Summary
Attitudes towards agricultural technology predispose farmers to accept; reject, or stop using the new technology after it has been adopted. For this reason, attitude assessment offers a valuable procedure to generate measurements of support to agricultural technology and data to predict farmers’ behavior. Measurements of support and predicted behavior can inform planning of extension activities and eventually contribute to improvement of rates of adoption of agricultural technology.

This paper is part of a larger research proposal on adoption of improved maize varieties and chemical fertilizers (NPK and urea) in the “Machipanda” and “Manica” administrative posts of the Manica district(1). The paper addresses attitude assessment and measurement of attitude-behavior relationship, and it is organized in three chapters: background, literature review, and methodology.

Chapter I: Background
In the country of Mozambique (Southern Africa), a relatively large proportion of the cultivated land (40.46 percent) is planted to maize (Zea mays L.) (Instituto Nacional de Estatistica (INE)) [National Institute of Statistics], 2001. Research at the household level in the rural areas of Mozambique, also suggests that maize is an important food crop for the Mozambican population. For example, (Sistema Nacional de Aviso Previo (SNAP)) [National Early Warning System], 2002; Abdula (2005), mention that maize is a basis of the diet for most families in Mozambique, and is grown by about 80 percent of rural households.

One of the main problems with maize production is its low average yield; generally less than 1.5 ton/ha (Jeje et al. 1998). The consequences of low maize yield include maize deficit and unmet needs. For the commercial year 2002/03, Mozambique experienced maize deficit of 22, 000 ton, and unmet needs of 15, 000 ton (Bias and Donavan, 2003).

On the other hand, research suggests that maize yield can be increased if farmers use improved maize production practices (i.e., improved seed, chemical fertilizer, and improved husbandry practices). Using data from various sources including the Ministry of Agriculture and Fisheries and the World Bank, Jeje et al. (1998) reported dramatic change in maize yield with use of improved maize production practices. According to Jeje et al. (1998), maize yield can be increased by 576 percent; i.e. from 1.5 ton/ha to approximately 6.5 ton t/ha.

Based on the belief that technology is available to help farmers increase maize yield, the public extension is implementing a national technology-transfer program - the DNER/SG 2000 extension program - for the dissemination of improved maize seed and chemical fertilizer (NPK and urea). The “Machipanda” and “Manica” administrative posts of the Manica district are among the target areas of the DNER/SG 2000 extension program.

Adoption studies conducted in this region, had addressed issues of: profitability of maize technology; the overall results of the DNER/SG2000 extension program; and the discontinuance of the technology after being adopted by the farmers. However, there is still knowledge gap in terms of the adoption history in the two administrative posts, i.e. knowledge is still needed to help understand how the diffusion of maize technology occurred in the “Machipanda” and “Manica” administrative posts.

Another adoption issue, which has not yet being studied in the region, is farmers’ attitudes towards agricultural technology, nevertheless, farmers’ attitudes are valuable input to improve the adoption rates and inform planning of extension programs (CIMMYT, 1993).

This paper is part of a larger research proposal on adoption of improved maize varieties and chemical fertilizers, in the “Machipanda” and “Manica” administrative posts. The paper addresses the attitude assessment and the relationship between attitude and behavior. The following questions will guide research:

1. What attitudes do farmers in the “Machipanda and “Manica” administrative posts, hold towards improved maize varieties and chemical fertilizer?
2. How adoption rates of improved maize varieties and chemical fertilizers varied throughout the time (years) in the “Machipanda and “Manica” administrative posts?
3. How farmers’ attitude towards improved maize varieties and chemical fertilizer is correlated to adoption of improved maize varieties and chemical fertilizer?
4. What is the relative contribution of farmers’ attitudes, for adoption of improved maize varieties and chemical fertilizer in the “Machipanda” and “Manica” administrative posts?

Chapter II: Review of literature
A. Definitions and theoretical views about attitudes
Oskamp and Schultz (2005), after reviewing several definitions of attitudes, define attitude as “a predisposition to respond in favorable or unfavorable manner with respect to a given attitude object” (p.9). Attitude object include situations, institutions, concepts or persons (Aiken, 2002).

Like Oskamp and Schultz (2005), Eagly and Chaiken (1993) emphasize attitudes as evaluative responses. Eagly and Chaiken (1993) define attitude as “an evaluative state that intervenes between certain classes of stimuli and certain classes of evaluative responses, which express approval or disapproval, favor or disfavor, liking or disliking, approach or avoidance, attraction or aversion towards the attitude object”(p. 3). According to Eagly and Chaiken (1993) evaluative responses are provided for verbal statements of beliefs, verbal statements of affect, and verbal statements concerning behavior towards the attitude object. This implies a tripartite view of attitude; i.e. attitude consist of three dimensions: the cognitive, affective and behavioral dimensions:

Eagly and Chaiken’s view about attitudes is similar to latent process viewpoint attributed to Defleur and Westie (1963) by Oskamp and Schultz (2005). Under the latent process viewpoint attitude is viewed as a process occurring within the individual. Attitude is used to explain the relationship between stimulus events and the individual’s responses. In this sense, an attitude is an intervening variable; i.e. a theoretical construct that is not observable in itself, but which mediates or helps to explain the relationship between certain observable stimulus events (the environment situation) and certain responses.

This study adopts the tripartite view of attitudes and the conceptualization of attitude provided by Oskamp and Schultz (2005), who state that “an attitude might be conceptualized as a summary of all of a person’s evaluative beliefs about, affective reactions toward, and behavioral responses to an attitude object” (p.14).

B. Attitude and behavior (A-B) relationship
According to Oskamp and Schultz (2005) attitude “as an intervening variable is a useful concept only if it conveniently summarizes, or predicts, or is related to patterns of actual behavior” (p. 266). This argument suggests that social researchers, who study attitudes, would produce more valuable results if in their studies include analysis of attitude-behavior (A-B) relationship.

Attitude-behavior relationship (A-B) is important for predicting the general tendency that the individual will engage in behaviors relevant to the attitude object (Eagly and Chaiken, 1993). From the point of view of extension educators, it is important that attitudes serve a predictive purpose with respect to farmer’s behavior if measurements of support to the use of improved maize production practices are to be useful tools in guiding program planning.

Agricultural extension agencies often assist farmers in improving maize productivity, through diffusion of conservation tillage practices, chemical fertilizers, and herbicide-tolerant maize varieties. The success of the program depends on farmer’s volitional behavior; i.e. farmers have to decide whether they will perform or not perform behaviors related to improved varieties (such as planting the seed, buying the seed, processing corn for animal feed, and cooking corn for consumption) and behaviors related to the conservation tillage (such the application of herbicide, and no mechanical tillage).

For monitoring program implementation, extension officers can assess farmers’ attitudes towards improved maize production practices; and generate measurements of support to improved maize varieties and conservation tillage practices. For these measurements to be useful tools in informing program planning, it is important that attitudes serve a predictive purpose with respect to farmers’ behavior. Indeed, extension educators’ ultimate goal is to change farmer’s behaviors; i.e. to change the way farmers till the soil and their preferences regarding the type of maize varieties to grow.

It is assumed that a change in the overt behavior (use of improved practices) will be consistent with the attitude held. In many cases, however, attitudes and actions are quite different (Rogers, 1995); but still it is essential to explore the association between attitudes and behaviors (A-B) and formulate hypotheses on the predictability of farmer’s behavior from attitudes.

Attitudinal relevance
Attitudinal relevance has been considered as one important factor in attitudes and behaviors (A-B) correlations (Kim and Hunter, 1993). Kim and Hunter, use the concept of “attitudinal relevance” (i.e. the degree of match between attitude elements and behavioral elements) for understanding A-B correlations. According to Kim and Hunter (1993) for the correlation (A-B) to exist, the attitude elements have to be relevant to behavior elements in the scale.

Using meta-analysis method Kim and Hunter (1993) tested the attitudinal relevance, through a review of one hundred and thirty eight studies selected from a total sample of 90,808 studies that addressed more than 20 behavioral topics. The studies were coded as low relevance, medium relevance, and high relevance based on how attitudes items represented behavior elements. If action and target in the attitude elements and in the behavior elements matched, the study was classified under high relevance. If only target in behavior and attitudes elements matched, the study was classified under medium relevance. If none of the two matched, the study was classified under low relevance. For studies that addressed multiple behaviors the base for classification was whether the behaviors were representative of general attitude.

Because this study will not explain the effects of the moderator factors of attitude behavior correlations, the study adopted the principle of attitudinal relevance to validate its conclusions about the existence or non-existence of A-B relationship.

Eagly and Chaiken’s descriptions of compatible measures to increase attitude-behavior correlations, helped clarify the notions of action and target used by Kim and Hunter. Eagly and Chaiken (1993) state that according to Fishebein and Ajzen, behavior has four elements: action, target, context and time. The action is the behavior itself. For example, a farmer is planting improved maize seed. The action is toward a target. For example, a farmer is planting improved maize seeds. The context refers to the place where the action is performed. For example, a farmer is planting improved maize seeds in his/her plot. The time refers when the action takes place. For example, a farmer is planting improved maize seeds in his/her plot this current cropping season.

For the present study the behavior (adoption of improved maize seed and chemical fertilizer) is measured by asking farmers whether they planted improved maize varieties and used chemical fertilizers in the cropping season that the survey is conducted. Farmers’ attitudes towards improved maize and chemical fertilizer are measured mainly by verbal statements of beliefs.

C. Research hypotheses
Several aspects explained in the literature review, namely: (i) direct experience, such as planting improved maize varieties and using chemical fertilizers; (ii) the attitudinal relevance; and (iii) the fact that attitudes and behavior will be measured at the same time, led to the following hypotheses about A-B relationship:

H1. Relatively strong positive correlations can be expected between adopters’ attitudes toward improved maize varieties and the adoption of improved maize varieties (behavior).
H2. Relatively strong positive correlations can be expected between adopters’ attitudes toward chemical fertilizers, and the adoption of chemical fertilizer (behavior).

Chapter III: Methodology
A. Research design
This study will use a descriptive and cross-sectional survey research design. The study will use an interview schedule for data collection. Data will be collected from a random sample of approximately 300 farmers in the Machipanda and Manica administrative posts of the Manica district, during February-April 2006. Data will be analyzed using the SPSS version 12.

Sample selection
The population of interest to this study consists of all farmers in the Machipanda and Manica administrative posts of Manica dstrict. Approximately 300 farmers will be selected randomly, using a multistage sampling. The multistage sampling will involve two levels, the level of administrative post and the level of locality or bairro.

Level of administrative post
Step I: Stratification of localities or bairros
The researcher will obtain the list of localities or bairros from the administration headquarters. Either the bairros or localities have to be classified as rural zones and not urban zones. Then, the researcher will assess the localities in terms of proportion of adopters and non-adopters. Those localities that have fairly similar proportion of adopters and non-adopters will be grouped together (stratification). From the groups of localities, the researcher will select randomly two localities. To help make tradeoffs between localities, other aspects such as proximity between localities, proximity of each locality to roads, input retailers, market for maize, and extension services (either the main office or the house of the extension agent), will be considered.

Level of locality or bairro
Step II: Stratification of households
The researcher will list all the households that exist within the localities or bairros selected in step I. This will be followed by the identification of adopters and non-adopters. Two groups (stratas) will be formed, the group of adopters and the group of non-adopters. A random sample will be selected proportionally from each group, to constitute a sample size of between 50 and 80 farmers.

If it reveals to be difficult to distinguish between adopters and non-adopters beforehand, the researcher will: list all the households within the bairro or locality. Select randomly a sample size which is much larger than the sample size needed (say we need 40 households and we select 80). Then keep interviewing and for each interview identify if the farmer is an adopter or non-adopter, and group accordingly. Then the researcher will stop interviewing the category (adopter or non-adopters) as soon as she has reached the number of respondents in that category.

Item analysis for the attitude scale
Item analysis will be performed for reliability and as a step in the process of constructing the attitude scale. Two methods will be used to perform the item analysis: item-total correlations (Oskamp and Schultz, 2005: p. 50) and the method of Cronbach’s alpha. The following steps will be followed for performing item analysis:

(i) Pre-testing of the interview schedule on 15 farmers in the Machipanda and Manica administrative posts.
(ii) Entering data and computing descriptive measures (Mean and Standard Deviation) for the items.
(iii) Comparing the items based on the Mean and Standard Deviation. Among items with similar averages, items with least spread measures are preferred for the final questionnaire.
(iv) Computing the correlation of respondents ‘scores on an item with their scores summed over all the items. Higher correlations indicate better item, and items with low (less than .25 or .30) or no correlation with the total score are discarded.
(v) Computing Cronbach’s alpha. For a reasonably accurate scale the value of Cronbach’s alpha ought to be at least .85. But in research practice, scales with smaller values are commonly used (Willock et al. 1999). This fact may be due to the threat to validity created by eliminating items in order to achieve high reliability index. Therefore, before an item is eliminated based on Cronbach’s alpha its validity will be analyzed on the basis of theory.

C. Modeling adoption
Adoption is measured quantitatively as a dichotomous response variable (adoption or no-adoption) subject to the influence of a number of continuous and or/categorical independent variables, including level of education, gender, equipment owned, and sources of information [CIMMYT (1993); Zegeye, Tadesse and Tesfaye (2001a, 2001b), Zegeye and Tesfaye, (2001), Zegeye and Haileye (2001)].

Adoption is coded as 1 and no-adoption is coded as 0 (Chatterjee, Hadi, and Price, 2000). The regression model aims at modeling the probabilities of adoption or no-adoption. According to Hosmer and Lemeshow (2000), in many fields, the logistic model is the standard method of analysis when the outcome variable is dichotomous. The cumulative distribution of normal curve (probit model) has also been used for modeling dichotomous response variables. However, the logistc model is considered simpler and superior to the probit model (Chatterjee et al. 2000). From a mathematical point of view, the logistic model is considered flexible and easily used function (Hosmer and Lemeshow, 2000).

The logistic model
The logistic model is described by the following formula (Chatterjee, et al. 2000; Hosmer and Lemeshow, 2000):

References
Abdula, D. C. (2005). Improving Maize Marketing and Trade Policies to Promote Household Food Security in Southern Mozambique. Unpublished M.Sc Thesis. Michigan State University. Department of Agricultural Economics.
Aiken, L. R. (2002). Attitudes and Related Psychosocial Constructs. Theories, Assessment, and Research. Sage Publications.
Bias, C. & Donovan, C. (2003). Gaps and Opportunities for Agricultural Sector Development in Mozambique. Research Paper 54E. East Lansing Michigan: Michigan State University. Department of Agricultural Economics.
Chatterjee, S., Hadi, A. S. & Price, B. (2000). Regression Analysis by Example. Third Edition.
Wiley Series in Probability and Statistics.
CIMMYT. (1993). The adoption of Agricultural Technology. A guide for Survey Design.
Mexico, D.F.:CIMMYT.
Eagly, A. & Chaiken, S. (1993). The Psychology of Attitudes. Harcourt Brace & Company. USA.
Hosmer, D. & Lemeshow, S. (2000). Applied Logistic Regression. A Wiley–Interscience Publication.
Instituto Nacional de Estatistica (INE) [National Institute of Statistics]. (2001). Censo Agropecuario. [Agricultural and Livestock Census]. Maputo. Mozambique.
Jeje, J. J., Machungo, C., Howard, J., Strasberg, P., Tschirley, D., Crawford, E. & Weber, M. (1998). What Makes Agricultural Intensification Profitable for Mozambican Smallholders? An Appraisal of the Inputs Sub sector
Oskamp, S. & Schultz, W. P. (2005). Attitudes and Opinions. Third Edition. Lawrence Erlbaum Associates, Publishers. Mahwah, New Jersey.
Kim, Min-Sun & Hunter, J. E. (1993). Attitude–behavior relations: A meta-analysis of attitudinal relevance and topic. Journal of Communication; 43, 1.
Willock, J., Deary, I. J., McGregor, M., Sutherland, A., Edwards-Jones, G., Morgan O., Dent, B., Grieve, R., Gibson, G. & Austin, E., (1999). Farmers’s Attitudes, Objectives, Behaviors, and Personality Traits: The Edinburgh Study of Decision Making on Farmers. Journal of Vocational Behavior 54, 5-36.
Rogers, E. M. & Havens, E. (1961). The Impact of Demonstrations on Farmers’ Attitudes Toward Fertilizers. Research Bulletin 898. Ohio agricultural Experiment Station. Wooster, Ohio.
Sistema Nacional de Aviso Previo (SNAP) [National Early Warning System] (2002). Database Maize in Mozambique: 1992/93-2000/2001. Source: Sistema de Aviso Previo Retrieved on 05/20/05, from www.fanrpan.org/download/vulnerability.pdf www.aec.msu.edu/agecon/fs2/mozambique/wps54e_fig2.pdf - Jun 25, 2005.
Tesfaye, Z. & Alemu, H. (2001). Adoption of Improved Maize Technologies and Inorganic Fertilizer in Northwestern Ethiopia. Research Report No. 40. Ethiopian Agricultural Research Organization.
Tesfaye, Z., Bedassa, T. & Shiferaw, T. (2001). Adoption of High Yielding Maize Technologies in Major Maize Growing Regions of Ethiopia. Research Report No. 41. Ethiopian Agricultural Research Organization.
Tesfaye, Z. & Shiferaw, T. (2001). Determinants of Adoption Maize Technologies and Inorganic Fertilizer in Southern Ethiopia. Research Report No. 39. Ethiopian Agricultural Research Organization.