Do Event-Contingent Diaries About Marital Conflict Change Marital Interactions?

Abstract

Recent increase in the use of diary measures has prompted questions about the effect completing diaries has on participants. After extensive training, married couples completed event-contingent diaries about their couple disagreements for 15 days, focusing on emotional and behavioral aspects of marital conflict. Serving as a control for placebo effects of participation, the diary group (n = 110) and a nonrandomized control group (n = 57) also completed brief daily checklists, with minimal training, over the reporting period. Before and after the reporting period, couples engaged in videotaped conflict resolution tasks that were coded for conflict behaviors and emotions expressed. ANOVAs comparing groups indicated that completing event-contingent diaries for 15 days had no apparent effects on observed husbands' and wives' expressed emotions and behaviors during marital interactions. At the same time, parental reports on the brief daily diaries indicated minimal differences between the groups in global measures of marital functioning. Husbands' self reports in the home indicated decreasing perceptions of marital quality over the recording period, suggesting husbands' reactivity to the diary recording. Thus, although no differences in conflict tactics emerged in the laboratory, self-reports in the home suggest potential reactivity effects.

1.1 Introduction

Within marital research, questions of ecological validity concerning traditional methods of measurement (e.g., global self-report questionnaires, laboratory observations) have prompted researchers to develop new methods that capture events as they occur in context (Bolger, Davis, & Rafaeli, 2003; Cummings, Davies, & Campbell, 2000; Green, Rafaeli, Bolger, Shrout, & Reis, 2006). A common criticism of questionnaire data is that participants must rely on recollection of events, feelings, and behaviors and that reliance on memory could affect the accuracy of the reports (Stone & Shiffman, 2002). At the same time, results based on laboratory observation may not generalize outside of the lab setting.

One proposed methodological advance is the use of diary methods. Diary methods include a variety of strategies, also known as event experience sampling (Csikszentmihalyi, & Larson, 1987), or ecological momentary assessment (Stone & Shiffman, 1994), that allow participants to record behaviors, emotions, and environmental factors as they occur in natural settings (Green et al., 2006; Laurenceau & Bolger, 2005). Recording events in this way decreases participants' reliance on memory, fostering reports that are more accurate. Diary methods also eliminate concerns about the generalizability of responses outside of the laboratory because participants document events in their natural contexts.

Recent increase in the use of paper and electronic diaries has called attention to the need for further study of the effects and quality of the data collected with diary methods (Green et al., 2006). Although diary measures provide unique information that cannot be captured in global self-report questionnaires and observational techniques, some worry that repeatedly filling out detailed assessments focusing on key emotional and behavioral dynamics could inadvertently affect the constructs researchers are trying to understand. Relatedly, the process of attending to significant details of interpersonal interactions may be seen as potentially therapeutic, that is, acting to change behavior. Being aware of and reporting on detailed aspects of one's experiences could have the unintended effects of experiencing events differently, reacting to experiences differently, or habituating to experiences (Bolger et al., 2003). For example, using electronic diaries, Marco, Neale, Schwartz, Shiffman, & Stone (1999) failed to find support for their theoretically and empirically based hypotheses that coping predicted stress-related mood changes. In the discussion of their null findings, the authors suggested that the process of self- monitoring might have increased participants' awareness and altered cognitions, actions, and emotions over the response period, thereby undermining relations between coping and stress-related mood changes. Despite this concern, few studies have examined the possibility that diary procedures designed to increase the intensity and accuracy of reports about family events could influence participants' emotions, behaviors and cognitions.

One recent study examined the effects that completing daily handheld computer diaries had on behavioral and motivational reactivity related to drinking behaviors. Hufford, Shields, Shiffman, Paty, & Balabanis (2002) found that completing daily diaries had a small effect on participants' motivational reactivity, and no effect on their behavioral reactivity. In other words, participants showed no change in their reports of drinking behaviors, but they showed significant differences in readiness to change their behaviors. Although there was no control group in this study, raising concerns about internal validity, the results provide initial evidence that completing daily diaries have minimal effect on behavioral reactivity.

The effects of writing about stressful daily experience such as chronic pain has been examined within health psychology suggesting limited reactivity to daily monitoring procedures. Aaron, Turner, Mancl, Brister and Sawchuk (2005) examined reactivity to completion of electronic diaries about pain in patients with chronic temporomandibular disorder. They found that although the patients believed the daily diaries had effects, analyses from the diaries themselves showed no change over time in reported pain. In a randomized study of pain monitoring, Stone et al., (2003) found limited reactivity in terms of self-report experience of pain in patients who experience chronic pain. Stone et al. (2003) also examined effects of different sampling densities (reporting 3, 6 or 2 times per day) finding minimal differences between groups.

The aim of the current study is to examine participant reactivity in the context of completing detailed and extensive daily diaries about marital conflict. In-home checklists and diaries have long been used in marital research to assess couples' daily interactions (Weiss & Perry, 1983) and as a therapy outcome measure (Margolin, 1990). We are not aware of any systematic study of behavioral reactivity in this context, although researchers have noted the possibility of participant reactivity (Margolin, 1990). If the process of closely monitoring one's behavior has the effect of changing behavior, cognitions, or emotions, researchers using diary methods must cautiously interpret data collected with these instruments.

Event contingent diaries call the most attention to the elements of family events, including marital conflicts, and thus are most likely both to be useful for overcoming limitations of retrospective report instruments (e.g., questionnaires) and to foster participant reactivity. Event-contingent diaries are especially useful in gaining information about clearly defined events as they happen naturally (Bolger et al, 2003). Using an event-contingent diary, participants are asked to complete the instrument following specific event(s) of interest. This methodology is especially informative if a researcher wants particular details of events, which might be missed or forgotten using questionnaire approaches (Bolger et al., 2003). On the other hand, event-contingent diaries draw intensive attention to the phenomena under study, thereby increasing the possibility of changing processes of responding, that is, participant reactivity. One hypothesis would be that training participants to identify specific behaviors and reactions to events would increase their reactivity to those events, potentially changing their interactions.

Reactivity has been defined as changes in the variable under study as a function of that variable being monitored or observed (Nelson, 1977). Hufford, et al. (2002) suggested a more general definition of reactivity as, “a source of unwanted variance in the dependent variable of interest” (p. 206). In the current study, participant reactivity refers to changes in the individual's responses to family or other psychologically meaningful events due to the repeated completion of diary records about the phenomenon over time. Thus, changes in responding in the home as well as behavioral changes in the laboratory were examined.

A viable alternative hypothesis is that any daily participation that requires simply attending to family events may result in participant reactivity, regardless of whether couples are engaged in completing detailed diaries of disagreements. A further advance in the current study is the inclusion of a control group that also completes home reports during this period. Accordingly, serving as a control for placebo effects of participation, a control group as well as the diary group was instructed to complete brief daily checklists, with minimal training, over the recording period.

The current study examines whether monitoring daily conflict behaviors and emotions as they occur in context affects participants' behavioral and emotional patterns of responding during subsequent conflict. Data presented here were drawn from a longitudinal study of marital conflict and family functioning and are presented here as a secondary data analysis. Both in-home checklists and laboratory interactions were used to assess change in marital variables as measures of reactivity.

1.2 Method

1.2.1 Participants

Participants were drawn from a longitudinal study examining the effects of specific marital conflict tactics on child outcomes. Married (n=162) and cohabiting non-married couples (n=5) who had been living together for at least 2 years and had a child between the ages of 8 and 16 years were recruited to participate. Recruitment strategies included letters sent home to schools, postcards send to families and fliers posted throughout the community, as well as newspaper, radio, and television advertisements. The current study involves a sub-sample of couples (n=167); n = 110 in the treatment condition, n = 57 in the control condition. Ninety-four percent of the couples were Caucasian, 4% were African American and 2% were biracial or other. Wives had a mean age of 37.64 years (SD = 5.88), and husbands' mean age was 40.15 (SD=6.41). Husbands' had a mean score of M = 110.61 (SD = 21.86) on the Short Marital Adjustment Task (SMAT; Locke & Wallace, 1959) and wives' had a mean score of M=113.62 (SD=23.23). Using a score of less than 100 by either spouse on the SMAT, 42% of the total sample was classified as maritally dissatisfied. Median income for both groups was in the range $40 000-$65 000.

1.2.2 Procedure and Measures

Couples came into the lab twice, with at least 15 days separating visits. In addition to completing multiple self-report questionnaires regarding various family and personal variables, at the first laboratory session, couples engaged in a Marital Conflict Resolution Task in which they were asked to discuss a topic that they typically had difficulty handling. These interactions lasted 7.5 minutes and were videotaped and later coded for specific behaviors (e.g. verbal anger, calm discussion, defensiveness) and emotions expressed (anger, positivity, sadness and fear) and degree of resolution achieved (see Du Rocher Schudlich & Cummings, 2003). Behaviors were coded on a 3-point scale indicating the degree to which each partner demonstrated the given behavior (0 = absence of the behavior and 2 = strong display of the behavior). Emotions and resolution were rated on 10-point scales (0 = absence of the emotion, 9 = strong display of emotion). After extensive training, the reliability of the rating of research assistants was high. Mean cronbach alphas for husbands' conflict tactics were .89 (range: .80-.99) and emotions were .89 (range: .84-.96). Mean alphas for wives tactics were .86 (range: .55-.99) and emotions were .88 (range: .74-.96). The alpha for degree of resolution, achieved was .90 for husbands and .91 for wives. The marital conflict literature provides support for the external validity of marital conflict responses in the laboratory (e.g., Gottman & Krokoff, 1989).

Following the marital conflict resolution task, couples in the diary group were trained to complete Marital Daily Records and Child Response Records (MDRs and CRRs; Cummings, Goeke-Morey, Papp, & Dukewich; 2002), an event contingent paper diary assessing aspects of marital conflict, including time and length, topics, tactics used, emotions expressed, resolution achieved, and attributions made, as well as children's responses to conflict. For the purpose of this study, marital conflict episodes were defined as any difference of opinion between partners. Research assistants trained husbands and wives to reliably recognize and report their own, their partners', and their child's emotions and behaviors during marital conflict episodes. The one-hour long training session included verbal and written descriptions of specific behaviors as well as presentation of video clips of actors simulating each behavior. Both husbands and wives demonstrated they could reliably identify the specific behaviors. Mean kappa for mothers was .88 (range .74-.99) and mean kappa for fathers was .85 (range= .69-96).

Couples completed the MDRs and CRRs separately following each episode of conflict for 15 days. Research assistants called couples throughout the recording period to remind them to fill out the MDRs and to answer any questions about the reporting procedure. After the recording period, couples returned to the lab with their completed diaries. They again participated in a conflict resolution observation in which they were videotaped discussing a topic that they did not handle well; these interactions were coded for conflict behaviors, emotions and resolution (Du Rocher Schudlich & Cummings, 2003). To check that couples were still able to identify conflict behaviors, participants watched a second set of training tapes and coded the behaviors. Reliability estimates indicated that couples were able to identify the conflict behaviors at the end of the reporting period (mean kappa for mothers was .83; mean kappa for fathers was .81). Additionally, utilizing a subset of diaries that were matched between husbands and wives during the recording period, HLM analyses indicate significant agreement between spouses on conflict tactics used in the home (See Cummings, Goeke-Morey & Papp, 2003), indicating that couples were able to reliably identify categories of responding during marital conflict in the home during conflicts as well as in the laboratory.

The dependent variables for analyses of the videotaped conflict resolution task precisely matched the emotions and behaviors coded during the event-contingent diaries. Thus, these analyses provided an exact test of whether the specific marital conflict behaviors that were recorded during event-contingent diaries but observed in the videotaped laboratory interactions were in any way affected by the completion of these diary records, which have informed multiple published studies (e.g., Cummings et al., 2002; 2003). The coding of these responses pre and post the completion of diary records has the distinct advantage of providing an objective record of any changes that is independent of any possible biases in spouses' self-reports.

In addition to the diary treatment group, we used a control group of couples who did not complete detailed, event-contingent diaries. Because the question of interest in this study is whether awareness of conflict processes and intensive diary reporting of conflict interactions alters the nature of couples' conflict, we determined that an appropriate comparison condition, which would control for simply the activity of completing home reports, was a brief, global, daily home report with minimal training. Accordingly, husbands and wives in both the diary treatment group and the control group received a brief instruction session (lasting approximately five-ten minutes) and completed brief, global daily checklists at home each day for 15 days. On these checklists, participants answered several questions about family relationships, including the number of marital conflict episodes (using the same definition as for the extensive diary procedure), and ratings of the emotional quality of the marital relationship for that day.

In order to control further for placebo effects brought on by researcher interest in the couple, the protocols for lab visits were identical for both groups, with the exception of the diary treatment (i.e., extensive training to be aware of conflict processes and provide detailed diary reporting of conflict interactions). That is, both the diary and control groups completed the same questionnaires and lab tasks (e.g. marital conflict resolution task) in the same order, both received some degree of training from a research assistant to record behavior and relationships at home, and both received reminder telephone calls during the reporting period. The minimal training, reporting time requirements, or global ratings made by the control group were not expected to make couples increasingly aware of their marital interactions. No mention was made during checklist training about specific conflict topics, behaviors, emotions, or cognitions, about conflict resolution, or about children's responses to marital conflict.

Although subjects were not randomly assigned to groups, we did not expect differences between the two groups on demographic variables or functioning (individual, marital). The control group was added after the onset of the study. Assignment to groups was not random but determined by the time the family entered the study. Independent t-tests show no significant differences between groups in marital satisfaction scores (Wives: t (165) = 0.75, p>.05; Husbands: t (165) = 0.28, p>.05) or in general psychological symptoms (Wives: t(165) = 0.67, p >.05; Husbands: t(165) = 0.90, p>.05), as measured by the general severity scale of the Symptom Checklist-90-R (Derogatis, 1994). Comparing the diary and control groups shows no significant difference in marital satisfaction classification for husbands (χ²(1) = 1.53, p > .05) or wives (χ²(1) = .25, p > .05).

1.3 Results

Means and standard deviations for conflict behaviors used and emotions expressed during the marital conflict resolution task at time 1 and time 2 are presented in Tables 1 and 2. Several behaviors including physical aggression towards an object, and physical aggression towards a person, occurred very infrequently or not at all during the interactions, thus these variables were excluded from additional analyses. As expected, t-tests of all behaviors and emotions revealed that no significant differences existed between the diary and control groups on any dimension at time one, reducing any concerns about possible effects due to nonrandom assignment. Mean differences ranged from 0.03 to 0.24 for emotions (ts (162) range: .07 - 1.36) and from .004 to 0.23 for conflict behaviors (ts (165) range: 0.04 -1.50).

Table 1.

Means and Standard Deviations of Husbands' Tactics and Behaviors Expressed

	Diary Group				Control Group
	Time 1		Time 2		Time 1		Time 2
	M	SD	M	SD	M	SD	M	SD
Emotions (range 0-9)
Positive	3.51	1.52	3.50	1.67	3.50	1.77	3.74	1.75
Angry	2.55	2.04	2.60	1.98	2.38	2.25	2.19	2.06
Afraid	0.17	0.70	0.11	0.50	0.26	0.79	0.19	0.66
Sad	0.37	1.26	0.55	1.49	0.14	0.61	0.38	0.90
Behaviors (range 0-2)
Defensiveness	0.25	0.64	0.33	0.69	0.36	0.74	0.19	0.51
Humor	0.16	0.46	0.14	0.46	0.22	0.56	0.16	0.45
Problem Solving	1.24	0.96	1.14	0.93	1.00	0.99	0.90	0.97
Nonverbal Hostility	0.21	0.54	0.30	0.67	0.22	0.56	0.45	0.73
Verbal Hostility	0.22	0.50	0.26	.060	0.22	0.56	0.17	0.50
Physical Affection	0.13	0.47	0.13	0.47	0.16	0.52	0.29	0.70
Support	0.32	0.68	0.31	0.66	0.45	0.78	0.31	0.63
Personal Insult	0.03	0.16	0.04	0.27	0.07	0.32	0.05	0.22
Physical Distress	0.04	0.27	0.05	0.28	0.00	0.00	0.00	0.00
Resolutiona	4.71	2.12	4.65	2.08	4.86	2.61	4.26	2.26
Composites
Positive	12.04	4.37	11.84	4.31	12.12	5.27	11.53	4.86
(range: 0-28)
Depressive	0.63	1.63	0.73	1.89	0.43	1.13	0.60	1.18
(range: 0-22)
Destructive	3.29	3.17	3.56	3.26	3.28	3.60	3.12	3.28
(range: 0-21)

^aRange for resolution : 0-9

Table 2.

Means and Standard Deviations of Wives' Tactics and Behaviors Expressed

	Diary Group				Control Group
	Time 1		Time 2		Time 1		Time 2
	M	SD	M	SD	M	SD	M	SD
Emotions (range 0-9)
Positive	3.45	1.44	3.67	1.63	3.48	1.63	3.67	1.59
Angry	2.55	2.20	2.98	2.21	2.50	2.55	2.71	2.14
Afraid	0.31	0.93	0.36	1.11	0.24	0.80	0.34	0.87
Sad	0.57	1.56	0.87	1.77	0.55	1.76	1.52	2.44
Behaviors (range 0-2)
Defensiveness	0.15	0.51	0.29	0.65	0.10	0.41	0.19	0.55
Humor	0.09	0.29	0.06	0.31	0.10	0.36	0.03	0.18
Problem Solving	1.38	0.88	1.11	0.93	1.12	0.98	0.98	0.98
Nonverbal Hostility	0.23	0.55	0.43	0.76	0.36	0.72	0.60	0.84
Verbal Hostility	0.30	0.60	0.34	0.69	0.24	0.60	0.26	0.61
Physical Affection	0.19	0.57	0.25	0.62	0.26	0.64	0.43	0.78
Support	0.35	0.70	0.32	0.66	0.52	0.84	0.40	0.70
Personal Insult	0.09	0.40	0.09	0.40	0.10	0.41	0.07	0.26
Physical Distress	0.08	0.34	0.09	0.40	0.09	0.39	0.24	0.63
Resolutiona	4.69	2.12	4.41	2.00	4.74	2.57	4.24	2.24
Composites
Positive	12.11	3.19	11.78	4.37	12.16	5.06	11.71	4.43
(range: 0-28)
Depressive	0.98	2.13	1.35	2.71	0.88	2.19	2.16	3.17
(range: 0-22)
Destructive	3.38	3.48	4.26	3.84	3.35	4.00	4.00	3.68
(range: 0-21)

^aRange for resolution : 0-9

To test for differences in change over time between the two groups, 2 × 2 ANOVAS were conducted separately for husbands and wives. Reflecting the focus on participant reactivity due to training and completion of event-contingent diaries about marital conflict, report of results is limited to instances involving main effects for group and interactions between time and group. Given the large number of analyses conducted, significant p-values were adjusted for the number of tests completed within each family of analyses. Within all time by group interactions, families of analyses were defined by emotions, behaviors, and composite scores resulting in critical p-values of .0125 for emotions (p = .05/4), .005 for behaviors (p =.05/10), and .0167 for composites (p = .05/3). Significant interactions between time and group would indicate the two groups are changing differently over the 15-day period. For husbands and wives, the results show no significant main effects for group and no significant interactions, indicating no differences in change between the diary and control groups in conflict tactics expressed in the laboratory interaction task. For husbands, F values (1, 162) for the interactions between time and group predicting emotions ranged from 0.01 to 0.47, and F-values (1, 162) for conflict behaviors ranged from .01 to 3.21. For wives, the interactions between time and group predicting emotions ranged from 0.01 to 0.03, and F-values (1, 162) for conflict behaviors ranged from .03 to 2.63. Power analyses indicate that the current sample was sufficiently large to detect even small effect sizes with a power of .80 (Maxwell & Delaney, 2004).

Composite scores were also created to allow for exploratory tests for differences between groups in larger categories of conflict strategies used. The assumption was that composite, macro-variables might possibly be more sensitive to differences between conditions by increasing the variability of dependent variables than micro assessment of marital conflict behavior. Positive, destructive, and depressive composites were created by summing across several emotions and behaviors previously identified as reflecting these three categories of conflict styles (DuRocher Schudlich & Cummings, 2003). For these analyses, all tactics and behaviors were included to maximize variability. The constructive composite includes positive emotion expressed, humor, support, physical affection, calm discussion, problem solving and resolution. The destructive dimension includes verbal and nonverbal hostility, defensiveness, pursuit, personal insult, physical aggression and anger. The depressive composite includes physical distress, withdrawal, sadness and fear. Similar to the previous results, 2 × 2 ANOVAs showed no significant interactions for husbands or wives. F-values (1, 162) for the composites ranged from .09 to 2.62.

1.4 Changes in Parent Report Daily Checklists

The brief daily checklists completed by both groups provided an ecologically-valid test of possible reactivity effects on responding occurring in the home. Hierarchical linear modeling (HLM; Raudenbush & Bryk, 2002) was employed to test for changes in responding to the global index of marital emotional quality, and number of interactions reflecting disagreements between partners over the recording period. Although these variables do not test specifically for reactivity in terms of change in marital conflict tactics and emotions, they provide valuable information about possible changes in marital conflict or functioning during the recording period. For these models time (i.e., number of days since the beginning of the reporting period) was used as the predictor in the level one equations and group membership (i.e., diary or control) was used as the predictor in the level two models. Constructing models in this way allowed for the examination of change in outcome variables as a function of time in the study. Including group membership at level two allowed for the examination of differences in trajectories in the outcome variables as a function of being in the diary group versus the control group. For tests of change over time in number of interactions between spouses, hierarchical generalized linear modeling was used with a Poisson distribution to take into account the non-normal distribution of count data.

Descriptive statistics for the checklist variables appear in Table 3. Conditional models were tested with group as the predictor (0 = control; 1 = diary) at level two for all four variables. Both husbands (Diary: γ = -.001, SE = .01, p > .05; Control: γ < -.001, SE = .01, p > .05) and wives (Diary: γ = -.002, SE = .01, p > .05; Control: γ = -.001, SE = .01, p > .05) in both groups reported no change over time in the probability of a disagreement occurring over the reporting period. For husbands, group status was a significant predictor of change in marital quality with husbands in the diary group reporting a decrease in the quality of the emotional relationship with their spouse over time (γ = -.04, SE = .01, p < .05); no change was observed in the husbands in the control group (γ = .02, SE = .01, p > .05). Wives reported no change in the quality of the emotional relationship with their husbands (Diary: γ = -.001, SE = .02, p > .05; Control: γ = .01, SE = .01, p > .05)

Table 3.

Descriptive Statistics for Checklist Variables

		Diary Group			Control Group
		M	SD	Range	M	SD	Range
Husbands
	Marital Quality	6.45	2.02	0-9	6.83	1.87	0-9
	Marital Interactions	0.86	0.98	0-8	2.37	2.23	0-17
Wives
	Marital Quality	6.75	2.16	0-9	7.06	1.71	0-9
	Marital Interactions	0.98	1.09	0-8	2.70	2.67	0-22

Discussion

With the increased use of diary studies in multiple subfields of psychology, as well as for studying family processes, the question of reactivity effects is an increasingly significant concern. The goal of the current study was to assess the effects of completing event-contingent diaries about marital conflict on husbands' and wives' emotional and behavioral reactivity during conflict. The current results suggest that completing highly detailed event-contingent diaries about marital conflict tactics and emotions during marital disagreements for two weeks has minimal effect on husbands' and wives' emotions and behaviors during subsequent marital interactions; however, recording events in this way may have implications for perceptions of marital quality over time.

Multiple emotions and behaviors were examined in marital conflict interactions in the lab prior to and following the diary reporting period, with no significant differences emerging between the group completing diaries and a control group. That is, training in observing one's partner's behaviors and emotions on psychologically meaningful aspects of an important category of interpersonal interaction (i.e., marital conflicts) did not systematically affect how marital partners argue in the laboratory after 15 days of recording daily interactions. Given the recent increased use of diary methodologies in social science research (Green at al., 2006) the present findings lend support for the use of these ecologically advanced tools, reducing possible concerns about overt behavioral or emotional reactivity. Of the multiple analyses completed, no significant differences in change over time emerged, suggesting that extensive training and 15 days of event-contingent diary completion did not result in systematic change in couples' emotions expressed and conflict behaviors used as observed in the laboratory interaction.

Thus, although the current study had a sufficiently large sample to detect small effects, no differences were detected between the two groups in observable responding in highly detailed coding of interparental interactions, supporting the use of highly trained participant reporters in marital research. Participants in the diary group were trained to report reliably on the emotions and conflict behaviors of themselves and their partners, and they remained reliable over the recording period as indicated by their high kappa values at the time 2 identification task. Despite their sustained sensitivity to recognizing these specific elements of marital conflict, they showed no differences in change over time for behaviors or emotions expressed during the conflict resolution task. These results should be interpreted cautiously recognizing that marital interactions in the lab are likely to be less sensitive to reactivity effects compared to actual interactions that occur in the home; however they provide initial evidence that sustained changes in interaction patterns did not occur as a results of the diary procedure.

Results from the in-home checklists provide additional evidence that overt changes in marital interactions did not change as a function of training to identify and record conflict behaviors. Both husbands and wives reported no change over time in the number of marital disagreements that occurred over the recording period. In a previous study with a different sample (n = 51) and applying identical diary procedures (Cummings, 1999), husbands and wives were asked if the recording procedure affected naturally occurring interactions in the home with many participants indicating it had no influence on their interactions. Although participants did indicate that they were more aware of and attentive to their disagreements, only one couple indicated even anecdotally they had fewer disagreements resulting from the diary procedure and no participants indicated that the number of disagreements increased.

Despite the lack of difference in changes in behavior in the laboratory marital interaction task and in daily reporting of conflict frequencies, differences in trajectories between the diary and control groups did emerge in subjective reports. Specifically, husbands in the diary group reported a decline in their perception of marital quality over time. One interpretation is that husbands were made more aware of the conflict and tension in the marital relationship by reporting in detail on specific interparental disagreements, leading to a relative decline in their perceptions of marital quality over time. By contrast, wives may already be more aware of the negative elements of marital conflict episodes than husbands, so that this experience in attending to the specifics of marital conflicts had relatively little impact on their perceptions of the marital relationship. The pattern of results is similar to previous studies showing changes in subjective feelings about the phenomenon under study, but little change in observer ratings and objective ratings of the same constructs (e.g., Aaron et al., 2005; Hufford et al., 2002).

The current results must be interpreted with recognition of the study's limitations. First, change in reporting of marital conflict tactics and emotions based on the diary measure itself could not be examined with the current data set, resulting in measures of reactivity that might be less sensitive to change. Examining change over time in the marital daily records would provide a more sensitive test of behavioral and emotional reactivity in the home. An additional concern with using paper diaries is that they can be filled in after the fact, or at the end of the day, potentially limiting participant reactivity (Bolger et al., 2003; Broderick & Stone, 2006). In addition, because the larger study was not designed to study participant reactivity, participants were not randomly assigned to groups. Finally, the brief, global measure of marital quality does not provide a psychometrically established assessment of marital satisfaction and adjustment, so the findings with regard to husband's perceptions of marital quality should be interpreted with extreme caution.

Consistent with previous studies demonstrating minimal effects of completing diaries (Aaron et al., 2005; Hufford et al., 2002), the current study suggests that completing event-contingent diaries in the context of marital disagreements, despite the fact that these events involve categories of emotional and behavioral responding that are linked with psychological adjustment and well-being (Cummings & Davies, 2002), does not appear to change the ways in which couples interact during conflict episodes. The results do not support anecdotally based opinions that completing diaries will change marital behavior. This study provides a relatively stringent test, because marital partners were highly and reliably trained on recognizing and reporting on interparental behaviors and emotions during conflicts over relatively long periods, marital partners demonstrated reliability in reporting in both home and laboratory assessments, the statistical power of analyses was adequate to detect even small effects, and a wide range of potentially changeable behaviors were systematically reported on by the couples. On the other hand, there was some evidence for change in in-home reports of relationship quality suggesting that researchers should be aware of potential changes in subjective feelings as a possible reactive effect to completing detailed diaries.