An Overview of the Risks and Protectors for Children of Separation and Divorce

5. Measures Used to Assess the Impact of Divorce on Child Adjustment

5.1  The Assessment of Reliability and Validity

For a measure to be a useful assessment tool, it must show good reliability and validity.  Reliability estimates provide an indication of the quality of the measure.  Do the individual items of the measure agree well with one another so that the measure is coherent (referred to below as internal consistency)?  If a person completes the measure on two separate occasions (with a short interval between assessments), are the person's answers similar (referred to below as test-retest reliability)?  Reliability estimates greater than .80 are considered good (Bakeman and Gottman, 1986).  Depending on the purpose for which the instrument is used, reliabilities slightly lower than this are sometimes acceptable.

Validity is the extent to which the measure actually measures what it says it does.  For example, does the measure of marital conflict discriminate between people who are happily married and those who are experiencing significant marital problems?  Does it predict marital breakdown over time?  Does a new test of marital conflict agree well with a previously established test?  Correlations between tests can range from -1.0 to 1.0 with values of zero representing no agreement and values closer to 1.0 representing good agreement.  Sometimes other indices that help us evaluate the effectiveness of the measure are also quoted.  For instance, with some of the measures described below, researchers assess stability over a long period of time, a year or more.  We would expect scores to show agreement over time, but obviously we would not expect this to be as high as the agreement shown for a one-week period.  In the following sections we report technical details for those people who will find such details informative.  We also provide less technical summaries.  Measures have been chosen for description if they have shown promising results with respect to reliability and validity.  Some measures have been investigated more fully than others, and these differences are noted.

5.2  Measures of Marital Adjustment and Satisfaction

The Short Marital Adjustment Test and the Dyadic Adjustment Scale are measures of general marital satisfaction, as opposed to measures of marital conflict.  They both include items that assess elements of marital conflict, but also items that assess non-conflictual elements of the marriage.

5.2.1 The Short Marital Adjustment Test

The Short Marital Adjustment Test (SMAT) was developed by Locke and Wallace (1959) to be a reliable, valid and brief parent-reporting measure of marital adjustment (see Appendix A for full measure).  Marital adjustment is defined by the researchers as the degree of "accommodation" between the dyad at a given time (Locke and Wallace, 1959).  Individuals are asked to rate various aspects of their marriage, such as their degree of happiness in the marriage and their degree of satisfaction with the decision to marry their spouse; the extent to which the couple agrees or disagrees on various issues such as finances, recreation, friends and sexual relations; the way in which conflicts are handled; and the degree of closeness between the couple (e.g. engaging in enjoyable activities together, confiding in one another).  In total, the measure comprises 15 items organized in a Likert-rating and multiple-choice format.

In the original study, Locke and Wallace (1959) reported a reliability coefficient of .90, based on a sample of 118 couples.  This demonstrates that the measure is internally consistent or that the items within the measure relate well to one another.  Furthermore, the measure has shown test-rest reliability scores of .75 over three weeks.  This suggests relatively good stability in the way that couples rate their marriage.  In addition, the measure has been found to clearly discriminate between couples that are well-adjusted and maladjusted (separated, divorced or in marital therapy).  For instance, 96 percent of well adjusted couples achieved scores of 100 or more on the SMAT, whereas only 17 percent of maladjusted couples achieved adjustment scores of 100 or more.  Therefore, a score of 100 was chosen as the cut-off so that individuals who obtained a score below 100 would be considered to have a maladjusted relationship with their partner and individuals with a score greater than 100 would have a well adjusted relationship with their partner.

5.2.2 The Dyadic Adjustment Scale

The Dyadic Adjustment Scale (DAS) constructed by Spanier (1976) is a widely used 32-item, self-reporting measure of marital or dyadic satisfaction (see Appendix B for full measure).  The SMAT measures various components of dyadic satisfaction, such as dyadic consensus (the extent of agreement/disagreement between the couple on various issues), dyadic satisfaction ("how often do you discuss or have you considered divorce, separation or terminating your relationship"), dyadic cohesion ("do you and your mate engage in outside interests together?") and affectionate expression ("do you kiss your mate?").

The DAS yields both individual component scores and a total dyadic adjustment score.  The reliability coefficients for these components of dyadic adjustment range from .73 (affectionate expression subscale) to .94 (dyadic satisfaction subscale).  Spanier (1976) reports the reliability coefficient for the total dyadic adjustment score as being .96.  In the original study, the measure was administered to a sample of 218 married persons and 94 divorced persons.  For each of the 32 items, the divorced individuals were found to differ significantly from the married individuals.  Therefore, the measure adequately discriminates between satisfied and dissatisfied couples.  In subsequent studies, the DAS was also found to reliably discriminate between distressed and non-distressed spouses (Eddy, Heyman and Weiss, 1991).  In a recent study that used the DAS, internal consistency for both mother and father reports was found to be .91 (Davis et al., 1998).  The DAS also correlates well with the SMAT.  The correlation between the SMAT and the DAS was found to be .86 for married individuals and .88 for divorced individuals.

5.3  Measures of Marital Conflict

Although numerous studies have used, and continue to use, general measures of marital adjustment/satisfaction to assess the association between unhappy homes and child well-being and marital conflict and child well-being (Davis et al., 1998), recent studies have found that specific measures are more appropriate in predicting child outcomes than general, global measures.  Specifically, several researchers have found that interparental conflict is a better predictor of child maladjustment than general marital/dyadic satisfaction in a non-clinical sample (Buehler et al., 1998; Davies and Cummings, 1994; Emery and O'Leary, 1982).  Although general measures of marital satisfaction, such as the SMAT and the DAS, include items that assess dyadic conflict, there are valid and reliable measures that only assess dyadic conflict.  The following is an overview of two of these measures, namely the Conflict Tactics Scale (CTS) and the Children's Perception of Interparental Conflict Scale (PIC).

Both the CTS and the PIC have demonstrated very good psychometric properties.  In addition, the CTS has been used to collect information from both parents and children and the PIC has been found to be a reliable and valid measure of parental conflict from a child's perspective.  Therefore, both parent and child reports of marital conflict have been well established.  The use of child or parent reports have their own individual advantages.  Using a parent report is advantageous because it is less intrusive for children.  Since child measures have only been developed for children in grade four or above (and are unlikely to be reliable for younger children), parent measures can also be used for young children.  The value of the child report lies in its stronger relationship to child outcomes.

5.3.1 The Revised Conflict Tactics Scale

The original CTS, developed by Straus (1979), has recently been revised to increase the scale's validity and reliability.  The scale is now known as the Revised Conflict Tactics Scale or CTS2 (Straus et al., 1996; see Appendix C for full scale) and like the CTS, it assesses the extent to which partners in a dating, cohabiting or marital relationship managed conflicts over a 12-month period by engaging in psychological and physical attacks on one another.

The CTS is the most widely used and validated scale of dyadic conflict.  Although many measures of dyadic conflict exist, the CTS offers the most concrete assessment of anger-based dyadic conflicts.  Straus et al. (1996) report that as many as 400 studies (collectively involving more than 70,000 participants) have used the CTS.  Within these 400 studies, numerous investigations were conducted to assess the measure's validity and reliability.  The CTS has been used with participants of diverse cultural backgrounds and in at least 20 countries, including Hong Kong, India, Japan and Spain.  The CTS has also been used as a diagnostic instrument in therapy to assess the severity of conflict and abuse within relationships (Straus et al., 1996).

Both the CTS and the CTS2 are based on conflict theory, which assumes that conflict is an inevitable part of all interactions.  Items on the CTS and the CTS2 probe for specific, concrete tactics used by members of a dyad to resolve a conflict.  Specific tactics range in severity from passive tactics, such as calm discussions, to forceful tactics, such as physical attacks.  Thus, two strengths of the CTS2 are that it focusses on concrete methods of conflict resolution and it assesses a range of conflict tactics.  In addition, the measure assesses both what the participant and their partner have done to manage conflicts.

The original CTS, comprising 18 items across three scales (reasoning, verbal aggression and violence), was criticized for being too short and thus not adequately assessing various types of conflict tactics.  To improve its validity and reliability, the CTS2 is comprised of 39 items across five scales.  Thus, the measure samples numerous conflict tactics and types of inter-partner abuse.

The five scales in the CTS2 are as follows:

Negotiation Scale
The negotiation scale comprises six items that assess actions taken to settle a disagreement through discussion.  The scale includes both cognitive (i.e. one partner suggests a compromise) and emotional (i.e. one partner asks the other how he/she is feeling) conflict management tactics.  On the CTS, the negotiation scale was called the "reasoning scale" and contained only three items.
Psychological Aggression Scale
The psychological aggression scale comprises eight items that assess the use of both verbal and nonverbal acts of aggression.  This scale was called the "verbal aggression scale" on the CTS, but was renamed because it includes aggression items that are nonverbal in nature (e.g. "stomped out of the room").
Physical Assault Scale
The physical assault scale includes 12 items that assess the use of physical violence to manage dyadic conflicts.  This scale was called the "violence scale" on the CTS and included nine items.
Sexual Coercion Scale
This scale was not on the CTS.  This new scale assesses an individual's use of coercion to compel his or her partner to engage in unwanted sexual activity  It includes seven items that probe a range of coercive acts, such as verbal insistence to physical force.
Injury Scale
This scale is also new with the CTS2.  It measures the injury inflicted by an individual's partner and includes six items that assess the degree of pain, tissue or bone damage, and the need for medical attention.

Unlike the CTS, the CTS2 differentiates between minor and severe acts of physical assault, sexual coercion and injury.  This distinction was made because research indicates that the etiology and treatment of occasional minor violence is quite different from that of repeated severe assaults (Straus et al., 1996).  In addition, items on the CTS2 are arranged in hierarchical order so that initial items depict socially acceptable ways of dealing with conflict (e.g. discussed issues calmly) and final items depict the most severe form of physical assault (used a knife or gun).

Like the CTS, the CTS2 was validated on a sample of college students.  This was done because research assessing the validity and reliability of the CTS had established that the factor structure of student data is very similar to that found in national and clinical samples.  Initial research assessing the psychometric properties of the CTS2 indicates that, like the CTS, it is highly reliable (coefficients range from .79 to .95).  Preliminary evidence of some types of validity has already been established (Straus et al., 1996).  Also like the CTS, the CTS2 can be completed by either parents or children.  Research using the CTS indicates adequate concordance between mothers' and fathers' reports, and between reports by children and parents (Johnston et al., 1989).  In addition, like the CTS, the CTS2 can be used to assess conflict within marital dyads (Jenkins, 2000) and post-divorce conflict (Johnston et al., 1987; Johnston et al., 1989).

5.3.2 Children's Perception of Interparental Conflict Scale

The Children's Perception of Interparental Conflict (PIC) scale, developed by Grych, Seid and Finchman (1992), is a 48-item, true-false measure that assesses children's perceptions of various aspects of marital conflict (see Appendix D for full scale).  The PIC includes the following three scales:  conflict properties (frequency, intensity and resolution of conflicts), perceived threat (child's feelings of threat and coping efficacy) and self-blame (child's perceptions that he or she is to blame for parental conflict).  Children rate parental behaviour on a three-point multiple-choice scale (1=false, 2=sort of true and 3=true).  Higher scores indicate increasingly negative forms of conflict or appraisal.

Internal consistency of the PIC has been found to be above that recommended for research instruments (ranges from .78 for self-blame scale to .90 for conflict properties scale), and test-retest correlations (ranges from .70 for conflict properties scale to .76 for self-blame scale) indicate that the PIC is reasonably stable over a two-week period (Grych et al., 1992).  The instrument was validated on 9- to 12-year-old children; it was found that school-age children can reliably report their perceptions of parental conflict.  In fact, stability of measurement over a 12-month period was adequate (.64, .51 and .47 for the three PIC subscales) and comparable to other well established measures using children's reports (ranges from .38 to .52) (Finchman, Grych and Osborne, 1994).  In assessing the PIC's validity, researchers found that children's reports of the frequency, intensity and resolution of conflict were consistently related to ratings of adjustment made by parents, children, children's teachers and peers.  In contrast, parental reports of conflict correlated only with parental reports of adjustment.  This suggests that a child report measure of parental conflict may be more effective as a screening tool than a parent report instrument when the goal is to target families whose children are adversely affected by conflict.  The PIC has been validated for use with college students (Bickham and Fiese, 1997) and it shows high stability and reliability.

5.4  Post-Divorce Measures

The two measures discussed in this section relate to conflict and family functioning following divorce.  Although both measures demonstrate promising psychometric properties, they are fairly new and are thus less well established than the PIC and the CTS.  In addition, both the PIC and CTS have been used to assess parental conflict following divorce and have been shown to be highly reliable and valid.

5.4.1 The Post-divorce Parental Conflict Scale

The Post-divorce Parental Conflict Scale (PPCS) is an 82-item, self-report measure that assesses type and level of parental conflict following divorce as reported by the child (Morris and West, 2000).  The measure is made up of three scales (verbal hostility, indirect hostility and physical hostility) and ratings are made on a 5-point Likert scale (1=the event never happened, to 5=the event happened everyday) according to the first year after parental divorce and the previous 12 months.  Items on the PPCS progress from low conflict and hostility to intense and violent conflict.  As well, children separately report behaviours exhibited by their mothers and fathers.  Example items include "My mother discussed issues calmly with my father" and "My father threw things at my mother."

Studies using the PPCS have shown good reliability for mother and father subscales (ranging from .80 to .90).  For instance, Morris and West (2000) found that for all subscales on the PPCS, internal consistency was .80 or greater.  In addition, the PPCS was found to have good correlations with the PIC (Morris and West, 2000).

5.4.2 The Divorce Adjustment Inventory Revised

The newly revised Divorce Adjustment Inventory or DAI-R (Portes, Haas and Brown, 1991) is a 42-item, parent report instrument that assesses child adjustment and family functioning after parental separation.  Five factors are assessed on the DAI-R:  conflict and dysfunction; favourable divorce condition and child coping ability; positive divorce resolution; external support systems; and divorce transitions.  Ratings are made on a 5-point Likert scale ranging from 1 (strongly disagree) to 5 (strongly agree).  Preliminary analyses demonstrate good internal consistency (ranges from .69 for the total scale score to .83 for the family conflict and dysfunction scale) (Portes, Smith and Brown, 2000).

5.5  Measurement of Childhood Disturbance

If a program or service delivery objective is to identify children who are likely to fare less well after a divorce, then a good strategy would be to use a well established conflict measure (like the CTS2 and/or the PIC) in conjunction with a child adjustment measure (the Achenbach Child Behaviour Checklist, CBCL; Achenbach and Edelbrock, 1991).  The measurement of childhood disturbance is well established.  Although numerous standardized measures exist, the parent form of the CBCL is frequently used in both research and clinical settings.  The CBCL is designed to record, in a standardized manner, the behaviour problems and competencies of children between the ages of 4 to 16 years, as reported by their parents.  The CBCL is a self-report measure that can be used with parents who have reading skills as low as the fifth-grade level or administered by an interviewer if necessary.

The CBCL yields a Total Behaviour Problem Score as well as a number of different problem profiles, such as an externalizing and internalizing profile.  The CBCL is widely used because of its sound psychometric properties, and studies indicate high coefficients of reliability.  Test-retest reliability of item scores has been found to be .95 with an inter-interviewer reliability of .95.  In terms of the measure's validity, reports indicate that the correlation between the CBCL Total Behaviour Problem Score and the total score of other instruments ranges from .71 to .92.  Studies also show that 116 of the 118 behaviour problem items on the CBCL were significantly associated with clinical status, and this also offers support for the measure's validity.  (Entire paragraph, Achenbach and Edelbrock, 1991.)

5.6  Purpose of Conflict Measures

5.6.1 Research Purposes

When conflict measures are used in research studies, researchers are interested in identifying patterns in data between variables, i.e. the association between parental conflict and children's conduct problems.  Measures used in research are not typically used to make decisions about individual children or families.

5.6.2 Program Assignment Purposes

To use a conflict measure to screen individual families into different kinds of programs, the psychometric properties of the measure must be well established.  Further, there must be demonstrated efficacy of the instrument with the population of interest.  It is beyond the scope of this review to outline all the steps that would be advisable before such research instruments are used as measures for assessing individuals.  We do, however, raise several issues below that are worthy of consideration.

  1. To use a conflict measure to assign families to different types of intervention, the measure must be able to discriminate between families that have transient conflict and those who show high conflict over time.  Although predictive validity has been established for the CTS and the PIC, it has not been established for post-divorce conflict measures.  The validity of the CTS and the PIC has also not been demonstrated for families undergoing the kind of major reorganization that accompanies divorce.

  2. To assign families to different types of intervention, cut-off points must be established.  This is usually done on the basis of the distribution.  For instance, the top 10 percent of the sample will be considered high conflict.  The cut-off point must be low enough to include those families who will benefit from the intervention in question and high enough to exclude people who do not need the intervention.  Cut-off points have not been established for most of the conflict measures.

  3. If the aim of screening families is to be able to offer services to families who would benefit from particular services, then the use of two instruments, such as a parental conflict measure and a child adjustment measure, should be considered.  As we have seen, not all children react adversely to parental conflict or divorce.  By screening families according to whether conflict is high within the family and whether children are already showing adjustment difficulties, children and families most in need of service could be targeted.