Anthropogenic Determinants of Success in Agricultural Education: The Case of Jimma College of Agriculture, Ethiopia

Bedassa Tadesse^* and Kidist G/Sellassie^**

Abstract: An analysis of the determinants of students' success in middle-level² agricultural training was carried out using secondary data obtained from the Office of the Registrar, Jimma College of Agriculture. The study used multivariate statistical tools involving Fisher's linear discriminant function and LOGIT qualitative response model. The results of the study indicate that sex, preference for agriculture, region, students' parental occupation and streams attended at the high school discriminate between successful and unsuccessful students as a group. Being from a rural area, however, does not give a student more chance to succeed in middle-level agricultural education. Furthermore, the study revealed that the probability of succeeding in agricultural education could not be predetermined on the basis of results in the Ethiopian School Leaving Certificate Examination (ESLCE³).

ACKNOWLEDGEMENTS

The financial grant for undertaking this research was provided by Jimma College of Agriculture. We are, therefore, highly indebted to the Research and Publications Office of the College, the Registrar's Office, Ato Million Abebe and W/Assiya Simeneh for their assistance. The anonymous referees of this journal also deserve our thanks.

1. INTRODUCTION

The declining level of the educational achievements of students in developing countries has been a matter of increasing concern to their governments in general and policy makers in particular. Especially in a country like Ethiopia, with less educated manpower, very few educational institutions and limited numbers of graduates at the tertiary level (degree and diploma), the educational achievement of students admitted to higher education institutions is of great concern.

The new education and training policy of the country puts emphasis on functional, purposeful and relevant education and training, and that it should enable Ethiopian children and youth to acquire a body of knowledge, and to develop a system of skills and relevant attitudes that would help them solve day to day problems in order to live a productive life, and to play an active and meaningful role in the development of their country (Azeb Desta, 1995). This positively ambitious and highly challenging requirement envisages making education a means of solving Ethiopia's problems of poverty, disease, and backwardness.

Agricultural training is one of the areas where such a change is desired so as to bring about a significant improvement. Much is expected from agricultural colleges, their graduates, and research institutions. The judicious selection of students for training is, therefore, of paramount importance. However, this is neglected in many developing countries (Watts, 1979) because, in large countries such as Ethiopia, it is very difficult and expensive to make people travel long distances for interviews. As a result, the selection of students for training is frequently based solely on academic results. Furthermore, given the limited seating capacity in the higher learning institutions of the country, pre-entry experience and other socio-economic characteristics which could affect a student's performance may not be insisted upon as it is not possible to place all the academically qualified candidates in study programs of their preference. In this regard, Omolo (1982) reported that due to ever-increasing facilities in urban schools, extension agents (trainees) are increasingly being drawn from non-agricultural areas.

However, the key to development lies ultimately in people and their creative capacity, technical knowledge, and their ability to adapt to new conditions. These vary with their background, experience and other anthropogenic characteristics (Fauchon, 1975). Further, Watts (1979) indicated that practical experience is indispensable for any one taking up a career in agriculture. For instance, for the livestock sector, it is essential to select students with the right dispositions and to choose those who have had some exposure to the special problems of the sector as well as having a keen interest in working with livestock. However, given the existing pattern of student admission to higher education institutions in Ethiopia, certain areas of training such as agriculture, which is likely to assign the candidate to a remote rural village, will be chosen by either a very small number of candidates or those whose academic achievement is relatively poor. This is because diploma-holders from universities and colleges reportedly lack the `right attitude' and prefer `white collar' jobs (Ibid).

Therefore, using students' pre-entry experience and related socio-economic characteristics as well as their attitudes to working with farmers in remote rural villages as mandatory selection criteria could help to draw the right candidates for training and producing enthusiastic and qualified agricultural extension agents. Furthermore, such selection criteria will reduce the number of dropouts and the attrition rate in different colleges for which the budget has already been allocated from the meagre financial resources of the country. This, in turn, enhances proper utilisation of the earmarked budget, the limited seating available and the potential for better manpower development.

To this end, an in-depth analysis of some of the anthropogenic characteristics of students enrolled in various institutions and graduates must be undertaken, their relation with their academic achievements in the college and their performance in the field must be investigated. Such quantitative as well as qualitative investigations, which require rigorous statistical analyses, however, are scanty. This study is, thus, an attempt to make a contribution in this direction. The first phase of the study was carried out at the Jimma College of Agriculture using data obtained from the Office of the Registrar on students admitted to the College during the past six years (1989/90-1994/95).

2. OBJECTIVES

The general objective of the study is to assess the success rate and socio-economic characteristics of students enrolled in middle-level agricultural training. More specifically, the objectives of the present study are:

1) to identify anthropogenic characteristics that differentiate between successful and unsuccessful students;

2) to assess the link between anthropogenic characteristics and student achievement; and

3) to examine factors determining the probability of student success in middle- level agricultural training.

3. HYPOTHESIS

In setting the above objectives, the following working hypotheses were applie

i) The anthorpogenic characteristics of Ethiopian students enrolled in middle- level agricultural training do not differ by achievement.

ii) A student with a rural background has a better chance of succeeding in agricultural training than one without a rural background.

4. METHODOLOGY

4.1 The Sample and Data:

A time series data on a list of 1419 students enrolled in the College over the past six years (1989/90 - 1994/95) was collected from the Office of the Registrar, Jimma College of Agriculture. Of these students, 378 students who did not properly complete the application form for admission were omitted. Therefore, the data applies to a group of 1041 students for whom complete information was available. Descriptive characteristics of all the selected students were identified and then a working sample of 316 students (constituting 30 percent of the group) was randomly selected for a detailed analysis. The selection was carried out using the probit maximum likelihood sample selection model by taking into account some major reference characteristics such as region, residence, sex, success rate and parental background.

The data were then subjected to a rigorous statistical analysis to examine the determinants of student success and identify factors that discriminate between successful and unsuccessful students.

4.2 The Empirical Models

a) Discriminant Function Analysis

Fisher's two-group linear discriminant function was applied to identify those anthropogenic variables that pose maximum possible separation between successful and unsuccessful students as follows:

L = ₁ X₁ + ₂X₂ + ₃X_{3 n}X_n

= A'X ..1

Where:

A' = Vector of 1 x m discriminate coefficients (latent roots ()) to be estimated

X = Vector of n x 1 anthropogenic variables characterising the population

L = Total discriminant score.

Considering the two groups of population with N₁ and N₂ samples, the vectors of their expected values can be estimated as:

x₁' = [ x₁₁ x₁₂ x₁₃ ... x_1p] ....................................1.2

X₂' = [ x₂₁ x₂₂ x₂₃ ... x_2p] ....................................1.3

with the grand mean vector of the two samples given as:

x' = [ x₁ x₂ x₃ ... x_p], ....................................1.4

The mean discriminant score values for the two groups in terms of their estimates could be obtained as:

L₁ = A' x₁ and ....................................1.5

L₂ = A' x₂ .................................... 1.6

Therefore, with the grand mean (L) given as:

L = A x, ...............................1.7

The difference in the mean values of the two groups is obtained as:

L₁ - L₂ = A' x₁ - A' x₂

= A' (x₁ - x₂)

The variance of L₁ - L₂ is equal to:

Var(L₁ - L₂) = Var(A' x₁) + Var(A'x₂)

= A' (_₁/N₁)A + A'(_₂/N₂)A ....................1.8

Where, _₁ and _₂ denote the variance-covariance matrices of the two populations.

Now assuming:

_₁ = _₂ = __xx ,

Var(L₁-L₂) = A'__xx A (1/N₁ + 1/N₂) ....................1.9

Since x is in general unknown, it can be estimated from the data as __xx and thus the maximum separation between the two groups of sample population can be achieved by solving the equation:

A = _^-1 _xx(x₁ - x₂) ......................1.10

The solution to the equation (1.10), i.e., the vector of coefficients A was arrived at using the computer package SPSS. After deriving the vector of coefficients A, i.e., the latent roots (_i), the discriminating power of the function was evaluated through Roy's theta (_) criterion, and the Mahalanobis's D² as shown by Marascuilo and Levin (1983).

b) Logistic Regression

A multinomial logistic regression (LOGIT) model was used to estimate the probability of success for a student enrolled in middle-level agricultural training and identify factors that have a detrimental effect on such a probability. The coefficients were estimated through a maximum likelihood method by a computer software package, Limdep 6.0.

Following Hosmer and Lemeshew (1989), let Y_j =1 if a student succeeds (graduates) and Y_j = O, otherwise (i.e., if academically dismissed). Then,

Y_j = f(Z_j) ........................2.0

Where:

Z_j = __o + _₁X₁ + _₂X₂ +... + __nX_n ........................2.1

Z_j is an observable index determined by a linear combination of various social, economic, demographic and other anthropogenic variables (X_i) characterising a student. Therefore, the probability (P_j) of a student enrolled in the college to succeed in middle-level agricultural training can be calculated as:

P_j = exp (__iX_ij + V_j)/(1+ exp(__iX_ij +V_j))

= exp (Z_j)/(1+ exp (Z_j)) ........................2.2

Where:

i = 0, 1, 2 ... n variables

j = 1,2,3 ... m observations

__i = coefficients to be estimated

X_ij = social, economic, demographic and other anthropogenic variables

P_j = probability to succeed

V_j = random error term

c) Statistical Tests

All statistical tests were conducted using t, F, Z and X² distributions evaluated at

P< 0.001, 0.01, or 0.05 levels of significance.

5. RESULTS

5.1 Anthropogenic Differences by Success Group

Descriptive statistics and differences in the anthropogenic characteristics of the total sample by success group are provided in Table 1.

5.1.1 Demographic Characteristics

It is evident from the table that with a mean age of 19.72 years at admission for all students, differences do not exist between successful and unsuccessful students. On the other hand, while students enrolled in the college thus were dominantly (87.3 percent) male, the proportionality test for a binomial sample distribution (here sex) reveal that the proportion of successful male students is significantly larger. This would mean that female students are not only small in proportion but also remain largely unsuccessful. It could be seen from the table that female students account for only 7.2 percent of successful and almost 30 percent of the unsuccessful candidates.

5.1.2 Socio-economic Background of Parents

No specific study has been undertaken on the impact of parental background, i.e., the social and economic background of parents, on student achievement in Ethiopia. Yet it is believed that it has an impact. To this end, the parental background and residence of the sample students were investigated. About 76.8 percent of the sample students had unemployed mothers and 54.7 percent had fathers engaged in farming. The study indicated that a significantly larger proportion (P < 0.05) of successful candidates had fathers engaged in farming. This might be associated with the practical experience which is indispensable for any one taking up a career in agriculture (Watts, 1979).

5.1.3 High School Background

Among the important pre-college characteristics considered in the present study are: high school stream enrolled in, ESLCE results and status of student at admission (whether regular, private or quota student). Taking into account the five major study streams in the high school, 79.4 percent of the candidates admitted were from agriculture and science streams. On the whole, 84.8 percent of them were regular students with a mean grade point average of 2.90 in the ESLCE. The success rate, however, did not differ significantly on any one of these pre-college variables.

5.1.4 Preference for Agriculture

It is believed that placement of a candidate on the basis of his/her preference may contribute to a better performance since a student might be psychologically well prepared to face even harsh conditions in areas of his/her choice. In other words, students may give more preference to disciplines in which they think they would perform better. However, under a condition where the final job placement for all fields is not uniform, preferences are not likely to based on ability to perform better or psychological preparedness. Nevertheless, the existing practices oblige students to rank all the areas available for higher studies. It is, therefore, likely that the candidate may give the least preference rank to a programme which he/she is not interested in. Despite this, there were candidates who never indicated any preference to join agriculture. Accordingly, while 84.8 percent were those who preferred to study agriculture at various levels, about 15 percent of students admitted to the college were those who never showed any preference for agriculture. However, a significantly (P < 0.05) larger proportion (20.09 percent) of the unsuccessful students was from those who preferred to study agriculture.

5.1.5 Academic Performance at College

The mean achievement of the candidates in terms of their GPA during each semester and the overall CGPA is provided in the last portion of Table 1. It is evident from the table that the achievements of successful and unsuccessful students differ widely. Therefore, to formulate appropriate strategies for reducing the attrition rate, the anthropogenic characteristics were further examined using Fisher's linear multiple discriminant function. The coefficients of the variables with the multivariate tests are produced in Table 2. All the multivariate tests used show that the discriminant function estimated produces maximum possible distance between the centroids of the two groups. In this regard, Roy's _ criterion (which reflects the squared multiple correlation coefficient) indicates 83.7 total explained variance. Mahalanobis's D² also shows strong evidence for rejecting the null hypothesis that the centroids of the two groups are not different. This implies that the centroids are far apart and the function has the ability to discriminate between the groups.⁴

Hence, the coefficients of the variables could be used to identify anthropogenic characteristics that discriminate between the groups. Accordingly, sex, preference for agriculture, region and parental occupation, followed by streams attended at high schools discriminate between successful and unsuccessful students as a group. The ESLCE result and residence least discriminate between the groups. Among all the characteristics, achievement variables, particularly, the 1^st and 2^nd semester GPAs possessed the largest discriminate coefficients. This would mean that successful and unsuccessful students, as a group, differ widely in their 1^st and 2^nd semester performances. This suggests that achievement variables differentiate students enrolled in middle-level agricultural training more than socio-economic characteristics.

On the other hand, although residence least discriminates between the groups, achievement and other anthropogenic characteristics were further investigated based on residential background of the sample population (Table 3) so as to examine the notion that rural students perform better in agricultural education. On the basis of this, the test for differences in semester achievements shows no differences except the 4^th semester GPA, which was higher for students from rural areas. Among the anthropogenic characteristics, differences were observed in terms of sex composition, parental occupation and preference to study agriculture. Moreover, a relatively larger proportion (92.8 percent) of rural students was male and 87.4 percent of them preferred to study agriculture. Despite these, the success rate does not differ by residence. That is, though a larger proportion of successful students belong to rural areas, the proportion of successful rural students was not significantly different from successful urban students when evaluated in terms of their respective total population.

Therefore, it could be concluded that despite differences in sex, parental occupation, preferences and the 4^th semester achievement, rural students do not perform better than others in agricultural education under a similar environment.

5.2 Determinants of the Probability of Success in Middle-level Agricultural Education

The final status of a candidate enrolled in the training program is based upon his or her cumulative grade point average (CGPA) all through the four semesters. Investigating what determines this cumulative achievement and thereby the chance of a student to graduate is, therefore, of paramount importance in formulating appropriate policy strategies so as to boost the success rate.

The probability of a candidate's success depends on a number of pre-college factors and situations within the institutions, and these are believed to influence achievements (Stones and Morris, 1972). These factors include demographic variables such as age and sex, and individual psychometric factors such as maturity, preparedness, realising ability and analytical capacity. Furthermore, pre-college academic background and parental occupation are among the variables that could influence performance and final success.

Some of these variables have a direct influence while others affect only performance and indirectly the achievement and success rate. The achievement probability that a candidate admitted to the college, therefore, depends upon these variables. The degree to which these variables affect (positively or negatively) success is thus investigated using dummy coded dependent variable (i.e., status) coded as 1, if successful and 0, otherwise. This was estimated using a Qualitative Response Model (QRM) known as LOGIT regression, the maximum likelihood method. Description of the variables used in the model and expectation of their influence is provided below.

5.2.1 Demographic Variables

5.2.1.1 Age at Admission (years) (X₁):  

Age is an indicator of maturity and thus it is positively correlated with analytical and understanding ability (Smith, 1991). However, with increase in age the capacity to memorise declines. College studies require both analytical and memorising ability. However, taking into account common observation, it is a priori expected that age at admission negatively influences achievements⁵ and thus the probability of success.

5.2.1.2 Sex (X₂):  

There is no theory that specifies differences in achievement by sex. However, many feel that the tedious and continuous field work and laboratory practice in agricultural education require strong physical endurance. So females are likely to achieve lower grades. Nonetheless, taking into account the fact that most (60 %) of the agricultural work in Ethiopia is being carried out by females, it is not possible to formulate any a priori expectation.

5.2.2 Socio-economic Background of Parents

5.2.2.1 Mother's Employment (X₃) & Father's Occupation (X₄):  

Among the important parental backgrounds of students, employment status of mothers (X₃) and occupation of fathers (X₄) were considered. It is likely that parental background affects the amount of support a student may receive and consequently performance through its psychological implications (example, stress). On the other hand, owing to their exposure to farming, one could expect that students with fathers engaged in farming perform better than others in agricultural training.

5.2.2.2 Residence (X₅):  

Residence could be considered as a proxy for reflecting the extent of exposure to practical agricultural operations and thus practical knowledge. As experience plays a key role in success, it is, therefore, a priori expected that rural students will perform better in agricultural education than others, and thus have a greater chance to succeed.

5.2.3 Pre-College Characteristics

5.2.3.1 Preference (X₆):  

Preference is a reflection of interest, preparedness and psychological readiness to undertake an activity. On the other hand, it may be based on ability (capacity), as assessed by the individual himself, and the nature of occupation presumed to be undertaken upon completion of the training (Borg et al., 1970). On these bases, it is expected that the preference variable would have a positive coefficient.

5.2.3.2 High School Stream (X₇): 

Being in the science and agriculture streams during pre-college studies exposes students to most of the basic courses offered during the I^st semester of their admission and the principles and practices of some of the applied agricultural subjects offered during the later semesters. It is, therefore, hypothesised that students from science and agricultural streams in their pre-college studies will perform better than others and thus have a higher probability to succeed in agricultural education institutions. Thus, X₇ is a priori expected to have a positive coefficient.

5.2.3.3 ESLCE (X₈):

 The ESLCE is a standard national examination set to assess achievement during pre-college studies and thus serves as a basic criterion for admission to higher studies. However, depending upon the seats available in various higher education institutions of the country and other governmental policies (such as gender issues), the cut-off point varies from year to year. On the assumption that the ESLCE is a standard examination, it is hypothesised that the higher the ELSCE result the more the probability of a student's success in college studies.

5.2.3.4 High School Status (X₉): 

The admission status of a candidate varies on the basis of whether or not the student comes directly from school, i.e., whether he/she is a regular or private candidate or not. A regular student with ESLCE results just below the cut-off point but with strong evidence of better achievement in a rural high school may be admitted as a quota student. Admission also used to be granted as a merit scholarship to those who participated in the national military service during the Derg⁶ regime. The variable X₉ was coded 1, if regular, and 0, otherwise, to compare regular students' success rate with others (who have not come directly from schools). On the bases of the time gap between their admission and their first ESLCE, and their lower ESLCE, it is a priori hypothesised that results for quota and military service candidates have a lower chance of success and thus a positive coefficient for X_9.

5.2.4 College Performances (X₁₀ - X₁₃)

Once candidates are admitted to the college they will be exposed to a similar environment and receive uniform treatment. However, they will undergo different courses, varying from basic pre-college courses in the 1^st semester to applied and special agricultural courses specific to their field of specialisation, during the 2^nd, 3^rd and 4^th semesters. Thus, each semester's grade point average (GPA) was taken (X₁₀-X₁₃) to evaluate the extent to which it determines the relative probability of the student's success. As they are all achievement variables that determine the cumulative GPA based upon which the status of a student is determined, they were all expected to have a positive coefficient. The most important issue to be investigated, however, was the relative size of their coefficients in affecting the probability of success.

The maximum likelihood estimates of the coefficients of the various social, economic, demographic and achievement variables affecting the probability of student success in middle-level agricultural training is provided in Table 4. With a highly significant (X² = 326 df = 12) log likelihood ratio of -176.59, it is evident from the table that the data fits the model very well. Moreover, the classification table, which compares predictions with actual observations, indicates an overall 94.04 percent correct reclassification.

Among the anthropogenic variables used in the model, sex, mother's employment status, preference for agriculture, and stream attended in high school were found significant. However, the mother's employment status and the student's preference to study agriculture had unexpected signs. Contrary to the a priori expectation, students with unemployed mothers had a greater chance to succeed in middle-level agricultural training. This may be because such students work with a strong determination when admitted to higher studies owing to the responsibility they shoulder to support themselves and their parents.

It is surprising that preference as an indicator of interest, preparation and psychological readiness to undertake an activity has not possessed the expected positive sign. This shows that the placement of students contrary to their preference had no impact on their probability to succeed. This indicates that preference ranks given by the candidates are merely based on the perceived job opportunities after graduation rather than on what they perceive about their own capabilities and their pre-college experience.

Regarding gender, it is evident from the table that the probability of student success in middle-level agricultural training differs significantly by sex. That is, males had a greater chance to succeed. Therefore, to increase the proportion of educated Ethiopian women in agriculture, ways and means should be sought to improve the performance and thereby the success rate of females enrolled in higher education.

With reference to the streams attended at high school, it is found that students who enrolled in agriculture and science streams had more chance to succeed in middle-level agricultural training. This indicates the relevance of placing candidates on the basis of their pre-entry major streams in high schools.

The coefficient of residence, though in line with the a priori exception, was not significant. Therefore, the hypothesis that a student with a rural background has more chance to succeed in middle-level agricultural training could not be supported.

All the achievement variables had the expected signs and the 1^st, 2^nd, and 4^th semester GPAs were found to be significantly determining the success status of students in order of importance. Therefore, college authorities should give special attention to advisory, guidance and counselling services, particularly during the first two semesters, so as to reduce the prevailing high (26.4 percent) attrition rate.

The coefficient of the ELSCE results bears an unexpected sign. However, it was not significant. Therefore, the predictive validity of the ESLCE result is poor. That is, given the chance, students who scored less than the cut-off ESLCE points used for admission could even succeed. At the same time, being a regular student does not also give a candidate a greater chance of success.

6. CONCLUSION AND IMPLICATIONS

Assessment of the anthropogenic characteristics of Ethiopian students enrolled in middle-level agricultural training was conducted using data obtained from the Office of the Registrar, Jimma College of Agriculture.

The study indicated that sex, preference for agriculture, region and parental occupation, followed by streams attended at high schools, discriminate between successful and unsuccessful students as a group. ESLCE results and residence least discriminate between the groups. Among all the characteristics, achievement variables, particularly, the 1^st and 2^nd semesters' GPAs, possessed the largest discriminate coefficients. This means that successful and unsuccessful students as a group differ widely in their 1^st and 2^nd semester performances. This suggests that achievement variables differentiate students enrolled in a middle-level agricultural training more than socio-economic characteristics.

The test for differences, if any, between students from rural and urban areas in terms of their success rate showed that despite their differences in sex composition, parental occupation, preferences to join agricultural programs and the fourth semester achievement results, rural students do not perform better than urban students in agricultural education under a similar environment. This shows that the selection of trainees for agricultural education need not necessarily be from rural areas. Given the opportunity, students from urban areas, too, perform adequately in agricultural education.

Assessment of the anthropogenic determinants of the probability of a student to succeed (graduate) in middle-level agricultural training revealed that sex, mother's employment status, preference to join an agricultural program, streams attended in high schools have a detrimental effect. Of all the achievement variables, 2^nd, 1^st and 4^th semester GPAs significantly determine the probability of success in order of importance. The ESLCE result, however, could not be taken as a deciding factor for success.

The findings of the study imply that:

1. College authorities should give special attention to advisory, guidance and counselling services, particularly during the first two semesters so as to reduce the prevailing high (26.4 percent) attrition rate.

2. Given the chance, students who scored less than the cut-off ESLCE points used for admission could also succeed.

3. Being a regular student does not give a candidate a better chance to succeed.

4. To raise the participation of educated Ethiopian women in agriculture, ways and means should be sought to improve the performance and thereby the success rate of female students enrolled in a middle-level agricultural training. This could be made possible through the arrangement of special classes, tutorials and case-specific treatments that may vary from one training institution to another.

NOTES

The Jimma College of Agriculture is the pioneer institution where research and higher level agricultural studies began in Ethiopia. It is located some 330 kilometres southwest of Addis Ababa. With an intake capacity of 500 students, the College offers diploma programs in Plant Sciences, General Agriculture and Animal Husbandry.

² Though not a standard representation, this was taken to mean a two-year (four semester) diploma program in agricultural education.

³ The Ethiopian School Leaving Certificate Examination (ESLCE) is a national examination offered to Ethiopian nationals. Results in the ESLCE are used for the admission and placement of high school graduates in any one of the colleges and universities of the country.

⁴ It is also customary to use Hotelling's trace (T²) and Wilks's Lambda () in assessing the significance of a discriminant function. Given the Eigen value and Roy's theta () reported on Table 2, those interested can compute T² as (n-2)2.777 & the Wilks's Lambda () as 1- following Marascuilo and Levin (1983).

⁵ It is true that the negative influence of age at admission on achievement is inconclusive as it varies with the nature of the curriculum, field and level of study, and the specific country facts. It is, therefore, imperative for readers of this article to be cautious with regard to its implication for other cases.

⁶ A council of military dictators that ruled Ethiopia from 1974 to 1991.

REFERENCES

Azeb Desta. 1995. Development of Problem Solving Skill in Teaching and Learning: Strategies and Techniques. Educational Journal. 2 (2):81-109.

Borg, W. R., P. Kelly. M. Gali. 1970. The Microcourse: A Micro Teaching Approach to Teacher Education. London: Macmillan.

Fauchon, J. 1975. Integrated Rural Development and Planning for Rural Communities. Training for Agricultural and Rural Development. FAO. 15:79-85.

Hosmer, D. W. and S. Lemehsew. 1989. Applied Logistic Regression. New York: John Wiley.

Marascuilo, A. and J. R. Levin. 1983. Multivariate Statistics in the Social Sciences: A Researcher's Guide. California; Cole Publishing Company.

Omolo, A. O. 1982. Outreach Programmes for Training Front Line Extension Workers: A Case Study in Kenya. Training for Agricultural and Rural Development. FAO. 26: 33-37.

Smith, C. M. 1991. Reinventing Student Teaching. Journal of Teacher Education. 42(2): 104-118.

Stones, E. and S. Morris. 1972. Teaching Practice: Problems and Perspectives London: Methuen.

Watts, R. 1979). The Importance of Practical Training in the Livestock Sector. Training for Agricultural and Rural Development. FAO,

Table 1: Descriptive Statistics of the Anthropogenic Characteristics of Students by Status Group

Anthropogenic Characteristics	Total Sample (n=316)	Unsuccessful (n = 78)	Successful (n=238)	Z- Value
1) Demographic Characteristics
a) Age at admission	19.72(1.93)	20.62(2.23)	19.42(1.73)	0.438
b) Sex (percent male)	0.873	0.705	0.928	-4.108**
2) Parental Background
a) Mother's employment (percent unemployed)	0.768	0.743	0.77	-0.603
b) Father's occupation (percent in farming)	0.547	0.397	0.596	-3.115**
c) Residence (percent rural)	0.588	0.461	0.630	-2.619**
3) High School Characteristics
a) ESLCE	2.902(0.201)	2.933(0.229)	2.892(0.189)	1.074
b) Stream (percent Academic Science and Agriculture)	0.794	0.769	0.802	-0.608
c) Status (percent Regular)	0.848	0.897	0.831	1.567
4) Interest
a) Preference rank to study Agriculture(percent preferred anyway)	0.848	0.910	0.827	2.043**
5) Academic Performance
a) 1st semester GPA	2.210(0.864)	1.195*0.471)	2.544(0.685	-11.36**
b) 2nd semester GPA	2.149(1.108)	1.202(0.471)	2.651(0.662)	-9.770**
c) 3rd semester GPA	2.061(1.222)	1.200(0.611)	2.645(0.687)	-6.316**
d) 4th semester GPA	2.672(11.486)	1.638(0.485)	3.500(0.130)	-4.820**
e) CGPA	2.265(0.861)	1.152(0.382)	2.631(0.629)	-17.52**

(Figures in parentheses are standard errors)