El objetivo del trabajo ha sido la construcción de una serie de funciones de utilidad que permitan pronosticar el nivel de satisfacción general que alcanzarán los sujetos que acceden a los programas de Escuelas Taller y Casas de Oficios, de formación ocupacional, patrocinados por el Instituto Nacional de Empleo (INEM). Para ello, se ha utilizado una muestra de sujetos de la provincia de Cádiz, que habían accedido a este tipo de programas, y se les aplicó el cuestionario MPS de Fernández Seara (1987), en su versión original y en una versión abreviada, y se les pidió que emitieran un juicio, en una escala del 1-10, acerca del nivel de satisfacción (NSG) alcanzado. Con los datos obtenidos se construyeron las distintas funciones de utilidad.


The original Spanish version of this paper has been previously published in Psicología del Trabajo y de las Organizaciones, 1999, Vol. 15. No 1, 45-62
...........
Correspondence concerning this article should be addressed to María Isabel Barbero García. Departamento de Metodología de las Ciencias del Comportamiento, Facultad de Psicología, UNED, Ciudad Universitaria S/N, Madrid, Spain. E-mail: ibarbero@cu.uned.es

This work forms part of a line of research aimed at the study of job satisfaction and the design of instruments and indicators to measure it, the objective being to facilitate decision-making in the personnel selection process for all activities within the framework of the Occupational Training Programmes.

The work is based on the assumptions of Behavioural Decision Theory, and particularly its perspective on choice under risk and uncertainty, since subjects on this type of programme select a particular course or speciality only on the basis of its name, without more information. Also, staff recruitment processes usually become decisions made under risk conditions, given that there is no empirical evidence as to whether the candidates chosen are the most suitable. This is particularly true when using procedures such as the interview, in which subjective aspects have significant influence.

An initial review of the field of Behavioural Decision Theory ( Slovic, 1972; Slovic, Fischhoff and Lichtenstein, 1977, 1984; Tversky and Kahneman, 1980; Payne, 1982; Macía, Barbero, Pérez-Llantada and Vila, 1990; White, 1990, etc.) suggests a conception of decisions as the result of a scale of inferred preferences that lead, in an objective way, to a particular choice; it also suggests focusing the problem on the measurement of the preferences on which subjects base themselves in order to resolve the state of ambiguity confronting them in a given situation.

The procedure followed in order to achieve the proposed objective is based on Savage’s Principle of Insufficient Reason (1954), in which the possible results are evaluated through subjective utilities multiplied by the probabilities, also subjective, assigned by subjects to the different events that lead to a random choice, where the preference scale does not play a rational role in the decision.

In accordance with this model, the results obtained permit the measurement of the degree of job satisfaction involved in a particular job and the subject’s appreciation of the utility of the psychosocial motivations associated with the job, as well as the prediction of the level of job satisfaction a subject will experience within the specific field of work.

Philosophical rationalism established a bridge between the senses and reason in order to explain man’s knowledge. Kant (1781) maintained that the exterior world sends man a set of disordered, formless and chaotic stimuli which, on being perceived, are organised in categories, thus constituting phenomenological knowledge. He claimed, in sum, that knowledge consists of what is presented to the subject and what the subject does with it, which he called intuitions.

The inclusion of intuitions, as the basis of human knowledge, made possible the development, in the field of mathematics, of the theory of expected utility ( Neumann and Morgenstern, 1944; Savage, 1954; Bernouilli, 1954) for explaining human decisions. Classic probability theory assumed that decisions were governed by what is known as mathematical expectation, understood as the expected value obtained by multiplying the possible results of an event by its objective probability of occurrence. Bernouilli observed the lack of empirical confirmation of this perspective, and reformulated it, substituting the objective values scale by a subjective one which he called a "utility scale". Neumann and Morgenstern, combining the utility and probability theories, developed an entire series of axioms to explain rational behaviour through the construction, by the subject, of a utility function that would take account of his/her preferences in the case of different results. By substituting objective values with the corresponding subjective ones, the subject establishes a scale of preferences that allows him/her to create a utility function, defined by axioms on preference-indifference relationships, and whose ultimate purpose is to maximise expected utility on the expectations that this process generates in the subject.

Savage (1954) completes Neumann and Morgenstern’s model by substituting the objective probabilities with subjective ones fixed by subjects according to their own criteria, thus obtaining the expected subjective utility model.

All of these perspectives are consolidated within psychology as the so-called Behavioural Decision Theory (BDT) aimed at the study of how decisions are made and how they should be made ( Coombs, Dawes and Tversky, 1981).

Josef Kozielecki (1981), in his book Psychological decision theory, offers a perspective halfway between psycholinguistics and Piaget, and which bases the study of decisions on human operational systems. Kozielecki considers the decision-maker as an information-processing system which, rather than describing the task, makes an internal representation of it and the risks it implies. "Decisions would be influenced by subjects" personal clearacteristig, their personality and their cognitive system. Through the analysis, based on Bayes’ conditional probability theory, of certain exercises carried out in situations of uncertainty, they reach the conclusion that subjects draw up a series of internal schemas based on cause and effect relationships, which leads them to the decision after they have, subjectively, assigned causality to it.

Carroll and Payne (1976) carried out an interesting study in which they analysed decision-making as an activity that relates personal perception to personal attribution and information processing. Their conclusion is that decisions should be studied from the perspective of psychology and its theoretical principles. Gul (1985) demonstrates the relationship between cognitive styles and decision-making, considering the former as mental categories that operate at different mental levels.

Mitchell and Beach (1976) and Zedeck (1977) maintain that Behavioural Decision Theory can be used as a measure of social motivations; Harrel and Stahl (1981) reaffirm this on demonstrating it experimentally, through the use of the T.A.T. in a job-choice situation.

Atkinson and Feather (1966) analyse the search for employment and, in consequence, choice of occupation, through Expectation-Value Theory, which maintains that young people initiate jobseeking when they perceive the possibility of finding a job and work is valued positively.

Thus, if job satisfaction is conceived as the degree of fit subjects experience between their needs, their expectations and the benefits of the job ( Mumford, 1976; Locke. 1976; Holland, 1985; Michalos, 1986), we can relate work motivation to the satisfaction offered by a particular job. If, in addition, we consider work motivation as the utility subjects find or expect from working, we can, including the level of job satisfaction, define a subjective utility function that explains the reasons why subjects work.

In the case of two or more utility functions being defined for the same group in different employment situations, it would be possible to analyse their evolution over time, allowing the function to predict the workers’ performance. If the measures for obtaining the utility function and its evolution were taken for workers and unemployed people, the predictive function could also serve as an instrument for choice of profession.

The subjects in this study are participants in the Escuela Taller y Casa de Oficios (Workshop School and Vocational Centre) programme sponsored by the National Institute of Employment (INEM). The Escuela Taller y Casa de Oficios (E.T./C.O.) programme was set up in 1985 on an experimental basis, under the control and management of the National Institute of Employment. Its aim is to improve the employment possibilities of jobless people under 25 (until 1994 over 25s were also allowed to participate) through training and work experience courses in the areas of: restoration and promotion of the artistic, historical, cultural or natural heritage; rehabilitation of urban or natural environments; improvement of citizens’ living conditions; as well as in other activities related to the social and general welfare of the public. Through the programme young people would acquire the professional skills and experience necessary to find employment in these areas.

The E.T./C.O. programmes may be sponsored by civil service departments, local councils, regional governments or institutions, national associations, non-profit making organisations, etc. Both the Workshop Schools and the Vocational Centres offer a six-month initial training phase, and another phase (more than one extra phase in the case of Workshop Schools) of training combined with work experience. The aim of these second phases is to provide learning and the opportunity to acquire qualifications and professional experience. The duration of the entire programme is one year in the case of the Vocational Centres and between one and a maximum of two years in the Workshop Schools.

An important aspect of these programmes that should be borne in mind is that only those who have failed to achieved the objectives of secondary school education are admitted as participants. Programmes are organised with the specific aim of providing participants with basic and professional training that will allow them to work or continue their studies in areas regulated by the Ley Orgánica del Sistema Educativo (LOGSE), the Spanish law governing education.

During the first stage of training, E.T./C.O. participants are entitled to receive the grant provided for in R.D. (Royal Decree) 631/93, of around 600 pesetas (slightly less than 4 Euros) per day. In the second stage they are taken on under an Apprenticeship Contract as long as they fulfil the statutory requirements.

In 1993, a total of 37,120 students throughout Spain began the initial training stage of the E.T/C.O programme. 1,030 (2.8%) of these were from the province of Cádiz. The sample used in this study was made up of 500 subjects with no previous work experience that had been accepted as participants on one of the E.T./C.O. in the province of Cádiz. The characteristics of the sample can be seen in Table 1.

Officially, participants had no prior professional experience, though some had worked in domestic tasks or illegally in family businesses.

The general design used for this study falls within the framework of non-experimental longitudinal designs. The procedure followed consisted in the application of two questionnaires, one biographical and the other of psychosocial motivation (Motivación Psicosocial, MPS), in two different phases, to a sample of subjects that had been admitted as students to E.T./C.O. Projects in different parts of Cádiz province.

The application of the questionnaires (abridged MPS and biographical) in the first phase was carried out in the students’ first month as participants in the Project (student phase). So that students answered more honestly, the application was carried out anonymously, and a code (known only to the student and the researcher) used for purposes of subsequent matching. The only identification data students were obliged to provide were biographical details (age, sex, level of education and the course being studied).

In the second phase, with subjects now having spent a total of nine months on the Project, the biographical questionnaire (with an extra item on level of satisfaction produced by participation in the Project) and the psychosocial motivation questionnaire (complete MPS) were applied once more.

Fernández Seara’s Psychosocial Motivation Scale (MPS), published by TEA in 1987, is made up of 173 items. These are divided into three sections: the first section (Item 1 to Item 126) requires a T/F (true/false) response, depending on the extent to which the subject is in agreement with the statement; the second (Items 127 to 147) and third sections (Items 148 to 173) have a similar answer mechanism, whereby the subject has to select from a series of items only those which s/he finds attractive, indicating the degree of that attraction. The second part is evaluated on a scale with three response categories: average (2), a lot (3), and very much indeed (4), while the third uses a scale with four response levels: some (1), average (2), a lot (3) and very much indeed (4). Another difference between the second and third sections is that the items of the second section focus on work-related aspects and those of the third on matters related to subjects’ life in general.

Using the principal components analysis technique, six basic psychosocial motivations are defined: Social acceptance and integration (SA), Social recognition (SR), Self-esteem/Self-concept (SC), Self-development (SD), Power (PO) and Security (SE).

The theoretical model used in the construction of the questionnaire recognises that motivations are dynamic and are expressed in different behavioural forms, objective and subjective. Therefore, five components are defined in which the different motivations are to be measured. These components are conceptualised as levels, and cover aspects such as: the arousal action of the behaviour (Level of Arousal -ARO-), subject’s cognitive anticipation with respect to the probability of success or failure in a given task (Level of Expectations -EXP-), the degree of effort the individual makes in order to achieve goals (Level of Execution -EXE-), the importance the subject places on external reinforcers (Value of Incentives -INC-) and the degree to which external incentives or reinforcers exceed what the subject considers to be fair or insufficient (Level of Satisfaction -SAT).

The MPS scale is adapted to the world of work, and the content of the items is related to aspects specific to that context. Since the subjects in the first study phase were students with no professional experience, some of the scale’s content had to be omitted so that the questionnaire would be more appropriate to the particular situation. Items corresponding to the following components were eliminated: Value of Incentives and Level of Satisfaction. In addition, all items referring to particular work situations were removed from the Arousal, Expectations and Execution components. The final abridged version of the MPS was made up of 89 items covering the six factors (basic psychosocial motivations) measured only in the Arousal, Expectations and Execution components.

The material was organised so that in the first phase of the study only the abridged MPS questionnaire, adapted to the situation of the students, was to be applied, and in the second phase the complete MPS questionnaire in its original form would be applied. A measure of General Satisfaction Level (GSL) would also be taken in the second phase.

Variables used were:

GSL: level of life satisfaction generated by participation in the Project.

First of all, using the data obtained in the second phase (complete MPS), a discriminant analysis was carried out. The criterion used for the selection of variables was Mahalanobis’ maximisation of squared distances ( Bizquerra, 1989), with General Satisfaction Level (GSL) being used as the grouping variable. Since the subjects within the GSL variable were distributed along a preference scale of 1 to 10, an excessive number of groups was generated for the discriminant analysis to be carried out. Given the possibility that there might not be a clear difference between two adjacent variables, it was considered appropriate to divide the sample into three groups and choose the two extremes, leaving out of the initial analysis the group including subjects with intermediate values. The number of subjects selected for the analysis altogether represented 37.2% of the total. There were 97 subjects within the DISSATISFIED category (scores of less than 6) and 89 subjects that fell into the SATISFIED category (scores of equal to 10), so that the two groups were of comparable size.

The results of this discriminant analysis show that only 8 of the 30 variables used (six psychosocial motivations multiplied by the five components in which each had been measured) discriminate significantly among the subjects. As the Stepwise method excludes from the analysis values that do not provide significant information in the discriminant function, the eight values included reflect the motivations that actually establish differences between the categories of satisfied and dissatisfied. In order to understand the acronyms used, note that the first three letters refer to the measurement component and the last two to the motivation. Thus, for example, AROSE would indicate the Arousal (ARO) component in the Security (SE) motivation.

In this table we can observe the Wilks’ Lambda value (quotient between determinant of the variability matrix within the groups and determinant of the total variability) and the least squared differences obtained by the Mahalanobis method (measure of distance between the two groups), as well as their respective significance levels.

Table 3 shows the discriminant function, which is a linear combination of the eight variables included and a constant, similar to the multiple linear regression equation.

This discriminant function is needed in order to assign the subjects to their appropriate groups. In our case, the discriminant function correctly classifies 80.65%.

However, as it was of interest to check the possibility of estimating, by means of the Work Utility Function, the degree of General Satisfaction a participant on this type of E.T./C.O. Project would experience, a calculation was made of the mean of the scores assigned (through the Work Utility Function) to subjects within the same level of the General Satisfaction variable (Table 4). For instance, the mean subjective utility of subjects in Category 10 of the General Satisfaction variable would be 0.82. Category 4 includes all subjects with a score of less than or equal to 4, since the number included is very small.

It can be observed how the mean scores estimated through the utility function change their sign from the 7 value onwards. The negative scores appear, according to the results of the discriminant analysis, in those cases in which subjects feel most motivated by the achievement of Security and Power, whilst when they are more attracted by Social Acceptance, Social Recognition and Self-development they present positive results (Table 5).

Graph 1 shows, on the y-axis, the mean work utility function scores for each value of the GSL (x-axis) variable. It can be observed that subjects who consider the Project to be useful with respect to their professional and personal interests (positive utility) feel satisfied, while those who do not consider it useful (negative utility) feel dissatisfied. This utility function, it will be remembered, was constructed with data obtained in the second phase, in which subjects had already undergone basic training and had progressed to the status of students-workers.

The ultimate objective of the study was to attempt to construct a utility function that would allow us to predict subjects’ job satisfaction in the different Projects on the basis of the data obtained in the first phase, when they were still students. However, in this phase there was no available measure of subjects’ job satisfaction level (the GSL variable), since they had no work experience.

The following steps were taken to resolve this problem:

This second function allows 72.58% of correct classifications to be obtained.

Despite the fact that this discriminant function has a predictive power of approximately 8% less than the first one, on correlating the values of the two functions a Pearson correlation coefficient of 0.75 is obtained, indicating an acceptable level of similarity between them.

In order to check the degree of similarity between the two work utility functions (second and third), and between these and the General Satisfaction variable, a procedure made up of different phases was followed. First, a distribution in centiles was obtained of all scores assigned, through the second utility function, to subjects in the second (worker) phase. Second, the centile value was found which, within the previous distribution, was obtained by each of the mean scores assigned through the second utility function to subjects included in the same level in the General Satisfaction variable. Third, the distribution in centiles of all the scores assigned through the third utility function to subjects in the student phase was found. Finally, we searched for all of those values assigned through the third utility function that corresponded to the centiles of the mean scores assigned through the second function to subjects included in the same level of the General Satisfaction variable. The results of this entire process can be seen in Table 8.

If we consider the positive values in the functions calculated for the worker phase (first and second utility functions) as an indication that subjects were satisfied, in the case of the third function, where all the values are negative, this indicator is reflected by the value –1.54.

When the mean scores assigned for the second and third utility functions are represented in terms of the levels of the General Satisfaction variable, we can observe a close correspondence between the two (Graph 3). Also, since there is a change of sign in the second function between percentiles 32 and 29, and given the similarity between the distributions, the value –1.54 in the third function can be considered as similar to the value 7 in the second function. It can therefore be accepted that the more the values obtained with the third function exceed –1.54, the greater will be the predicted degree of job satisfaction; and the reverse –the lower these values, the lower the predicted degree of job satisfaction.

Although it is clear that one of the main motivations that explain why subjects work is salary ( Bravo and cols., 1993), the results obtained from the different analyses allow us to conclude that level of Job Satisfaction is also strongly influenced by aspects of work related, at a given moment, to needs of a social nature (social recognition, social acceptance and security) and to an increase in future employment possibilities (self-development). These conclusions are supported by various studies ( Hackman and Oldman, 1976; Peiró, 1984; Wosy, 1989; Peiró and cols., 1987, 1990, 1993, 1994; Sanchís, 1991; MOW, 1991; Salanova and cols., 1993.)

For those who are unable to maximise the utility of the speciality (negative utility in terms of Decision Theory), their remaining on the programme can only be attributed to the economic incentive, so that the Project becomes merely a temporary means of subsistence.

With respect to the third utility function, we venture to suggest that all scores higher than –1.54 obtained in subjects without work experience would be indicative of their future degree of dissatisfaction in this employment context.

Finally, we believe that the three utility functions obtained may serve to predict and explain job satisfaction levels, by means of the appreciation of the utility subjects attach to the actual work done, in a context with such specific characteristics as those of Occupational Training programmes.