Methodology

The Puntos Centinela (PC) 2017 covers the analysis of three main subjects, related with the monitoring and evaluation of the operation of PROSPERA´s components. These subjects were defined as the cornerstones for the study: 1) perception about the inputs for the program´s operation, 2) the management for the operation and production of aids and services, and, 3) the satisfaction of the beneficiaries about the delivery of services and aids of the program.

In addition, the study includes other topics of interest related with the socioeconomic context of the beneficiary population, such as: indigenous population, poverty perception, and social deprivations, among others.

Within the analysis dimensions that PC considers, the methodology is focused in the measurement of eight attributes:

1. Sufficiency: Measures if the inputs for the execution, management, reception and delivery of the aids and services are sufficient to cover all the beneficiaries´ needs.
2. Quality: Measures if the necessary inputs, the management of key processes and services of PROSPERA have the operative characteristics to fulfill the objectives of the Program from the point of view of the beneficiaries and PROSPERA personnel.
3. Usefulness of the information: Measures if the beneficiaries, service providers and PROSPERA staff perceive that they possess clear, adequate, necessary and official information about the Program.
4. Knowledge: Measures the level of knowledge of the beneficiaries, services providers and PROSPERA staff, about rights, actions and the aids and services of the Program.
5. Accessibility: Measures the level in which the beneficiaries have access to attention, information and the aids and services that the Program offers, if they fulfill the necessary requirements and the means are available.
6. Conditioning: Measures if the aids and services of PROSPERA were delivered without being associated to a negative stimulus or a political or economic booster.
7. Opportunity: Measures the level of satisfaction of the beneficiaries with the timeliness of the attention, information, aids and services received.
8. Satisfaction: Measures the level of satisfaction of the beneficiaries with the attention, information and the delivery of aids and services. Likewise, it measures the level of satisfaction of PROSPERA personnel with the implementation and management tasks that are part of their daily work.

To achieve its objectives, PC 2017 considered the application of ten questionnaires designed to capture the perception about the Program in ten analysis units, according with the following classification:

 STATISTIC SAMPLE OF PC 2017

The design of the statistic sample of PC 2017 was: probabilistic, multi-stages, stratified and by clusters. It was a probabilistic design because each possible sample has the same probability of being chosen. The sample consisted of two phases. In the first one, a locality sample was obtained. In the second stage, for each selected locality, a sample of households that are covered by PROSPERA was obtained.

Additionally, the statistic sample was stratified in two dimensions: urbanization of the locality and the belonging to indigenous localities. The stratification of localities was performed using directly the classification provided by PROSPERA.

The grounds for performing a stratified sampled laid on that it allows grater consistency in the results, since the individuals within each stratum are much more similar than the individuals in general, this provides the prior estimates of the mean of the population with a lower standard error than the one that can be obtained from a random unrestricted sample of the same size.

The population was divided into 32 mutually exclusive regions (states); each region was projected independently, in such a way that the information could be analyzed by region. The localities were considered as clusters.

Unlike past editions, PC 2017 was characterized for having advantages over several aspects:

• The 2017 sample has a statistical representation nationally for the analysis units. Additionally, for the census of cardholder beneficiaries with co-responsibility counted with statewide representativeness and for the cardholders without co-responsibility had an error of 15%.
• Additionally. PC 207, considered stratums of urbanization and indigenous presence for the sample of beneficiaries.
• PC 2017 was statistically representative at national with statistical basis by stratum. This means, that the sizes of the samples varied according to the number of beneficiaries with or without co-responsibility, that exist in the population of each stratum.
• Additionally, PC 2017 allowed the monitoring all the chain of supports granted by PROSPERA to the beneficiary families because the sample of cardholders served as the deterministic axis of the other analysis units. It means, that if a family resulted in a statistic sample, also the sons and daughters will have scholarships; just as schools, vocals and health units that provided services.
• Finally, the sample design of PC 2017 eliminated the conceptualization errors of “units of sampling” existent in previous editions; During the 2016 analysis, the term “unit of sampling” was mistakenly used and must be changed for the term unit of analysis, to avoid confusions. In some cases the unit of analysis is the sampling unit, but not in all cases.
  
  

 SCHEME OF SAMPLE SELECTION

Due to the characteristics of the statistic sample, the sampling of cardholders registered under the schemes With or Without Co-responsibility was carried out in two stages, and served as the deterministic axis of the other analysis units. Firstly, during the stage of selection of clusters. The Primary Units of Sampling (UPM for its acronym in Spanish) were the localities. Secondly, the Secondary Unit of Sampling (USM for its acronym in Spanish) was constituted by the identification of the other analysis units. In this way, the sample of Scholarship holders of High school that belonged to the families of the cardholders in the sample. Similarly, the vocals were selected depending on the localities in which the cardholders in the sample were located.

For the primary schools, junior high and high schools and their personnel, the units of sampling were the schools that provide services to beneficiary families in the localities that were sampled. If the locality had only one school then it was sampled; but, for localities with more than one school, a random sample of 5% of the schools in the locality was performed.

Similarly, for the medical units and its staff, the sampling units were the medical unit that provides services to the beneficiary families in the localities that were sampled. If the locality had only one medical unit then it was sampled; however, for localities with more than one medical unit, a random sample with the 40% of the medical units in the locality was performed.

In the case of PROSPERA`s staff and the Settlement Institutions staff, the analysis unit was the sampling unit. This is to say that a whole sampling was performed.

In conclusion, the sample and its disaggregation were obtained through probed statistical methodologies. In the first place, a stratification of the population was performed. Then, a cluster sampling (localities) was performed within each stratum and, subsequently a random sampling of the beneficiary households was performed within the clusters. Therefore, once the beneficiary households were selected, it was possible to find evidence for the other analysis units.

 SAMPLE COMPOSITION

As it was explained, for the edition of PC 2017, the classification of the cardholders with or without co-responsibility served as the deterministic axis of the samples of the other analysis units. Thus, the methodological design that characterized PC 2017, based on the stratifications of household characteristics allowed to generate a sample with the same tendency as the whole beneficiary population of the Program. The sampling distribution by stratum for cardholders with or without co-responsibility is shown in the following tables:

The sample design allowed to have nationally statistically representative results for the other analysis units. With that, the representativeness of the sample allowed that the results of PC 2017 could be considered as robust to al population levels. The distribution of the sample by state was the following:

 REPRESENTATIVENESS

Due to the characteristic and the design that the statistical sample had, the results that were obtained from the interviews may be understood as robust. Therefore, the statistical sample that was used provided a representation of the results of the evaluated universe.

In this sense, the following table presents the size of the sample and the representativeness obtained by the Analysis Unit, in the different stratums and to a national and statewide level. It should be noted that when talking about representativeness with statistical basis, is at a confidence level of 95%.