General description
Events and parameter estimates

General description

Being a model for the study of childlessness, the main event of RiskPaths is the first pregnancy (which is always assumed to lead to birth). Pregnancy can occur at any point in time after the 15th birthday, with risks changing by both age and union status. The underlying statistical models are piecewise constant hazard regressions. With respect to fertility this implies the assumption of a constant pregnancy risk for a given age group (e.g. age 15-17.5) and union status (e.g. single with no prior unions).

For unions, we distinguish four possible state levels:

• single
• the first three years in a first union
• the following years in a first union
• all the years in a second union

(After the dissolution of a second union, women are assumed to stay single). Accordingly, we model five different union events:

• first union formation
• first union dissolution
• second union formation
• second union dissolution
• the change of union phase which occurs after three years in the first union.

The last event (change of union phase) is a clock event - it differs from other events in that its timing is not stochastic but predefined. (Another clock event in the model is the change of the age index every 2.5 years) Besides unions and fertility, we model mortality--a woman may die at any point in time. We stop the simulation of the pregnancy and union events either when a women dies, or at pregnancy (as we are only interested in studying childlessness), or at her 40th birthday (since later first pregnancies are very rare in Russia and Bulgaria and are thus ignored for this model).

At age fifteen a woman becomes subject to both pregnancy and union formation risks. These are competing risks. We draw random durations to first pregnancy and to first union formation. There are two additional competing events at this stage-mortality and change of age group. (As we assume that both pregnancy and union formation risks change with age, the risks underlying the random durations only apply for a given time period--2.5 years in our model--and have to be recalculated at that point in time.)

In other words, the 15th birthday will be followed by one of these four possible events:

• the woman dies
• she gets pregnant
• she enters a union
• she enters a new age group at age 17.5 because none of the first three events occurred before age 17.5

Death or pregnancy terminates the simulation. A change of age index requires that the waiting times for the competing events union formation and pregnancy be updated. The union formation event alters the risk of first pregnancy (making it much higher) and changes the set of competing risks. A woman is then no longer at risk of first union formation but becomes subject to union dissolution risk.

Events and parameter estimates

First pregnancy

As outlined above, first pregnancy is modeled by an age baseline hazard and relative risks dependent on union status and duration. The following Table 1 displays the parameter estimates for Bulgaria and Russia before and after the political and economical transition.

Table 1

	Bulgaria	Russia
15-17.5	0.2869	0.2120
17.5-20	0.7591	0.7606
20-22.5	0.8458	0.8295
22.5-25	0.8167	0.6505
25-27.5	0.6727	0.5423
27.5-30	0.5105	0.5787
30-32.5	0.4882	0.4884
32.5-35	0.2562	0.3237
35-37.5	0.2597	0.3089
37.5-40	0.1542	0.0909

Table 1 before and after the political and economical transition

	before 1989 transition		10 years after transition: 1999+
	Bulgaria	Russia	Bulgaria	Russia
Not in union	0.0648	0.0893	0.0316	0.0664
First 3 years of first union	1.0000	1.0000	0.4890	0.5067
First union after three years	0.2523	0.2767	0.2652	0.2746
Second union	0.8048	0.5250	0.2285	0.2698

The data from Table 1 is interpreted as follows in the model. As long as a woman has not entered a partnership, we have to multiply her age-dependent baseline risk of first pregnancy by the relative risk "not in a union". For example, the pregnancy risk of a 20 year old single woman of the pre-transition Bulgarian cohort can be calculated as 0.8458*0.0648 = 0.05481. At this rate of ? =0.05481:

• The expected mean waiting time to the pregnancy event is 1/ ? = 1/0.05481 = 18.25 years;
• The probability that a women does not experience pregnancy in the following 2.5 years (given that she stays single) is exp(-?t) = exp(-0.05481*2.5) = 87.2%.

Thus at her 20th birthday, we can draw a random duration to first pregnancy from a uniform distributed random number (a number that can obtain any value between 0 and 1 with the same probability) using the formula:

RandomDuration = -ln(RandomUniform) / (;

As we have calculated above, in 87.2% of the cases, no conception will take place in the next 2.5 years. Accordingly, if we draw a uniform distributed random number smaller than 0.872, the corresponding waiting time will be longer than 2.5 years, since
-ln(RandomUniform) / ( = -ln(0.872)/0.05481 = 2.5 years. A random draw greater than 0.872 will result in a waiting time smaller than 2.5 years-in this situation, if the woman does not enter a union before the pregnancy event, the pregnancy takes place in our simulation.

To continue this example, let us assume that the first event that happens in our simulation is a union formation at age 20.5. We now have to update the pregnancy risk. While the baseline risk still stays the same for the next two years (i.e. 0.8458), the relative risk is now 1.0000 (as per the reference category in Table 1) because the woman is in the first three years of a union. The new hazard rate for pregnancy (applicable for the next two years, until age 22.5) is considerably higher now at 0.8458*1.0000 = 0.8458. The average waiting time at this rate is thus only 1/0.8458 = 1.18 years and for any random number greater than exp(-0.8458*2)=0.1842 the simulated waiting time would be smaller than two years. That is, 81.6% (1 - 0.1842) of women will experience a first pregnancy within the first two years of a first union or partnership that begins at age 20.5.

First union formation

Risks are given as piecewise constant rates changing with age. Again we model age intervals of 2.5 years. These are the rates for women prior to any conception, as such an event would stop our simulation.

Table first union formation

	before 1989 transition		10 years after transition: 1999+
	Bulgaria	Russia	Bulgaria	Russia
15-17.5	0.0309	0.0297	0.0173	0.0303
17.5-20	0.1341	0.1342	0.0751	0.1369
20-22.5	0.1672	0.1889	0.0936	0.1926
22.5-25	0.1656	0.1724	0.0927	0.1758
25-27.5	0.1474	0.1208	0.0825	0.1232
27.5-30	0.1085	0.1086	0.0607	0.1108
30-32.5	0.0804	0.0838	0.0450	0.0855
32.5-35	0.0339	0.0862	0.0190	0.0879
35-37.5	0.0455	0.0388	0.0255	0.0396
37.5-40	0.0400	0.0324	0.0224	0.0330

The parameterization example given in Table 2 has the following interpretation: the first union formation hazard of Bulgarian women of the first cohort is 0 until the 15th birthday; afterwards it changes in time steps of 2.5 years from 0.0309 to 0.1341, then from 0.1341 to 0.1672, and so on. The risk is highest for the age group 20-22.5--at a rate of 0.1672, the expected time to union formation is 1/0.1672=6 years. A women who is single on her 20th birthday has a 34% probability of experiencing a first union formation in the following 2.5 years (p=1-exp(-0.1672*2.5)).

Second union formation

A woman becomes exposed to the second union formation risk if and when her first union dissolves. As a difference to the first union formation which is based on age, this process does not start at a fixed point in time but is triggered by another event (first union dissolution). Accordingly, the time intervals of the estimated piecewise constant hazard rates refer to the time since first union dissolution.

Table second union formation

	before 1989 transition		10 years after transition: 1999+
	Bulgaria	Russia	Bulgaria	Russia
<2 years after dissolution	0.1996	0.2554	0.1457	0.2247
2-6 years after dissolution	0.1353	0.1695	0.0988	0.1492
6-10 years after dissolution	0.1099	0.1354	0.0802	0.1191
10-15 years after dissolution	0.0261	0.1126	0.0191	0.0991
>5 years after dissolution	0.0457	0.0217	0.0334	0.0191

Union dissolution

Both first and second unions can dissolve, with such processes starting at the first and second union formations, respectively. As the sample size is very small for the modeling of the second union dissolution event we do not distinguish the before and after transition cohorts for this event.

	before 1989 transition		10 years after transition: 1999+
	Bulgaria	Russia	Bulgaria	Russia
First year of union	0.0096	0.0380	0.0121	0.0601
Union duration 1-5	0.0200	0.0601	0.0252	0.0949
Union duration 5-9	0.0213	0.0476	0.0269	0.0752
Union duration 9-13	0.0151	0.0408	0.0190	0.0645
Union duration >13	0.0111	0.0282	0.0140	0.0445

 

	Bulgaria	Russia
First 3 years of union	0.0371	0.0810
Union duration 3-9	0.0128	0.0744
Union duration 9+	0.0661	0.0632

Mortality

In this sample model, we leave it to the model user to either set death probabilities by age or to "switch off" mortality allowing the study of fertility without interference from mortality. In the latter case, all women reach the maximum age of 100 years. If the user chooses to simulate mortality, the specified probabilities are internally converted to piecewise constant hazard rates (based on the formula -ln(1-p) for p<1) so that death can happen at any time in a year. If a probability is set to 1 (as is the case when age=100), immediate death is assumed.

When installed on a computer, Modgen integrates itself into the (required) Microsoft Visual Studio C++ environment. The visual components of Modgen are a separate toolbar as well as additional items under the Tools and Help menus of Visual Studio. Modgen also appears as an option in the file dialog box for creating a new project as well as in the dialog box for adding a file to an existing project.

Figure 1 displays a screenshot of the programming interface as it appears after opening the Modgen application 'RiskPaths.sln'. The Modgen toolbar consists of several icons for running Modgen, accessing help, opening the BioBrowser tool, and switching the language (between English and French).

Figure 1: The programming interface

Modgen code is organized into several files, each with the file extension .mpp. As can be seen in the Solution Explorer window (Figure 1), RiskPaths consists of eight .mpp files grouped in the "Models (mpp)" folder. These are the essential files of RiskPaths, i.e. the files containing all Modgen code written by the model developer.

When invoking the Modgen tool (which can be accessed from the toolbar, or from the first item under the "Tools" menu), these .mpp files are translated into C++ code. Thus Modgen acts as a pre-compiler, creating one .cpp source code file for each .mpp file and putting the resulting .cpp files in the "C++ Files" folder. The Modgen tool also adds model-independent C++ code components to the "C++ Files" folder; these additional files^{Footnote 1} should not be changed by the model developer and are essential in order to use the C++ compiler to build the Modgen application.

The model parameters are contained in one or more .dat files organized in a folder labelled "Scenarios". These files are loaded at runtime and contain the actual values assigned to the parameters.

When running the Modgen tool, Modgen - like the C++ compiler - produces log output that is displayed in the Output window. Any error messages are also displayed in this window, and clicking on a particular error message leads you directly to the corresponding Modgen code that produced the error.

Two steps are required to create a Modgen application from the Visual Studio environment. First, Modgen has to translate the Modgen code in the .mpp files; this is done when invoking the Modgen tool. Second, the resulting C++ application has to be built and started. This can be done in one step by selecting "Start Debugging" in the "Debug" menu or by clicking the corresponding icon on the toolbar.

Notes

PDF Version (PDF, 161.4 KB)

• Getting started
• Essential components of a BioBrowser session
• The BioBrowser menu and toolbar
• How to use BioBrowser: The Basics
• State selection and navigation
• Browsing and changing the underlying chart data
• Display and output options
• Appendix: Creating a new Modgen database file

Introduction

BioBrowser, the Modgen Biography Browser is a stand-alone software product which supplements the Modgen language used for dynamic longitudinal microsimulation modeling.  BioBrowser allows the analyst to graph the microdata generated by the model.  Its purpose is to aid in uncovering possible algorithmic errors in the model, or to study some particularly interesting cases with respect to the specified Modgen model.

Microsimulation models written in the Modgen language generate synthetic lifetimes of individual actors.  Each actor is defined as a set of states which describe the characteristics of the actor.  For example, an actor could be a male individual whose attributes are described through the following states: age, sex, marital status, and health status.  The values of these states change as the actor progresses through his lifetime.  In our example, the individual’s age would change on each birthday while the marital status would change at the time point at which he/she was married, divorced, etc. 

BioBrowser is a tool which allows the analyst to graphically examine the characteristics and attributes of an actor over the course of his/her lifetime.  BioBrowser can graphically present one or many states for one or many simulated lifetimes.  In this way, BioBrowser complements the other reporting features inherent in Modgen which are designed to provide detailed cross-sectional information on a collection of actors at a given reference time or state.

The graphical representations produced by BioBrowser originate from a special database file which is the product of a Modgen model simulation run.  Once this file has been created there exists a variety of possible graphics which the analyst can create with BioBrowser.  The specifications of these graphics are controlled by the user through drop-down menus and options.  Therefore, an analyst with a limited knowledge of the Modgen modeling environment can create an impressive array of longitudinal graphics showing the characteristics of the actors at different points in time.  All of these graphical representations can be saved for editing at some future time and/or routed to a printer or clipboard.

Specifying the contents of the database file requires the analyst to have some knowledge of the Modgen simulation environment.  A section below describes the components of Modgen with which one must be familiar to successfully create a database file.  Further details can be found in the Modgen Developer’s Guide.

Contents of BioBrowser

The Modgen Biography Browser (BioBrowser) installation package consists of a sample database file demo(trk).mdb and a sample biography file demo.bbr.  These files are referred to extensively to provide worked examples of the browser.  The nature of these files is discussed in How to Use BioBrowser: The Basics.

System Requirements and User Feedback

BioBrowser has been tested on Windows XP and does not have substantial requirements for CPU, disk or memory.

Users with questions or problems with any aspect of this software are welcome to contact the development team at microsimulation@statcan.gc.ca.

Modgen components

Before using BioBrowser, it is important for the analyst to understand some of the essential components of Modgen.

database (.mdb) files
These files are created by Modgen during the simulation phase of the model.  They contain the raw data necessary to construct the graphical representation created by BioBrowser.  Although the database files can be read by BioBrowser, BioBrowser can never modify the contents of these files.  All BioBrowser sessions begin by opening a pre-existing database file.

dominant actors
These elements are at the core of any Modgen simulation exercise.  Dominant actors are usually persons or households which are created at the beginning of the simulation process and undergo changes to their characteristics as they proceed through their lives.  Dominant actors are defined by their characteristics (states) and by the events which transform their states.

non-dominant actors
Modgen simulates one case at a time where a set of dominant actors undergoes changes to its states.  One possible change to a person actor’s state is a marriage or a common-law union.  When this event has occurred, Modgen generates an appropriate spouse.  This spouse, another person actor, is termed a non-dominant person actor.  Once created, non-dominant actors undergo the same possible events as the dominant actors of the same type.  Non-dominant actors are linked to their dominant actor.

tentative actors
The process of generating a non-dominant actor in Modgen involves generating a sequence of potential candidates.  The candidates who are not chosen are termed tentative actors since they have no links to any of the dominant actors in the model.

states
These elements define the characteristics of the actors over the span of their lifetimes.  Examples of states might include age, employment status, or educational attainment.  States can be scalars or arrays.

Before beginning to use BioBrowser, a database file needs to be created using Modgen. If you want to examine states which are not in the database, a new Modgen simulation must be run and a new database file needs to be created.  A sample database demo(trk).mdb was included with this software package.  For more information on creating new databases in Modgen please refer to the Modgen Developer's Guide or, for a quick overview/refresher, see Appendix: Creating a new Modgen database file.

BioBrowser components

BioBrowser takes the database and creates graphics of the characteristics of the actors.  In addition to the above Modgen concepts, there are other concepts which relate specifically to BioBrowser.

biography (.bbr) files
These files contain the graphical representations which the analyst has created during a BioBrowser session.  The biography files can be created, saved, and edited by the analyst during a BioBrowser session.

display band
The graphical display of a state or linked actor.

filter
The criteria used to narrow or refine the set of actors to be used in a biography..

navigation band
A type of display band which also includes a set of buttons which allows the user to go from the display bands of one actor to another and add new states to the biography.  The buttons resemble the control buttons on the front of a CD player.

Menu commands

The BioBrowser menu bar contains a set of standard menus available in most Microsoft Office applications, as well as some application specific commands.  Some of the same functions may be available as Toolbar buttons or through keyboard equivalents.

Pop-up menus:  Some commands are only available from pop-up menus or by double-clicking on the chart area over the display bands of the desired state.  Use the right mouse button click to access the pop-up menus. These menus will differ depending on whether or not the state is a simple state or a linked actor. For simple states such as “employed” below, the following commands are available:

For the filter tracking band and linked actors, access is also provided to the navigation band commands, as shown below:

The toolbar

The toolbar provides quick access to the most frequently used menu items and commands in the BioBrowser application. Each button is described by a Tool-Tip or status bar description. If you have a small screen at low resolution you may choose not to display the Toolbar. Choose Tools/Options and Click on the View Toolbar Option.

Icon	Description	Menu equivalent
	Create new biography	File / New
	Open saved biography	File / Open
	Save biography	File / Save
	Print active biography	File / Print
	Copy active biography to clipboard	Edit / Copy
	Undo last add	Edit / Undo Last Add
	Show or hide grid lines	Format / Grid Lines
	Show or hide guide lines	Format / Guide Lines
	Show or hide navigation bands	Format / Navigation Bands
	Change background colour	Format / Background colour
	Change chart colour	Format / Chart Colour
	Invoke BioBrowser Help	Help / Contents

Formatting the chart area

When a state is added various defaults for chart presentation are used depending on its Modgen state type.  At present, five chart types are permitted:

table of four different chart types

Line	Level	Horizontal Bar	Point

Event

BioBrowser recognizes the following Modgen state types and plots them by default according to their state type:

State Type	Default Chart Type
Integer	Level
Long	Level
Floating Point	Line
Double	Line
Time	Line
Logical	Horizontal Bar
Classification	Event
Range	Level

The Line style draws one line between 2 adjacent points, whereas the Level style draws two lines (a vertical then a horizontal) between 2 points.  The Horizontal Bar, although most appropriate for logical, classification and range type of Modgen states can be used on all states.  For continuous states such as float or double, the horizontal bar uses colour interpolation from a start and end colour defined by the user.  No legend is available for Horizontal Bar.

The default colour for Line, Level and Point plots is blue.  The default colours for Horizontal Bar are white and gray.  You can control the line thickness, band width and point size for the biography but not at the level of a single state.  At present, these settings are global to the biography window.  All such settings are saved with the biography.

Double-click on a chart within a biography, or right click on it and select State Properties from the pop-up menu, to re-format it for chart type and colour.  A format string for the Y-Axis labels where appropriate can be changed at this time as well.  The following dialog box is used to set or adjust these properties:

If the chart type selected is Horizontal Bar a second colour will be presented for selection. For logical states, these will be the False and True colours. For all other states types these will be used as a start colour and end colour in a colour interpolation process.

Setting and saving display options

Use the Format menu commands to change display options for the active biography window. The View menu, which includes the ToolBar and Status Bar display options, are application global.

The Tools/Options menu can be used to set and save session defaults and display default options for new biographies. The Options dialog box consists of three tabs: General, Chart Defaults and Axes Defaults. The General options will take effect immediately, whereas the two Default tabs are used only with new biography creation. The OK button will save these defaults to your application ini file.

The General tab below sets and saves session defaults used at application startup. To change them during the session, use the View and Tools menus.

The chart defaults used for new biography creation consists of the following display options. To change these options for an already open biography, use the Format menu.

Axes defaults for new biography creation are set and saved within the third options tab. Axes properties for an open biography can be set by double clicking the axes area of the chart window or by using the Axes Properties  command from the Format menu.

Note that all display options and axes properties currently in effect for the open biography are saved with the biography file.

Sending a biography to the printer or clipboard

To print a biography, use File/Print or the Print button on the toolbar.

The printed biographies are sized to fit the page while maintaining their aspect ratio. The orientation used will depend on the aspect ratio of the biography window being printed, i.e., if the window is wider than tall, landscape will be used.

To send a biography to the clipboard, use Edit/Copy, Ctrl-C or the Toolbar copy button

What can simulation add to statistical analysis?
Desired features of a RiskPaths microsimulation model

What can simulation add to statistical analysis?

Before we can answer the question of what simulation can add to statistical analysis, we first need a good understanding of what the statistical results presented in the previous section reveal. The estimation results for the two countries and two cohorts allow us to study similarities and differences between the countries, as well as the changes in parameters over time separately for each of the individual processes. We see a remarkable similarity in parameters across the two countries especially for the pre-transition cohorts. Bulgaria differs from Russia basically only in the three times lower union dissolution risks and the slower speed of second union formation. Accordingly, comparing the pre- and post-transition cohorts, we find dramatic changes in most processes. The risk of first births was halved in the first three years of the first union with no later recovery, although the parameters stayed relatively unchanged after three years in a union. Also, in second unions, fertility dropped by more than 50%. The biggest difference between the two countries after the transition is in first union formation--rates halved in Bulgaria but stayed stable in Russia. For first union dissolution we see the opposite picture--union dissolution risks increased by around 40% in Russia while staying almost unchanged in Bulgaria.

These are typical examples of insights we can gain by single process analysis. We have separated a complex system into its component processes and studied the changes within those processes. In the case of fertility we have introduced relative risks--we study how certain factors (here, different union statuses) influence a single process. This is a very typical analytical question; scientific literature is rich of this kind of research.

The power of microsimulation unfolds when we study various processes simultaneously. Even in our very simple demographic example, results are difficult to interpret when we are interested in the effect of changes in single processes on aggregate outcomes. For example, what is the effect of Russia's 40% increase in union dissolution risks on childlessness? The effect will depend on fertility out of unions and in second unions as well as the speed of second union formation. The relative risk of fertility is higher in second unions than after three years in the first union, but second union formation takes time (during which fertility is very low) and not all women enter a second union. Do these effects cancel themselves out or does union dissolution affect fertility - and in which direction? Such questions invite us to use microsimulation for sensitivity analysis. How do aggregate outcomes change in response to the change of a single parameter? Note that we now have moved analysis from the level of a single process to an analysis of system behaviour.

A comparison of the two cohorts invites a further type of system analysis--what is the relative contribution of the change in single processes to the aggregated outcome? Comparing the two simulated cohorts we see that childlessness has increased considerably in both countries but even more so in Bulgaria. We can use microsimulation to decompose the contributions of the changes in the various processes to the aggregate change. How much would childlessness have changed if only fertility parameters changed? What is the contribution of changes in union formation? Has the increase in union dissolution risk contributed to the increase in childlessness in Russia? Of course, the aggregate change is not the simple arithmetic sum of partial effects. Some process changes might have a stronger or weaker effect in the presence of changes in other processes. For example, the effect of the change in fertility in second unions will heavily depend on the likelihood of being in a second union which is subject to first union formation and dissolution risks. Microsimulation can help us to identify and better understand such interactions.

Looking at the post-transition cohort, we have already entered the domain of predictions. As data were collected 14 years after the transition, in reality no post-transition cohort has gone through its whole reproductive period. Thus, for cohort measures like childlessness, the assessment of consistency with other data sources is limited to a comparison with other projections. But we can also use our model for predictions under alternative assumptions on future changes in processes. We might have a theory that leads to the assumption that only parts of the observed changes are of a permanent nature (e.g. caused by cultural change) while others are transitory (e.g. resulting from economic crisis, therefore reversible with economic recovery). What would happen if fertility rates moved back to their initial values while slower (later) union formation persisted--or vice versa? Such an analysis can produce surprising results, as it is not always a reversal of the process which initially had the biggest overall impact that will generate the biggest opposite effect.

Are there policy implications? While our model is of course too simple for policy analysis, it does not require much imagination to see how microsimulation can support policy making.

• In many cases, the studied phenomena are of direct policy relevance. Fertility decline will for example impact the sustainability of social security systems. A good demographic projection model can therefore produce valuable data input for subsequent good-quality planning. Microsimulation is the tool to combine separate statistical models into projection models.
• Events simulated in a microsimulation model can also be policy targets themselves. A government might aim at influencing fertility. This is possible if policies exist which are capable of influencing the modeled processes. However, we first have to be able to understand the individual contribution of those processes to the aggregated outcome we aim to change; thus we need microsimulation. If we can attach price tags on such policies, we are also able to use microsimulation to find the most cost-efficient policy mix - and to study possible side effects. (Socialist Russia and Bulgaria actually had a set of powerful policies in place, such as bachelor taxes and privileged access to housing for young married couples. The price tag for regulating individual life choices turned out to be rather high.)
• Microsimulation allows us to complement the models resulting from statistical analysis with detailed policy scenarios and economic accounting models. It provides a very natural tool for policy simulation, as policies are defined at the individual or micro level. This leads to applications which integrate demographic and economic modeling.

Desired features of a RiskPaths microsimulation model

Input: Parameter tables, scenarios, and simulation settings

Even being a very simple model, RiskPaths has around 130 parameter values which users should be able to set and store conveniently. We would expect these parameters to be well-organized in the microsimulation application, appearing as easy-to-access (or navigate) labelled tables which could be read or modified as required

When using a model we typically create different scenarios, i.e. different parameterizations of the model. We need to be able to save these scenarios so that certain simulations can be reproduced in future. Scenarios contain all parameter tables and, ideally, supplementary text descriptions or notes that outline the specific changes embedded in each scenario. Additionally, scenarios should include scenario settings, such as the number of simulated cases (given that RiskPaths is a case-based model), A large sample size will reduce Monte Carlo variation but comes at the cost of slower simulation runs. If we are only interested in broad aggregates, then smaller sample sizes might suffice. On the other hand, a detailed analysis of rare events or a detailed breakdowns of results (e.g. by age groups) would require large samples. Additionally, users might not wish to produce all available output. Narrowing down the desired output can again speed up simulations but also leads to a more concise and focused presentation of results according to user needs.

All of the above (parameter tables, descriptive notes, number of cases, choice of output to produce) is part of a scenario. For our RiskPaths applications, we would expect all this information to be stored together for a given scenario and we would expect it to be easily retrieved, viewed, and modified.

Output and output views

Microsimulation models can produce output on two levels: micro and macro. A microsimulation application could conceivably write all individual level characteristics and all their changes over time into a file and leave it to the user to analyse the resulting data file with statistical software. In the RiskPaths case, this would lead to a file storing the dates of all simulated events that occur over the simulated life course of each single individual. Only six events can happen in a simulated life, so each data record would contain at most six variables: four union formation / dissolution events, conception, and death. For more complex applications, file size and complexity could be enormous.

As well as such a longitudinal file, we might also be interested in cross-sectional output, recording the states of all individuals at a certain point in time. While the use of such a file is rather limited when simulating a single cohort, it would resemble a cross-sectional survey or population census in a population model.

Usually, a model user will not be interested in micro files per se but in the analysis that is performed on them. The user will typically aggregate data and produce summary indicators and tables. If model developers already know how simulated data will or should be analyzed, such measures and tables can already be calculated and produced within a microsimulation application run. In this case, users would not need to run additional statistical routines; they could see results immediately after a simulation was performed. In our RiskPaths model, output does not exceed a small number of tables and summary indicators which we expect to be produced within the application. We are interested in age-specific fertility rates, childlessness, the mean age at first conception, first conception by union status, and some mortality measures.

Just as with parameter tables, aggregated model output also requires organization. We might want to present some summary measures of one or several related behaviours together in a table and we surely want to order table output in a meaningful way. Additionally, as with parameters, we would expect table results to be labelled for easy reading and understanding.

Because all microsimulation results are subject to Monte Carlo variation, aggregated numbers are only one view of the results. We might also be interested in getting distributional information on each table value. Such information would help us to set an appropriate population size sufficient for a desired level of result precision.

A special type of micro-data output is the graphical display of individual careers. This can be a helpful feature, as it provides users with a window to the simulated individuals, and thus a way to see the operation of the statistical models. This can also be useful for model developers as it supports model debugging. Since RiskPaths is a training tool, we are interested in displaying how individual biographies result from statistical processes. Thus, besides life course events, we might also want to see how the risks of the alternative events change over time and life course situations.

User interface and documentation

So far, we have formed expectations about the content, display, and organization of model input and output data. From the user perspective, do we just have to add a start button to complete the microsimulation application? Almost all contemporary software applications contain help files. As users of microsimulation models, we should expect access to detailed online help, not only on the use of the modeling software itself but also on the model's specific elements and the interrelationships amongst those elements.

The user interface
Parameter tables
Performing a simulation run
Table output: aggregates and distributions
Model help and documentation
Graphical output of individual histories

In the remainder of this discussion we provide a quick explorative tour of the visual interface provided by Modgen for the RiskPaths model. To run RiskPaths, both the Modgen Prerequisites application and the RiskPaths executable have to be installed on your computer. As is true for all Modgen applications, RiskPaths contains a help system including documentation on the (model-independent) Modgen user interface and the actual RiskPaths model itself. Accordingly, in the description below, we concentrate on the central steps of running RiskPaths, leaving it to you to explore the model and software in depth with the assistance of the detailed help files.

The user interface

All Modgen applications have the same graphical user interface (Figure 1) which consists of the following parts:

• a menu bar and a toolbar to administer and run scenarios, as well as to get help
• a selection window containing a hierarchically grouped list of all model parameters and output tables
• a frame in which all corresponding parameters or tables can be displayed

When starting the RiskPaths.exe application, the selection window and table frame are empty, as we first have to load (or create) a simulation scenario. To do so, follow the following steps:

• Open the simulation scenario 'base.sce'. This can be done by clicking the 'Open' button or by selecting 'Open.' from the 'Scenario' menu.
• Choose the scenario settings--the settings dialog box can be accessed by clicking the 'Settings' button or by selecting 'Settings.' from the 'Scenario' menu. Specify a small number of simulated cases (e.g.10,000) so that your first model runs quickly. Also, ensure that 'MS Access tracking' is switched on. This will allow you to view individual biographies using the BioBrowser tool that comes with Modgen
• Save your scenario under a new name by selecting 'Save as.' from the 'Scenario' menu.

Parameter tables

Users of a Modgen application have control over all parameters contained in the model's parameter tables. An individual parameter table can be selected by clicking its list entry in the selection window. The table is then displayed in the display frame in which it can also be edited. Modgen parameter tables can have any number of dimensions, ranging from a parameter with a single checkbox to parameters with numerous characteristics or dimensions (e.g.region, sex, age, time).

Performing a simulation run

Click the 'Run/resume' button or select 'Run/resume' from the 'Scenario' menu. The progress of the simulation is displayed in a progress dialog box. A small sample of 10,000 actors takes around 20 seconds to run. After the model run is complete, all output tables will have been updated by Modgen.

Table output: aggregates and distributions

Simulation results are written to predefined output tables. Note that the values displayed in the output table represent only one of several possible views on the results. By right-clicking a table, a properties sheet for the table can be accessed. Among other things, this allows the display of distributional information (standard errors and the coefficient of variation) of all simulated values. Table contents can also be copied and pasted. You have the choice to copy the table as displayed, or all dimensions of the table at once (if there are more than two dimensions).

Model help and documentation

As is true with all Modgen applications, RiskPaths provides help files of various types. Two are related to Modgen itself--a general user guide for the visual interface plus release notes for Modgen. The other help files are model-specific. All Modgen applications contain a detailed encyclopaedic model documentation file. This documentation is automatically created from properly commented code.

Graphical output of individual histories

The Modgen Biography Browser (BioBrowser) application is a tool for the graphical display of individual life courses. This view on the simulation results is especially useful for model debugging. In order to use the tool, the tracking feature has to be switched on in the scenario settings. The list of variables to be tracked also has to be declared by the model developer in the model code via a tracking statement. Modgen than tracks all changes of those variables included in the tracking statement for a sample of simulated actors (where the size of this sample is specified as one of the scenario settings).

To display biographies created by RiskPaths, just start the BioBrowser application and load the tracking-file of your simulation scenario, e.g.Base(trk).mdb.