Larry H. Ludlow
Rose M. Alvarez-Salvat
Boston College
Lynch School of Education
Journal of Personnel Evaluation in Education, 2001, 15:2, p. 111-119
Running Head: Spillover in the academy
Address: Larry H. Ludlow, Ph.D. , Boston College, Lynch School of Education, 140 Commonwealth Avenue, Chestnut Hill, MA, 02467-3813.
Work: (617) 552-4221.
Fax: (617) 552-1981.
Email: Ludlow@BC.EDU
Spillover in the academy: Marriage stability and faculty evaluations
Abstract
The literature on the relationship between work and family has shown that there is a "spillover"" effect between both domains. In particular, research that has investigated the influence of home environment on work has shown that family instability affects work satisfaction and performance. This study investigates the spillover between family and work by examining the link between marital status and work performance across the three phases of marriage, divorce, and remarriage. Specifically, this article links marital status and work performance through a longitudinal analysis of a set of university teaching evaluations. A polynomial regression model was fit to the data and a cubic curve through the three periods of marriage, divorce, and remarriage was statistically significant. Implications of the study and areas for future research are discussed.
Key Words: Spillover effect, instructor evaluations, longitudinal analysis, work and family
Spillover in the academy: Marriage stability and faculty evaluations
For most, if not all of us, it is a fact that work and family contribute greatly to our identity, self-concept, and life satisfaction (Axelrod, 1999). Furthermore, we understand there exist a myriad of interactions between work and family that lead to life satisfaction (Blustein, 1997). The research on relationships between work satisfaction and marriage stability, in particular, is extensive and is primarily located within the literature on marital satisfaction, work identity and satisfaction, and dual career couples (Blair, 1998; Gaesser & Whitbourne, 1985; Ray, 1990). These studies have shown how experiences in one arena of an individual's life have an impact upon experiences in another arena of one's life. This interchange, or "spillover", between arenas or domains suggests that career and family lives are entangled with one another and that to understand strain in one domain it is essential to have information on both facets of an individual's life.
Most of the spillover studies have investigated how work or
career satisfaction affects one’s
personal life. Benin and Nienstedt (1985), for example,
examined how job satisfaction affects marital happiness
and global happiness. They found that job satisfaction influenced marital happiness
and the effects of job satisfaction and fulfillment interacted
with the effects of marital happiness in producing global
happiness. Barnett (1994) found that the relationship between work and marital
quality was significant for both men and women. As job
role quality decreased, psychological distress increased
and marital quality decreased.
Gaesser & Whitbourne (1985) investigated worker identity and concluded that an intrinsically based work identity (salary, security, relationship with co-workers, and the work setting and hours) reduces the energy available for commitments outside of one's work responsibilities. Perrewe, Hochwarter, and Kiewitz (1999) examined the relationship between value attainment and work-family conflicts and concluded that work interference with family had a negative relationship with attaining one’s values.
In contrast to these work-to-home spillover studies, research has examined how family obligations, marriage satisfaction, and family environment lead to satisfaction or dissatisfaction at work. Greenhaus & Beutell (1985), for example, found that family size and family conflicts were positively related to work conflicts. In addition, Barnett (1993) found that, for working men and women in dual careers, positive experiences in their partner or parent role buffered the effects of negative job experiences on psychological distress.
Chiu (1998) also examined the role conflict between family and work. This study indicated that family role overload due to demands and/or obligations from one’s family life, led to difficulty in dealing with other role activities, such as work. Eagle, Icenogle, Maes, and Miles (1998) also found that certain family situations were more likely to result in work-family interrole conflict. Family expectations did not directly affect time spent at work but did increase emotional strain and psychological fatigue, which in turn had a negative effect upon job performance.
Although these studies examined how various general family issues impact work quality and work satisfaction, less research has been done on how the specific circumstances of divorce and marital discord impact one's performance and happiness at work. Some studies have shown how divorce and marital discord affect the level of stress and daily responsibilities in one’s life, such as increased responsibilities at home, less concentration at work, and increased feelings of stress (Huddleston & Hawkings, 1991; Hope, Rodgers, & Power, 1999; Wheaton, 1990). In addition, Peterson and Gonzalez (2000) explain that work performance can suffer for both men and women during the process of divorce and long after the divorce has taken place. Some of the reasons for marital failure listed in a study of divorced men and women were infidelity, falling out of love, emotional or financial problems, communication problems, and job conflicts (Albrecht, Bahr, & Goodman, 1983). However, minimal research has been done specifically on how such antecedents and consequences of divorce impact work satisfaction and performance.
In addition to the lack of research on the relationship between divorce and work, research on the relationship between remarriage and work satisfaction has also been minimally investigated. Some research on remarriage has indicated inconclusive findings where remarriages have been found to be less, equally, and more satisfying than first marriages (Albrecht, Bahr, & Goodman,1983). However, recent research has shown individuals are more satisfied with their new marriage than their previous marriage and remarried individuals are more likely to express overall satisfaction with their lives (Buunk, & Mutsaers, 1999; Demo, & Acock, 1996).
To build upon the previous research on the spillover between family and work we hypothesize that work satisfaction and performance can be linked to marital satisfaction through all three phases of marriage, divorce, and remarriage. Specifically, the purpose of this research is to propose and test a model linking the relationship between marital status and work performance through a longitudinal statistical analysis of a set of university teaching evaluations. In addition, this study contrasts two statistical approaches to modeling the spillover linkage between marital status and work performance. The statistical models differ in their characterization of the three marital experiences as either discrete, independent stages or as continuous phases of an interwoven process. Furthermore, in contrast to the typical group-level cross-sectional research design, this study employs a single-subject time-series design.
Method
The data come from a single university professor’s course evaluations from the years 1983 to 2000. The evaluations were conducted at the end of each semester. The courses range from undergraduate child development and research methods courses, to graduate introductory and advanced statistics courses and seminars. They include fall, spring, and summer courses and ranged in size from four to 52 students. All the courses were offered through a school of education at a university in the New England area. There are a total of 78 course evaluations.
The specific questions that students answer are fairly typical of this type of evaluation. They ask about the percent of time students spent on the course relative to other classes, the extent to which principles and concepts were understood, availability of the instructor outside of class, instructor enthusiasm, instructor subject matter knowledge, and extent to which factual information was acquired. Most questions are scored from 1 to 4, where 4 is the most favorable rating. The university-generated evaluation reports include an overall percent "excellent", "very good", "good", "acceptable", and "poor" rating of the professor for each class. Class enrollment was also available.
Of particular interest to the present analysis is the coding of the professor’s marital status over this time period (i.e., married, separated/divorced, and re-married). He was married when the evaluations began (1983), separated and eventually divorced after the 44th evaluation (1992), and re-married after the 62nd evaluation (1997). The hypothesis that prompted this research was that there was a statistical relationship between the teaching ratings and his marital status. Specifically, it was hypothesized that the percent of "excellent" teaching evaluation ratings suffered during the tumultuous period of separation and divorce and that those same ratings improved with re-marriage. Furthermore, the hypothesis is decidedly directional. That is, home life (as defined by the marital stages) influenced work performance (as defined by the evaluations), not the converse.
Results
An ordinary least squares (OLS) hierarchical regression was performed (SPSS 9.0). The percent "excellence" ratings served as the outcome variable. Marital status was the primary predictor. It was dummy coded to contrast the married vs. separated/divorced conditions ("DUM1") and the re-married vs. separated/divorced conditions ("DUM2"). The dummy vectors could have been constructed differently but the percent of variance they accounted for would have been exactly the same (Pedhazur, 1997).
The comparisons between the mean ratings for the three marital conditions could have been tested though a one-way analysis of variance but in order to maintain continuity between the different statistical analyses the regression approach was taken. Since class size was correlated to evaluation ratings in these data (r=-.523. p<.0005) it was considered a confounding variable whose influence should be partialled prior to testing the marital conditions. This decision was based on the fact that the variable was beyond the control of the professor and was independent of marital status. The full regression model is, therefore, equivalent to an analysis of covariance testing for differences between the marital conditions after removing confounding variance attributable to class size (Huitema, 1980).
The full model accounted for a statistically significant percent of variation in the excellent ratings (R2=.301, F=10.324, p<.0005). The effect due to class size was statistically significant (b = -.897, t = -4.925, R2 = .273, p<.0005) but the unique effect due to the marital conditions was not (R2 change = .027, p=.251). In other words, there was no statistically significant difference between the mean evaluation ratings for the three marital conditions after partialling the effect attributable to class size.
At this point it is important to recognize that all that this statistical model tested was whether or not the mean ratings differed significantly or not. This means that the model treated the effects of marriage, separation/divorce, and re-marriage as though they were discrete, independent time periods. That is, the model assumed the effects of marital status on work (through the evaluations) would be bounded by the start and end of each period. The model does not say anything about effects upon work that might be evident prior to separation/divorce and prior to re-marriage. Furthermore, the model treated the evaluation data as if they were a linear function across all three time periods--regardless of whether or not the slope of that function within individual time periods was, in fact, flat, positive, or negative.
Those issues are addressed in the next analysis. Specifically, an OLS polynomial regression model was fit to the data. This model tested the extent to which there was a positive slope to ratings in the married period, a flat or negative slope in the separated/divorced period, and a shift back to a positive slope in the re-married period.
The polynomial components were based on the sequence order of the evaluation ratings. The linear component consisted of the sequential evaluations numbered from 1-78 = " SEQ1 ", the quadratic component was SEQ12 (or " SEQ2 "), and the cubic component was SEQ13 (or " SEQ3 "). As before, class size was entered first, followed by the linear, quadratic, and cubic components testing the trend of excellence ratings over time.
Figure 1 contains the cubic curve fit to the data from the first course to the last. The "percent excellent ratings" are given a different symbol within each period. The vertical dashed lines mark the transition points from married to separated/divorced and separated/divorced to re-married. Note how the curve starts off with a positive slope in the married period at the beginning of the professor’s tenure, then takes a negative dip in slope prior to the onset of the separated/divorced period, and then reverses direction and takes a positive slope prior to and continuing in the remarried period. Not only do the slopes differ in the three periods but the slopes change direction prior to the formal transition points.
Insert Figure 1 about here
The results of the regression are detailed in Tables 1, 2, and 3. The full model is statistically significant (R2=.352, F=9.635, p<.0005). The effect due to class size is the same as before (b=-.897, t= -5.011, R2=.273, p<.0005,). The polynomial tests, however, differ from the dummy coded tests performed previously. As a set they accounted for statistically significant variance (R2 change = .078, p=.043). Each individual predictor was statistically significant: linear effect (b=2.244, t=2.204, p=.024), quadratic effect (b=-6.996E-02, t=-2.436, p=.017), and cubic effect (b=6.133E-04, t=2.550, p=.013).
Insert Table 1, 2, and 3 about here
Various model assumptions were tested at this point. The Durbin-Watson statistic testing for a lag-one autocorrelation was not statistically significant. The residuals fit to a normal curve were reasonably normally distributed. The conditional variances of the residuals were reasonably homoscedastic. The two largest studentized residuals were due to two courses with higher than expected percent excellence ratings.
Discussion
The literature on work and family relationships indicates that the family domain has a great impact upon the work domain (Gaesser, & Whitbourne, 1985; Greenhaus, 1988; Greenhaus, & Beutell, 1985). Family stress, family conflict, and family role overload can became more than a problem with the family, it can have a negative impact upon work performance and job satisfaction. Furthermore, despite the limited research done in the area of divorce and remarriage and its relationship to work satisfaction, it is understood that divorce and remarriage may have a direct positive or negative effect upon work performance and satisfaction. There has not, however, been any previous research that has posed and statistically tested a longitudinal model of how family life can affect, or be detected in, work performance over the periods of marriage, separation/divorce, and re-marriage.
Thus, the purpose of this study was to explore the relationship between family and work through a statistical model that tested the extent to which change in marital status was related to changes in work performance. More specifically, the relationships between the periods of marriage, divorce, and remarriage and one aspect of a professor’s work performance were examined through two competing statistical models. Both models tested the hypothesis that teaching evaluation ratings would suffer during the period of divorce, and then improve with remarriage. The polynomial regression model, however, proved more powerful in revealing the trend in the ratings over the three marital periods.
The overall results, then, support the spillover model, showing that changes in one’s marital status may be reflected in one’s performance at work, specifically for a faculty professor teaching at a university. When the marital situation was most stressful, teaching evaluations suffered. When the marital environment improved, so did the teaching evaluations. Furthermore, these phases of marriage and their effects on work performance were amenable to single-subject statistical modeling. This fact is important because it means that an investigation of extrinsic work conditions, e.g., marital status as a proxy for marital satisfaction, upon teaching is not restricted to a qualitative case-study or clinical analysis.
Some limitations to the study and areas for future research are worth noting. This analysis did not address the specifics underlying the changes in marital status. Certainly the changes in teaching evaluation ratings were not a simple consequence of change in marital status. The ratings were, in part, a function of many factors that led to the changes in marital status. Those factors, however, were not available for measurement and subsequent inclusion in the present analysis. Future research can more intensively examine how specific factors that lead to marital satisfaction or discord have a direct negative impact upon work satisfaction and performance.
Nonetheless, the statistical methodology and results are relevant for at least two interested parties: counseling psychologists and faculty evaluators. The field of counseling psychology has an investment in understanding what factors lead to overall life satisfaction. Personal and vocational counseling have often been artificially separated from one another, underestimating the impact that both domains have on one another (Blustein, in press). The present study, in addition to many others, shows that there is a spillover effect between marriage and work satisfaction. Blustein and Spengler (1995) acknowledge this interaction and encourage a domain-sensitive approach to intervention as a way to understand the full array of human experiences. Their approach aims to understand the importance and interdependence of a client’s work and family life, and its ultimate impact on life satisfaction. By understanding the interrelated facets in one’s life and intervening in these areas, an individual is seen as a whole person with many influencing aspects.
The present findings are also relevant for faculty evaluators, e.g. deans, invested in the optimal performance of their faculty. By understanding the interdependence of work and family satisfaction, evaluators would be able to provide guidance and support when faculty experience difficulty in either domain. This acknowledgement of how family may influence work may allow evaluators to intervene appropriately and prevent ultimate job loss.
Finally, this study has particular relevance for the faculty evaluation literature. These longitudinal data and their analysis suggest a statistical means of examining how an individual’s teaching evaluations may be affected not only be intrinsic work conditions such as class size but also by factors extrinsic to the work environment. The significance of such an extrinsic, longitudinal analysis should be evident when we consider the influence that teaching evaluations have upon high-stakes decisions around tenure, promotion, merit, and increments. Those deliberations often treat instructor evaluations as though they occurred in discrete time-frames (i.e. a semester at a time) that were independent of factors external to the classroom.
The present analysis illustrates that it is possible to study and interpret teaching evaluation ratings not only as a function of the specifics of the class environment but also as a longitudinal function of the personal situation of the instructor teaching the course. Furthermore, the present study illustrates that powerful statistical analyses of student teaching evaluations may be performed usefully at the instructor-level where evaluations count the most. This finding is important because it goes against the observation that aggregated cross-sectional results are often contradictory and ambiguous in terms of the extent to which teaching evaluations by students serve any useful function (Aleamoni, 1999; Wachtel, 1998).
References
Albrecht, S.L., Bahr,, H.M., & Goodman, K. L. (1983). Divorce and remarriage: Problems, adaptations, and adjustments. Westport: Greenwood.
Aleamoni, L.M. (1999). Student rating myths versus research facts from 1924 to 1998. Journal of Personnel Evaluation in Education. 13(2), 153-166.
Axelrod, S.D. (1999). Work and the evolving self: Theoretical and clinical considerations. Hillsdale, NJ: Analytic Press.
Barnett, R. C. (1994). Home-to-work spillover revisited: A study of full-time employed women in dual-earner couples. Journal of Marriage and the Family, 56(3), 647- 656.
Barnett, R. C., Marshall, N. L., Raudenbush, S. W., & Brennan, R. T. (1993). Gender and the relationship between job experiences and psychological distress: A study of dual-earner couples. Journal of Personality and Social Psychology, 64(5), 794- 806.
Benin, M. H., & Nienstedt, B. C. (1985). Happiness in single- and dual-earner families: The families of marital happiness, job satisfaction, and life cycle. Journal of Marriage and the Family, 47(4), 975- 984.
Blair, S. L. (1998). Work roles, domestic roles and marital quality: Perceptions of fairness among dual-earner couples. Social Justice Research, 11(3), 313- 335.
Blustein, D.L. (in press). The interface of work and relationships: Critical knowledge for 21st century psychology.
Blustein, D.L. (1997). A context-rich perspective of career exploration across the life roles. Career Development Quarterly, 45, 260-274.
Blustein, D. L., & Spengler, P. M. (1995). Personal adjustment: Career counseling and psychotherapy. In W.B. Walsh & O.S. Osipow (Eds.), Handbook of vocational psychology: Theory, research, and practice (pp. 295-329). Hillsdale, NJ: Lawrence Earlbaum Associates.
Buunk, B.P. & Mutsaers, W. (1999). Equity perceptions and marital satisfaction in former and current marriage: A study among the remarried. Journal of Social and Personal Relationships, 16(1), 123- 132.
Chiu, R. K. (1998). Relationships among role conflicts, role satisfactions and life satisfaction: Evidence from Hong Kong. Social Behavior and Personality, 26(4), 409-414.
Demo, D. H. & Acock, A. C. (1996). Singlehood, marriage, and remarriage: The effects of family structure and family relationships on mother’s well-being. Journal of Family Issues, 17(3), 388- 407.
Eagle, B.W., Icengogle, M.L., Maes, J.D., and Miles, E.W. (1998). The importance of employee demographic profiles for understanding experiences of work-family interrole conflicts. The Journal of Social Psychology, 138(6), 690-709.
Gaesser, D. L. & Whitbourne, S. K. (1985). Work identity and marital adjustment in blue-collar men. Journal of Marriage and the Family, 47(3), 747- 751.
Greenhaus, J.H. (1988). The intersection of work and family roles: Individual, interpersonal, and organizational issues. In E.B. Goldsmith (Ed.), Work and family: theory, research, and applications (pp.23-44). Newbury Park: Sage.
Greenhaus, J.H., & Beutell, N.J. (1985). Sources of conflict between work and family roles, Academy of Management Review, 10(1), 76- 80.
Hope, S., Rodgers, B., & Power, C. (1999). Marital status transitions and psychological distress: longitudinal evidence from a national population sample. Psychological Medicine, 29(2), 381-389.
Huddleston, R.J., & Hawkings, L.D. (1991). The effect of divorce on daily routine. Family and Conciliation Courts Review, 29(2), 150- 159.
Huitema, B. (1980). The analysis of covariance and alternatives. NY: Wiley.
Pedhazur, E. (1997) (3rd ed.). Multiple regression in behavioral research. NY: Harcourt.
Perrewe, P.L., Hochwarter, W.A., and Kiewitz, C. (1999). Value attainment: An explanation for the negative effects of work-family conflict on job and life satisfaction. Journal of Occupational Health Psychology, 4(4), 318- 326.
Peterson, N. & Gonzalez, R. C. (2000). The role of work in people’s lives: Applied career counseling and vocational psychology. Australia: Brooks/Cole.
Ray, J. (1990). Interactional patterns and marital satisfaction among dual-career couples. Journal of Independent Social Work, 4(3), 61- 73.
Wheaton, B. (1990). Where work and family meet: Stress across social roles. In J. Eckenrode, & S. Gore (Eds.), Stress between work and family: The Plenum series on stress and coping (pp.153-174). New York: Plenum.
Wachtel, H.K. (1998). Student evaluation of college teaching effectiveness: A brief review. Assessment and Evaluation in Higher Education, 23, 191-210.
Figure 1. Cubic relationship between excellence ratings and marital status
|
Model
|
R
|
R square
|
Adjusted R square |
Std error of the estimate
|
Change Statistics
|
Durbin-Watson | ||||
|
R square change
|
F change
|
df1
|
df2
|
Sig. F Change
|
||||||
| 1 | .523 (a) | .273 | .264 | 19.4483 | .273 | 27.851 | 1 | 74 | .000 | |
| 2 | .593 (b) | .352 | .315 | 18.7534 | .078 | 2.862 | 3 | 71 | .043 | 2.050 |
| a Predictors: (constant), CLASS ENROLLMENT | ||||||||||
| b Predictors: (constant), CLASS ENROLLMENT, SEQ1. SEQ2. SEQ3 | ||||||||||
| Model |
Sum of Squares
|
df
|
Mean Square
|
F
|
Sig
|
R square change
|
||
| 1 | Regression | 10534.075 | 1 | 10534.075 | 27.851 | .000(a) | ||
| Residual | 27989.367 | 74 | 378.235 | |||||
| Total | 38523.442 | 75 | ||||||
| 2 | Subset Tests | SEQ1, SEQ2, SEQ3 | 3019.483 | 3 | 1006.494 | 2.862 | .043(b) | .078 |
| Regression | 13553.558 | 4 | 3388.390 | 9.635 | .000(c) | |||
| Residual | 24969.884 | 71 | 351.689 | |||||
| Total | 38523.442 | 75 | ||||||
a. Predictors: (constant), CLASS ENROLLMENT
b. Tested against the full model
c. Predictors in the full model: (constant), CLASS ENROLLMENT, SEQ1, SEQ2, SEQ3.
|
Model
|
Unstandardized coefficients
|
Standardized Coefficients
|
t
|
Sig.
|
||
|
B
|
Std. error
|
Beta
|
||||
| 1 | (constant) | 55.244 | 4.552 | 12.137 | .000 | |
| CLASS ENROLLMENT | -.933 | .177 | -.523 | -5.277 | .000 | |
| 2 | (constant) | 35.910 | 10.150 | 3.538 | .001 | |
| CLASS ENROLLMENT | -.890 | .178 | -.498 | -5.011 | .000 | |
| SEQ1: Linear | 2.244 | .975 | 2.204 | 2.302 | .024 | |
| SEQ2: Quadratic | -6.996E-02 | .029 | -5.521 | -2.436 | .017 | |
| SEQ3: Cubic | 6.133E-04 | .000 | 3.579 | 2.550 | .013 | |