Corps de l’article

A wrongful dismissal claim is a common-law recourse for non-union employees in Canada to seek compensation in the event of employer-initiated terminations other than for just cause. Wrongful dismissals cover a broad range of situations, from terminations due to employee performance to terminations arising purely out of economic necessity. Such a dismissal is described as “wrongful” because the employer is said to have committed a breach of an implied employment contract term, namely, the indefinite term of hiring. In such events, the employer is required to pay severance compensation, the major component of which is the reasonable notice period or wages in lieu of notice. Severance compensation has been a topic of growing importance as many organizations strive to gain competitiveness by downsizing and restructuring. For obvious reasons, the determination of reasonable notice periods is important to employers, legal professionals, human resource (HR) practitioners and employees. However, systematic research on it has been meagre and what is “reasonable” remains questionable. This study will examine the determinants of reasonable notice periods for two groups of decision-makers: legal and HR professionals. Although most severance compensation decisions do not end up in litigation, legal decisions are important as they influence HR decisions. As such, the analysis of them forms the first part of the study. The second part examines the HR decisions from a multi-theoretical approach as such decisions are likely more complex due to various other forces at work, including economic and social ones. Finally, a comparison is made between the legal and HR perspectives to see how they are similar to or different from each other.

Literature review and prior research

Legal literature has recognized a number of factors used by judges to determine notice periods. According to Levitt (1992), the list consists of over 100 factors. On the other hand, HR manuals or consultant reports often suggest some simple rule-of-thumb formulae such as one month per year of service for management employees. The two extremes have caused much confusion for decision-makers. Some authors in the legal literature have helped to qualitatively identify the more important factors. The most commonly adopted factors came from the landmark case, Bardal v. Globe and Mail Ltd. ([1960] Ontario Weekly Notes 253 at 255), which states:

There could be no catalogue laid down as to what was reasonable notice in particular classes of cases. The reasonableness of the notice must be decided with reference to each particular case, having regard to the character of the employment, the length of service of the servant, the age of the servant and the availability of similar employment, having regard to the experience, training and qualifications of the servant.

Other factors that have been cited as potentially important include the economic climate, the industry norms, the hiring circumstances, employee performance and mitigation factors (Harris 1980; Mole 1990; Levitt 1992; Sproat 1995). Such literature, however, helps little in quantifying decisions.

On the quantitative side, there have been only a handful of studies done. McShane (1983), McShane and McPhillips (1987), Wagar and Jourdain (1992) all found length of service, employment status in terms of either job status or salary and year of the decision positively related to the notice period. Some of these studies also indicate age, special hiring circumstance and a poor labour market were associated with a longer notice. In a relatively small sample study in Ontario, Liznick (1987) only found length of service, job status and salary to be significant predictors. Harris (1989: 4–43–4–49) reported a study by Fisher and Goodfield on their computer database categorizing employees by occupation, age and length of service. It confirmed a higher award for more recent cases and rebutted the common belief of the “rule-of-thumb” formula, although there appeared some evidence the 1-month-per-year rule applied more to cases between 8 to 13 years of service. Rights Associates’ 1992 survey on 402 HR practitioners across Canada confirmed that 98% of the respondents used years of service in the calculations of severance followed by 74% using position, 63% using salary, 60% using age, and 48% using the reason for termination (Raices 1992). Four years later, Rights Associates (1996) confirmed that these factors continued to serve as the bases for severance compensation decisions in organizations. Apparently some general formulae were adopted by most of the 378 organizations surveyed with the 1-month-per-year of service formula most commonly applied to department head levels and above.

While the Rights Associates’ studies have provided some useful information to HR practitioners, the confirmation that a factor was used in a decision did not indicate the weight attached to it. On the positive side, the studies have shed light on the possibility that the factors considered by HR practitioners may not be the same as those of the legal professionals. For example, labour market conditions, generally said to be an important factor from the legal perspective, were not considered by the majority of HR respondents as a significant factor, while reason for dismissal, not commonly viewed as a prominent factor in legal notice period decisions, was taken into account by almost half of the participants in the first study and slightly less in the second.

The scanty systematic empirical research, limited number of factors used in each study, inconsistencies in some of the factors’ significance, apparent shifting trends identified in these studies, possible differences between the legal and HR perspectives and the lack of a theoretical framework for understanding the decisions imply that a more up-to-date study on both the legal and HR perspective is needed. It is to this end that the current study is directed.

The legal perspective

An analysis of the legal decisions for severance compensation is important to provide a reference point for legal professionals as well as a basis for the comparison with the HR decisions that will be captured in the second part of the study.

Court decisions on wrongful dismissal are based on common law, in which the abidance by legal precedents and legal principles are paramount. Therefore, it is not surprising that there is a general lack of a theoretical framework for explaining such decisions. Nonetheless, prior research and relevant legal literature provide an understanding of the possible decision criteria that inform our hypotheses. For example, the occupational level, salary, and length of service of an employee are believed to relate to his or her specialization and employability. The implicit assumption is there are fewer positions available at the more senior and specialized levels, making it more difficult to find comparable jobs at those levels. To a certain extent, industry differences may also be related to the nature of the employment and employability. Older employees are believed to encounter more difficulties in finding alternative employment. It is also more difficult to find employment in a tight labour market situation. The more difficult it is for the employee to find a comparable job, the longer the notice period should be. While hiring circumstances may not be directly related to employability, it is assumed that employees lured into new employment or relocating to take up employment would have incurred additional costs that need to be compensated in the event of termination. As an employee is expected to perform satisfactorily in his or her job, some judges have considered an employee’s job performance to be a factor in the compensation. Mitigation is another factor the judges would likely consider as the need for the claimant to mitigate the damage seems to be a generally adopted legal principle.

It seems that courts may consider more than just what is fair to the employee in granting the compensation as they have to balance between the rights and the efficiency paradigms. The former puts a premium on the employee’s dignity and autonomy while the latter on the employer’s freedom to pursue profitability (England 1995). For example, under the efficiency paradigm, a lower award would be favoured in times of organizational economic difficulty.

Hence, from the legal literature and prior research, we can establish the following hypotheses.


All else equal, the notice period awarded by courts is related to the following factors in the direction shown in brackets:


employee’s length of service  (+)


employee’s age  (+)


employee’s occupational level  (+)


employee’s salary  (+)


special hiring circumstances, i.e., relocation or enticed away from secure employment  (+)


good employee performance  (+)


bad employee performance  (–)


diligent mitigation efforts by employee  (+)


lack of mitigation efforts by employee  (–)


unemployment rate  (+)


unfavorable labour market condition (this factor is case-specific, unlike the unemployment rate factor)  (+)


poor company financial performance  (–)


industry differences:

Construction  (–)

Government/quasi-government  (+)

Control factors that are included are employee gender and year of the decision. Although gender has not been proven statistically significant in prior studies, other studies on legal issues involving arbitration decisions (e.g., Bemmels 1988a, 1988b, 1991) have found that the gender of the employee could be an important factor. As for the year of decision, most prior empirical studies indicated a positive relation with the notice period, but this significant finding has received little explanation and might well be due to other factors such as the unemployment situation. Further, it is understandable that (a) the notice period cannot forever continue to rise without a good reason; (b) courts tend to place an upper limit on the notices; and (c) the interest of the employer might be receiving increasing attention. The year of the decision was therefore included in the analysis to determine if the previous findings of significance are still valid.

Data and Methodology

A total of 132 cases on wrongful dismissal in Alberta decided between 1970 and 1996 are included in the current analysis. This time period was chosen because of the lack of Alberta published reports before that time. These cases are reported in various publications including the Alberta Report, Alberta Law Report, Canadian Cases on Employment Law, Western Weekly Report, as well as the Unreported Alberta Decisions held in the University of Alberta Law Library. In locating these cases, the Canadian Abridgement (2nd ed. Vol. R14A September 1996 reissue and R14A Supplement by Carswell Thomson Professional Publishing), an index with case summaries and reference to the published reports, was used as the main source of reference. As a supplement, the Alberta Decision, which gives summaries of selected published and unpublished cases, was also used. The cases were read and coded by two independent researchers to enhance reliability. The descriptions of the explanatory variables are given in Table 1. Linear regression analyses were performed, with the notice period (NOTICE) in number of months as the dependent variable.

Table 1

Explanatory Variables for Regression Analyses: Legal Part

Explanatory Variables for Regression Analyses: Legal Part

Table 1 (suite)

Explanatory Variables for Regression Analyses: Legal Part

While UNEMPLOYMENT refers to the general unemployment situation, LABOUR MARKET looks at the case-specific employment difficulties, recognizing the possibility that the general unemployment conditions may not always reflect the situations for certain locations or occupations. The correlation between the two in this study is non-significant (at 0.115).

-> Voir la liste des tableaux

Findings and Discussions

The results of the regression analyses, including a full model and two reduced models, are shown in Table 2.

Consistent with prior studies, age, length of service, and salary level were found to be significant factors. The findings confirm that the longer the length of service, the greater the age or the higher the salary, the longer the notice period awarded. The employee’s occupational level was also found to be significant in the partial models, indicating that a higher occupational level is associated with a longer notice period. In the full model, the coefficient is still in the direction predicted. It is noted that the inclusion of the dummy variable for the construction industry in particular reduced the significance of occupational level. Since construction had a relatively larger proportion of employees in the lower occupational groupings, when a relatively short notice period was given, it could be due to the employee being in the construction industry or in the lower occupation levels. As such, the inclusion of the industry variable affected the occupational level variable. In view of the small number of cases for each industry, past research findings that support the occupational level variable, and the fact that the re-running of the regressions without the salary variable (SALARY), another indicator for the employment status’ always gives rise to very significant coefficients for the occupational level, it would not be advisable to exclude the occupational level factor in favour of the industry variables. Despite the correlation between occupational level and salary, both variables are significant in the final reduced model (column c). The results, therefore, support H-A1 to H-A4, which propose a positive relation between the notice period, and length of service, age, occupational level, and salary respectively. As for H-A5 on the presence of enticement into employment or relocation on hiring to positively relate to the notice period, the coefficients are not even in the direction predicted. This is probably due to the small number of cases involved in such hiring circumstances, chance occurrence, and the likelihood that such hiring factors are important in cases only where a termination occurs within a very short time frame of hiring.

Table 2

Regression Results: Legal Part (Dependent Variable: NOTICE)

Regression Results: Legal Part (Dependent Variable: NOTICE)

Full model with the industry variables.


Partial model with all variables except the industry variables. (The relatively small sample size has made it necessary to reduce the number of variables used, and since the number of cases per industry was small and industry variables had not been proven significant in previous studies, they were taken out to run this partial model.)


Reduced model with only the most significant variables: AGE, LENGTH OF SERVICE, OCCUPATIONAL LEVEL, SALARY, and LABOUR MARKET.


p < 0.001 (1-tailed)


p < 0.05 (1-tailed)


p < 0.001 (2-tailed)


1. The total number of cases is 132. Listwise deletion reduced N to 75 (excluding one outlier) due to missing values mainly for age or salary. Further analyses were done using pairwise deletion and estimation method (with age and salary estimated from the other variables) to increase the size of N and for cross-validation. The results of these two supplementary methods are comparable to the listwise deletion approach with the exceptions of LABOUR MARKET, IND_CONSTRUCTION and IND_GOVT which were not found to be significant in the supplementary methods (but the signs were consistent with the listwise deletion method).

2. Although some employee characteristics variables, e.g., LENGTH OF SERVICE, OCCUPATIONAL LEVEL, SALARY, and AGE are expected to be correlated, with the highest correlation among them at 0.58 (between OCCUPATIONAL LEVEL AND SALARY), collinearity diagnostics in the regression did not indicate any major multicollinearity problems (condition numbers involved are less than the benchmark value given in Jobson 1991: 283).

-> Voir la liste des tableaux

H-A6(a) and (b) predict that the notice award will be related to employee performance. There is no statistically significant evidence at the 5% conventional level to show the relation although the GOOD PERFORMANCE coefficients had a positive sign. As there is a moderate positive correlation between this variable (p-value = 0.079, two-tailed) and the notice period, and the regression results (column b) give p-values of less than 0.10 (one-tailed), future studies may wish to focus on this variable. With an apparent shift in organizations towards efficiency, if the courts share this organizational view, this factor may gain significance over time. As for bad performance, there is no indication that this will lead to the lowering of the notice period. In fact, the direction of the effect is opposite of the effect predicted. This may reflect the courts’ general attitude against taking the middle-of-the-road approach. That is, employee performance should only affect the decision of whether there is just cause or not, and once just cause has not been proven, the notice decision should be independent of employee performance. This is in line with the Supreme Court decision (Dowling v. Halifax (City), 33 Canadian Cases on Employment Law (2d) 239) that ruled out the role of “near cause” in notice period decisions. It should also be noted that there were only a few cases of poor performance involved in these decisions as cases of poor performance that fell within the scope of just cause had to be excluded. Hence, further corroboration with a larger sample of poor performance cases may be needed.

H-A7(a) and (b) relate to the effects of mitigation efforts on the notice period. Neither good mitigation efforts nor a lack of mitigation efforts was found to be significantly related to the notice period. As the coefficients of good mitigation efforts actually have a sign contrary to prediction, it is likely that good mitigation efforts do not lead to any increase in notice. The small number of cases involving special mitigation efforts, however, makes any findings tentative.

H-A8(a) specifically looks at the relation of the unemployment rate on the notice period. Despite the common belief that the notice period should be related to the difficulty of finding alternative employment and that unemployment rate is a good indicator of such a difficulty, unemployment rate was not found to be a significant factor. Its direction was as that predicted. It shows that judges probably did not take note of the unemployment rate to any great extent. As the overall unemployment rate for the province might be regarded as too general, judges might tend to rely on case-specific situations as presented by the parties, such as how many comparable jobs had been advertised and how many jobs the employee had unsuccessfully applied for. In this regard, it is not surprising to find support for H-A8(b). As predicted by this hypothesis, the variable LABOUR MARKET, which captures the case-specific labour market situation as revealed by the judges’ comments and other evidence presented is significant. Yet, as compared with the prior research work, which suggests that this factor was gaining significance, the findings here seem to suggest otherwise. LABOUR MARKET was only found to be marginally significant. As the labour market condition is often associated with the economy, it is likely that a poor labour market exists alongside a business downturn. In the future, if a shift from the rights paradigm (which emphasizes labour rights and autonomy) to the efficiency paradigm (which emphasizes organizational efficiencies and profits) occurs, LABOUR MARKET is likely to become a less significant factor.

Related to the efficiency discussion is H-A9, which involves the bad organizational performance factor. If an organization is in financial trouble, it is expected that there might be some relief in its severance obligation, as some court decisions were known to have taken such a position. Although this factor does not reach statistical significance, its coefficients are in the direction predicted. No previous research has included this factor before. It would be interesting to see if there is any developing trend in this area.

H-A10 hypothesizes that there are industry differences for the notice decision. More specifically, for the construction industry where there are seasonal fluctuations and employment is usually viewed to be of a less permanent nature, the results suggest the notice period could be lower. Conversely, a higher notice award was associated with government and quasi-government organizations that are usually fairly large, and unlike private organizations, are less likely to have ability-to-pay problems. While the findings highlight some possible industry differences, in view of the limited number of cases for each industry, the results would require further corroboration.

The year of the decision factor is not significant when the unemployment rate is included. However, when the unemployment rate is excluded, the coefficient for the year of the decision decreased substantially, though not to the extent of being statistically significant. This raises doubt on whether the significance of the year of the decision variable found in prior studies might have been due to the omission of the unemployment rate factor. Probably, it is the macro-economic unemployment situation that matters rather than the year of the decision per se.

Similar to prior research, gender of the employee was found to be non-significant. In terms of possible discrimination against females in the court decisions, the positive sign of the coefficients (which means the notice period is more in favour of the female gender) does not confirm the existence of sex discrimination.

Overall, the models were able to explain the NOTICE variance very well, with an adjusted R2 ranging from 0.75 to 0.79. Analyses with interaction variables did not reveal significant interaction effects.

In sum, similar to prior research, length of service, age, salary, and occupational level were identified as significant factors. The labour market condition is marginally significant whereas the unemployment rate is not. Contrary to previous studies, the year of the decision is not a critical factor. There also seem to be industry differences especially for the construction industry, and government and quasi-government organizations. The consistent directions of the factors relating to good employee performance and organizational performance suggest that these factors should be further examined as there is the potential that they may gain significance with a possible shift towards efficiency concerns. Overall, there are a number of interesting variables being studied here but due to the limited number of cases, future research along these lines is necessary. For the present study, the 5 factors in the reduced model are the most significant factors determining the notice period decisions in Alberta for the period under study. These factors are therefore incorporated into the construction of the survey for the second part of this study.

The human resource perspective


If the determination of notice period is a complex decision from the legal perspective, it is bound to be more so in an organizational setting from the HR perspective. Mitroff (1983: xii) considers that real life problems in organizations have many dimensions, many forces are at work and many different values in conflict. It is unlikely that one single theory is sufficient to explain the phenomenon. Therefore, four approaches are used to look at the notice period determinants: legal, economic or financial, social, and individual decision-makers’ characteristics. It should be noted that theories under each approach may not be mutually exclusive. Since this research is exploratory in nature, it is important that the hypotheses are set not only with theoretical support, but also based on qualitative information supplied by HR practitioners in real organizational settings. The latter information was obtained by semi-structured interviews with 13 HR practitioners in various industries, including manufacturing, construction, oil and gas, consulting, mining, financial services, retail, public utility, biochemical, and forestry. They had HR experience ranging from 3 to over 40 years, with the majority at 10 or more years and holding senior positions.

Under the legal approach, it is expected that HR decisions will be influenced by legal decisions as practitioners tend to avoid costly and time-consuming litigation. All practitioners interviewed agreed that court decisions were important considerations. Thus, factors found to be significant predictors in the court decisions are hypothesized to be significant determinants too for HR decisions.

H-B1: All else being equal, (a) length of service, (b) age, (c) occupational level, (d) salary, and (e) labour market condition are critical factors under the HR perspective.

Other than considering the legal implications, HR practitioners are likely to have economic and financial concerns. Economically rational decision-makers would view severance compensation as an economic exchange and pursue cost minimization in the transaction. Swift (1983) suggests that the severance compensation offered by organizations may be the expected value of the court settlement less the litigation costs to the employee. Hence,

H-B2: All else being equal, the notice period awarded by the HR practitioners will be shorter than that awarded by the courts in a given circumstance.

Risk and expected values are integral elements in economic decisions. It can be expected that if the perceived risk of litigation is high and practitioners can have a choice of giving a higher or lower award, they would opt for the higher award to reduce the risk of potential litigation. Thus,

H-B3: All else being equal, the notice period awarded by HR practitioners will be longer when the perceived risk of litigation by the employee is high.

Numerous studies have found that organizational decisions, in particular decisions relating to the compensation policy, are determined in part by the organization’s ability to pay (e.g., Lentz 1998; Levine 1993; Nay 1991; Werner and Gemeinhardt 1995; Young and Kaufman 1997). Most HR interviewees agreed that this financial factor would become more important when many layoffs were involved. Therefore,

H-B4: All else being equal, the notice period awarded by HR practitioners will be shorter when the organization’s financial situation is tight.

According to the human capital theory, length of service, salary and occupational level are good indicators of the firm-specific investment made by the employee in anticipation of a higher income in the future employment relationship. Early termination of the relationship results in a loss of the expected yield and, hence, a need for compensation. As such, this theory can lead to the same hypotheses as H-B1(a), (c) and (d). Moreover, since women tend to spend less time in the labour force due to their child-care and housework responsibilities, they may have less incentive to invest in work skills (Becker 1975). This theory assumes the human capital accumulation for women to be generally less than that of men, and the level of compensation for their termination should be accordingly lower. Therefore, an additional hypothesis generated from this theory is:

H-B5: All else being equal, notice periods awarded by the HR practitioners will be longer if the employee’s gender is male.

This hypothesis can also be generated from the perspective of pure gender bias against women in general. In the HR interviews, all the interviewees said that they, themselves, would not use gender as a determining criterion. Yet, some believed that gender might affect some other practitioners’ decisions. One interviewee said he believed that there was systematic bias out there against woman.

Using a social approach, the equity principle or the need principle under the social justice theory may be at work in severance compensation decisions (Deutsch 1975). Generally, the equity principle looks at the contributions made. Thus, indicators of contributions such as length of service, occupational level, and salary (with the latter two believed to reflect the rewards associated with contributions), mentioned in the H-B1 can also be explained using this theory. Moreover, to the extent that equity theory places emphasis on the differentiation between actions that are within or without an individual’s control, it makes sense to provide a lower award to a person whose fault leads to his or her termination than to a person terminated for reasons beyond his or her control, such as organizational restructuring. Thus, reason for termination may be a factor influencing the severance notice decision.

Most HR practitioners interviewed were aware that there is always a range to the notice decision, be it in civil courts or in organizations. Within the bounds of consistency and legal obligations, many of them would try to give a more generous award to employees leaving for non-performance related reasons. “I’ll go the extra mile” or “I’ll bend over backwards for them [the employees]” were the words used by some practitioners referring to layoffs of people who “just happened to be in the wrong place at the wrong time.” Rights Associates have also shown in their studies that one of the factors affecting severance compensation is the reason for termination (Raices 1992; Rights Associates 1996). Hence,

H-B6: All else being equal, notice periods awarded by the HR practitioners will be shorter for employees terminated for performance than those terminated for non-performance related reasons.

Under the need principle of justice, rewards should be given according to what can meet the basic needs of the individual. From this perspective, the personal circumstances of the individual terminated from employment should be taken into consideration. In the HR interviews, some saw the organization as having a moral obligation to take care of their employees. As such, they would take employee hardship into consideration in the severance compensation. Therefore, it is expected that:

H-B7: All else being equal, the notice periods awarded by HR practitioners will be longer when the terminated employee is perceived to suffer a high degree of hardship resulting from the termination.

The need principle also predicts that poor labour market conditions and age should be positively related to the notice period as it would be more difficult for the employee to find alternative employment when the market is tough and when they are older. Thus, these echo H-B1(b) and (e).

As an organization is a social system, one can expect HR practitioners’ decision-making to be largely influenced by the organizational culture, such as the organization’s concern for its people (Jennings and Wattam 1998). It is proposed that where the organization’s concern for its members is high, HR decision-makers are more inclined to help employees even if it may mean a higher cost to the organization. In the interviews with HR practitioners, there was the consensus that the organization’s culture should and would determine the direction of severance compensation decisions. Thus,

H-B8: All else being equal, the notice periods awarded by HR practitioners will be longer if the organization has a high concern for its employees and staff relations.


A policy-capturing approach is used to find the decision criteria of HR practitioners when awarding notice periods (Slovic and Lichtenstein 1971). A survey was constructed with 30 hypothetical scenarios (arranged in different orders in four versions to avoid bias due to the ordering) and the respondents were asked to render a decision for each scenario. The use of common hypothetical scenarios allows for a consistent comparison among decision-makers and avoids the unnecessary concerns of providing actual sensitive and confidential employee information. Each scenario contained 11 variables (see the second to twelfth variables in Table 3) with values assigned to create different combinations. In making the decisions about notice periods, respondents were asked to assume that they were an HR consultant advising in a general non-union setting where there were no prior specific contractual or policy constraints.

For variables that were likely to be naturally correlated, namely occupational level, salary, age and length of service, the values of the variables were sampled from the Alberta wrongful dismissal court cases analyzed in the previous part. This was done by doing a cluster analysis on these variables and sampling from each cluster. The values for the other variables were randomly assigned and confirmed to be not significantly correlated with each other.

As equity is in the eyes of the beholder, different individuals may make different decisions based on their background, experience and characteristics. It is therefore important to control for the decision-maker characteristics. The survey therefore captures such information, including the respondent’s gender, age, HR experience, position level, level of involvement in severance decisions, industry type, size of organization, HR professional designation, and main severance compensation objectives (to avoid litigation, to help employees, or to be fiscally accountable to the organization). A summary of the variables used in the survey, including both the scenario variables and respondent characteristics variables is given in Table 3.

Table 3

Variables Used: HR Part

Variables Used: HR Part

Table 3 (suite)

Variables Used: HR Part

-> Voir la liste des tableaux

The survey respondents were the members of two human resources professional associations in Alberta, the Human Resources Management Association of Edmonton (HRMAE) and the Human Resource Institute of Alberta (HRIA). A total of 1,021 surveys were sent, with instructions stating that only individuals with responsibilities for or who had informed knowledge on severance decisions should respond. One hundred and forty-nine completed surveys were returned representing a response rate of 15%. The low response was not surprising as a large number of the recipients were in HR specializations other than severance compensation and had little knowledge of severance decisions.

Findings and Discussions

After eliminating the outliers and the non-usable responses, 139 responses were used in the aggregate regression analysis. As the collinearity diagnostics for this set of analyses indicated a strong collinearity between occupational level and salary, inclusion of only one of such variables was deemed to be sufficient. Analyses were, therefore, done with the two variables included separately. The results were very similar whether occupational level or salary was included. Table 4 gives the regression results (when occupational level instead of salary was used) with only the scenario variables found significant in the predicted direction included.

The R2 for this regression is 0.55 meaning that the model was able to explain 55% of the variance. A model including the respondent characteristics (not shown here) gives an R2 of 0.567. This means that most of the variance was actually explainable by the scenario variables. Inclusion of the respondent variables did not change the effects of the scenario variables. AGE, LENGTH OF SERVICE, and OCCUPATIONAL LEVEL have positive coefficients, indicating that a higher value for these variables was associated with a longer notice period, that is, the older the age, the longer the service, or the more senior the occupational level, the longer the notice period. Thus, H-B1(a), (b), and (c) are supported. (When salary was included instead of occupational level, it also was a significant factor, providing support for H-B1(d) as well.) A poor company financial situation (COMPANY FINANCE dummy coded 1) was associated with a lower notice award, lending support for H-B4, whereas a poor labour market condition (LABOUR MARKET dummy coded 1) was associated with a higher award, supporting H-B1(e). The negative REASON coefficient means termination due to restructuring (REASON = 0) was associated with a longer notice as compared with termination due to performance-related reasons (REASON = 1), supporting H-B6. GENDER is also significant in this aggregate regression analysis, indicating that female terminated employees may be awarded longer notice periods as compared with male employees. With the limitation that t-statistics in a repeated measures design may be inflated, caution must be exercised in determining if GENDER, the t-statistics of which being the smallest among other significant variables, is really a significant variable. While further research on this variable will be needed, the evidence helps to reject H-B5 that proposes that the notice period would be in favour of male employees. As for H-B2 that predicts the HR awards to be lower than the court awards, a discussion will be provided in the later comparison section. Among this list of variables, LENGTH OF SERVICE contributes greatest to the variance. The explanatory power of this variable measured by the difference in R2 between this model and the model without this variable is 0.31 (31%). All other variables were able to explain only one or two percent of the notice variance.

Table 4

Regression Results with 7 Significant Scenario Variables Only: HR Part (Dependent Variable: NOTICE)

Regression Results with 7 Significant Scenario Variables Only: HR Part (Dependent Variable: NOTICE)

N = 4110 (after removing cases of missing values for NOTICE)


With one respondent answering a set of 30 scenarios, the observations are not totally independent and the standard errors of the Ordinary Least Squares (OLS) regression coefficients tend to be underestimated. However, the OLS regression estimates are still unbiased. Especially when the regression is used for prediction purposes, the OLS is still a good and valid method to use. One method suggested to handle this repeated measures concern is introducing dummy variables to block within-person variance while determining the remaining variance explained by the scenario variables (Tyler and Steensma 1995). This controls for each subject’s idiosyncratic contribution to the overall regression and thus should yield accurate standard error estimates (Judge and Bretz 1992). In the current analysis, the inclusion of dummy variables did not change the significance of any coefficients in the model, confirming that any correlation of error terms here did not pose a problem in the interpretation of the regression results.

-> Voir la liste des tableaux

There is no evidence to support H-B3 or H-B8 that the risk of litigation or a company culture in favour of caring for employees was related to a longer notice period. The data did not point in the hypothesized direction. As for H-B7 that predicts a relation between the notice period and personal hardship of the employee, the results were insignificant when occupation level was included but significant when salary was included. This inconclusive finding may be an area for future investigation.

Among the individual characteristics used as control variables, the respondents’ years in HR (RES_HRYEAR), size of their organization (RES_SIZE), whether their industry was manufacturing (RES_MANU), and whether the avoidance of litigation was their main severance compensation objective (D_LITIGATE), are significant at the p < 0.001 level. From these results, it can be concluded that notice period was positively related to the length of service in HR and the size of the organization. The manufacturing industry was associated with a shorter notice period. When avoidance of litigation is the primary severance compensation concern, the notices awarded were higher as compared with respondents whose main objective was to help the employee or to be fiscally responsible to the organization.

Overall, the important determinants of notice period decisions relating to the scenario variables are length of service, occupational level or salary, age, labour market conditions, the company’s financial situation, the reason for termination, and possibly, the gender of the employee. The most prominent determinant is the length of service. The results show that various theoretical perspectives, namely, the legal, economic/financial, and social, are operative, confirming that HR decisions are more complex than simply following the legal perspective. In the following section, a comparison will be made between the legal and HR perspectives.

Comparison of legal and HR perspectives

A comparison of the HR practitioners’ decision model with that of the legal perspective shows that some critical factors are quite consistent, with length of service no doubt a very important explanatory variable under both perspectives. Age and labour market conditions are found to be significant in both legal and HR models. Employment status, in terms of either salary or occupational level, is also a critical determinant under both perspectives.

One purpose of this research was to determine if the legal and HR perspectives deviate from each other. In this study, poor company performance was not found to be statistically significant under the court decisions although it was in the direction predicted. This factor probably only comes into the picture in a near-bankruptcy situation. In the HR decisions, poor company financial situation was a significant determinant that may reduce an average notice period decision by almost a month.

Poor employee performance was also found to be insignificant in court decisions. This empirical finding, together with the Supreme Court decision in 1998 (Dowling v. Halifax (City), 33 Canadian Cases on Employment Law (2d) 239), confirms that “near cause” plays no role in the legal notice period decisions. However, for the HR practitioners, not only was the performance-related termination situation found to lower their notice period decisions, as expressed by some interviewees, some actually felt that it should be the case. A quite common view from the HR interviewees was that it was fair and justifiable to help the employees terminated for non-performance-related reasons more than those terminated for some fault of the employee’s own, as long as it did not jeopardize the overall consistency of the application of their severance policies.

The other variable that was found to be significant under the HR decisions but not in the court decisions was the gender of the employee. However, as suggested in the HR analyses section, the significant finding needs to be corroborated. Nonetheless, it would be advisable for the HR practitioners to ignore the gender effect as decisions based simply on gender difference are discriminatory, which contravenes human rights legislation.

After comparing which variables are significant, the next question is whether there is a difference in the overall magnitude of the notice decisions. This will be dealt with using a general descriptive approach, followed by a statistical analysis.

Using the reduced model in column (c) in Table 2 as the legal model and the model in Table 4 as the HR model, the notice periods awarded under the two perspectives for different occupational groupings in an average situation are shown in Table 5. In this comparison, the individual characteristics variables for the HR perspective were not included because (a) they have very limited explanatory power, (b) they do not significantly affect the coefficients of the scenario variables, and (c) it is better to compare apples with apples, i.e., the individual decision-maker characteristics variables were not outlined in the court cases.

It can be seen that other than for the lowest occupational grouping, the notice awards given by the courts were generally higher than those given by the HR practitioners. For the middle and lower levels without special situations adversely affecting the HR decisions, the awards under the two perspectives seem more comparable. However, for the most senior occupational level, the deviation was as large as 4 to 6 months for an average case. There are some indications that the overall HR decisions for senior occupational level employees were less generous, but for the most junior occupational level, HR practitioners might be a little more lenient. This could be due to the recognition of the employees’ needs or that the practitioners just tended to avoid granting the upper and lower end extremes.

Table 5

Comparison of Legal and HR Awards Using Average Situations by Occupational Levels

Comparison of Legal and HR Awards Using Average Situations by Occupational Levels


1. The top part of the table provides the average age, salary and service descriptions of the real court cases by occupational level. Such information was used to calculate the awards under the two perspectives (except salary is not used for the HR model) as shown in the bottom part of the table.

2. As the gender of the employee was found to be significant in the HR model, the average of the notice for a male and for a female employee was used in calculating the HR awards for comparison purpose.

-> Voir la liste des tableaux

In the next segment, a more sophisticated statistical procedure will be used to compare the magnitudes of the awards between the two perspectives as well as the coefficients of the common significant variables.

Since there are more significant variables under the HR model than the legal model, these extra variables, namely, GENDER, REASON, and COMPANY FINANCE, were first controlled for. That is, a regression analysis was run with NOTICE as the dependent variable and these three variables as the independent variables. The residuals were saved in the database spreadsheet. The constant term of the regression model was then added back to the residuals to form the revised NOTICE, which then controlled for the effects of the three independent variables. Since the HR model uses only occupational level and not the salary variable (which is highly correlated with occupational level), it is probably justifiable for comparison purposes to exclude the salary variable from the legal model as well. The two databases (court decision database and HR survey database) were then combined, using common variable names for length of service, occupational level, labour market conditions and age. A dummy variable HR_DUMMY was introduced with a “1” coded for HR survey cases and “0” for court cases. The model gives an R2 of 0.495. The HR_DUMMY has an unstandardized coefficient of –4.323 (p < 0.001), indicating that for an average case, the HR notice decision was about 4.3 months lower than the court decisions. This gives support to H-B2 which expects the notice awarded by HR practitioners to be shorter. This deviation in the magnitude of the awards can be attributed to a few reasons. First, it is possible that not all HR practitioners are fully aware of the common law decision criteria. Second, there are always financial and budget constraints with which HR decision-makers have to be concerned. Also, it is common knowledge that not all employees paid lower than an average court award in similar situations will litigate. The employees concerned may not even know of the general court settlement amounts and if they do, there are still the questions of time and legal costs involved. So, there is the possibility that HR practitioners purposely give awards lower than what the court may give, but not to the extent that the disgruntled employees would seek litigation. This finding lends support to the cost-benefit optimization theory under the economic or financial approach.

A model with the slope shifters (Jobson 1992), which are interaction variables created using the HR_DUMMY and other variables, however, did not result in any significant slope shifting effects, meaning that there is insufficient evidence to conclude statistically significant differences in the weights attached to the criteria from the two perspectives.

The implications of the difference in the award magnitude for the two perspectives mean that employees are often the ones to lose. They are the party with limited bargaining power because they may still need a good reference from their past employer and may not be comfortable pursuing litigation. In terms of legal costs, as compared with an organization’s entire budget, such costs may be insignificant. However, for the terminated employees, the amount of legal costs will likely come out of their final severance pay. Even if they know they can get more in court, it may not be worthwhile to pursue such a course.

For the HR practitioners, the findings indicate that it may be necessary for them to review their decision criteria. As most of the practitioners want to make reasonable and consistent decisions, it is important that they justify the difference between their notice decisions and the legal decisions. If HR practitioners’ objectives in severance compensation are to be fair to the employees and not be involved in unnecessary litigation, perhaps their awards should be revised upwards.

On the other hand, the difference could also mean that HR practitioners might view the court decisions as overly generous. It is also possible that they would prefer the court also to take into consideration the organization’s financial situation and the reason for termination. Whether organizations will lobby for legal changes more in line with the efficiency paradigm and the “employment at will” concept, and how far their influence may go is not known. Future research, especially longitudinal studies, should determine if the extent of deviation between the two perspectives increases or decreases over time, and why.


In comparing the legal and HR perspectives, it is obvious that HR decisions involved not only legal considerations, but also economic or financial concerns as well as equity issues in the social and moral sense. That may explain why the notice periods tended to be shorter for organizations with a poor financial situation, and longer for employees terminated for restructuring. Practitioners probably believe they have a greater moral obligation to help employees with good performance, who are terminated for reasons beyond their control. Another important finding in this comparison is that, overall, HR awards were less than the legal award decisions. This lends support to the hypothesis that HR practitioners may try to minimize the transaction cost and may pay less to terminated employees than the court, knowing that it may not be worth the employee’s time and money to pursue litigation. However, the true reason for this deviation should be explored further in future research. If the difference is purely due to ignorance of the general common-law settlements, then this study should provide a good reference for practitioners. In any case, it is important that the decision-makers, both legal and HR professionals, be able to justify their decisions and that they be consistent in the application of their decision criteria. They should also look at each other’s decisions to determine if a difference should exist, and what the implications of any deviations are. This study is believed to be the first in this area comparing the two perspectives. More work is certainly required in the long run to understand the deviations and to see if the perspectives are converging or diverging over time.