as a guest, you do still need to go through an authentication process as if you were a named user. This can seem a little cumbersome, but at present it is the only mechanism available to us. So, to login as a guest:

Login for registered users

If you login to the GEMDE portal as a named user, there are several functions available to you which are not available at the 'guest' level. These are to upload new records and data resources; to edit and revise existing resources; and to provide comments and user feedback on resources through the 'expert ratings' option.

To login as a named user, you first have to contact the GEMDE project group to request a personal account. Then, you need to go through a Shibboleth authentication process . At the time of writing (June 2010) we have a transitory arrangement in place whereby you need to manually request a personal account by email, then register as part of the 'National e-Science Centre' (i.e. rather than your own institution). In the longer term we do plan to use institutional login accounts, to make the process simpler. So, for the time being, to login as a named user:

Here are some graphics illustrating the steps involved:

Filling in the entry page at guest level
(a) Click on the link to the portal (b) Entering details for guest level access
Filling in the entry page at named user level Image of the portal when logged in as a registered user
(c) Entering named account details (d) Image of the portal when logged in as a registered user

Using the browse facility

'Browsing' through GEMDE refers to searching through resources registered at GEMDE in a structured way. The browse interface works in a fairly conventional way on the GEMDE system. When you go to the browse page, you will be presented with lists of categories and subcategories, which you can click accordingly to take you to further information on relevant resources.

There are a couple of important points to note however. These can often cause confusion the first time you use the Browse system:

Using the search facility

You can search for resources in GEMDE using free-text entries into the 'search' tab. By and large, the search engine works in the standard ways (described below). As with browsing, however, it is important to understand that two different types of resources can be searched (i.e. MIRs - ethnic Minority Information Resources; and MUGs - ethnic Minority Unit Groups).

Initial submission of data on a MUG or MIR

If you're a registered user, you can enter data to the GEMDE system. See access instructions to arrange user registration if required.

Please do consider adding your own data into the GEMDE system. The GEMDE portal is intended to be a collaborative, research oriented resource to support the sharing and distribution of data related to the analysis of ethnicity in the social science research community. If you have generated data - e.g. a definition of ethnic categories; a recode of an ethnicity measure; or some relevant aggregate statistics about ethnic groups - then GEMDE would be a good place to deposit it and distribute it for the wider benefit of the research community.

Don't be hesitant!The forms for initially submitting data to GEMDE are intended to be short and fairly self-explanatory, but in any case you can't do much too damage if you initially enter something wrong:

The images below try to illustrate the steps involved in depositing data at GEMDE

Follow the 'deposit' link to submit a data resource on the GEMDE portal
You'll need to decide whether your resouce is a MUG or a MIR... (MUG=Minority Unit Group, i.e. a listing of categories or taxonomy; MIR=Minority Information Resource, i.e. other information about the categories of a MUG)
As an illustrative example, these images show us uploading data transcribed from a research paper by Khattab (2009). This data features both a MUG and a MIR. The MUG us a list of ethno-religious groups, supplied as a MS Word file. The MIR is a table of statistics about those grous, supplied as an MS Excel file
To upload the MUG fill out the form...
...taking care to attach the data file...
...before clicking on 'upload MUG'.
Likewise to upload the MIR fill out the form and attach the file, then click upload...
Note that as soon as the resources are uploaded, they are visible to search terms submitted to other users...
..but we strongly encourage you to edit/update the submissions, for instance in order to nominate the categories which will be picked up under 'browse' searches

Further editing and metadata provision

It is very valuable to us if after completing an initial submission of a resource, you go back to it under the 'edit data resource' tab and add further information about it.

Entering and using user-ratings

Tools for discerning the quality of data resources are obviously valuable in a resource provision such as GEMDE. We are deliberately 'pluralistic' in accepting most resources into the system without substantial quality checks. We do however have 'expert' and 'user' ratings systems available to help navigate according to opinions on quality.

Using the 'live data analysis' facility

The GEMDE portal includes a facility for running queries on selected micro-data files with relevant information on ethnicity. The files have been stored securely on the GEMDE server and the facility doesn't allow you to access the data or to run custom analyses on them. However, it is possible to submit information requests about relevant distributional features of the sample.

The examples apply to the UK since the 1970's onwards, and analyses focus upon important correlates of ethnic groups - age, gender and socio-economic measures such as education. We put this resource up because we consider it is very important for researchers to have a basic idea of correlations between ethnicity and these factors, and how they change by time and the age of a sample.

As a guest or registered user, you can follow the link to 'microdata' and identify the query you would like to run, according to your specification from a choice of a few parameters about the data you're interested. The output from the query will be a graph depicting the data, and it is also possible for you to access he log file from the analysis, including the statistics, which was used to generate the graph (in the R language). The current queries avaialable (we hope to add more) cover:





WORKING WITH DATA ON ETHNICITY: MOST USEFUL THINGS

A consistent form of feedback we've recieved during the development of GEMDE is that most users would simply like a fairly firm steer on what they should be doing with measures of ethnicity in a particular context. This often isn't what GEMDE offers - the GEMDE portal services concentrate upon allowing researchers to find out about all sorts of different approaches that might be taken with data on the topic, in a rather pluralistic way. After spending some time working in the topic domain, though, there are certain principles or resources we would tend to highlight to people as particular useful to know about or as good practice to follow. So, with apologies for introducing another daft acronym, here our our 'MUTs':

  • Simplifying categories: Although most of us do it intuitively, it is rarely ideal to collapse together different ethnic categories just because they are sparsely represented. If you do do this, try to use the same collapse as has been done in another cited reference. Alternatively, consider a device like effect proportional scaling in a dimension of difference as a feasible alternative to grouping many categories together. There are resources on both previous collapsed measures, and effect proportional scaling, deposited as 'MIRs' within GEMDE
  • Comparative research: There have already been some efforts to think systematically about comparing measures of ethnicity across time or between countries. In the UK, Bosveld et al (2006) is a good staring point. For cross-national research, a paper by Lambert(2005) is available via on the GEMDE portal.
  • Multivariate analysis: Consider age differences! In many countries, there are substantial differences in age profiles between ethnic groups (due to immigration cohort differences) so socio-economic analysis of ethnic differences need to be careful to fully control for age differences in order that they don't conflate age or life-course stage effects with ethnicity effects.
  • Statistical modelling: Consider interaction terms! A lot of the time, the substantive analysis of interest is actually an interaction between an ethnicity difference and some other factor (e.g. the hypothesis is that the effects of education operate differently for different ethnic groups, so an interaction term could naturally test for this).
  • Statistical modelling: Since most measures of ethnicity have quite skewed distributions, it is valuable to use a 'quasi-variance' style approach to reporting parameter estimates for different ethnic group effects, because of the 'reference category problem' (for explanation see our own QV webpages, which include some illustrative worked examples).

Last updated 18/DEC/2010, by Paul Lambert