Predictors of Attrition Intentions among Tutors in Colleges of Education in Ghana ()
1. Introduction
In the Education Sector, tutors are seen as the pivot in achieving educational goals as they are mandated to train teachers who will implement the curriculum at the pre-tertiary level. The success of a country’s education policy, therefore, depends on its ability to recruit and retain its tutors. However, there is a growing concern that Colleges of Education are facing challenges of tutor attrition which thwarts the efforts of training teachers. The Colleges of Education Weekly Journal (2021), in its special report on the rate of attrition in the Colleges of Education, indicated that as of May 2021, over 140 tutors across the 46 public Colleges of Education have left the Colleges after these institutions were upgraded to tertiary status to offer a four-year Bachelor of Education (B.Ed.) degree programme in 2018. These tutors left to seek greener pastures at the technical and traditional universities in the country, mainly after further studies in Doctor of Philosophy (Ph.D.) or Master of Philosophy (M.Phil.) programmes. To such people, it appears mobility is influenced by poor remuneration and conditions of service that do not match the workload. Even though both assistant lecturers and tutors hold similar qualifications (M.Phil., which is the minimum requirement needed to teach in a tertiary institution in Ghana), and they both teach undergraduate students, conditions of service skewed largely in favour of those teaching in the universities. Teachers who experience burnout show high levels of depersonalization from the job, low morale, and emotional exhaustion. Once the negative feelings become too overwhelming, it is hard for teachers to feel their value in the profession, and so they leave (Vesely, Saklofske, & Leschied, 2013).
The implications of teacher attrition intention are profound. If left unaddressed, high levels of attrition intention can lead to actual turnover, which disrupts the continuity of education, affects student achievement, and imposes significant financial burdens on educational systems due to the costs associated with recruiting and training new teachers (Ronfeldt, Loeb, & Wyckoff, 2013). For instance, tutor attrition rates in Colleges of Education in Ghana have reached an alarming proportion (COE Weekly Journal, 2021). According to the News Desk of Colleges of Education Weekly Journal (2021), information received from most Principals of Colleges of Education shows that between September 2020 and June 2021 there have been, on average, three (3) tutors from each College who have left the Colleges of Education to take up appointments elsewhere Evidence on teacher attrition rate within a six-academic year period from all public Colleges of Education in Ghana is shown in Table 1.
Table 1. Attrition rate of tutors in Colleges of Education in Ghana.
ACADEMIC YEAR |
ATTRITION RATE |
2016/2017 |
34 |
2017/2018 |
26 |
2018/2019 |
43 |
2019/2020 |
29 |
2020/2021 |
76 |
2021/2022 |
65 |
TOTAL |
273 |
Source: Field survey (2026).
Source: Field survey (2026).
Figure 1. A graph of tutor attrition rate in Colleges of Education.
The trend of attrition of tutors in the Colleges of Education, as shown in Table 1 and Figure 1, revealed a generally increasing rate of attrition within the six academic-year period (2016/2017 - 2021/2022) in all Colleges of Education in Ghana. Compared with the period from 2016 to 2018, the rate of attrition from 2020 to 2022 was much lower in the Colleges of Education. The total turnout rates for the entire period under consideration revealed a growing trend of tutor attrition in the Colleges of Education in Ghana.
With respect to the rising rate of tutor attrition in Colleges of Education, the question one may attempt to ask is, what are the causes of tutor attrition intentions in the public Colleges of Education in Ghana? Could it be that some of the predictors, like remuneration, progression, physical conditions, motivation, workload, and many others, are working against tutor retention in Colleges of Education? Could it also be that the geographical locations of these Colleges of Education are contributory factors in tutor attrition? Answers to these posers are of great interest to this study. Hence, the study on predictors of attrition intention among tutors in Colleges of Education in Ghana.
Objectives of the Study
Specifically, the study sought to:
1) Find out the factors predicting attrition intentions of tutors in Colleges of Education in Ghana.
2) Assess the level of experience of tutors and how that influences tutor attrition intention in Colleges of Education.
3) Examine the gender difference in tutor attrition intention in Colleges of Education.
Hypotheses
The following hypotheses guided the study:
H0: There is no statistically significant influence of predictors of tutor attrition intention on tutor attrition in Colleges of Education.
H1: There is a statistically significant influence of predictors of tutor attrition on tutor attrition intention in Colleges of Education.
H0: There is no statistically significant difference between gender and attrition intention among tutors in Colleges of Education.
H1: There is a statistically significant difference between gender and attrition intention among tutors in Colleges of Education.
H0: There is no statistically significant influence of tutor experience on the rate of tutor attrition intention.
H1: There is a statistically significant influence of tutor experience on the rate of tutor attrition intention.
2. Methodology
A quantitative approach, together with a descriptive cross-sectional survey design, was used in this study. The researcher measured the variable of interest as they had occurred and there was no conscious attempt to manipulate any of the variables that were measured in the study. A descriptive cross-sectional survey design was considered in this study because it provided the researcher with information about the prevalence of tutor attrition intentions in Colleges of Education in Ghana. The target population for the study was all tutors in the 46 public Colleges of Education in Ghana. However, an inclusion criterion was established to exclude public Colleges of Education that were less than ten years in existence in the study because it appears tutor enrolment in such Colleges were less, with few tutors who were qualified to teach in tertiary institutions. Therefore, the prevalence of tutor attrition intention might not be much felt in such Colleges. The Colleges that were more than ten years in existence and were included in the study were 37 Public Colleges of Education in Ghana with about 1424 full-time teaching staff. In generalising findings from this study therefore, the inclusion and exclusion criteria should be taken in consideration in order to avoid bias.
A stratified sampling technique was used to group the 37 public Colleges of Education into four zones; thus, Northern zone, Brong-Ahafo-Ashanti zone, Accra-Volta-Eastern zone and Western-Central zone. These groupings were done taking into consideration the concentration of public Colleges of Education in the country with regard to their locations in the various regions of Ghana.
For proportional representation of Colleges of Education in the study, table of random numbers generator was then used to select 4 Colleges from each zone. Therefore, a total number of 16 Colleges of Education were considered in the study with approximately 706 tutors. The researcher proportionately selected tutors from each stratum to form a sample size of 400 tutors. However, at the end of data collection, the researcher realized a retention rate of 361 tutors. Therefore, the data retention rate of 361 was used for analyses in the study. A structured questionnaire was adapted and used for data collection on the predictors of attrition intention. The items were adapted from Nyamekye (2018), which were used for data collection in a study on the relationship between human resource management practices and intention to quit among administrative staff of the University of Cape Coast with a Cronbach’s Alpha of 0.823. It consists of 61 items under eight subscales: Remuneration, Social/Employee relations, Workload, Physical environment and Safe Working Environment, Motivation, Progression/Development, Social Identity, and Prestige. The items were measured using a scale of 1 to 5 where 1 = Strongly Disagree, 2 = Disagree, 3 = Undecided, 4 = Agree and 5 = Strongly Agree. Multiple regression, Independent Sample T-Test, and One-Way ANOVA were used to test hypotheses 1, 2, and 3, respectively.
3. Results
Hypothesis One
H0: There is no statistically significant influence of predictors of tutor attrition intentions on tutor attrition in Colleges of Education.
H1: There is a statistically significant influence of predictors of tutor attrition intentions on tutor attrition in Colleges of Education.
Hypothesis one sought to find out the predictive strength and direction of the predictors of attrition intention. These predictors included low remuneration, poor employee relations, excessive workload, poor physical environment and safe working conditions, low motivation, unclear progression, gender, teaching experience, level of qualification, social identity, and prestige. A multiple linear regression was run to find out the influence of these variables on tutor attrition intention.
Table 2. Multiple Regression of the influence of predictors of attrition on tutor attrition intention.
Variables |
B |
SE B |
Β |
t |
p |
Constant |
61.586 |
8.670 |
|
7.103 |
0.000 |
Remuneration |
−0.345 |
0.098 |
−0.177 |
−3.533 |
0.000 |
Employee Relation |
0.130 |
0.101 |
0.065 |
1.278 |
0.202 |
Workload |
0.312 |
0.169 |
0.093 |
1.846 |
0.066 |
Physical environment and safe working conditions |
−0.664 |
0.114 |
0.297 |
−5.847 |
0.000 |
Motivation |
−0.079 |
0.204 |
−0.020 |
−0.386 |
0.700 |
Progression |
−0.194 |
0.139 |
−0.071 |
−1.392 |
0.165 |
Social identity |
0.161 |
0.258 |
0.031 |
0.623 |
0.533 |
Prestige |
0.047 |
0.129 |
0.018 |
0.362 |
0.718 |
Source: Field Survey (2026). p = 0.000, r = 0.381, F = 7.473, DF = (8, 352). The coefficient of determination, denoted as R-squared (R2) = 0.145.
Table 2 shows that a significant regression equation was found F (8, 352) = 7.473, p < 0.01. The coefficient of determination, denoted as R-squared (R2), r2 was 0.145. This indicates that, overall, the predictors of attrition intention had a significant influence on tutor attrition with the exception of factors such as employee relations, workload, motivation, progression, social identity, and prestige which did not significantly predict tutor attrition intention. The model accounts for 14.5% of the variation in tutor attrition intention with an effect size of 0.17 (Cohen’s F), indicating a small effect size. The effect size of 0.17, measured using Cohen’s F, suggests a small effect. Therefore, the null hypothesis “There is no statistically significant influence of predictors of tutor attrition on tutor attrition intention in Colleges of Education” was failed to be rejected.
Hypothesis two
H0: There is no statistically significant difference between gender and attrition intention among tutors in Colleges of Education.
H1: There is a statistically significant difference between gender and attrition intention among tutors in Colleges of Education.
Research hypothesis two sought to find out gender differences in tutor attrition intention in Colleges of Education. An independent samples t-test was conducted to find out if there was a statistically significant gender difference in tutor attrition intention in Colleges of Education. The independent variable was gender with two categories: males and females, while the dependent variable was tutor attrition intention among tutors in Colleges of Education. Table 3 presents the results of the independent samples t-test.
Table 3. Gender differences in tutor attrition intention in Colleges of Education.
Gender |
N |
M |
SD |
T |
Df |
p |
Effect Size |
Male |
265 |
37.6189 |
10.31888 |
−0.078 |
145.660 |
0.938 |
0.000 |
Female |
96 |
37.7292 |
12.36334 |
|
|
|
|
Source: Field survey (2026). N = 361, p = 0.938.
Table 3 presents the results of an independent t-test examining gender differences in tutor attrition intention in Colleges of Education. The t-test compares the means and standard deviations of attrition intention scores between male and female tutors. Levene’s test of homogeneity of variance revealed that the variances for the two groups (Males and Females) were not equal (p = 0.000). Results of the t-test revealed that there was no statistically significant difference in tutor attrition intention scores for Males (M = 37.6189, SD = 10.31888) and Females (M = 37.7292, SD = 12.36334), t (145.660) = −0.078, p = 0.938 (two tailed). Therefore, based on the results of the t-test, there is no evidence to suggest a significant gender difference in tutor attrition intention in Colleges of Education. Both male (M = 37.6189, SD = 10.31888) and female (M = 37.7292, SD = 12.36334) tutors appear to have similar levels of attrition intention. Therefore, the null hypothesis “There is no statistically significant gender difference in tutor attrition intention in Colleges of Education” was failed to be rejected.
Hypothesis three
H0: There is no statistically significant influence of tutor experience on the rate of tutor attrition intention.
H1: There is a statistically significant influence of tutor experience on the rate of tutor attrition intention.
Hypothesis three sought to find out the influence of tutor experience on tutor attrition intention. The independent variable, tutor experience was measured on five levels: 0 - 5 years, 5 - 10 years, 10 - 15 years, 15 - 20 years, 20 - 25 years and 25 - 30 years. The dependent variable was tutor attrition intention. A one-way analysis of variance was conducted to find out if tutors attrition intention is influenced by their teaching experience.
Table 4. Descriptive statistics for experience and attrition intention.
Years |
N |
M |
SD |
Std. Error |
95% Confidence
Interval for Mean |
Minimum |
Maximum |
Lower Bound |
Upper Bound |
0 - 5 |
83 |
34.5060 |
11.71550 |
1.28594 |
31.9479 |
37.0642 |
18.00 |
54.00 |
5 - 10 |
32 |
43.8125 |
11.34343 |
2.00525 |
39.7228 |
47.9022 |
12.00 |
57.00 |
10 - 15 |
55 |
34.3818 |
6.55882 |
0.88439 |
32.6087 |
36.1549 |
19.00 |
43.00 |
15 - 20 |
87 |
38.9310 |
12.28138 |
1.31670 |
36.3135 |
41.5486 |
12.00 |
55.00 |
20 - 25 |
73 |
41.3699 |
8.16583 |
0.95574 |
39.4646 |
43.2751 |
16.00 |
55.00 |
25 - 30 |
31 |
33.1290 |
9.80388 |
1.76083 |
29.5329 |
36.7251 |
17.00 |
53.00 |
Total |
361 |
37.6482 |
10.88224 |
0.57275 |
36.5218 |
38.7746 |
12.00 |
57.00 |
Source: Field survey (2026).
Table 4 presents the means and standard deviations of the various levels of tutor experience in relation to tutor attrition intention. An observation of the means revealed that tutors with 5 - 10 years’ (M = 43.8125, SD = 11.34343) experience had a higher attrition intention than all other years of experience. However, an observation of the means alone is not enough to justify differences in observed mean differences; hence, the ANOVA analysis was needed to make such conclusions. However, it is important to check the Levene’s test for further decisions on whether to read the ANOVA table or the Welch table (Table 5).
Table 5. Levene’s test of homogeneity.
|
Levene Statistic |
df1 |
df2 |
Sig. |
Based on Mean |
9.147 |
5 |
355 |
0.000 |
Source: Field survey (2026).
The reported significance level (Sig.) is 0.000, indicating that the observed differences in variances among the groups are statistically significant. This suggests that the assumption of equal variances was violated; hence, a robust test of equality of means was conducted. The Welch’s test is used when the assumption of equal variances across groups is violated, and it provides a more reliable comparison of means in such cases. In such cases, the ANOVA table is not read, but Welch is read. However, the ANOVA table is produced here to enable readers to appreciate the effect size (Table 6).
Table 6. ANOVA table.
|
Sum of Squares |
Df |
Mean Square |
F |
Sig. |
Effect Size |
Between Groups |
4409.634 |
5 |
881.927 |
8.191 |
0.000 |
0.1036 |
Within Groups |
38222.688 |
355 |
107.670 |
|
|
|
Total |
42632.321 |
360 |
|
|
|
|
Source: Field survey (2026).
Table 7. Robust tests of equality of means.
|
Statistica |
df1 |
df2 |
Sig. |
Welch |
10.075 |
5 |
128.749 |
0.000 |
Source: Field survey (2026). a. Asymptotically F distributed.
The Welch table shows the results of robust tests of equality of means using the Welch’s test. The test statistic value is 10.075, and the degrees of freedom for the numerator and denominator are 5 and 128.749, respectively. The significance level (Sig.) is reported as 0.000, indicating that the observed differences in means are statistically significant. Thus, tutor attrition intention differs for tutors based on their teaching experience. To identify the mean differences among the groups, a Games-Howell post hoc test was run (Table 7).
Table 8. Games Howell post hoc test.
(I) Years of teaching |
(J) Years of teaching |
Mean Difference (I-J) |
Std. Error |
Sig. |
95% Confidence Interval |
Lower Bound |
Upper Bound |
0 - 5 |
5 - 10 years |
−9.30648* |
2.38216 |
0.003 |
−16.3268 |
−2.2861 |
10 - 15 years |
0.12421 |
1.56070 |
1.000 |
−4.3890 |
4.6374 |
15 - 20 years |
−4.42501 |
1.84048 |
0.161 |
−9.7309 |
0.8809 |
20 - 25 years |
−6.86384* |
1.60221 |
0.000 |
−11.4907 |
−2.2370 |
25 - 30 years |
1.37699 |
2.18041 |
0.988 |
−5.0290 |
7.7830 |
5 - 10 |
0 - 5 years |
9.30648* |
2.38216 |
0.003 |
2.2861 |
16.3268 |
10 - 15 years |
9.43068* |
2.19162 |
0.001 |
2.8973 |
15.9641 |
15 - 20 years |
4.88147 |
2.39891 |
0.336 |
−2.1823 |
11.9453 |
20 - 25 years |
2.44264 |
2.22137 |
0.879 |
−4.1639 |
9.0492 |
25 - 30 years |
10.68347* |
2.66863 |
0.002 |
2.8286 |
18.5384 |
10 - 15 |
0 - 5 years |
−0.12421 |
1.56070 |
1.000 |
−4.6374 |
4.3890 |
5 - 10 years |
−9.43068* |
2.19162 |
0.001 |
−15.9641 |
−2.8973 |
15 - 20 years |
−4.54922 |
1.58614 |
0.053 |
−9.1340 |
0.0356 |
20 - 25 years |
−6.98804* |
1.30215 |
0.000 |
−10.7568 |
−3.2193 |
25 - 30 years |
1.25279 |
1.97045 |
0.988 |
−4.6088 |
7.1143 |
15 - 20 |
0 - 5 years |
4.42501 |
1.84048 |
0.161 |
−0.8809 |
9.7309 |
5 - 10 years |
−4.88147 |
2.39891 |
0.336 |
−11.9453 |
2.1823 |
10 - 15 years |
4.54922 |
1.58614 |
0.053 |
−0.0356 |
9.1340 |
20 - 25 years |
−2.43883 |
1.62700 |
0.665 |
−7.1356 |
2.2580 |
25 - 30 years |
5.80200 |
2.19869 |
0.102 |
−0.6521 |
12.2561 |
20 - 25 |
0 - 5 years |
6.86384* |
1.60221 |
0.000 |
2.2370 |
11.4907 |
5 - 10 years |
−2.44264 |
2.22137 |
0.879 |
−9.0492 |
4.1639 |
10 - 15 years |
6.98804* |
1.30215 |
0.000 |
3.2193 |
10.7568 |
15 - 20 years |
2.43883 |
1.62700 |
0.665 |
−2.2580 |
7.1356 |
25 - 30 years |
8.24083* |
2.00349 |
0.002 |
2.2973 |
14.1843 |
25 - 30 |
0 - 5 years |
−1.37699 |
2.18041 |
0.988 |
−7.7830 |
5.0290 |
5 - 10 years |
−10.68347* |
2.66863 |
0.002 |
−18.5384 |
−2.8286 |
10 - 15 years |
−1.25279 |
1.97045 |
0.988 |
−7.1143 |
4.6088 |
15 - 20 years |
−5.80200 |
2.19869 |
0.102 |
−12.2561 |
0.6521 |
20 - 25 years |
−8.24083* |
2.00349 |
0.002 |
−14.1843 |
−2.2973 |
Source: Field survey (2026). *The mean difference is significant at the 0.05 level.
The post hoc test revealed that there was a significant mean difference between 0 - 5 years and 5 - 10 (p = 0.003). An observation of the means revealed that individuals with 5 - 10 years (M = 43.8125, SD = 11.34343) had a higher attrition intention than individuals with 0-5 years (M = 34.5060, SD = 11.71550). There was a significant mean difference between 0 - 5 years and 20 - 25 (p = 0.000). An observation of the means revealed that individuals with 20 - 25 years (M = 41.3699, SD = 8.16583) had a higher attrition intention than individuals with 0 - 5 years (M = 34.5060, SD = 11.71550). Furthermore, there was a significant mean difference between 5 - 10 years and 10 - 15 years (p = 0.001) and 25 - 30 years (p = 0.002). An observation of the means revealed that individuals with 5 - 10 years (M = 43.8125, SD = 11.34343) had a higher attrition intention than individuals with 10 - 15 years (M = 34.3818, SD = 6.55882) and 25 - 30 years (M = 33.1290, SD = 9.80388). There was a significant difference between 10 - 15 years and 20 - 25 years (p = 0.000). An observation of the means revealed that individuals with 20 - 25 years (M = 41.3699, SD = 8.16583) had a higher attrition intention than individuals with 10 - 15 years (M = 34.3818, SD = 6.55882). The findings suggest that tutor attrition intention varies based on their years of teaching experience in Colleges of Education (Table 8).
4. Discussion
Hypothesis one revealed that there is no statistically significant influence of predictors of tutor attrition on tutor attrition intentions in Colleges of Education. There are some differences between the findings of the current study and the reviewed studies by Chang et al. (2019), who conducted a systematic review of 29 empirical studies on predictors of tutor attrition in online learning environments. They found that job satisfaction, workload, communication, support, and technology-related factors were commonly studied predictors of attrition intention. In relation to this study, the finding that factors such as remuneration, physical environment, and safe working conditions significantly impact tutor attrition intention aligns with the emphasis on job satisfaction, workload, and support found in Chang et al.’s review.
Again, Liao, Armstrong, and Rounds (2008) identified three significant predictors of tutor attrition: dissatisfaction with training, workload, and pay. Dissatisfaction with training had the strongest negative effect. Although the current study did not specifically mention dissatisfaction with training, the emphasis on remuneration (pay) as an important factor aligns with Liao et al.’s findings. Further, Wang, Luo, and Zhang (2020) found that job satisfaction, work overload, and perceived organizational support were significant predictors. The current study’s focus on remuneration, physical environment, and safe working conditions as significant predictors aligns with Wang et al.’s finding that job satisfaction and work overload significantly affect attrition. The emphasis on support in Wang et al.’s study corresponds to the broader concept of support found in the current study.
Hypothesis two revealed that there is no statistically significant difference between gender and attrition intentions among tutors in Colleges of Education. The study finding contradicts studies conducted by Cha and Cohen-Vogel (2011), Addi-Raccah (2005), Deutsch and Yao (2014), and Eddy, Brownell, and Wenderoth (2014), who found that there are gender disparities in tutor attrition rates. These studies consistently showed that female tutors had higher attrition rates compared to male tutors. The reasons for attrition were also found to differ between male and female tutors, with female tutors often citing personal reasons, work-family conflict, or dissatisfaction with their tutoring experience.
The findings from hypothesis three suggest that tutor attrition intention varies based on their years of teaching experience in Colleges of Education. Comparing these findings with the previous studies, Chambers Mack et al. (2019) also found that teaching experience played a significant role in tutor attrition. They reported that teachers with fewer years of experience were more likely to leave the field. These findings align with the current study, indicating that attrition rates may be higher among tutors with less experience.
On the other hand, finding from the current study, which found higher attrition intention among tutors with 5 - 10 years and 20 - 25 years of experience, contradicts the study of Maré and Mutezo (2021), who found that tutor experience influenced tutor attrition rates in online learning environments. They reported that tutors with less experience and less support were more likely to leave their positions.
5. Conclusion and Recommendations
The combined predictors of attrition intentions, including low remuneration, poor employee relations, excessive workload, poor physical environment and safe working conditions, poor motivation, unclear progression, gender, teaching experience, level of qualification, social identity, and prestige, collectively contribute to tutor attrition intention. While low remuneration and perceiving the poor physical environment as safe were significant negative predictors, other factors did not significantly predict attrition intention. The researcher recommends that the Ministry of Education (MOE), in collaboration with the Ghana Tertiary Education Commission (GTEC), should conduct a comprehensive review of the salary structure in Colleges of Education to address remuneration concerns and consider salary increases or additional incentives to improve tutor satisfaction.
There was no significant difference between male and female tutors in terms of attrition intention, indicating that gender does not play a role in tutor attrition intention in Colleges of Education. It is therefore recommended that Principals and administrators of Colleges of Education should ensure equal treatment, opportunities, and career advancement prospects for both male and female tutors. They should implement and monitor policies that promote gender equity and prevent any potential gender biases or inequalities.
Furthermore, tutor attrition intention varied based on years of teaching experience and level of qualification. Therefore, Colleges of Education and donor agencies like Transforming Teacher Education and Learning (T-TEL) should establish mentorship programmes and career advancement opportunities specifically designed for tutors with 5 - 10 years and 20 - 25 years of experience. They should provide guidance, support, and professional growth opportunities to retain experienced tutors.