(PDF) The Influence of Landmark Judgments and Statutory Changes on the Family Litigation Explosion: A Citation Network Analysis

Canadian Journal of Family Law Canadian Journal of Family Law

Volume 34 Number 2

8-31-2022

The In;uence of Landmark Judgments and Statutory Changes on The In;uence of Landmark Judgments and Statutory Changes on

the Family Litigation Explosion: A Citation Network Analysis the Family Litigation Explosion: A Citation Network Analysis

Joseph Hickey

Lyndsay Campbell

Jörn Davidsen

Follow this and additional works at: https://commons.allard.ubc.ca/can-j-fam-l

Part of the Family Law Commons, and the Law and Society Commons

Recommended Citation Recommended Citation

Joseph Hickey, Lyndsay Campbell, and Jörn Davidsen, "The In;uence of Landmark Judgments and

Statutory Changes on the Family Litigation Explosion: A Citation Network Analysis" (2022) 34:2 Can J

Fam L 141.

The University of British Columbia (UBC) grants you a license to use this article under the Creative Commons

Attribution- NonCommercial-NoDerivatives 4.0 International (CC BY-NC-ND 4.0) licence. If you wish to use this

article or excerpts of the article for other purposes such as commercial republication, contact UBC via the

Canadian Journal of Family Law at cdnj;@interchange.ubc.ca

THE INFLUENCE OF LANDMARK

JUDGMENTS AND STATUTORY CHANGES

ON THE FAMILY LITIGATION EXPLOSION:

A C ITATION NETWORK ANALYS I S

Joseph Hickey, Lyndsay Campbell & Jörn Davidsen*

Family law in many countries has changed radically since

the 1960s. However, despite family law’s central

importance, few detailed quantitative analyses of the

relationship between legal developments (landmark

judgments and statutory changes) and the amount and

subject of family litigation have been made. We examine

this relationship using a unique dataset of citations among

Canadian family law judgments from all levels of the court

hierarchy. The network analysis draws attention to

significant changes in law and legal practice over time. Not

only did litigation increase overall, but the number of

judgments involving multiple legal issues grew

dramatically in the mid-1990s, signaling the increasing

complexity of litigation surrounding family breakdowns.

We probe this emergent co-occurrence of legal issues using

citation network analysis and find clear links to the

jurisprudential changes introduced through the landmark

1992 judgment Moge v Moge and the 1997 Federal Child

Support Guidelines.

* Joseph Hickey: Complexity Science Group, Department of Physics and

Astronomy, University of Calgary, corresponding author:

joseph.hickey@ucalgary.ca; Lyndsay Campbell: Faculties of Law and

Arts (History), University of Calgary; Jörn Davidsen: Complexity

Science Group, Department of Physics and Astronomy, University of

Calgary and Hotchkiss Brain Institute, University of Calgary.

CANADIAN JOURNAL OF FAMILY LAW [VOL. 34, 2022]

142

INTRODUCTION

In Canada, as in other western countries, family relations

and family structures have changed dramatically since the

1960s. Litigation volumes have generally increased

relative to population, families’ legal needs have

increasingly gone unmet, and family litigation has

become a major site of stress in the legal system.

1

The

underlying drivers of these developments are presumably

mainly social and economic, as individuals—especially

women, for themselves or their children—turned to the

legal system to address the consequences of family

breakdown. However, the role of legal change such as

key judgments and statutes also deserves attention. Have

changes in law aimed at supporting the interests of

women and children contributed to an increase in

litigation or a change in the type of issues being litigated?

1

Federal Judicial Center, “Federal Judicial Caseloads, 1789-2016: Trial

Court Caseloads since 1870” (no date), online: Federal Judicial Center

<www.fjc.gov/history/exhibits/graphs-and-maps/trial-court-

caseloads-1870>; Legal Services Corporation, Documenting the

Justice Gap in America: The Current Unmet Civil Legal Needs of Low-

Income Americans, 2d ed (Washington, DC: Legal Services

Corporation, 2007); Beverley McLachlin, “The Legal Profession in the

21st Century” (August 14, 2015), online: Supreme Court of Canada,

<www.scc-csc.ca/judges-juges/spe-dis/bm-2015-08-14-eng.aspx>;

Russell Engler, “Connecting Self-Representation to Civil Gideon:

What Existing Data Reveal about When Counsel Is Most Needed,”

(2010) 37:1 Fordham Urb LJ 37; Lua Kamal Yuille, “No One’s Perfect

(Not Even Close): Reevaluating Access to Justice in the United States

and Western Europe” (2004) 42:3 Colum J Transnat’l L 863; Asher

Flynn & Jacqueline Hodgson, eds, Access to Justice and Legal Aid:

Comparative Perspectives on Unmet Legal Need (Oxford, UK: Hart

Publishing, 2017).

THE INFLUENCE OF LANDMARK JUDGMENTS

143

Has the justice system responded adequately to these

jurisprudential changes?

Only a handful of studies have quantitatively

examined family litigation patterns before and after legal

change. Scholars have examined the effect of changes in

child custody laws on the number of family law cases and

the issues involved in Oregon,

2

Colorado

3

and Italy

4

; and

the effect of changes to child protection agency

intervention thresholds on the volume of child welfare

cases in Ontario.

5

The scarcity of such studies and their

2

Douglas W Allen & Margaret Brinig, “Do Joint Parenting Laws Make

Any Difference?” (2011) 8:2 J Empirical Leg Stud 304.

3

Jessica Pearson, Paul Munson & Nancy Thoennes, “Legal Change and

Child Custody Awards” (1982) 3:1 J Fam Issues 5.

4

Guido de Blasio & Daniela Vuri, “Effects of the Joint Custody Law in

Italy” (2019) 16:3 J Empirical Leg Stud 479.

5

Nico Trocmé et al, “What is Driving Increasing Child Welfare

Caseloads in Ontario? Analysis of the 1993 and 1998 Ontario

Incidence Studies” (2005) 84:3 Child Welfare 341. A substantial body

of literature explores the relationship between legal changes in family

law and social and economic outcomes. On the effect of fault vs no-

fault divorce laws on divorce rates, see Stéphane Mechoulan, “Divorce

Laws and the Structure of the American Family” (2006) 35:1 J Leg

Stud 143. On the effect of criminal sanctions, including incarceration

for failure to pay child support, on payment collection rates, see

Richard Lempert, “Empirical Research for Public Policy: With

Examples from Family Law” (2008) 5:4 J Empirical Leg Stud 907. On

the effect of changes in divorce laws on the labour force participation

of women, see Betsey Stevenson, “Divorce Law and Women’s Labor

Supply” (2008) 5:4 J Empirical Leg Stud 853. Such studies do not

generally analyze the relationship between legal changes and litigation

patterns.

CANADIAN JOURNAL OF FAMILY LAW [VOL. 34, 2022]

144

limited scope highlight the urgent need for further

quantitative studies about the relationship between legal

change and litigation patterns. One way to fill this

fundamental knowledge gap is to use data sources that give

a broad, system-wide view of family litigation patterns,

while also possessing the detail necessary for probing

beyond surface-level correlations between legal changes

and litigation frequencies. We adopt such an approach

here, using a unique citation dataset consisting of

judgments in all types of family law cases from all

Canadian common-law jurisdictions and levels of court over

several decades. We provide a high-level view of the

litigation patterns of different types of family law cases and

then drill down to examine, in more detail, how changes in

litigation patterns in particular areas of family law are inter-

connected with one another and related to key legal

developments. We apply quantitative analysis to a large

dataset to first show major changes in the amount and

complexity of Canadian family litigation that have not yet

been studied holistically, and then use our quantitative

analysis tools to search for likely reasons for these changes.

Our data is a citation network, a set of nodes and

links. The nodes represent documents—judgments—and

the links represent judges’ citations to previous cases.

Studies of legal citation networks have proliferated in

recent years, but most of these have focused only on the

highest level in the court hierarchy, such as the US

THE INFLUENCE OF LANDMARK JUDGMENTS

145

Supreme Court,

6

the European Court of Justice,

7

the

Supreme Court of India,

8

or the International Criminal

Court.

9

Some recent studies using citation data have

considered multiple levels of court hierarchies.

10

Studies by

6

James H Fowler & Sangick Jeon, “The Authority of Supreme Court

Precedent” (2008) 30:1 Social Networks 16; Michael J Bommarito II

et al “Distance Measures for Dynamic Citation Networks” (2010)

389:19 Physica A: Statistical Mechanics and its Applications 4201;

EA Leicht et al, “Large-Scale Structure of Time Evolving Citation

Networks” (2007) 59:1 European Physical J B—Condensed Matter

and Complex Systems 75; Tom S Clark & Benjamin Lauderdale

“Locating Supreme Court Opinions in Doctrine Space” (2010) 54:4

American J Political Science 871; Ryan Whalen, “Modeling Annual

Supreme Court Influence: The Role of Citation Practices and Judicial

Tenure in Determining Precedent Network Growth” in Ronaldo

Menezes, Alexandre Evsukoff & Marta C González, eds, Complex

Networks. Studies in Computational Intelligence, vol. 424 (Berlin:

Springer, 2013) 169.

7

Atieh Mirshahvalad et al, “Significant Communities in Large Sparse

Networks” (2012) 7:3 PLoS One e33721, DOI:

<doi.org/10.1371/journal.pone.0033721>; Jens Frankenreiter, “The

Politics of Citations at the ECJ—Policy Preferences of E.U. Member

State Governments and the Citation Behavior of Judges at the

European Court of Justice” (2017) 14:4 J Empirical Leg Stud 813.

8

Andrew Green & Albert H Yoon, “Triaging the Law: Developing the

Common Law on the Supreme Court of India” (2017) 14:4 J Empirical

Leg Stud 683.

9

Fabien Tarissan & Raphaëlle Nollez-Goldbach, “Temporal Properties

of Legal Decision Networks: A Case Study from the International

Criminal Court,” in Antonino Rotolo, ed, Legal Knowledge and

Information Systems: JURIX 2015: The Twenty-Eighth Annual

Conference (Amsterdam: IOS Press, 2015) 111.

10

Matthew P Hitt, “Measuring Precedent in a Judicial Hierarchy” (2016)

50:1 Law & Soc’y Rev 57; Michael J Nelson & Rachael K Hinkle

“Crafting the Law: How Opinion Content Influences Legal

CANADIAN JOURNAL OF FAMILY LAW [VOL. 34, 2022]

146

Matthew P Hitt and by Michael J Nelson & Rachael K

Hinkle use simple citation counts in regression analyses to

assess the influence of particular cases from the US

Supreme Court and appellate (federal and state) courts, but

neither study makes use of the network structure of the

data.

11

Krzysztof J Pelc presents a measure of the

importance of cases within the network of panel and

appellate body rulings in World Trade Organization

disputes and investigates its correlation with the

commercial value of the disputes, to test the hypothesis

that nations litigate in low commercial value cases in

order to secure useful precedents for future high value

cases.

12

John S Liu et al., studying a citation network of

legal decisions in trademark dilution cases from the US

Supreme Court and the Circuit Courts of Appeal, find a

network method to be capable of identifying judgments of

higher than average importance (as assessed by legal

experts) when the treatment of the citations is included in

the analysis.

13

A “vertical analysis” of citation networks, which

considers judgments at all levels of court in a specific

area of law, promises different insights about the structure

Development” (2018) 39:2 Justice System J 97; Krzysztof J Pelc, “The

Politics of Precedent in International Law: A Social Network

Application” (2014) 108:3 Am Political Sci Rev 547; John S Liu et al,

“Citations with Different Levels of Relevancy: Tracing the Main Paths

of Legal Opinions” (2014) 65:12 JASIST: J. Assoc. Information

Science & Technology 2479.

11

Hitt, supra note 10 and Nelson & Hinkle, supra note 10.

12

Pelc, supra note 10.

13

Liu et al., supra note 10.

THE INFLUENCE OF LANDMARK JUDGMENTS

147

and evolution of the law, such as by demonstrating not just

which cases are most authoritative but how particular

judgments structure the cases lawyers bring subsequently.

Our citation network consists of Canadian family law

judgments from all levels of the court hierarchy

(provincial trial-level including dedicated “family law”

courts, provincial appellate-level, and Supreme Court of

Canada (SCC)) and the citations among these judgments.

Our analysis reveals well-known changes in Canadian

law and suggests a role for legal doctrine in shaping

the dramatic increase in family litigation since the 1968

federal Divorce Act was enacted.

14

Five main time

periods emerge, shaped by landmark cases and key

legislation. In the 1990s, the volume of family litigation

increased sharply, as did judgments featuring multiple

legal topics, such as spousal support and property

division. Our analysis suggests that the SCC’s judgment

in Moge v Moge

15

in 1992 and the enactment of the Federal

Child Support Guidelines (FCSG) in 1997

16

have led to a

significant increase in judgments involving multiple legal

topics—that is, in more complex cases. In particular, we

find evidence that Moge may have led to the use of unequal

division of family property as a means of providing

financial compensation to former spouses in provinces

14

RSC 1970, c D-8.

15

Moge v Moge, [1992] 3 SCR 813, 99 DLR (4th) 456 [Moge].

16

SOR/97-175 [FCSG]. These guidelines were, and are, mandatory for

divorces, which are brought under the federal Divorce Act, but all the

provinces and territories except Quebec have generally incorporated

them into legislation that governs family breakdowns that do not

involve applications for divorce.

CANADIAN JOURNAL OF FAMILY LAW [VOL. 34, 2022]

148

where the property division legislation was sufficiently

flexible. Notably as well, following the implementation of

the FCSG, cases primarily concerned with custody and

access but that also raised financial issues such as child

support, spousal support, or division of family property

became just as common as cases in which the financial

matter was primary and custody secondary.

Interpreting new divorce legislation passed in

1986,

17

Moge revised and expanded the factors to be

considered in awarding and revising spousal support. In

doing so, Moge set out the new governing principle that the

Divorce Act was intended to promote the equitable sharing

of the economic consequences of marital breakdown,

superseding an approach that promoted a “clean break”

between divorcing spouses. The FCSG introduced a new

framework for determining child support amounts, with

mandatory quanta (varying by province), if payers’

incomes did not exceed $150,000; however, several

important factors—including the amount of time a child

spent with a parent—could justify varying these standard

amounts.

18

In section III, we examine the relationship

17

Divorce Act, 1985, SC 1986, c 4.

18

On the FCSG, see Kristen Douglas, Divorce Law in Canada (Ottawa:

Library of Parliament, Parliamentary Information and Research

Service, 2008); Carol Rogerson, “Child Support under the Guidelines

in Cases of Split and Shared Custody” (1998) 15:2 Can J Fam L 11;

Nicholas Bala, “A Report from Canada’s ‘Gender War Zone’:

Reforming the Child-Related Provisions of the Divorce Act” (1999)

16:2 Can J Fam L 163; Tina Maisonneuve, “Child Support under the

Federal and Quebec Guidelines: A Step Forward or Behind?” (1999)

16:2 Can J Fam L 284; Department of Justice Canada, Children Come

First: A Report to Parliament on the Provisions and Operation of the

THE INFLUENCE OF LANDMARK JUDGMENTS

149

between these legal changes and proliferating, increasingly

complex litigation, and discuss possible explanations that

emerge from our data. We lay the groundwork for this

analysis by describing our data and methods and then

providing, in section II, a high-level overview of what a

citation network clustering analysis shows about the

evolution of Canadian family law from 1980-2015.

I. MATERIALS AND METHODS

A. DESCRIPTION OF THE DATA

1. Source of court decisions

Our citation network data consists of judgments from the

Canadian “Cases and Decisions” database of WestlawNext

Canada (WLC). We use the names and dates of judgments,

the citations between judgments, and the legal topics in

each judgment.

The vast majority of cases in the database also have

associated with them information from two other, formerly

print, sources: the Canadian Abridgement Digests (CAD)

and Canadian Case Citations (CCC).

19

The CAD are short

Federal Child Support Guidelines (Ottawa: Department of Justice

Canada, 2002); Douglas W. Allen & Margaret F Brinig, “Child

Support Guidelines: The Good, the Bad, and the Ugly” (2011) 45:2

Family LQ 135.

19

WLC asserts that the CAD “covers virtually every case reported in

Canada since 1803, and every unreported case received from the courts

since 1986”, except Quebec civil law cases: “A Short Guide to the

Canadian Abridgment in Print and on WestlawNext Canada” (Toronto:

Carswell, 2014). According to WLC, the online version of the CCC

CANADIAN JOURNAL OF FAMILY LAW [VOL. 34, 2022]

150

summaries of judgments, organized hierarchically by topic

and sub-topic. Family law (“FAM”) has twenty highest-

level topics, including custody and access, division of

family property, and child protection, each of which is

broken down into sub-topics, sub-sub-topics, and so on.

“FAM.III.3.c.i”, for example, means “Family law; III.

Division of family property; 3. What constitutes property;

c. Pension benefits; i. Income from pension.” WLC also

contains the CAD as a separate source, so that a researcher

interested in income from pensions, for example, can

readily find the relevant law.

20

The CCC information provides the case’s

(“KeyCite Canada”) contains the full history of all Canadian reported

cases back to 1867, plus judicial treatments of cases that fall within the

following coverage: selected reported cases before 1977, all reported

cases since 1977, and all unreported cases since 1986 (“KeyCite

Canada,” (2010), online (pdf): WestlawNext Canada,

<westlawnextcanada.com/dynamicdata/attacheddocs/lawsource/laws

ource_keycitecanada.pdf>. Spot-checking suggests that the coverage

in WLC is substantially comprehensive, although perhaps exaggerated

with respect to the decisions of the lowest levels of court. We do not

have a means of determining how many family law judgments are

omitted from our dataset due to being absent from the CAD. However,

a comparison against court statistics (see section II.A) shows that our

dataset captures the broad movements in number of judgments over

our period of interest (1980-2015). Thus, while our data does not cover

the full universe of Canadian family law decisions, it is sufficient for

the purpose of investigating the relationship between legal

developments and the amount and subject of family litigation.

20

As the CAD summaries were written by legal editors, the topics in our

dataset were assigned by humans and not by an algorithm. The

available old paper editions of the CAD reveal that the CAD

classification scheme underwent little change from 1990 to 2018,

which covers the period of focus of our analysis in section III.

THE INFLUENCE OF LANDMARK JUDGMENTS

151

i

subsequent legal “treatment”, whether it was “Followed”,

“Considered”, “Referred to”, “Distinguished”, or “Not

Followed” by each case that cited it. For a given judgment

in the WLC database, the topic information comes from the

CAD, and the treatment information comes from the CCC.

2. Court decisions used in this study

Our dataset consists of all (59,514) of the judgments in the

Canadian “Cases and Decisions” database that have at least

one lowest-level CAD topic belonging under the heading

“Family law (FAM),” plus the citations among all such

judgments, up to December 31, 2015. Having found all

these judgments, we obtained the citations among them by

searching for every judgment in the set that had been cited

by at least one other judgment in the set. In the citation

network we constructed from this data, the nodes represent

judgments and the links represent citations from newer to

older judgments.

Fig. 1 shows a small citation network. The number

of links (citations) that each node (judgment) receives from

later nodes is called the node’s “in-degree,”

k

iin

,

where the

subscript i indicates a particular node, i. The number of

links that a node makes to older nodes is the node’s “out-

degree,”

k

iout

.

The total of a node’s in-degree and out-

degree is its degree,

k

i

=

k

iin

+k

iout

.

[…]

CANADIAN JOURNAL OF FAMILY LAW [VOL. 34, 2022]

152

Figure 1: Schematic example of a citation network

Note: Example of a citation network with eight nodes and ten links. All

links are directed toward the past. The node labelled “4” has degree

k

4

= 5

, in-degree

k

4in

= 5

, and out-degree k4out = 3.

16,422 of the 59,514 judgments in our dataset have

no citation links to other family law judgments (they have

degree k = 0). Our citation network therefore contains

43,092 nodes with k > 0. Table 1 shows the proportion of

judgments issued by the courts of each province, plus the

three territories and the Federal Court of Canada, in: i) our

full dataset (k ≥ 0) of 59,514 judgments (third column of

table 1); and ii) our citation network (k > 0) with 43,092

judgments (fourth column of table 1). Quebec is poorly

covered in our data. Otherwise, the ordering of the other

nine provinces in table 1 follows their ordering by

decreasing population, except that Manitoba has a slightly

THE INFLUENCE OF LANDMARK JUDGMENTS

153

larger population than Saskatchewan.

21

Table 1: Percentage of Judgments by Region

Region

Abbrev.

Citation

network

(k >0)

Full

dataset

(k ≥ 0)

Region’s % of

non-QC Cdn

pop’n in 2001

Ontario

ON

32%

34%

50%

British Columbia

BC

24%

23%

17%

Alberta

AB

10%

13%

Saskatchewan

SK

9%

4%

Nova Scotia

NS

8%

7%

4%

Manitoba

MB

5%

New Brunswick

NB

5%

3%

Newfoundland &

Labrador

NL

3%

2%

Prince Edward

Island

PE

1%

Quebec

QC

1%

-

Territories

TT

1%

<1%

Federal court

FE

<1%

-

The citation network contains 150,748 links

between pairs of nodes. Each link has one of the following

“treatment” values, which indicates how the later judge

treated the earlier case: “Followed” (19% of links),

21

For population data, see “Population estimates, quarterly,” (June 22,

2022), online: Statistics Canada

<https://doi.org/10.25318/1710000901-eng>.

CANADIAN JOURNAL OF FAMILY LAW [VOL. 34, 2022]

154

“Considered” (45%), “Referred to” (32%),

“Distinguished” (3%), or “Not Followed” (0.5%). We refer

to these treatment values as F, C, R, D, and N. When, for

example, we exclude the “Distinguished” (D) and “Not

Followed” (N) links from our analysis, we use the label

“FCR” to specify the links we are using. We refer to the F

links as “positive”, the C and R links as “neutral”, and the

D and N links as “negative”.

B. CLUSTERING METHODOLOGY FOR

CITATION NETWORKS

This study uses two “clustering” methodologies to explore

the citation network. Clustering methods find groups—

clusters—of nodes that are similar to each other.

22

In this

study, we apply two different clustering methods.

23

The

first groups nodes into clusters based on their temporal

22

Santo Fortunato & Darko Hric, “Community Detection in Networks:

A User Guide” (2016) 659:11 Physics Reports 1.

23

The PhD thesis Joseph Hickey, A Complex Systems Study of Social

Hierarchies and Jurisprudence (PhD Dissertation, University of

Calgary, 2019) [unpublished] contains extensive mathematical

descriptions of the two clustering techniques, specification of the

parameter values used where applicable, and demonstrations of the

robustness of the clustering results as applied to our data. In applying

the clustering methods, we do not distinguish between the different

types of citations (positive, neutral, and negative) discussed in section

I.A—in our research applying the clustering methods to this and other

datasets, we did not find a significant advantage to distinguishing

between the different types of citations, and found the results to be

largely robust even when excluding or changing the relative weights of

certain types of citations. However, the different types of citation links

are used in some of our analyses that probe the temporal clusters that

emerge from our data, section II.

THE INFLUENCE OF LANDMARK JUDGMENTS

155

citation behaviour, i.e. the year(s) when the cases they cite

were decided. This method illuminates how legal issues

and authoritative judgments rise and fall in prominence

over time. The second method groups nodes together based

on how closely linked they are by their citations; that is, it

identifies cases that have similar topics by recognizing

their shared jurisprudential lineages and connections. This

methodology permits us to examine the evolution of sub-

areas of law over time.

E.A. Leicht et al developed the temporal clustering

method and applied it to a legal citation network of US

Supreme Court judgments.

24

In this method, each cluster is

defined in terms of the probability that a node belonging to

the cluster links to a node issued in a particular year in the

past. For example, judgments issued in the 1980s that only

cite past judgments issued in the 1970s would tend to be

grouped into one cluster, and judgments issued in the 1990s

that only cite judgments from the 1980s would be grouped

into a different cluster. The clusters can therefore

correspond to different time periods (or “epochs”) in the

evolution of the citation network.

25

24

Leicht et al, supra note 6.

25

Ibid. Technically, in applying this method, one infers the parameters of

a statistical model of the probability that a node belongs to a given

cluster given its pattern of citations to nodes issued in particular past

years. Given a specified number of clusters, the inference procedure

produces two sets of parameters: the share of nodes belonging to each

cluster, and a “citation profile” for each cluster, which is a function

equal to the probability that a citation made by a node belonging to a

particular group extends to a node issued in a particular past year. This

“citation profile” is plotted in fig. 4C and discussed in Section II.C.

The statistical inference is done using the “expectation-maximization”

algorithm. In our study, the choice of five clusters was not arbitrary:

CANADIAN JOURNAL OF FAMILY LAW [VOL. 34, 2022]

156

The topical, or thematic, clustering method was

developed by Martin Rosvall and Carl T Bergstrom.

26

Conceptually, the method is similar to file compression

techniques where one tries to represent a set or a stream of

data in a form that requires less information. This method

rests on the fact that judgments on related topics cite

common precedents: a judgment can be imagined as

embedded in a relatively dense set of connections to other

judgments on one or more similar topics. One can imagine

a person walking from node to node (judgment to

judgment) on the citation network by randomly following

the links between nodes. The nodes are then grouped into

clusters so as to minimize the amount of information

required to describe the path of this “random walker”. The

clusters obtained by this procedure can be quite distinct for

citation networks, as recent studies have demonstrated for

both scientific and judicial citation networks.

27

We make no assumptions about what level of court

we ran the algorithm assuming each of 2, 3, 4, 5, and 6 clusters, and

then used the number of clusters (5) that maximizes a statistical

quantity (the “expected log-likelihood”) central to the inference

method. Further technical details, including a description of the

expectation-maximization procedure, are available in Hickey, supra

note 23.

26

Martin Rosvall & Carl T Bergstrom, “Maps of Random Walks on

Complex Networks Reveal Community Structure” (2008) 105:4

Proceedings of the National Academy of Sciences 1118.

27

See Lovro Šubelj, Nees Jan van Eck & Ludo Waltman, “Clustering

Scientific Publications Based on Citation Relations: A Systematic

Comparison of Different Methods” (2016) 11:4 PLoS ONE e0154404;

Mirshahvalad et al, supra note 7.

THE INFLUENCE OF LANDMARK JUDGMENTS

157

may produce the most influential judgments, so we apply

both clustering methods to the entire dataset, irrespective

of the judgment’s level in the court hierarchy.

Unsurprisingly, though, judgments of higher-level courts

are often both highly-cited and highly-connected (or

“central”) according to network measures,

28

such that these

judgments have a large influence on the structure of the

clusters that emerge in our analysis, and they feature

prominently in the following sections.

II. EVOLUTION OF CANADIAN FAMILY LAW,

1980-2015

This section examines quantitatively the high-level

features of the evolution of Canadian family law over

recent decades. We begin by showing the disproportionate

rise in family litigation, relative to population growth, in

the 1990s. We then show how the preeminence of

particular legal topics in family law have changed over

time, with cases involving multiple legal topics becoming

much more common beginning in the mid-1990s. To

position these changes in the broad evolution of Canadian

28

Network centrality measures are mainly used to rank the importance of

judgments. Applied to our dataset, we find a strong correlation between

rankings obtained using several different network centrality measures,

such that we focus on the simplest such measure—the total number of

citations received by a judgment, kin (also called the “in-degree

centrality”)—when we wish to rank-order judgments (as in table 2).

Since clustering is based on grouping together (rather than rank-

ordering) judgments, it can give information about broad or system-

level patterns in a citation network that would not be observed by

considering centralities alone. An analysis of the network centralities

of the judgments in the dataset presented in this article is contained in

Hickey, supra note 23.

CANADIAN JOURNAL OF FAMILY LAW [VOL. 34, 2022]

158

family law, we apply temporal citation network clustering,

which identifies five “epochs” and highlights the

importance of the two key legal developments (Moge and

the FCSG) that we examine in detail in section III.

A. INCREASE IN LITIGATION VOLUME

Canadian commentators have identified a crisis of access

to justice, often connecting it to the rise of self-represented

litigants.

29

Information gathered by Statistics Canada

suggests that the issue is not an increase in divorce and

family breakdown but instead an increase in prolonged

litigation with repeated applications to court, which are

especially prevalent in cases involving children and child

support.

30

Most Canadian courts do not publish records of

litigation volume, but fortunately the Provincial Court of

British Columbia has reported the number of new family

29

McLachlin, supra note 1; John-Paul Boyd, “‘Family Justice in Canada

Is at a Breaking Point’ Redux” (8 March 2019), online: Slaw

<www.slaw.ca/2019/03/08/family-justice-in-canada-is-at-a-breaking-

point-redux/>; Julie Macfarlane, “The National Self-Represented

Litigants Project: Identifying and Meeting the Needs of Self-

Represented Litigants: Final Report” (May 2013), online (pdf):

National Self-Represented Litigants Project (NSRLP)

<representingyourselfcanada.com/wp-content/uploads/2015/07/nsrlp-

srl-research-study-final-report.pdf>.

30

Mary Bess Kelly, “Family Court Cases Involving Child Custody,

Access and Support Arrangements, 2009/2010” (29 March 2011),

online: Statistics Canada Juristat <www150.statcan.gc.ca/n1/pub/85-

002-x/2011001/article/11423-eng.htm>; Mary Allen, “Family Law

Cases in the Civil Courts, 2012/2013” (28 April 2014), online:

Statistics Canada Juristat < https://www150.statcan.gc.ca/n1/pub/85-

002-x/2014001/article/13005-eng.htm>.

THE INFLUENCE OF LANDMARK JUDGMENTS

159

law cases and applications covering the full range of family

law litigation (including applications for orders for

divisions of family property, child support, spousal

support, custody and access, child protection, and other

issues) received by its courts over most of our time period.

This is shown in fig. 2.

Figure 2: Number of new family law cases and

applications in the BC Provincial Court

Note: Number of new family law cases and applications (including

“subsequent applications” in cases initiated in a preceding year) in

the BC Provincial Court, as reported by the court

31

or the BC Ministry

31

Provincial Court of British Columbia, “Annual Reports” at 2000-2001,

2002-2003, 2005-2006, 2008-2009, 2010-2011, 2011-2012, and 2015-

2016, online: Provincial Court of British Columbia Court Reports

CANADIAN JOURNAL OF FAMILY LAW [VOL. 34, 2022]

160

of the Attorney General (MAG).

32

Data points connected by line

segments come from the same annual report. Data points for “BC Prov

Crt 2002/03” are covered by “MAG 2002/03.” Discrepancies between

data from different annual reports arise from delays in data entry or

changes to the method of extracting data.

33

Fig. 2 shows a dramatic increase in the number of

new cases and applications brought annually, beginning

about 1994 and peaking about 2008. This increase—from

about 17,500 to about 42,500 per year, or about 250%—

outstrips BC’s population growth of about 130% from

1991 to 2006.

34

A similar increase is seen in our data, as

shown in fig. 4a, further below.

B. EMERGENCE OF MULTI-TOPIC JUDGMENTS

IN THE 1990S

In addition to the increase in the amount of litigation

occurring in the 1990s, our data also reveal the rapid

emergence of more complex, multi-topic judgments

(judgments with two or more legal topics) in the 1990s.

<www.provincialcourt.bc.ca/news-reports/court-reports>.

[www.provincialcourt.bc.ca/Archive].

32

British Columbia Ministry of the Attorney General, “Annual Report”

(Victoria, BC: Ministry of the Attorney General, 1987-1992, 2002,

2003).

33

See Provincial Court of British Columbia, “Annual Report 2011-

2012”, supra note 31 at appendix 3.

34

Government of British Columbia, “2016 Census: Population and

Dwelling Counts” (8 February 2017), online: BC Stats

<www2.gov.bc.ca/assets/gov/data/statistics/people-population-

community/population/pop_census_2016_highlights_population_dwe

llings.pdf>.

THE INFLUENCE OF LANDMARK JUDGMENTS

161

The judgments in our data bear at least one topic

from the CAD classification scheme. Fig. 3 shows the

twenty most frequently-appearing topics in our citation

network. The grey curves show the total number of

judgments with each topic in each year. The solid black

curves show the number of judgments with only that topic,

and the dashed black curves show the number with that

topic and at least one other (the multi-topic judgments). In

each panel, the solid and dashed black curves add up to the

grey curve.

Panels (fig. 3a-g) show the change in the seven

most frequently-appearing topics from 1970 to 2015: child

support, spousal support, custody, family property, costs,

child protection and domestic settlements. Before the mid-

1990s, when a judgment involved the topic of child

support, spousal support, custody, family property, or

domestic settlements, that topic tended to be the only one

in the case; however, after the mid-1990s, these topics

began more often to co-occur with one another and with

other topics (in fig. 6a-d and g, the dashed and solid black

curves cross).

35

The rise of multi-topic judgments occurred

simultaneously with the overall increase in family

litigation. This is a key feature of the data that we examine

in more detail in section III. The per-year numbers of

judgments with less frequently-occurring topics (fig. 3h-p)

remained steady or decreased over time, with the exception

of restraining orders (fig. 3m) and relationship of parent

35

The emergence of multi-topic judgments is not mainly due to an

increase in judgments about costs. Of all judgments with precisely two

topics, only 12.5% have costs as one of the topics.

CANADIAN JOURNAL OF FAMILY LAW [VOL. 34, 2022]

162

and child (fig. 3n), which increased.

36

Figure 3: Frequencies of legal topics in family law over

time

Note: The 20 most frequently occurring legal topics in the citation

network. The x-axes show time, from 1970-2015. The y-axes show the

number of judgments: associated with the indicated topic (grey curve);

associated with the indicated topic and no other topic (solid black

36

Hickey, supra note 23 discusses other changes in family litigation

revealed in fig. 3.

THE INFLUENCE OF LANDMARK JUDGMENTS

163

curve); associated with the indicated topic and at least one other topic

(dashed black curve). The highest level of the CAD classification

contains 20 topics, one of which, “support,” is sub-divided into sub-

topics for spousal support and child support. We consider spousal

support and child support to be highest-level topics, giving 21 topics

in total. The least frequently occurring topic, “habeas corpus involving

children,” is not shown but is included in our analysis.

C. TEMPORAL “EPOCHS” AND CITATION

FREQUENCIES OF LANDMARK JUDGMENTS

Temporal clustering of the citation network reveals five

epochs characterized by citations to thirteen SCC

judgments, listed in table 2, including the ten most highly

cited judgments in our data date from 1987 to 2006. The

other three lie outside of this time window: the most cited

post-2006 judgment and the two most cited pre-1987

judgments. We call these thirteen judgments landmarks.

Fig. 4b shows the five clusters that emerge.

37

As

can be seen, each cluster dominates for several years, in

that over half of the judgments issued in that time period

belong to it. After a while, a new dominant cluster emerges.

The five epochs are the five time periods in which the

majority of judgments belong to a particular cluster.

Fig. 4c shows the statistical quantity (the “citation

37

The number of clusters (five) was chosen using a quantitative criterion

(maximization of the expected log-likelihood of the statistical model)

and confirmed by several robustness checks. In particular, the main

divisions between temporal epochs observed in the statistically optimal

clustering (five clusters) are maintained even when using the method

to extract four or six clusters. These points are described in detail in

Hickey, supra note 23.

CANADIAN JOURNAL OF FAMILY LAW [VOL. 34, 2022]

164

profile”) that is primarily responsible for defining the

clusters. A peak in a cluster’s curve indicates that the

judgments in the cluster frequently cite one or more

judgments issued in the year of the peak. For example, the

peak at 1999 in the curve for the fourth (star markers)

cluster indicates that judgments in cluster 4 often cite one

or more judgments from 1999, in that case Bracklow.

38

Similarly, the peak at 2006 for the fifth cluster (circle

markers) indicates that judgments in cluster 5 frequently

cite at least one judgment from 2006, in fact D.B.S.

39

The

peaks in fig. 4c are earlier than the maximum values for the

corresponding clusters in fig. 4b because the citations

accumulate after a judgment is issued. The five epochs that

emerge from the temporal clustering therefore reflect

widespread judicial attention to a judgment or a group of

judgments issued in a certain, earlier year.

The temporal clustering method does not directly

identify key statutory changes, but transitions between

clusters often emerge from such developments when

judgments interpreting new legislation become landmarks.

The first epoch ends with the enactment of various

provincial matrimonial property statutes in the later 1970s,

which enhanced wives’ rights to property when the

marriage broke down and superseded previous common-

law approaches.

40

The shift from cluster 2 to 3 followed

major changes in the Divorce Act of 1986, the first since

38

Bracklow v Bracklow, [1999] 1 SCR 420, 1 DLR (4th) 577 [Bracklow].

39

DBS v SRG, 2006

SCC

37, [2006] 2 SCR 231 [DBS].

40

Julien D Payne & Marilyn A Payne, Canadian Family Law, 3rd ed

(Toronto: Irwin Law, 2008) at 564.

THE INFLUENCE OF LANDMARK JUDGMENTS

165

1968. The Federal Child Support Guidelines (FCSG) were

implemented in 1997, which marks the transition from

cluster 3 to 4. Amendments to the FCSG effective in 2006

precede the transition from cluster 4 to 5.

41

[…]

41

While the landmark judgments most cited by each cluster (peaks in fig.

4c) occur after the important statutory change, the judgments that

belong to a given cluster will also often cite judgments that pre-date

the change. For instance, the y-axis values in fig. 4c for cluster 4 (star

markers) and cluster 5 (circle markers) are greater than zero for 1992-

1996, before the FCSG were implemented: judgments from these years

are still cited by judgments belonging to clusters 4 and 5, although not

as frequently as the judgments responsible for the peaks in fig. 4c for

those clusters.

CANADIAN JOURNAL OF FAMILY LAW [VOL. 34, 2022]

166

Figure 4: Number of family law judgments per year

and five temporal epochs

Note: a) Number of family law judgments issued per year: regardless

of whether the judgment makes or receives a citation (solid black

curve) (k ≥ 0); that make or receive at least one citation (grey curve)

(k > 0). b) The fraction of judgments belonging to each of the five

temporal clusters, in each year, t (for any t, the five y-axis values sum

to 1). c) The probability,

Θ

r

(t),

that a judgment belonging to cluster r

(where r ranges from 1 to 5) makes a citation to a judgment issued in

year t.

Θ

r

(t)

is normalized such that it sums to 1 for a given cluster, r,

across all years t, i.e.,

!

"

!

#

$

%

& '

"

!"#

"

!$%

.

Peaks in a given coloured

THE INFLUENCE OF LANDMARK JUDGMENTS

167

curve in panel (c) indicate that the judgments belonging to the

corresponding cluster (depicted in panel (b)) have a high probability

of citing one or more judgments that were issued in the year of the

peak. Dashed vertical lines in all panels indicate years of significant

legislation. 1978 saw the beginning of a two-year wave of new

provincial matrimonial property acts. The federal Divorce Act was

overhauled in 1986. The FCSG came into effect in 1997 and were

modified in 2006.

[…]

CANADIAN JOURNAL OF FAMILY LAW [VOL. 34, 2022]

168

Figure 5: Citations received by landmarks from

judgments in temporal clusters 2 to 5

Note: Each panel shows how often each of the thirteen landmark SCC

judgments listed in table 2 is cited by the judgments belonging to a

particular temporal cluster. Vertical bars show the number of

“Followed” citations; e.g., in (a) Pettkus (positioned at 1980 on the x-

axis) received approximately 100 “Followed” citations from

judgments in temporal cluster 2. Closed circles indicate the number of

“Distinguished” or “Not Followed” citations, and open circles the

number of “Considered” or “Referred to” citations (right y-axis). Of

the two 1987 landmarks, Pelech is on the left and Richardson on the

THE INFLUENCE OF LANDMARK JUDGMENTS

169

right, and of the two 1999 landmarks, Bracklow is on the left and

Hickey on the right.

Figure 6: Scaled citation frequencies of landmark

judgments

Note: Number of non-“negative” (FCR) citations received per year t

by the thirteen landmark judgments listed in table 2, scaled by the total

number of FCR citations made by all judgments issued in year t. The

y-axis therefore represents the “share” of FCR citations made by all

judgments in year t that went to a given landmark. This scaling allows

a comparison of landmark citation frequency behaviour over several

decades during which the number of nodes (judgments) and the

number of links (citations) in the citation network increased.

Landmarks are listed in chronological order in the legend. Dashed

vertical lines indicate the same changes in family legislation as in fig.

4. A curve’s colour indicates the cluster in fig. 5 that is most strongly

characterized by the landmark.

Temporal clusters tend to be determined not by one

landmark judgment but by a certain combination of them.

Certain legal issues and ways of approaching them define

each epoch. Fig. 5 shows the salience of different sets of

CANADIAN JOURNAL OF FAMILY LAW [VOL. 34, 2022]

170

judgments in the four epochs from 1980 to 2015, in terms

of the number of times the landmark SCC judgments are

cited by the judgments in clusters 2 through 5.

42

Cluster 2,

which spans most of the 1980s, features references to

Pettkus v Becker (1980) and Rathwell v Rathwell (1978),

cases that addressed the division of property.

43

The cases

Pelech v Pelech (1987) and Richardson v Richardson

(1987) are often cited by judgments in cluster 3, but also

receive a significant number of “negative”

(“Distinguished” and “Not Followed”) citations from

judgments in this cluster.

44

These cases articulated a “clean

break” approach to spousal support that was initially

influential but was superseded five years later by the

dominant case in cluster 3, Moge v Moge (1992), which

reinterpreted the principles behind spousal support in light

of the new 1986 Divorce Act.

45

The second-most cited case

in cluster 3, Gordon v Goertz (1996), concerned a parent’s

wish to change an order for custody and access in order to

move to Australia with the child. The epoch corresponding

to cluster 4 (spanning most of the 2000s) shows the

continued importance of Gordon, the persistence of

references to Moge, and the appearance of Hickey, which

evaluated revising intertwined orders for child and spousal

42

Cluster 1 predates our period of interest, from 1980 to 2015.

43

Pettkus v Becker, [1980] 2 SCR 834, 117 DLR (3d) 257 [Pettkus];

Rathwell v Rathwell, [1978] 2 SCR 436, 83 DLR (3d) 289 [Rathwell].

44

Pelech v Pelech, [1987] 1 SCR 801, (1987) 38 DLR (4th) 641 [Pelech];

Richardson v Richardson, [1987] 1 SCR 857, 38 DLR (4th) 699

[Richardson].

45

Moge, supra note 15.

THE INFLUENCE OF LANDMARK JUDGMENTS

171

support.

46

The fifth epoch featured the 2006 landmark

DBS, which concerned retroactive orders for child support.

The open circles in fig. 5d—for “Considered” and

“Referred to”—stayed high for Moge and Hickey: even

when courts were not actually following these cases they

still referred to them.

[…]

46

Hickey v Hickey, [1999] 2 SCR 518, 172 DLR (4th) 577 [Hickey].

CANADIAN JOURNAL OF FAMILY LAW [VOL. 34, 2022]

172

Table 2: Set of Thirteen SCC Landmark Judgments

Landmark

Citation

F

C

R

D

N

Total

Moge

[1992] 3

SCR 813

937

987

516

19

0

2459

Gordon

[1996] 2

SCR 27

1428

441

265

16

1

2151

Bracklow

[1999] 1

SCR 420

704

571

376

16

0

1667

DBS

2006 SCC

37

1062

328

200

8

0

1598

Willick

[1994] 3

SCR 670

307

357

242

4

0

910

Young

[1993] 4

SCR 3

215

374

237

6

0

832

Pelech

[1987] 1

SCR 801

183

338

104

59

0

684

Hickey

[1999] 2

SCR 518

188

223

226

0

637

Richardson

[1987] 1

SCR 857

137

316

83

40

5

581

Miglin

2003 SCC

24

304

152

103

17

0

576

THE INFLUENCE OF LANDMARK JUDGMENTS

173

Pettkus

[1980] 2

SCR 834

196

199

92

21

0

508

Kerr

2011 SCC

10

169

89

50

2

0

310

Rathwell

[1978] 2

SCR 436

88

140

37

8

0

273

Note: The thirteen landmark judgments consisting of the ten most cited

judgments in the citation network, plus the two most cited pre-1987

judgments (Rathwell and Pettkus) and the most cited post-2006

judgment (Kerr). The columns F, C, R, D, and N indicate the number

of “Followed”, “Considered”, “Referred to”, “Distinguished”, and

“Not Followed” citations received by each judgment. The last column

shows the total number of citations received by each judgment.

Fig. 6 shows patterns in how frequently the

landmark judgments were cited favourably or neutrally

(FCR citations). Many of the landmark cases were initially

cited frequently, with citations dropping off as new

landmarks took their places. Pelech and Richardson had

short-lived high impact, while Moge had a longer-lived

high impact. Bracklow had a longer lasting but lower

impact, similar to Pettkus. Other judgments that did not

peak and decline, such as Young and Gordon, have a longer

duration of impact. A quantitative comparison of the level

and duration of impact of the landmarks is done by fitting

an exponentially decaying function

y

=

Ae−t/τ

to the

curves in fig. 6. Table 3 lists the seven landmarks for which

a reasonable fit of this function can be obtained, along with

the corresponding values of the parameter

τ.

This

parameter quantifies the rate of decline of the landmark’s

citation frequency, where a larger value of

τ

corresponds

CANADIAN JOURNAL OF FAMILY LAW [VOL. 34, 2022]

174

to a slower decline. Judgments with a large peak value in

fig. 6 therefore have a high level of impact, and judgments

with a large value of

τ

have a long duration of impact.

47

Table 3: Decay Constants of Landmark Citation

Frequencies

Landmark τ

Rathwell (1978) 9.1 ± 0.8

Pettkus (1980) 11.5 ± 0.8

Pelech (1987) 4.4 ± 0.2

Richardson (1987) 6.7 ± 0.3

Moge (1992) 12.8 ± 0.7

Bracklow (1999) 23 ± 2

DBS (2006) 14 ± 4

Note: Values of τ obtained from least-squares fits (Levenberg-

Marquardt algorithm)

48

of the function ln(y) = ln(A) − t/τ, for seven

landmarks with decaying citation frequencies. Error values are equal

to the square-root of the variance estimate for the fitted parameter τ.

Figs. 5 and 6 point to important transitions in

Canadian family law. The first epoch ends at the point that

provincial legislation provided remedies that made

47

Although the citation frequency for Willick has a peak and decline, it

is not included in table 3 because the exponential function does not fit

well. The citation frequency decreases rapidly but then levels out as of

2002, whereas an exponential function would continue to decay after

2002.

48

William H Press et al, Numerical Recipes: The Art of Scientific

Computation, 3rd ed (Cambridge, UK: Cambridge University Press,

2007).

THE INFLUENCE OF LANDMARK JUDGMENTS

175

previous common law approaches less relevant. Murdoch

v Murdoch,

49

Rathwell, and Pettkus drew attention to the

legal vulnerability of wives whose names did not appear on

land titles and raised the spectre of a proliferation of trust-

based litigation. New provincial legislation rendered this

type of litigation less important for legally married

spouses, but the doctrines of resulting and constructive

trust remained relevant for legally unmarried couples.

The transition from the second to the third epoch

begins with the overhaul of the federal Divorce Act in 1986.

The first SCC judgments following the 1986 changes were

the Pelech trilogy issued in 1987: Pelech, Richardson, and

Caron.

50

In these cases, the Court adopted a “clean break”

approach to spousal support claims, refusing to reopen

support agreements that did not provide for long-term

support unless a radical change in circumstances had taken

place. Initially highly cited, these cases were largely

superseded by Moge v Moge (1992). Moge marks a turning

point in Canadian family law, as it not only revised the law

on spousal support but established the broader principle

that the Divorce Act was intended to promote the equitable

sharing of the economic consequences of marriage and

marital breakdown. The transition from the second to the

third epoch is therefore marked by the short-lived but high-

impact Pelech trilogy cases followed by the spike in

citations to Moge, which had equally high impact but a

longer duration and dominates the third epoch as shown in

fig. 5. Moge’s longevity is evident in its high decay

49

[1975] 1 SCR 423, 41 DLR (3d) 367.

50

Caron v Caron, [1987] 1 SCR 892, (1987) 38 DLR (4th) 735 [Caron].

CANADIAN JOURNAL OF FAMILY LAW [VOL. 34, 2022]

176

constant, τ (see table 3).

Citations to Willick (1994) and Gordon (1996) also

characterize the judgments in temporal cluster 3. Fig. 6

shows that Willick, like Moge, peaked and then tapered off.

Moge, however, has remained highly cited: the open circles

in figs. 5c and d show that it continued to be “considered”

and “referred to,” even when judges were more

conspicuously “following” other cases. While principles

articulated in Willick—an application to vary child

support—continued to be relevant in spousal support

variation cases, Willick undoubtedly became less important

with the passage of the FCSG in 1997, hence its rapid

decay from its peak in fig. 6.

51

The fourth temporal epoch, in the 2000s, is marked

by the persistence of Moge, references to Gordon (SCC,

1996), the appearances of Bracklow

52

(fig. 5c taller bar)

and Hickey (shorter bar) in 1999, and the beginning of the

rise in citations of D.B.S. Bracklow did not constitute a

transition point because it elaborated on the Moge

principles in the context of a spouse who had become

unemployable due to illness. About 60% of judgments that

cite Bracklow also cite Moge.

53

Hickey (1999), likewise,

51

Willick v Willick, [1994] 3 SCR 670, 119 DLR (4th) 405 [Willick]. See

also fig. 5.

52

Bracklow v Bracklow, [1999] 1 SCR 420, 1 DLR (4th) 577 [Bracklow].

53

Carol Rogerson predicted that because Bracklow offered no particular

test but made spousal support extremely discretionary, it was unlikely

to be as influential as Moge or Pelech. This hypothesis is borne out in

our data. See Carol Rogerson, “Spousal Support Post-Bracklow: The

THE INFLUENCE OF LANDMARK JUDGMENTS

177

followed Moge and Bracklow in speaking of the equitable

division of the economic consequences of marriage, this

time in the context of an application to vary both spousal

and child support in the face of inflation and changes in the

children’s needs. Hickey’s ruling that inflation constitutes

a significant change in circumstances for the purposes of

reassessing support awards has evidently remained

consistently important, but in a limited way.

As fig. 6 shows, Gordon (1996) and Young (1993)

maintain fairly constant scaled citation frequencies over

time. Both were difficult, fact-driven cases concerning

changes to custody and access orders. Gordon concerned

moving a child to Australia. In Young, the children and

their mother, the custodial parent, objected to the father’s

religious instruction.

54

Young appears as a lesser landmark

in the data (see fig. 5). Citations to Gordon have been

frequent but have not risen or fallen dramatically:

apparently it reflected an increasingly significant problem

for former spouses and their children. Evidently the

judicial reasoning in Young and Gordon, as in Hickey, had

not been superseded by case law or legislation by 2015,

when our dataset ended. The beginning of the fourth epoch,

then, is defined most significantly by the appearance of the

FCSG

.

The case whose appearance, in fig. 6, coincides

with the beginning of the fifth epoch is DBS, issued in

2006. This judgment decided four appeals pertaining to

Pendulum Swings Again?” (2001-02) 19 Can Fam LQ 185.

54

Young v Young, [1993] 4 SCR 3, 108 DLR (4th) 193 [Young].

CANADIAN JOURNAL OF FAMILY LAW [VOL. 34, 2022]

178

retroactive orders for child support pursuant to the FCSG

or the new provincial guidelines passed in their wake. The

SCC specified the principles governing retroactive child

support orders and differences between the federal and

provincial legislation. The combination of the release of

DBS and modifications to the FCGS accounts for the

transition to the fifth epoch.

55

III. CHANGES IN LAW AND THE 1990S FAMILY

LITIGATION SURGE

In the third epoch, both the volume of family litigation and

the number of judgments with multiple legal topics

increased rapidly. The 1992 SCC landmark Moge and the

1997 Federal Child Support Guidelines emerge as key. In

this section we examine the changes in family law that

underlay the temporal clusters to which this methodology

draws attention.

A. MOGE’S INFLUENCE: PROVINCIAL

DIFFERENCES AND CONNECTIONS BETWEEN

SPOUSAL SUPPORT AND FAMILY PROPERTY

Moge is our citation network’s most significant landmark:

it is the most highly cited judgment (see table 2), it has

significant impact as measured by its peak in fig. 6, long

55

SOR/2005-400, published December 14, 2005 and effective May 1,

2006. These amendments modified the FCSG table amounts, the

definition of “extraordinary expenses”, the manner of comparing

household incomes under the undue hardship provisions, and the

method of determining the incomes of foreign residents. See Payne &

Payne, supra note 40 at xxv.

THE INFLUENCE OF LANDMARK JUDGMENTS

179

duration of impact as measured by the decay constant τ in

table 3, and it received roughly as many “neutral” citations

as the dominant judgments of the fourth and fifth epochs.

The surge in family litigation in fig. 4 began around

late 1990, when leave to appeal was granted from the

judgment of the Manitoba Court of Appeal.

56

In granting

leave, the SCC signalled its willingness to reexamine the

law on spousal support—to reconsider such factors as the

dependencies developed over a long “traditional” marriage

and the need to sever the financial ties between people who

had been divorced for some time. The Court heard the case

in April 1992 and issued its decision in mid-December.

As shown in fig. 7, the rise in the number of family

law cases in each province between 1992 and 1994

correlates with the number of citations to Moge in 1994

(this holds true regardless of whether Ontario and British

Columbia are included).

57

While correlation does not

necessarily

imply causation, such a proportionality can

arise if there is a causal link between Moge and the increase

in number of judgments.

[…]

56

Moge v Moge (1990), 70 DLR (4th) 236, 64 Man. R. (2d) 172 (CA);

“Docket 21979, Andrzej Moge v Zofia Moge” (18 July 1994), online:

Supreme Court of Canada <https://www.scc-csc.ca/case-

dossier/info/dock-regi-eng.aspx?cas=21979>.

57

fig. 7 includes cases with (k ≥ 0).

CANADIAN JOURNAL OF FAMILY LAW [VOL. 34, 2022]

180

Figure 7: Citations of Moge vs absolute increase in

judgments, by province

Note: Number of citations of Moge made by judgments issued in 1994

versus the size of the absolute increase in number of judgments in our

full dataset (k ≥ 0) from the year of the release of Moge (1992) to the

year of the mid-1990s peak in family law judgments (1994), by

province. Correlation coefficient r = 0.97 (all data points), r = 0.75

(BC and ON excluded).

Fig. 7 is not scaled for population. Unsurprisingly,

Prince Edward Island (PE) had both the smallest increase

in number of judgments and the fewest citations of Moge.

However, in 1994 Ontario had almost three times BC’s

population but far fewer family judgments and citations of

Moge. Alberta (AB) had more than 3.5 times the

population of New Brunswick (NB), but New Brunswick

had more family judgments and more citations of Moge.

THE INFLUENCE OF LANDMARK JUDGMENTS

181

Moge’s interplay with provincial statutes deserves closer

examination.

The provinces with the most citations of Moge

before 1998 (when the FCSG introduce confounding

effects) are BC, Ontario, New Brunswick and Alberta. Fig.

8 shows, for these provinces, the number of cases that cite

Moge and pertain to at least one of family property, spousal

support, child support, custody and access, or domestic

settlements.

Figure 8: Topics of judgments that cite Moge, by

province

Note: Filled (grey) bars show the number of pre-1998 judgments that

cite Moge and that have the topic indicated on the x-axis, where “FP”

is family property, “SS” is spousal support, “CS” is child support,

“C” is custody and access, and “DS” is domestic settlements. Open

(black) bars show the number of judgments with the indicated topic

that also have the topic of spousal support.

Spousal support, of course, is a dominant concern

in these cases, but some cases not tagged as pertaining to

spousal support also cite Moge. The marked differences

that fig. 8 reveals between BC and Ontario call for

attention. BC had almost as many family property

judgments as judgments about spousal support; Ontario

had—both absolutely and relatively—far fewer family

property judgments. BC also had far more cases that

concerned only family property but that cited Moge.

CANADIAN JOURNAL OF FAMILY LAW [VOL. 34, 2022]

182

The large number of family property judgments in

BC appears to be due at least in part to BC courts applying

Moge principles to make unequal divisions of property in

order to provide lump-sum spousal maintenance. Of the

pre-1998 judgments in our dataset that cite Moge and have

the topic of family property, the four most highly cited are

from BC. In two of these cases, the BC Court of Appeal

relied on Moge in granting lump-sum maintenance in the

form of unequal property division, pursuant to a statutory

provision that permitted doing so if equal division would

be unfair given various factors, including the duration of

the marriage, the extent to which the property was acquired

by one spouse through inheritance or gift, each spouse’s

need to be economically self-sufficient, and “any other

circumstances relating to the acquisition, preservation,

maintenance, improvement or use of property or the

capacity or liabilities of a spouse.”

58

While the role of

Moge in justifying lump-sum maintenance through unequal

property division in BC has been noted before, it does not

appear to have been previously examined empirically.

59

58

Family Relations Act, RSBC 1979, c. 121, s. 51; Lodge v Lodge (1993),

79 BCLR (2d) 360, 48 RFL (3d) 365 (CA); Toth v Toth (1995), 13

BCLR (3d) 1, 17 RFL (4th) 55 (CA). The third case, Young v Young

(1990), 50 BCLR (2) 1, 29 RFL (3d) 113 (CA), [1993] 4 SCR 3, mainly

concerned child custody and access, although property division and

spousal support were addressed. Moge was cited at the SCC level

regarding the overall economic impact of divorce on women. In the

fourth case, between unmarried parties, Peter v Beblow, [1993] 1 SCR

980, 101 DLR (4th) 621, the Court expressed its agreement with Moge

principles while granting a constructive trust remedy.

59

See Rogerson, supra note 53.

THE INFLUENCE OF LANDMARK JUDGMENTS

183

The BC Family Relations Act broadly defined the

property to be divided and invited courts to consider many

factors justifying unequal division. Ontario, however, more

narrowly defined the property to be divided (excluding

most gifts and inheritances, for example, and property

brought into the marriage). The province likewise required

equal division unless it would be “unconscionable,” having

regard to various factors, many of which had to do with one

partner’s bad faith with respect to the assets.

60

Unconscionability is a higher threshold than unfairness,

and since fewer contentious assets tended to be divided in

Ontario, the statutory invitation to consider Moge

principles appears to have been less compelling than in BC.

Of the pre-1998 Ontario cases that cite Moge, division of

family property was not addressed in the two that were

themselves most cited, Elliot v Elliot and Robinson v

Robinson, which suggests the lack of a linkage between

Moge and property division in Ontario.

61

While the

categories of property subject to division differed in New

Brunswick and Alberta, both provinces’ statutes had lower

barriers to unequal property division than in Ontario (“not

just and equitable” in Alberta,

62

and “inequitable” in New

Brunswick

63

); similarly, the number of judgments

involving the topic of division of family property relative

to the number of judgments involving spousal support is

60

Family Law Act, RSO 1990, c F-3, ss. 4 & 5.

61

Elliot v Elliot (1993), 15 OR (3d) 265, 106 DLR (4th) 609 (CA);

Robinson v Robinson (1993), 15 OR (3d) 485, 107 DLR (4th) 78

[Robinson].

62

Matrimonial Property Act, RSA 1980, c. M-9, ss. 3-8.

63

Marital Property Act, SNB 1980, c M-1, ss. 1, 3, 6-9.

CANADIAN JOURNAL OF FAMILY LAW [VOL. 34, 2022]

184

larger in these provinces than in Ontario.

In summary, Moge preceded a significant increase

in the number of family law judgments. In most provinces,

the size of this increase correlates strongly with the number

of 1994 judgments citing Moge. Moge also preceded the

mid-1990s emergence of judgments with multiple legal

topics. Fig. 8 shows provincial variation in the

development of the post-Moge jurisprudence, including a

large number of BC judgments concerning family property.

BC judges appear to have relied on Moge to divide property

unequally to compensate for economic disadvantages

arising from marriage. In Ontario, spousal support and

property division tended to be decided separately, probably

largely because the legislation did not render unequal

property division as ready a tool in addressing spousal

support. Moge principles also influenced child support law.

In the next section, we examine the role of the Federal

Child Support Guidelines in the emergence of multi-topic

judgments involving child support.

B. INFLUENCE OF THE FEDERAL CHILD

SUPPORT GUIDELINES: CO-OCCURRENCE OF

CHILD SUPPORT WITH OTHER LEGAL TOPICS

The legal developments of the 1990s led to the tying

together of multiple legal topics, indicating an increase in

the complexity of family litigation. Before exploring the

co-occurrence of child support with other topics, however,

we consider what may appear an anomalous feature in the

data: the decline and rise in child support judgments in the

mid-1990s.

Passed as a regulation pursuant to the Divorce Act,

THE INFLUENCE OF LANDMARK JUDGMENTS

185

the Federal Child Support Guidelines came into effect on

May 1, 1997 and changed the law substantially with the

goal of making child support more standardized. A table

defined amounts of child support based on the paying

parent’s income, and additional rules took account of other

circumstances. Whereas child support awards made before

May 1997 were deductible by the paying parent and

taxable to the recipient, amounts awarded after May 1,

1997 were neither deductible nor taxable.

64

Our dataset seems to hint at how family lawyers

anticipated and responded to the FCSG. Between about

1994 and 1997, the number of judgments citing no other

judgments (k = 0)—usually trial court decisions—dropped,

while those that did cite other judgments stayed about the

same (fig. 4a). These k = 0 cases then multiplied after the

FCSG came into force. Likewise, cases concerning child

support and custody increased in the early 1990s, dropped

around 1997, and then jumped up dramatically in 1998,

growing less dramatically to 2015 (figs. 3a and c). In the

transition period from 1994 to 1999, the number of child

support judgments issued monthly varied considerably, as

shown in fig. 9. This was the period when the guidelines

were being discussed, anticipated, and then finally known

and applied—especially the table amounts. In March 1996,

the federal budget outlined the guidelines, explaining when

they would come into effect and their tax implications.

64

The Income Tax Budget Amendments Act, 1996, SC 1997, c 25, ss 8-

12; Government of Canada, Budget 1996: The New Child Support

Package (6 March 1996) (Chair: Paul Martin); Faye L Woodman, “Tax

Aspects of the New Child Support Guidelines: One Year Later” (1998)

15 Can Fam LQ 221.

CANADIAN JOURNAL OF FAMILY LAW [VOL. 34, 2022]

186

Until early 1997, Canadian courts continued to issue about

35 judgments per month on child support. The number of

child support judgments issued began to decline as the

Divorce Act amendments were passing into legislation

between November 1996 and February 1997.

65

A nadir of

three judgments for all of Canada was reached in August

1997.

66

The FCSG regulation passed on April 8 and took

effect on May 1, 1997, with no complicated transitional

provisions: orders issued after May 1, 1997 were made

under the new regulations. The 1997 decline in the number

of child support judgments issued may reflect hard-to-

measure factors such as lawyers waiting until they

understood the implications of the regulations before

applying for new orders, which then would have taken time

to be processed by courts. Early 1998 brought a huge

increase in the number of judgments, which climbed to a

new post-guidelines high of about 65 judgments per month,

with seasonal fluctuations. These effects are likely the

result of uncertainties about how the new rules would apply

and the need to navigate the new regime’s focus on proving

and calculating payers’ incomes, rather than simply

65

House of Commons Journals, 35-2, No 101 (18 November 1996) at

861–862 (third reading); Senate Journals, 35-2, No 71 (13 February

1997) (third reading); An Act to Amend the Divorce Act, the Family

Orders and Agreements Enforcement Assistance Act, the Garnishment,

Attachment and Pension Diversion Act and the Canada Shipping Act,

SC 1997, c. 1 [Divorce Amendment Act, 1997].

66

An annual low in July or August is normal in our data, but the February

1997 level is much lower than the levels in February 1995 and February

1996

.

THE INFLUENCE OF LANDMARK JUDGMENTS

187

assessing children’s expenses.

67

Judges also may have

taken advantage of the winter recess to complete their

judgments on the new issues these cases presented.

In sum, then,

the

spike in judgments in early 1998

probably arose from a combination of strategic,

institutional, and legal factors. But after the spike, the

number of applications heard in court stayed high.

[…]

67

Hypotheses such as parties seeking tax advantages from the new

regulations, courts introducing case management practices, or lawyers

for unmarried couples waiting for corresponding provincial legislation

to amend the rules applicable in these cases do not seem to account for

the surge in litigation in 1998, as when they occurred at all, they

occurred unevenly across the provinces over subsequent years.

CANADIAN JOURNAL OF FAMILY LAW [VOL. 34, 2022]

188

Figure 9: Monthly number of child support judgments

before and after FCSG

Note: Number of judgments (k ≥ 0) with the topic of child support, and

number of judgments (k ≥ 0) with the topics of both child support and

custody and access. The dashed lines indicate the publication date of

the 1996 Federal Budget (Mar. 6, 1996), the date the Federal Child

Support Guidelines came into effect (May 1, 1997), and the beginning

of the 1998 taxation year (Jan. 1, 1998).

In section III.A, we identified different effects in

different provinces of Moge’s linkage of spousal support to

family property. The FCSG were of course federal rules,

and their impact on co-occurrence is appreciable across the

country. Before 1994, only about 5-10% of judgments on

custody and access, spousal support and family property

also involved child support (see fig. 10). Since 1997, more

than 30% of cases on these topics have involved child

THE INFLUENCE OF LANDMARK JUDGMENTS

189

support as well. The co-occurrence of custody and access

with child support initially increased around March 1996,

when the FCSG were announced in the federal budget, with

the news that the guidelines would adjust support amounts

in cases of split or shared custody.

68

Concern about how

this would actually work seems reflected in the

preeminence of citations to Levesque (AB CA, 1994) in

judgments from 1994 to 1997 involving both child support

and custody and access.

69

Levesque apportioned support between two

income-earning parents, considering the costs of custody

and access and adjusting for income tax consequences.

Even before the FCSG, then, Levesque anticipated the

coming changes and called for judges to consider proposed

custody arrangements more carefully. Cases involving the

co-occurrence of child support with spousal support (solid

black curve) and family property (grey curve) also began

to rise in 1995. These increases may be due to the

combined influence of Moge, which provided that an

equitable sharing of the economic consequences of

marriage and divorce “can be achieved in many ways: by

spousal and child support, by the division of property and

assets or by a combination of property and support

entitlements”,

70

and Willick (1994), which closely followed

the Moge approach.

68

Government of Canada, supra note 64.

69

In judgments with these topics, Levesque v Levesque (1994), 20 Alta

LR (3d) 429, 116 DLR (4th) 314 (CA) [Levesque] was the most-cited

case in 1995 judgments (4 citations), the second most-cited in 1996 (7

citations—Moge had 9), and the most-cited in 1997 (6 citations).

70

Moge, supra note 15 at para. 46.

CANADIAN JOURNAL OF FAMILY LAW [VOL. 34, 2022]

190

Figure 10: Co-occurrence of various topics with child

support, before

and

after FCSG

Note: Percentage of judgments (k ≥ 0) with the topic indicated in the

figure legend that also involve the topic of child support. The dashed

line is at 1997, the year the FCSG came into effect.

In the remainder of this section, we use the thematic

clustering analysis introduced in section I.B to probe how

the implementation of the FCSG in 1997 linked together

child support and custody and access (shown in fig. 10).

This method finds clusters of judgments with similar

topics, based purely on the citation structure of the

network. For example, the largest cluster (thematic cluster

1) contains many judgments with the topic of custody and

access, while clusters 2 and 3 contain many judgments with

the topics of child protection and child support

THE INFLUENCE OF LANDMARK JUDGMENTS

191

respectively. The clustering results allow us to infer which

issue in a multi-topic judgment is dominant and which are

secondary.

71

For example, if a judgment that has both the

topics of custody and access and child support is assigned

by the thematic clustering method to cluster 1, then we

infer that the judgment’s dominant topic is custody and

access, with child support secondary. Similarly, a different

judgment that has the same two topics but that is assigned

to cluster 3 would have child support as its dominant topic,

and custody and access secondary. Fundamentally, this

method works because a judge with a difficult custody and

access issue but a straightforward question of quantum of

support will tend to cite more cases (thus generating more

links in our dataset) while wrestling with custody and

access: our method will identify the judgment’s dominant

topic as custody and access with child support secondary.

The method identifies not the issue that was necessarily

more important to the parties but the one that rested on

more analysis of past cases.

Examining the set of judgments involving custody

and access and at least one other topic reveals an

approximately equal number of judgments in which: (i)

custody and access is dominant and child support or

71

We first apply the thematic clustering method, using only the citation

structure of the network. Then, to verify our results, we examine the

judgments in each cluster. It turns out that each cluster is indeed

dominated by a certain topic or set of topics, which, unsurprisingly,

align with the topics assigned by the CAD editors but are not

determined by them (and they do not purport to identify dominant or

secondary topics in cases). The thematic clustering method and our

dominant/secondary topic inference are explained in extensive detail

in Hickey, supra note 23.

CANADIAN JOURNAL OF FAMILY LAW [VOL. 34, 2022]

192

another financial matter is secondary, and (ii) a financial

matter (child support, spousal support, or division of family

property) is dominant with custody and access secondary.

72

The significant presence of custody and access as a subject

of litigation in cases that are primarily about financial

matters invites an examination of the aspects of the FCSG

that may have inspired or facilitated this phenomenon.

Table 4 lists the ten most cited judgments in which

child support is the dominant topic and custody and access

is secondary. Three of these judgments, Green, Froom and

Spanier, concerned section 9 of the FCSG, which allows

for a reduction of the paying parent’s child support

obligation if the parent cares for the child at least 40% of

the time (the so-called “40% threshold”).

73

These cases

explored aspects of the operationalization of these

provisions, such as how to reduce the amounts and whether

to use hours or days in the calculation. In two other

judgments in table 4,

Omah-Maharajh

and

Dergousoff

,

high-income fathers sought to adjust custody and access

terms in order to reduce their payments under the FCSG.

74

The other five judgments in table 4 are more strictly about

72

See Hickey, supra note 23 for further details.

73

FCSG, SOR/97-175, s 9; Green v Green, 2000 BCCA 310, 75 BCLR

(3d) 306 [Green]; Froom v Froom (2005), 11 RFL (6th) 254, 194 OAC

227 [Froom]; Spanier v Spanier (1998), 52 BCLR (3d) 343, 40 RFL

(4th) 319 (SC) [Spanier].

74

Dergousoff v Dergousoff, (1999), 177 Sask R 64, 48 RFL (4th) 1 (CA)

[Dergousoff]; Omah-Maharajh v Howard, 1998 ABQB 81, 58 Alta LR

(3d) 236 [Omah-Maharajh] (in which the father argued that his high

access costs should be considered an “undue hardship” that would

justify reducing the child support amount).

THE INFLUENCE OF LANDMARK JUDGMENTS

193

financial calculations and do not speak much to the

relationship between child support and custody and

access.

75

In these second five cases, the secondary topic of

custody and access appears to be more distinct from child

support than in the first five.

[…]

75

See Tauber v Tauber (2000), 48 OR (3d) 577, 187 DLR (4th) 1 (CA);

Cornelissen v Cornelissen, 2003 BCCA 666, 21 BCLR (4th) 308;

Homsi v Zaya, 2009 ONCA 322, 65 RFL (6th) 17; Meade v Meade

(2002) 31 RFL (5th) 88 (ON SCJ); Titova v Titov, 2012 ONCA 864,

29 RFL (7th) 267.

CANADIAN JOURNAL OF FAMILY LAW [VOL. 34, 2022]

194

Table 4: Ten Most Cited Judgments with Dominant

Topic Child Support and Secondary Topic Custody and

Access

Judgment Citation Court kin Income

Green v

Green

2000 BCCA

310

BC-Ap

100

82,000

Homsi v Zaya

2009 ONCA

322

ON-Ap

83

19,000∗

Tauber v

Tauber

(2000) 48 O.R.

(3d) 577

ON-Ap

58

2,500,000

Dergousoff v

Dergousoff

(1999) 177

Sask R. 64

SK-Ap

44

179,000

Meade v

Meade

(2002) 31 RFL

(5th) 88

ON-Tr

41

53,000∗

Titova v Titov

2012 ONCA

864

ON-Ap

26

98,000

Froom v

Froom

(2005) 11 RFL

(6th) 254

ON-Ap

26

46,000

Omah-

Maharajh v

Howard

1998 ABQB 81

AB-Tr

23

219,000

Spanier v

Spanier

(1998) 40 RFL

(4th) 329

BC-Tr

19

93,000

Cornelissen v

Cornelissen

2003 BCCA

666

BC-Ap

18

522,000

THE INFLUENCE OF LANDMARK JUDGMENTS

195

Note on table 4: The column “Court” indicates province and court

level (trial, “Tr”, or appellate, “Ap”) for the judgment. The last

column shows income of the parent paying child support, rounded to

nearest $1000.

()

Income imputed by judge for purpose of calculating child support

amounts.

The eleven most cited cases in which custody and

access is dominant and child support secondary appear in

table 5. We consider eleven rather than ten judgments,

because the most cited judgment, Talbot v Henry

(1990),

pre-dates the FCSG.

76

As table 5 shows, in the ten post-

FCSG cases in which custody and access was dominant

and child support secondary, the median income of the

paying parent ($37,000) was significantly lower than in the

cases in which the dominant and secondary topics were

reversed (median $95,500, in table 4). As one would expect

from a reliable clustering method, these cases centred on

the appropriateness of the actual custody and access

arrangements, with financial matters secondary. Five of

these cases primarily concerned ordering shared parenting

when the parents were uncooperative or experiencing

considerable conflict.

77

Another two cases considered

terminating the father’s access in situations of even greater

76

Talbot v Henry, [1990] 5 WWR 251, 84 Sask R 170 (CA) [Talbot].

The case concerned the test for a change of circumstances that would

justify reopening a child support order, a matter rendered moot by the

FCGS.

77

Lawson v Lawson (2006) 81 OR (3d) 321, 214 OAC 94; Ladisa v

Ladisa (2005), 193 OAC 336, 11 RFL (6th) 50; Stewart v Stewart

(1994) 41 BCAC 213, 2 RFL (4th) 53 [Stewart]; Ursic v Ursic (2006),

32 RFL (6th) 23, 149 ACWS (3d) 103 (ON CA); Baker-Warren v

Denault, 2009 NSSC 59, 277 NSR (2d) 271.

CANADIAN JOURNAL OF FAMILY LAW [VOL. 34, 2022]

196

conflict.

78

The three other post-FCSG judgments in table 5

addressed moving with the child to another jurisdiction.

79

[…]

78

Jennings v Garrett (2004), 5 RFL (6th) 319, [2004] OTC 460 (SCJ);

Dixon v Hinsley (2001), 22 RFL (5th) 55 (ON CJ).

79

Falvai v Falvai, 2008 BCCA 503, 86 BCLR (4th) 47 [Falvai]; One v

One, 2000 BCSC 1584, 81 BCLR (3d) 315; JAD v LDD, 2010 NBCA

69, 364 NBR (2d) 200.

THE INFLUENCE OF LANDMARK JUDGMENTS

197

Table 5: Eleven Most Cited Judgments with Dominant

Topic Custody and Access and Secondary Topic Child

Support

Judgment

Citation

Court

kin Income

Talbot v Henry

[1990] 5

WWR 251

SK CA

100

Unstated†

Lawson v Lawson

(2006), 81 OR

(3d) 321

ON CA

89

66,000∗

Ladisa v Ladisa

(2005), 193

OAC 336

ON CA

85

26,000∗

Stewart v Stewart

(1994), 41

BCAC 213

BC CA

70

0†

Falvai v Falvai

2008 BCCA

503

BC CA

68

60,000†

Ursic v Ursic

(2006), 32

RFL (6th) 23

ON CA

65

56,000

One v One

2000 BCSC

1584

BC SC

56

54,000

Jennings v Garrett

(2004), 5 RFL

(6th) 319

ON SCJ

50

41,000∗

JAD v LDD

2010 NBCA

69

NB CA

47

33,000

Baker-Warren v

Denault

2009 NSSC

59

NS SC

(FD)

47

20,000∗

Dixon v Hinsley

(2001), 22

RFL (5th) 55

ON CJ

43

21,000∗

Note: *Income imputed by judge for purpose of calculating child

support amounts.

†No income was stated in Talbot; in Stewart, the judge found the father

CANADIAN JOURNAL OF FAMILY LAW [VOL. 34, 2022]

198

had “no significant income”; in the trial judgment of Falvai (2008

BCSC 79, [2008] B.C.W.L.D. 1796), the judge suggested the father

was capable of earning $30/hour but adjourned the question of child

support without imputing an income ($60,000 is our estimate based on

full time work at $30/hour: this allows calculation of a median income

of the cases in the table).

We have described the Moge reasoning that support

and family property are intrinsically connected and must be

taken together in addressing the economic consequences of

the marriage and its breakdown. The data show the

interrelationship of spousal support and family property

especially in BC, where the legislation, read alongside

Moge, invited judges to provide “lump-sum maintenance”

by apportioning the recipient spouse a larger share of the

family property. Similarly, child and spousal support

became explicitly linked under the FCSG, which required

judges to consider them together and reduce spousal

support in favour of child support if the payer lacked the

means to provide both.

80

The FCSG’s 40% threshold

linked child support and custody so that sometimes

applications to increase child support were countered with

applications to increase access time or to consider access

costs. The cases in table 5 suggest that the FCSG may have

created more litigable issues in custody and access battles,

such as the imputation of income to a parent resisting

financial disclosure.

81

Cases involving children, and

particularly access and support, have been shown to remain

in the court system longer than other family cases.

82

80

Divorce Amendment Act, 1997, s. 15.3.

81

See e.g. Coadic v Coadic, 2005 NSSC 291, 237 NSR (2d) 362.

82

Allen, supra note 30; Kelly, supra note 30.

THE INFLUENCE OF LANDMARK JUDGMENTS

199

C. THE INCREASE OF FAMILY LITIGATION IN

THE 1990S

The underlying causes of the great increase in family

litigation since the early 1990s are presumably largely

social and economic. In many Western countries, including

Canada, the 1990s were a period of austerity in which

social services were cut back drastically and the economy

transformed by globalization.

83

In Canada, social services

received massive cuts to federal funding shortly after the

implementation of the Canada-USA Free Trade Agreement

of 1989 and again after its successor, the North American

Free Trade Agreement of 1994.

84

During a serious

economic recession in the early 1990s, unemployment

rates were above 10% in much of the country.

85

Between

1989 and 1994, Ontario lost 189,000 manufacturing jobs,

which accounted for 80% of the province’s job losses.

86

Although the unemployment rate decreased in the mid-

1990s, part-time and temporary work grew considerably.

87

83

Christophe Guilluy, No Society (Paris: Flammarion, 2018); Savina

Gygli et al “The KOF Globalisation Index—Revisited” (2019) 14:3

Rev of International Organizations 543.

84

John W Foster & John Dillon, “NAFTA in Canada: The Era of a Supra-

Constitution,” in Karen Hansen-Kuhn & Steve Hellinger, eds, Lessons

from NAFTA: The High Cost of “Free Trade” (Ottawa: Canadian

Centre for Policy Alternatives, 2003) 83 at 92-94.

85

Dave Gower, “Canada’s Unemployment Mosaic in the 1990s” (1996)

8:1 Perspectives on Labour & Income 16.

86

Department of Finance, “The Economy 1994” (March 1994) online

(pdf): Newfoundland and Labrador Finance

<https://www.gov.nl.ca/fin/files/archives-e1994-1994-med.pdf>.

87

Leah F Vosko, Nancy Zukewich & Cynthia Cranford, “Precarious

CANADIAN JOURNAL OF FAMILY LAW [VOL. 34, 2022]

200

These economic conditions undoubtedly led to greater

stress on households and family relationships, and

probably to more family breakdowns. Indeed, the number

of lone parent families increased sharply from 1990 to

1994.

88

With less state funding supporting families directly

and indirectly, separated spouses were left to turn (with or

without success) to their former partners for support for

themselves and their children, a shift Brenda Cossman has

called “reprivatization”.

89

Increasing self-representation in

Canadian courts, partly due to the shortage of state-funded

legal aid,

90

has probably contributed to more time-

Jobs: A New Typology of Employment” (2003) 15:4 Perspectives on

Labour & Income 39.

88

Sharanjit Uppal, “Employment Patterns of Families with Children” (24

June 2015), online: Statistics Canada

<https://www150.statcan.gc.ca/n1/pub/75-006-x/2015001/article/

14202-eng.htm>.

89

Brenda Cossman, “Family Feuds: Neo-Liberal and Neo Conservative

Visions of the Reprivatization Project,” in Brenda Cossman & Judy

Fudge, eds, Privatization, Law, and the Challenge to Feminism

(Toronto: University of Toronto Press, 2002), 169 at 169.

90

On cuts to legal aid in the 1990s and the impact on family litigants, see

Katie Davis, “Legal Aid in Canada: Resource and Caseload Statistics

2002/03” (February 2004), online: Statistics Canada Canadian Centre

for Justice Statistics <

https://www150.statcan.gc.ca/n1/en/catalogue/85F0015X2003000>;

Andrea Taylor-Butts, “Justice Spending in Canada, 2001/01” (October

2002), online: Statistics Canada Juristat

<https://www150.statcan.gc.ca/n1/en/catalogue/85-002-X20020118

430>. Lawyers’ fees may also have increased in the 1990s. Canadian

Lawyer magazine’s annual survey of lawyers’ fees shows a roughly

130% increase from 2005 to 2015 in the national average hourly rate

charged by lawyers 5 and 10 years post-call, after adjusting for

inflation. See Michael McKiernan, “The Going Rate” (June 2015),

THE INFLUENCE OF LANDMARK JUDGMENTS

201

consuming litigation, with more interlocutory judgments

and more applications to reopen existing arrangements.

Major legal changes also occurred in the early to

mid-1990s. The Pelech trilogy’s “clean break” approach to

relationship breakdown—with its optimistic assumption

that former spouses would soon be financially self-

sufficient—declined in these hard times and was

superseded by the Moge approach of compensating for the

economic disadvantages experienced by spouses who took

care of the home during the marriage. Moge was the

beginning of a wave of legal developments that allowed for

the reopening of orders. First, the threshold for changing a

spousal support order was adjusted by Moge from a

“radical change” (Pelech trilogy) to a “material change”.

This change probably contributed to the sharp increase in

number of judgments from 1992 to 1994 (fig. 4a, black

curve). Two years after Moge, in 1994, Willick elaborated

on the Moge test for “material circumstances” and

indicated that larger support payments would be awarded.

This signalling and the impending FCSG may have

temporarily pushed litigation rates downward, by

encouraging potential payers—largely men—to negotiate

rather than contest support claims in court, which may have

contributed to the crash in the number of judgments from

online (pdf): Canadian Lawyer Magazine <

https://www.canadianlawyermag.com/staticcontent/images/canadianl

awyermag/images/stories/pdfs/Surveys/2015/CL_June_15_GoingRat

e.pdf>; Kirsten McMahon, “The Going Rate 2005” (September 2005),

online (pdf): Canadian Lawyer Magazine

<https://www.canadianlawyermag.com/staticcontent/images/canadian

lawyermag/images/stories/pdfs/Surveys/2005/04June%20-

%20Legal%20Fees%20Survey%20.pdf>.

CANADIAN JOURNAL OF FAMILY LAW [VOL. 34, 2022]

202

1994 to 1997 (fig. 4a, black curve). This hypothesis could

be investigated further by studying the rate of male-

initiated versus female-initiated cases before and after

Willick. Overall, though, conflicts over family resources

moved into courtrooms. The FCSG (1997) allowed for the

variation of pre-existing child support orders and, going

forward, permitted variation orders whenever any change

in circumstance took place that would call for a different

table amount, or, when table amounts were inapplicable

(such as for high-income spouses), when any change took

place to “the condition, means, needs or other

circumstances of either spouse or of any child who is

entitled to support.”

91

Having care of a child more than 40

percent of the time, for example, reduced a paying parent’s

out-of-pocket expenses, and former step-parents and others

found themselves with unexpected and unwelcome support

obligations which they challenged. In Gordon (1996), the

Court’s approach to amending custody and access

arrangements when the custodial parent wished to relocate

did little to reduce the uncertainty in these situations and

may well have led to more litigation.

92

Taken together

these changes probably contributed to the rise in variation

applications and judgments.

The increased complexity of the calculations and

considerations introduced to improve fairness may be

another factor leading to more judgments.

93

For example,

while the FCSG aimed at “improving the efficiency of the

91

FCSG, SOR/97-175, s 14.

92

Gordon, at paras 26-48.

93

Allen, supra note 30; Kelly, supra note 30,

THE INFLUENCE OF LANDMARK JUDGMENTS

203

legal process by giving courts and spouses guidance in

setting the levels of child support orders and encouraging

settlement,”

94

researchers have noted substantial

contestation over aspects of the FCSG such as the

determination of parents’ incomes, special expenses, undue

hardship, and responsibility for children over the age of

majority.

95

As well, to accompany the FCSG, the federal

government set aside substantial funds to assist provinces

and territories in establishing and improving their

programs and procedures to help parents obtain and vary

child support orders.

96

Along with other measures for

increasing access to justice, these initiatives presumably

led to more attempts to obtain child support and an increase

in the number judgments—and the load on courts.

IV. CONCLUSION

The volume of family litigation in Canada increased

dramatically in the 1990s. This was also a time of great

dynamism in family law as courts, urged on by advocates

of necessary family law reform, worked to reshape the law

to reflect the costs and benefits of marriage and its

breakdown more fairly, all against a backdrop of

widespread austerity and a reduction of state support for

families. This article has quantitatively examined the

relationships between two phenomena: the increase in the

amount and complexity of family litigation in Canada that

occurred in the 1990s, and changes in the law (landmark

94

FCSG, SOR/97-175, s 1(c).

95

Rogerson, supra note 18.

96

Department of Justice Canada, supra note 18.

CANADIAN JOURNAL OF FAMILY LAW [VOL. 34, 2022]

204

court judgments and key statutes) that occurred around this

time. While our analysis does not prove causation, and

social and economic factors are likely the main drivers

underlying litigation patterns, our analysis empirically

demonstrates intriguing correlations between legal changes

and changes in the amount and type of family litigation.

These results in the family law domain are analogous to

effects of judicialization in the political sphere that

sociolegal theorists have described

97

and, further,

demonstrate how probable relationships can emerge

through the holistic study of a dataset covering a very broad

set of Canadian family law judgments.

To obtain a high-level view of the evolution of the

law over several decades, we applied temporal clustering

to our citation network. This approach identified five major

epochs which, although defined purely based on the link

structure of the citation network, begin and end at points in

time that correspond closely to the dates of key statutes.

The third epoch contains the key 1990s changes in amount

and subject of litigation revealed in our data. We therefore

focus on the legal developments that define this epoch in

our search for possible explanations of the 1990s litigation

changes. The third epoch began shortly after the 1986

overhaul of the federal Divorce Act and declined with the

1997 enactment of the Federal Child Support Guidelines.

During this epoch two sets of landmark judgments laid out

opposing visions of the law of spousal support. The first,

the Pelech trilogy (1987), advocated a “clean break”

approach, emphasizing that separation agreements should

97

See e.g. Martin Shapiro & Alec Stone Sweet, On Law, Politics, and

Judicialization (Oxford: Oxford University Press, 2002).

THE INFLUENCE OF LANDMARK JUDGMENTS

205

be final and, usually, that both spouses would achieve self-

sufficiency. The second set, centred around Moge

and

followed by Willick, reduced the barrier to re-opening pre-

existing orders for spousal and child support and sought to

compensate spouses (usually wives) for economic

disadvantages flowing from their role in the marriage.

Both Moge and the FCSG instructed courts to

examine a wider range of circumstances and allowed for

the reopening of orders. Each preceded the rapid

emergence of judgments with multiple legal topics. Moge

appears to have led to the use of unequal division of family

property as a means of providing financial compensation to

former spouses in provinces where the property division

legislation was sufficiently flexible, especially in British

Columbia. The FCSG is associated with the emergence of

judgments that combined child support, in various ways,

with division of family property, spousal support, and

custody and access. With the goal of fairer, more holistic

results, Moge and the FCSG appear to have increased the

complexity of litigation by inviting multiple legal issues to

be argued together. For example, thematic clustering

shows that custody and access issues are litigated just as

often in cases where the dominant issue is financial (such

as child support, spousal support, or division of family

property) as in cases where the dominant issue is custody

and access, whereas previously these issues were mainly

litigated independently.

The findings of our study raise several questions for

future research. Tables 4 and 5 suggest that, for example,

cases in which custody and access is the dominant issue

may arise in different economic circumstances from cases

in which child support dominates. A close textual analysis

CANADIAN JOURNAL OF FAMILY LAW [VOL. 34, 2022]

206

of these judgments would allow further exploration of how

the FCSG affected family litigation and how the

characteristics of these two groups of judgments evolved

over time, as the legal system adapted to the FCSG regime

and to post-FCSG statutory revisions and new landmark

judgments.

A broader avenue for future research relates to the

argument that legal reform can “open the floodgates” to

litigation. Statutory changes can clearly lead to surges in

new cases, as occurred for divorces in Canada immediately

following the revisions to the Divorce Act in 1986;

98

however, anticipated surges do not always occur, as in the

case of Ireland’s 1996 constitutional amendment allowing

divorce.

99

Our results (fig. 7 in particular) suggest that

landmark judgments can also contribute to surges in new

cases, because they both raise new legal questions and open

new strategic angles for litigants. Further qualitative and

quantitative analysis—possibly of a comparative nature—

could illuminate the mechanisms and suggest methods for

addressing the inherent human and financial “transaction

costs” of law reform, to reduce the costs they externalize to

individuals and families and secure the improvements in

access to justice that are the goal of law reform efforts.

Driven by underlying social and economic forces

98

Statistics Canada, “A fifty-year look at divorces in Canada, 1970 to

2020” (9 March 2022), online: The Daily

<www150.statcan.gc.ca/n1/daily-quotidien/220309/dq220309a-

eng.htm>.

99

Jenny Burley & Francis Regan, “Divorce in Ireland: The Fear, the

Floodgates, and the Reality” (2002) 16:2 Int J Law Pol Fam 202.

THE INFLUENCE OF LANDMARK JUDGMENTS

207

and prevailing views about how these should be addressed,

family litigation has increased, with concomitant costs to

individuals and the legal system. Our data and analysis

suggest that landmark family law judgments and key

legislation have widened the invitation to use courts to

increase—or sometimes decrease—spousal and child

support to address the resources of those involved. With

lawyers and without them, and against a broader backdrop

of austerity measures, individuals have responded by

turning to the courts, contributing to the increase in family

litigation since the early 1990s. This study has highlighted

some specific ways in which legal changes in Canada

appear to be connected to increased amount and

complexity of family litigation; going forward, reformers

seeking to reduce pressure on families and the legal system

should examine these examples further and search for

direct causal links, to help anticipate the possible effects of

future statutory or precedential changes.

CANADIAN JOURNAL OF FAMILY LAW [VOL. 34, 2022]

208

The Influence of Landmark Judgments and Statutory Changes on the Family Litigation Explosion: A Citation Network Analysis

Abstract

Recommended publications

A Complex Systems Study of Social Hierarchies and Jurisprudence

S1 Appendix

Dispozitivne presude u sporovima o uzdržavanju djece

S2 Appendix