Migration Analysis Tools

Indicies ← Prev  Next → Spatial Data Manipulation Tools

Introduction

Background

This project provides a set of generic analytical tools for the interrogation of inter-regional migration flows matrices. The project also provides migration matrices for each intercensal period from 1976-81 to 2006-11 using consistent geographical boundaries which are not otherwise available for census time series analysis. The project delivers analytical tools in two key areas: First, tools to compute statistical metrics that measure five key components; (1) intensity of migration; (2) distance of migration; (3) migration connectivity; (4) migration impact, and; (5) direction of migration. Second, tools to cartographically and graphically explore and depict the dynamics of these statistical measures. Both key areas generate secondary data in the form the derived statistical measures, graphs and mapping that can be exported for reporting purposes and/or form inputs into further statistical analyses.

The analytical tools are:

  • Crude Migration Intensity
  • Crude Migration Intensity – In-Migration
  • Crude Migration Intensity – Out-Migration
  • Aggregate Net Migration Rate
  • Net Migration Rate
  • Migration Effectiveness Index
  • Circular Mean
  • Circular Spread
  • RayleighTest
  • Linear Mean
  • Origin-Destination Directions
  • Origin-Destination Distances

The migration and Journey to Work data can be analysed using the tools provided, and the results then downloaded as csv files for further manipulation by users using any software they prefer.

Classifications

The origins and destinations used in the migration dataset are Temporal Statistical Divisions (TSDs), which are aggregations of Statistical Local Areas (SLAs) or SA2s in the case of 2011 census data. The boundaries of SLAs can vary between census years, and SLA boundaries differ from SA2 boundaries, so the TSDs were created to provide a set of boundaries that are as consistent as possible to permit analysis of migration over an extended timeframe, including comparisons between intercensal periods.
The population of the migration dataset is all persons who changed their place of usual residence during an intercensal period (provided that their place of usual residence as recorded at the census could be coded to a valid TSD) classified by:

  • Intercensal period (seven periods, from 1976-81 through to 2006-2011)
  • Origin TSD (usual residence at the start of the intercensal period)
  • Destination TSD (usual residence at the end of the intercensal period)
  • Age Group (in five year age groups 5-9 through to 75+)
  • Sex

The population of the Journey to Work dataset is all persons who were employed in the week prior to census night, classified by:

  • Census Year (2001 or 2006)
  • Origin SLA (Place of Usual Residence at the time of the Census)
  • Destination SLA (Place of Work at the time of the Census)

Definitions

The Australian Internal Migration (AIM) database was developed by the Queensland Centre for Population Research (QCPR). This database holds internal migration flows between 69 Temporally consistent Statistical Divisions (TSDs), captured at each Australian Census since 1976. Data are derived from the Census 5 year migration question which asks where the respondent lived 5 years ago. The response to this question can be matched against the respondent’s current place of residence and other population characteristics. AIMS data are available by 5 year age groups and sex. This is a unique database that makes possible a comparison of migration flows within Australia over a long time period. This can be used to detect changes in migration patterns for the population as a whole and for specific age and sex groups, such as young persons, persons nearing retirement or the aged population. Issues such as changing patterns of movement out of rural areas to urban locations, or from inner city to middle or outer city regions, or from inland to coastal areas can be examined for specific population age groups over time using this database.

Temporal Statistical Divisions (TSDs) were created to divide Australia into a set of regions with consistent boundaries in order to permit analysis of migration over an extended timeframe, including comparisons between intercensal periods. TSDs were originally derived using 1996 Census Statistical Divisions with an additional breakdown of regions within the mainland capital cities based loosely on the planning regions defined or employed by government planning agencies. TSDs were formally defined as aggregations of Statistical Local Areas (SLAs). SLA boundaries vary between census years, and it is not possible to achieve a precise geographical match over time. TSDs in earlier and subsequent years were therefore defined to match the 1996 TSD boundaries as closely as possible using GIS overlays to establish lookup tables corresponding to each Census. For the 2011 Census, the regions were comprised of SA2s rather than SLAs, reflecting the change in statistical geography in Australia from the ASGC to the ASGS. The key feature of TSDs is that their boundaries are treated as constant over an extended period, which enables analysis of migration over a much longer period than would be possible using standard census geography. This makes it possible to detect long term changes in migration patterns that might otherwise be overlooked. For further information, see Blake et al (2000).

The Crude Migration Intensity (CMI) measures the proportion of the population who changed their place of usual residence over a defined time interval. It indicates the overall level or incidence of population mobility within a population. Using migration data from the Census, it is calculated as a probability, defined as the total the total number of persons who changed their region of residence between the start and end of the observation interval, divided by the population at risk of making such a move (the start of period population),expressed as a percentage. Region-specific intensities can also be calculated separately for in-migration to or out-migration from a region. For further information, including precise definitions of the population at risk, see Bell et al (2002), Rees et al (2000).

The Net Migration Rate (NMR) is computed as the balance of net migration flows to and from a given region, expressed as a percentage of that region’s starting population. It provides a measure of the impact of migration in modifying the size of the population in that region. See Bell et al (2002), Plane & Rogerson (1994)

The Aggregate Net Migration Rate (ANMR) is the system-wide equivalent of the NMR, computed across all regions in the system by expressing the sum of the positive net migration balances (summed across all regions) as a percentage of the system-wide starting population. The ANMR provides an overall measure of the impact of inter-regional migration in changing the distribution of population between regions within the settlement system. See Bell et al (2002).

The Migration Effectiveness Ratio (MER) is computed as the net migration balance of a region expressed as a percentage of the sum of the inwards and outwards flows to and from that region. It can assume values varying from -100 to +100 and indicates the efficiency with which migration operates as a mechanism to redistribute population to or from that region. See Bell et al (2002), Plane & Rogerson (1994).

The Migration Effectiveness Index (MEI) is the system-wide equivalent of the MER, computed as the sum of the absolute net migration balances (summed across all regions) and expressed as a percentage of the gross flows between all regions. The MEI provides an overall measure of the efficiency with which migration is operating to redistribute population between regions. See Bell et al (2002).

The Circular Mean is a measure of location. It is the mean or ‘average’ direction of migration flow. If an origin zone is thought of as the centre of a circle, then the directions of migration from that zone to other zones are treated as points on the circumference of the circle, and for each direction an angle from due North is calculated. As the smallest angle (0 degrees) and the largest angle (359 degrees) are actually adjacent to each other on the circumference rather than the maximum distance apart (which would be the case if they were treated as linear data), a circular mean is used in place of a linear mean. In calculating the circular mean each direction is weighted by the number of persons making that move. For further information, see Corcoran et al (2009), Fisher (1993).

The Circular Spread is a measure describing the level of dispersion in migration flows. Low values indicate a low level of spread while higher values indicate a wider degree of dispersion. In calculating the circular spread each direction is weighted by the number of persons making that move. For further information, see Corcoran et al (2009), Fisher (1993).

The Rayleigh Test is an inferential statistical method for testing if a particular group of migration flows is uniformly distributed in all directions, or if there is generally a preference for one direction. For further information, see Corcoran et al (2009), Fisher (1993).

Journey to Work (JTW) is the term used to describe the results of comparing the respondent’s place of work with their place of residence. The data on place of work are derived from responses to the census question which asks about the address of the respondent’s workplace in their main job held the previous week. These responses are coded to destination zones using indices provided by State transport authorities. The destination zones can be aggregated to higher geographic levels, such as SLAs or SA2s. From these data it is possible to analyse flows of employed persons from home to work, which can be matched against other population characteristics such as age, sex, occupation or industry. For transport authorities it is also useful to match these data against the responses to an additional census question which asks about the method(s) used to get to work. This comparison enables analyses such as the use of public transport, the extent of decentralisation of employment, the size and characteristics of daytime populations in the CBD or elsewhere, and the efficiency of transport flows. JTW data can be analysed in a similar way to migration data, as they involve an origin and a destination. Distance and direction of JTW flows are two of the measures provided in this database, which includes origin and destination data at SLA level for total persons, from both the 2001 and 2006 Censuses. The 2001 SLAs have been concorded to 2006 boundaries to make the data comparable over time. For measurement purposes the centroid of each SLA has been used as a reference point. In future there is potential to significantly enhance the value of the data with the addition of population characteristics such age, sex, occupation, industry and method of transport.

Running the Tools

This section provides a user tutorial for the AURIN migration analysis e-Tools.

This tutorial provides an example of migration intensity metrics usage and of circular statistics application to migration directional data.

In this tutorial interregional migration data will be selected within Queensland, the migration intensity metrics and circular statistics for migration direct will be compared for Males and Female migrants.

Area and Data Selection

The dataset we will be dealing with is Australian Internal Migration Dataset. The regions in this dataset are encoded using the temporal statistical division geography; this is a nonstandard geography specifically designed for migration analysis over long time-scales.

Select the AIM dataset as follows

  • Ensure the Area is set to Australia
  • Click + Dataset under the Data panel
  • Select Temporal Statistical Division as your Aggregation Level and click Search
  • There should be one dataset returned as in the following image. Select the dataset named TSD of inter-regional migration flows matrices
[Click to Enlarge]

[Click to Enlarge]

Select 2001-06 as your Census Period, 30-34 as your Age Group, and Females as your Gender. Click Add to add the dataset to your Data panel, and then repeat the same again for Males. You should now have two datasets sitting in your Data panel.

Intensity Metrics

Running the Tool

We will now apply the migration intensity metrics to both datasets and compare using a choropleth.

First we need to compute the Migration intensity metrics for each dataset.

  • Click Tools → Migration Analysis → Intensity – All Metrics to bring up the Intensity Metrics parameter input window
  • Enter the parameters as shown in the image below for Females
  • Click Add and Run
  • Repeat for Males

Once both workflows have completed two new datasets should have been added to the Data panel. These will be named as Output: Intensity-AllMetrics mm-dd hh:mm, you should rename the two datasets to easily distinguish between the male and female data

Visualising the Intensity Outputs

We can visualise the migration intensity metrics on a map using AURIN’s mapping tools. In this section we will map the Inbound Crude Mirgation Intensity for Females of age 30-34, as just computed. You first need to download the dataset as a CSV, and re-upload it, making sure you specify the Temporal Statistical Divisions as your level of aggregation

  • Click the Maps,Charts and GraphsMap Visualisations → Choropleth button in the Visualise panel.
  • Enter the Parameters as you shown in the image below and click Add and Display
[Click to Enlarge]

[Click to Enlarge]

Your female migration choropleth should look something like the image below

[Click to Enlarge]

[Click to Enlarge]

We will now use a static scatterplot to visualise the relationship between inbound and outbound migrations for 30-34 year old women in the 2001-2006 period.

To do this open the scatterplot tool (Tools → Charts  → Scatter Plot) and enter the parameters as shown below, click Add and Run.

[Click to Enlarge]

[Click to Enlarge]

Once you have run the tool click the Display button that appears. Your scatterplot should look something like the image below

[Click to Enlarge]

[Click to Enlarge]

Circular Statistics

We will use the data set selection from the previous section to compute the circular mean and circular spread of migration direction across Queensland. Since we are using the same dataset as above, we will just jump straight into using the circular statistics tool

Running the Tool

  • Click the Tools → Migration Analysis → Circular – All Statistics
  • Enter the Parameters as shown in the image below
  • Click Add and Run
[Click to Enlarge]

[Click to Enlarge]

Repeat these steps for the Male TSD dataset as well.

Visualising the Intensity Outputs

We can visualise the circular statistics metrics on a map using AURIN’s mapping tools. In this section we will map the Inbound Directional Spread for Females of age 30-34, as just computed. You first need to download the dataset as a CSV, and re-upload it, making sure you specify the Temporal Statistical Divisions as your level of aggregation

  • Click the Maps,Charts and GraphsMap Visualisations → Choropleth button in the Visualise panel.
  • Enter the Parameters as you shown in the image below and click Add and Display
[Click to Enlarge]

[Click to Enlarge]

Your map should look something like the image below

[Click to Enlarge]

[Click to Enlarge]

References

Bell M., Blake, M., Boyle, P., Duke-Williams, O., Stillwell, J., and Hugo, G. (2002) ‘Cross-national comparison of internal migration: issues and measures’. Journal of the Royal Statistical Society: Series A Statistics in Society, 165(3), 435-464.
Blake, M., Bell M. and Rees, P., (2000) ‘Creating a Temporally Consistent Spatial Framework for the Analysis of Inter- Regional Migration in Australia’. International Journal of Population Geography, 6, 155-174.
Corcoran, J., Chhetri, P., and Stimson, R. (2009) ‘Using circular statistics to explore the geography of the Journey-To-Work’. Papers in Regional Science, 1 119-132.
Fisher, N. I., ‘Statistical analysis of circular data’. Cambridge University Press, 1993.
Plane D. & Rogerson P., ‘The geographical analysis of population with applications to planning and business’. Wiley 1994.
Rees, P., Bell M., Duke-Williams, O., Stillwell, J., and Blake, M. (2000) ‘Problems and solutions in the measurement of migration intensities, Australia and Britain compared’. Population Studies, 54, 207-222