S. McGrath
 Department of Education

 J.W. Bulcock
 Faculty of Education

 A. Ponder
 Lakehead University
 Winter 1991

 They said it couldn't be done.  So I smiled and I went right to it.  I tried that thing that couldn't be done and I couldn't do it.


 A recent study of teacher leave patterns by McGrath (1990) and the Institute for Educational Research and Development in Newfoundland and Labrador provided descriptive data on leave use by teachers, mainly sick leave, by board, by geographic region and for the Province, as a whole (see Ponder and Bulcock, 1990).  The second, and perhaps more interesting part of the study, attempted to examine the relationship between certain personal and situational variables and teachers' use of sick leave. That is, to what extent does knowledge of the variables allow us to predict teacher absenteeism?

 The model used for this study was a modification of the one proposed by Steers and Rhodes (1987) which hypothesizes that work attendance is largely a function of two sets of variables, an employee's motivation to attend and a employee's ability to tend.  The revised model is presented in Figure 1. It incorporates many of the variables that are generally accepted as affecting employee leave use and is the only model found that includes the variable, ability to attend.

 This latter is considered to have direct application to the study because it recognizes that no matter how motivated an employee may be towork, attendance may be contingent on the ability of the employee to do so.  Illness, disability or other situational constraints beyond the control of the teacher sometimes result in short term absence from work.

 There were two primary sources of data for this study.  One consisted of information available from various files of the Department of Education; the other consisted of information gleaned through a questionnaire administered to a sample of elementary teachers on the Avalon Peninsula (n = 756).  The sixteen independent variables were subdivided intopersonal traits and situational variables as follows:


Age (E)
Sex (E)
Marital Status (E)
Education (E)
Experience (total) (Q)
Health (Q)
Accumulated unused sick leave (E)
Sick leave as an entitlement (Q)
Experience in present school (Q)


Size of school (E)
Place of residence (Q)
Distance to work (Q)
Coverage by substitutes (E)
internal coverage by staff (Q)
Urban/Rural community (E) 
School Board (E)1

 A full discussion of the related research pertaining to each of the variables and a rationale for its inclusion in the study is contained in McGrath (1990).

 Space does not permit the inclusion of the total questionnaire.  However, in order to give the flavour of the instrument, two samples have been included.  For example, teacher perceptions of sick leave asan entitlement was measured partly by the item:

 Like any other entitlement or employee benefit sick leave should be used.  Respondents could indicate, on a four point scale (strongly agree, mostly agree, mostly disagree, strongly disagree), this agreement or disagreement with the statement. 

 Similarly, perceptions of health were covered by the item: 

 Would you say your health is:

 (a) excellent
 (b)  good
 (c) fair
 (d) poor

 The data obtained from Department of Education files was essentially factual information.  The information from the two sources were then merged into a single file for analysis.  That is, the extent to which the independent variables listed above were associated with teacher sickleave use, was determined.  Table 1 and Figure 2 show the results of the analysis. 

 Figure 1.  Major influences on Teacher on Leave Usage

The independent variables found to be associated with sick leave usage included gender, marital status, unused accumulated sick leave, perceptions of health and sick leave as an entitlement.  However, collectively they explained only ten percent of the variance.

Discussion and Conclusions

 When a total of sixteen separate independent variables explain only ten percent of the variance (ninety percent remains unexplained) some serious questions need to be asked.  First, were the appropriate variables included in the model?  That is, have some important ones somehow been omitted.  The answer may have to be "hopefully not".  This study included more variables than had been contained in any study heretofore.  It investigated many of the same ones utilized in prior studies plus others which the researchers felt could be significant, e.g., sick leave as an entitlement.  Thus it is possible that some other critical variables have been omitted but the researchers are at a loss to say just what they might be.  Secondly, important variables may have been improperly measured.  But once again the researchers adhered to accepted measurement practices, though the problem of attempting to balance the numbers of items on the questionnaire with the limitations imposed by the overall length of the instrument was a difficult one.  Finally, the reasons teachers take leave may be idiosyncratic; that is, there are numerous reasons why teachers are absent (aside from being genuinely sick) but they are randomly distributed across the population.

 No clear patterns emerge.  Although the study breaks newground with respect to teacher leave use in the Province, it adds little to what has already been found by the somewhat limited research in the field in general.  Perhaps longitudinal studies or raising the level of aggregation to, for example, the school level might yield more definitive results.  At the same time, this may be, at best, optimistic speculation. Perhaps another approach to the problem might be more appropriate. That is, possibly teachers with high sick leave usage and those with low usage could be identified and small samples could be interviewed in-depth to attempt to determine factors which influence high or low rates of absenteeism.  Traditionally studies of leave patterns have been conducted through the use of survey research and data banks.  Perhaps the nuances of absentee patterns are too subtle to be revealed by the broad focus inherent in such research practices.

 Finally, in any analytical study, the distinction between significance and importance needs to be made.  In this particular study, the greatest amount of variance accounted for by any single significant variable was teachers perception of their own health, yet the total variance accounted for was only ten percent.  The authors leave the determination ofthe importance of their findings to the reader.


 Multiple Regression Parameter Estimates,
 Integrated Model



Age = Teacher's/respondents age, sex = gender, Mar = Marital status,EXP = years of teaching experience, YRSSCH = Teaching experience in presentschool, SLD = Accumulated unused sick leave, Health = Teachers' perceived health status, ENT 1 & 2 = Sick leave as an entitlement, FTTCHRS =School size, U/R = Urban or rural community of school, COV 1 = Absencecovered by substitute teacher, COV 2 - Internal coverage or filling inby other teachers or staff during an absence, RES = place of residenceof teacher, DIST = Travel distance to school.

Personal Traits

Situational Factors

Figure 2.  Parameter estimates for the Integrated Model*, (N =756)

*Standardized partial beta coefficients above the paths, t-values in parentheses below the paths; all non-significant paths were omitted; t-values greater than or equal. to 2.0 are significant at the <.05. 


 McGrath, Samuel (1990).  Relationships between selected variablesand teacher initiated leave during the 1987-88 school year in Newfoundland. Unpublished master's thesis, Memorial University, St. John's.

 Steers, Richard M. and Susan R. Rhodes (1987).  Major influenceson employee attendance:  a process model.  Journal of appliedpsychology, 63 (August), 391-407.


1. The letter (E) indicates that the data came from Department of Educationfiles, the letter (Q) identifies data derived from the questionnaire.