Mobile Edition
Print
Get the Mag
Weekly UpdatesAbstract Competitiveness in most developed countries is increasingly measured in terms of the quality of goods and services and to a lesser extent in terms of prices. It is crucial to know the opinion of the consumer with regards to the goods and services. The paper uses data from satisfaction surveys on local events celebrated during 2006 in the Maresme region (Barcelona). The effects of each explanatory variable on the satisfaction of individuals are analyzed using an ordered logit model for the Likert variable and a logit model for the binary variable. It is observed that, although generic valuations are very similar, the effects of some explanatory variables can be very different, since each indicator presents specific features.
Keywords Satisfaction * Consumer * Indicator
Introduction
In an economic context, competitiveness in most developed countries is measured increasingly in terms of the quality of goods and services and to a lesser extent in terms of prices. In this context, it is crucial to know the opinion of the consumer with regards to the goods and services. Therefore, any integrated system of quality will incorporate knowledge of the consumer's satisfaction with their own product as one of its most relevant aspects.
This increasing importance is not disconnected or exempt from activities in the public sector and, in particular, with the evaluation of local events; the economic impact of which is especially relevant in many cases. On the other hand, when we try to assess the satisfaction of a leisure event, that is, an event which attracts visitors from other towns who decide to dedicate their spare time to see the event, there are many possibilities of satisfaction indicators to use: indicators of direct or indirect satisfaction, binary or ordered.
At this point, it is necessary to know the consequences of the indicator of satisfaction chosen, as well as the change that can take place in the importance of the explanatory factors of the satisfaction depending on which indicator has been chosen. This is the main objective of this research that uses data from satisfaction surveys carried out on local events which took place during 2006 and 2007 in the Maresme region (Barcelona). Furthermore, the description of the explanatory factors of the satisfaction of these events is an additional objective of the paper, which is structured as follows. Firstly, there is a brief summary of the Economic Literature on the evaluation of satisfaction; in the Empirical Models section the empirical models used are introduced, in the Data section the database of the paper is described, while the Results section presents the main results. The study finished in the Conclusions section with some brief conclusions.
Economic Literature
The area of economic literature with most contributions on the satisfaction analysis of a product or service is to be found in market research. In this field, there is a considerable amount of literature with well developed and contrasted models of consumer satisfaction, such as the expectations-value model estimated by Mittal et al. (2004) using expectations-dissatisfaction and quality-dissatisfaction models by Oliver (1980) among others. (1) The theory of equity between the costs and benefits of consumers was introduced by Oliver and Swan (1989) and pinpointed in terms of the determinants of satisfaction, by Heskett et al. (1997). Finally, Latour and Peat (1979) suggest the theory of the ideal standard (Sirgy 1984) and Tse and Wilton (1988), the theory in which dissatisfaction is the result of the current situation and is independent of expectations. It is in this area where studies have also appeared which analyze the relationship between the satisfaction of a given set of aspects about the service with the global satisfaction of the aforementioned (Oliver 1997).
In the local public services framework, there are also studies that measure the relationship between the indicators of quality of public services with taxes and municipal expenditures (Beck et al. 1987) and studies which measure the relationship between objective indicators of certain public services such as the police and parks with the subjective judgments of citizens (Brudney and England 1982; Brown and Coutler 1983). More recently, the levels of citizen satisfaction regarding the same public services were compared between towns (Miller and Miller 1991; Kobayashi and Miller 2000). In this area, even though some studies have insisted on methodological aspects, most of them are an extension of more generic methodological questions related to social sciences and the research through surveys (Folz 1996; Kobayashi and Miller 2000).
Finally, the economic literature of local events has focused on its economic impact (Crompton and Mckay 1994; Kim et al. 1998) but not on the satisfaction of the individuals who attend these events, nor on the study of the best way to measure it. In Spanish economic literature, this gap becomes even more apparent because the economic literature about local events that studies the importance of the indicator of satisfaction chosen is very limited. Both questions constitute the two main characteristic features of the study that follows.
Empirical Models
The econometric models which will be used in this section depend on the type of indicator to be used.
Firstly, the most direct measure to capture the satisfaction of an individual is obtained when the survey includes a question in which the satisfaction is valued by means of a coded variable from 1 to 10 according to a Likert scale. In this case, using a multinomial logit model (2) does not consider the fact that the dependent variable reflects an order. Also, a regression model would consider the difference between an evaluation from 7 to 8 in the same way as one from 9 to 10, when, in fact, these numbers are only an order in the classification of evaluation.
To analyze this type of data, ordered logit models are used. Thus:
y * = [beta]X + [epsilon]
Where X are explanatory factors, [epsilon] not observable factors and y* a latent variable which is not observed because, given that they are only given J possible answers, an individual will choose the one that best reflects their satisfaction regarding the event. What is observed is, for example,
y = 0 if y * [less than or equal to]0
y = 1 if 0 [less than or equal to] y * [less than or equal to] [[mu].sub.1]
y = 2 if [[mu].sub.1] [less than or equal to] y * [less than or equal to] [[mu].sub.2]
...
y = J if [[mu].sub.[J-1]] [less than or equal to] y *
If we assume that [epsilon] has a logistic or normal distribution in all the observations, which is a generalization of the binomial logit model, then once we have obtained the log likelihood-function, it is optimized.
However, when the survey measures satisfaction from a binary indicator coded as 0 and 1, for example, "0 = Not satisfied" and "1 = Satisfied", the estimation of models which explain the fact that the visitor chooses between one or another option must be carried out with discrete choice models or qualitative response models.
In these models, there is supposedly a continuous latent variable and not observable which, in our case, would be consumer satisfaction. It is named y and satisfies:
y * = [beta]X + [epsilon]
where X are explanatory factors, [epsilon] not observable factors and [beta] parameters. However, we do not observe this latent variable, but only if the person is satisfied or not. Therefore, our observation is:
y = 0 if y * [less than or equal to] 0
y = 1 if y * > 0
The probability of the event y = 1 is:
Prob (y * > 0) = Prob ([beta]X + [epsilon] > 0)
If the distribution is symmetrical, as it is in the normal (probit models) and the logistics distribution (logit models)
Prob (y * > 0) = Prob ([epsilon] < [beta]X) = F([beta]X)
In both logit or probit models and ordered models, the marginal effects of the explanatory variables on the probabilities are not constant, that is, they are different from the coefficients of the model. The reason for this result is that the marginal effect depends on the point of the distribution analyzed. In the case of binomial logit models, only the sign of the coefficient is interpretable. Thus, a positive sign indicates an increase in the probability that y is equal to 1, and a negative sign, a decrease. In the ordered models, based on the estimation of [beta], the sign of the marginal effect can only be known with certainty when it is the first option (opposite to [beta]) and the last one (same sign that [beta]), but we cannot easily know what is happening in the intermediate zone. (3)
Data
The data is from satisfaction surveys on local events which took place during 2006 and 2007 in the Maresme region (Barcelona). (4) The event, an exhibition of floral decorations, can be considered as an event which visitors can visit frequently. We collected a total of 583 responses from the visitors to this local event. Through the survey we can define different satisfaction indicators that are grouped as follows:
Direct indicator
Evaluation (5): variable that takes values from 1 to 10 depending on the degree of visitors satisfaction with regard to the event (10 being indicative of the highest satisfaction).
Table 1 shows the descriptive statistic for this ordered indicator. We can see that the average visitor satisfaction was 7.67, which despite the range between 2 and 10 does not show a great dispersion given the value of the standard deviation.
Indirect qualitative indicators
Recommendation: dummy variable that takes the value 1 if the visitor recommends the event and 0 otherwise
FEED