FOI in Practice: Analysis of the Mexican FOI System

Washington D.C., March 20, 2008 – In celebration of Sunshine Week, the National Security Archive's Mexico Project publishes today a new study of Mexico's transparency law: "FOI in Practice: Measuring the Complexity of Information Requests and Quality of Government Responses in Mexico."

The study represents the first comprehensive analysis of the Mexican freedom of information law: what information requesters have sought and how the government has responded.

The authors analyzed the quality of government responses in relation to the complexity of FOI requests sent through Mexico's electronic information system from June 12, 2003 to April 30, 2006. After examining 1,000 information requests and corresponding government responses, the authors concluded that in 76% of the cases the government responses satisfied the requests of the user during the first three years of the law's existence. Nevertheless, the results also demonstrated that the most complex FOI requests were more difficult for public officials to answer, and received satisfactory responses in only 57% of the cases analyzed.

The findings serve as a warning about Mexico's need to improve the capacity of government agencies to respond to more complex requests for information as requesters become increasingly sophisticated in their demands over time.

FOI in Practice: Measuring the Complexity of Information Requests and Quality of Government Responses in Mexico

This study is an analysis of the quality of government responses in relation to the complexity of FOI requests sent through Mexico's electronic information system from June 12, 2003 to April 30, 2006. After examining 1,000 information requests and corresponding government responses, the authors of the study concluded that in 76% of the cases the government responses satisfied the requests of the user during the first three years of the law's existence. Nevertheless, the results also demonstrated that the most complex FOI requests were more difficult for public officials to answer; those classified as "highly complex" received satisfactory responses in only 57% of the cases analyzed. The findings serve as a warning about Mexico's need to improve the capacity of government agencies to respond to more complex requests for information as requesters become increasingly sophisticated in their demands over time.

This paper could not have been completed without the contributions and insight of Cristina Galíndez, Patrick Ball and José Luis Marzal.

Mexico's freedom of information law (Ley Federal de Transparencia y Acceso a la Información Pública Gubernamental - LFTAIPG) was signed by President Vicente Fox on June 10, 2002, and took effect one year later, guaranteeing the public's right to request and receive information from the government. It represents one of the most important human rights achievements accomplished during the Fox administration, (note 1) and was due in large part to the civil society actors who conceived the law and convinced Mexico's political leaders that it was necessary. It was a major step forward in the country's transition to democracy, and continues to play a vital role in fighting corruption and forcing increased government openness and accountability.

When the transparency law went into effect, the government's first objective was to ensure that every federal agency (dependencia pública)adopted the measures required to be able to respond to citizens' requests for public information. Agencies created new "liaison offices," or Unidades de Enlace, to serve requesters. Today, almost five years later, liaison offices are well established, operating with their own staffs under internal regulations that have formalized their procedures. As a result, openness advocates have shifted their attention from monitoring government compliance with the changes mandated by the law to evaluating how well the law is working. One key to measuring the law's impact is the quality of responses that public officials provide citizens exercising their right to know.

Government responses to information requests are determined in part by the law itself, which sets certain standards that must be met in order to avoid sanctions imposed by Mexico's information oversight commission (Instituto Federal de Acceso a la Información Pública—IFAI). These standards include the timeliness of agency responses (a 20-day requirement), the type of information subject to the law, and the exceptions granted an agency to justify a denial. Although the standards help guarantee a minimum level of compliance by agencies, they are insufficient for measuring the quality of agency responsiveness to FOI requests.

One way to measure the quality of responsiveness is to examine the degree to which the responses satisfy the original requests for information. In order to do so, it is first necessary to examine the characteristics of the requests themselves. What type of information are people seeking from the Mexican government? How complex is it? To what extent does it tax the capacity of government agencies to find and release it? Once the characteristics of the requests are better understood, it becomes possible to design measures to evaluate the satisfaction given by government's responses.

The following paper aims to take a look at the quality of government responses through a statistically representative sample of the FOI requests and responses available to us through the Mexican Sistema de Solicitudes de Información (SISI). Requests and responses were analyzed according to a qualitative framework designed specifically for this study. With this methodology, we were able to get a broad picture of the type of information being requested and the overall satisfaction provided by government agencies. Through our examination of 1,000 examples we were also able to identify common patterns and practices in requests and responses. We hope our findings will be useful to openness advocates and government authorities alike as they take stock of what the federal transparency law has accomplished to date, and how it can be improved and become more responsive in years to come.

The National Security Archive undertook this study with more than twenty years of experience in using the U.S. Freedom of Information Act (FOIA). Since the Archive's inception in 1985, Archive analysts have sent over 37,000 requests for information to the U.S. government. The idea for the study came from the Archive's Mexico Project after several years of consulting the SISI database of requests and responses. We found that requests filed through the Mexican Transparency Law were different from those normally sent through the FOIA. One of the most notable differences we found was that in many cases Mexican requesters did not ask for documents, but for data (such as agency budget figures, salaries, or staff levels, for example), and correspondingly most of the government responses were not provided in the form of official records. In this manner, we saw that the SISI system did not generally result in the declassification of government records; rather, officials were creating new information to meet the demands of the user (something forbidden in the U.S. law). We also noticed several cases where government responses were completely unrelated to the information petitioned in the corresponding FOI request – a disconcerting observation, which we felt deserved further examination.

We considered different approaches to analyze Mexico's access-to-information system for this study. Our first problem was selecting the sample data of FOI requests and the corresponding responses from government bodies. Our initial idea was to choose a particular month each year and focus on five agencies that received the most requests. However, we concluded that this method would not be statistically representative and would not allow us to draw conclusions about the whole body of requests and responses. We then contacted Patrick Ball, a scientist, statistician and director of the Human Rights Program at the Benetech Initiative who has conducted statistical analyses for several large-scale human rights projects, including truth commissions and human rights investigations in Latin America. With the help of José Luis Marzal, Director of Coordination between IFAI and federal agencies, we extracted the requests and response data from SISI in a format that Patrick could work with to get statistically representative samples.

The requests analyzed in this study were those filed through SISI as of April 30, 2006, totaling 81,336. Thus the conclusions provided in this paper are representative to the requests received through SISI from June 12, 2003 to that date.

After the random sample was identified, we contacted Cristina Galíndez, an independent consultant, who developed a qualitative methodology that looked at several variables: including the type of information that users were requesting (documents or data), the level of complexity of the request, and the extent to which the government's responses "satisfied" the requests (that is, resulted in the information originally sought).

As we began reviewing actual requests and responses, however, we quickly realized that our initial measurements of "satisfaction" were too limited. Not all government responses could be characterized as simply "satisfactory" or "not satisfactory." For example, a request that resulted in a denial because the information was classified for national security reasons would represent an "unsatisfactory" outcome from the requester's perspective, but would be the only response an agency could legally make. Accordingly, our consultant created new subcategories to represent exceptional cases, such as: classified information, information that is already publicly available, information that does not exist, and others. A detailed description of the methodology and its specific limitations can be found below.

In the course of our study, we looked at 1,000 requests and responses and then classified them using our methodology. Our qualitative analysis did not include a category for the date a request was filed (i.e., it was not possible to determine changes in the type or complexity of requests over time).

Finally, it is worth noting that there are elements in the information process that we could not factor into the methodology for this study. When considering the nature of agency responses, for example, there are tangible elements that can easily be measured, such as what documents were sent to a requester. However, there are intangible aspects of the service provided as well, because the performance of civil servantsdepends on the time and resources available to carry out their duties. There are many variables involved behind the scenes in the agencies that influence the kinds of responses agencies give, including their own budgets, staffs, and levels of training. In other words, this study cannot draw conclusions about the reasons for good/poor responses; we can only draw our analysis from the tangible elements that are part of the information process.

Mexico's electronic system for filing information requests is unique, in that it allows citizens to view every freedom of information request, as well as the corresponding agency responses. The only exceptions are requests that reveal personal information, or datos personales. The electronic request system (SISI)allows requesters to search for and review agency responses on the subject of their interest, and to tailor their requests based on similar information that has already been released. The transparency of the system also makes studies of this kind possible – something not feasible under the Freedom of Information Act, since the U.S. government has never provided universal access to all federal requests and responses in the same way.

When a user sends an information request to the Mexican government, they are given a number that corresponds to their request and the agency response. The user can then enter the SISI website and check the status of their request at any time. Once the agency responds, the material granted is posted on-line and made publicly available.

In order to carry out our study, we had to look at a significant percentage of the requests and agency responses made during the period examined (2003-06), using the ID numbers provided to the users. José Luis Marzal of Mexico's information oversight commission, IFAI – which serves as an ombudsman's office that monitors agency compliance with the transparency law – provided the IDs of every information request and agency response on file for the period covered by the study, in the form of an Excel worksheet.

At the start of our study, our universe of data consisted of 81,336 requests sent to the 252 government bodies (dependencies) between June 12, 2003 and April 30, 2006. In order to carry out this study, we first needed to find a way to examine a statistically representative sample of the requests – one which contained a proportionally accurate representation of each agency subject to the law. To accomplish this we turned to an expert in statistical analysis, Patrick Ball, Chief Scientist and Director of the Human Rights Program at the Benetech Initiative (which includes the Human Rights Data Analysis Group), who specializes in creating technology solutions for NGOs and human rights organizations. IFAI provided Patrick Ball with the "universe" of SISI requests for the period covered by our study and Patrick in turn extracted a random sample of data, organized in a form that we could use.

Once he had the data, Patrick broke it up into 400 distinct blocks – called "replicates" – each containing a sample of 200-250 requests taken from the total 81,336 requests from the period. The 400 replicates were not strictly "random"; each replicate contained records from the 252 government agencies in proportion to the number of requests the agencies received. Since many of the larger agencies received a much higher volume of information requests, a greater percentage of their records were factored in to each one of the replicates. Equally, the smaller agencies were factored in using a proportional amount of their information requests and government responses. In this manner, the sample data used in this study was proportionally stratified to represent every government body, and statistically representative of the entire universe of information requests and government responses for the period covered.

To carry out the study, we analyzed 1,000 requests from four different replicates: requests 2-1001 from replicates 1 thru 4. Our margin of error for conclusions drawn from the sample was +/- 3.1 percent. (note 2)

In order to carry out our evaluation of the quality of government responses to SISI requests, we had to develop a set of criteria to categorize the FOI requests as well as the official responses provided by government liaison offices. To do this, we came up with tangible measurements that could be used by an outside observer with access to the SISI system.

The study first focused on the involvement of the users and the important role they play in the outcome of the overall information process. Citizens who make information requests help determine agency responses at the outset, because they define the form of information sought – by soliciting a datum, a set of data, an official document, or a group of official documents. The quality of the responses will then depend in part on whether the liaison official fulfills the guidelines established by the user. If a citizen asks for a set of data, the quality of the response will be satisfactory only if the information offered is the corresponding set of data. By the same token, if the citizen asks for an official document, the response will be satisfactory if the requester receives the corresponding official document. In the case where the requester asks for a datum and the response provides a group of data that includes the requested information, then the response will be considered to have exceeded the expectations of the user.

Based on this line of reasoning, this study sought to analyze Mexico's information process based on two dimensions: the level of complexity of the FOI requests, and the quality of the responses provided by government bodies. The following is a description of the empirical guidelines used to measure these dimensions.

The first dimension that we examined was the complexity of the information request and the capacity of the agencies to respond based on the level of complexity. Not all the requests required the same amount of effort in searching for the corresponding information. To analyze the complexity of the requests, this study used the following criteria to examine the characteristics of requests that were the most simple to respond to, and those that resulted in a more complex search:

The second dimension of the requests that we examined was the type of information sought. Once we established the type of information requested, we could compare it to the type received. We looked at four different categories:

In order to evaluate the quality of government responses to SISI requests, the study followed the following steps:

a.) Assign a value to the requests based on the type of information described in the solicitud (one for a datum, two for a set of data, three for an official document, and four for a group of official documents).
b.) Assign a value to the corresponding government response using the same number system, based on the type of material provided by the information liaison officer.
c.) Compare the difference between the value of the of the agency response to the value of the request. The quality of the agency response is then determined by subtracting the number assigned to the response from the number assigned to the request. Using the results, the quality of the response is determined by the following scale:

d.) In cases where the agency provides material in response to a request that does not correspond to the information defined by the user, this response will be assigned a value of (NS – información no satisfactoria) and will be included with the non-satisfactory responses.
e.) In cases where the information solicited is already available in the public realm through media outlets, published information, public archives, the internet, or on the agencies website, the response is classified as public information (IP – información pública). The responses that provide clear and precise instructions indicating where to locate the information include an asterisk (*) and are categorized as satisfactory. The IP responses that do not provide clear instructions do not comply with the law's regulations, and are therefore classified as non-satisfactory.
f.) There are also cases where the agency withholds or denies the requested information. They are obviously unsatisfactory from the perspective of the requester, but are still legitimate responses under the rules established in the Transparency Law. Since there was no way to determine with absolute certainty if these responses are justified unless and until they are appealed, they were classified in this study as satisfactory. We read each and every request and response that fell within this category, and a detailed explanation of our findings appears below (see section entitled "The Responses"). The cases are as followed:

To determine the agency responses with respect to the complexity of the information requests this study used the following guidelines:

a) Classify the requests with respect to its level of complexity, based on the categories established above.
b) Establish the level of satisfaction in respect to the response to the information request
c) Group the requests together based on their category and determine the percentage of satisfactory and non-satisfactory responses based on each group.

In terms of the type of information sought by requesters surveyed, the majority asked for government data rather than official records. Ten percent of requesters sought a specific datum, while in 61% of cases, citizens petitioned agencies for a set of data. Only 18% of requests sought the release of official government documents: 9% asked for a single document and 9% asked for a group of documents. (Out of our 1,000 sample requests, 11% – 110 in total – sought personal data and could not be examined.)

Turning to the complexity of requests, the majority of requests was of low to medium complexity and did not place great demands on the corresponding agencies. Twenty percent of the cases were low complexity requests, soliciting information already available in the public domain through the government-published Diario Oficial de la Federación or the Internet. Almost half of them, 45%, were of medium complexity, relating to relatively permanent information such as employee salaries and agency infrastructure. Thirty percent were high complexity, relating to an agency's day-to-day operations. Surprisingly, only 5% of the requests sought information of "very high complexity," such as the agency's performance and program results.

Examining the question of overall requester satisfaction, we found that in 76% of the cases, the government's response satisfied the original request or exceeded the requester's expectation. Most of the responses exceeding user expectation (9%) were cases where the requester received official documents in response to a simple request for data. Requesters were unsatisfied with government responses in 24% of the cases, specific examples of which can be found in the next section.

In sum, the study makes clear that when citizens sought information of low, medium and high complexity, Mexico's government agencies were able to satisfy requesters or exceed their expectation 80% or more of the cases during the period under examination. But when requesters tried to obtain information of very high complexity, the satisfaction level fell to 57%.

Looking at the chart below, we see that the majority of total requests for information in every category of complexity sought the release of data rather than documents; the most obvious example is found in requests of medium complexity, which sought data in 91% of the cases. Requests for information of "very high complexity," on the other hand, asked for copies of government records in 41% of the cases.

The results of this study clearly demonstrate that most of the requesters are using Mexico's federal FOI law to seek data from government agencies and are not requesting copies of official documents. An overwhelming 71% of users asked for either an isolated number or fact, or a set of data. The requested data ranges from simple information related to statistics collected by an agency (e.g. request for the number of mobile telephone lines in the state of Nuevo Léon sent to the Federal Telecommunications Commission– ID 409) to complex questions regarding the operation of government agencies (e.g. request for the number of labor lawsuits and settlements paid by PEMEX ID 483).

In terms of the level of complexity of the requests, we found that in 45% of the cases the users wee interested in information related to characteristics of the organization (medium complexity) such as the number of employees and their salaries (e.g. request for a number of staff positions at Secretaría de Seguridad Pública and their salaries - ID 831) or an agency's infrastructure and equipment (e.g. request for a list of buildings that Secretaría de Relaciones Exteriores has leased - ID 759). Requesters in 30% of the cases asked for more complex information regarding the operation of the agency including copies of government contracts (e.g. requests for Pemex contracts – ID 737 and ID 738). Requests classified under the low complexity category represented 20% of the total, where users asked for information regarding norms and regulations, many of which are published in the Diario Oficial de la Federación (e.g. request for specific DOF about automobile taxes - ID 2). The category with the least number of requests was that of very high complexity with only 5% of the total cases analyzed falling in this category (e.g. request for Fideicomiso de Riesgo Compartido's policies regarding lost checks - ID 402).

We also found that in many cases, citizens asked the government to create or tailor information specifically for their requests. The law specifically addresses this issue, stating that agencies should not create new information for the user: Article 42 of the Ley Federal de Transparencia y Acceso a la Información Pública Gubernamental states that, "dependencies and entities are only subject to release documents in their archives." However, in practice agencies often go out of their way to meet the users' demands and in some cases officials even fill out questionnaires from requesters (e.g. request about salary raises sent to Centro de Teconología Avanzada A.C that read: "please fill out the annexed table" - ID 932). Although such responses represent outstanding efforts on the part of some Unidades de Enlace officers, government officials should comply with the law and provide the information to users as it was originally created. Given the exponentially increasing number of requests received by the federal government each year (the number of requests received in 2007 is almost double of the total of 2004), the burden posed by these special requests can only increase over time as well.

Another interesting finding of this study is that out of the total number of requests analyzed, 18% of requests sought information that was already publicly available. In 75% of the cases where the information was already publicly available, agency officials provided the user with specific instructions on where to locate the requested information, which was generally available on the internet (e.g. ID 778 is a request sent to SAGARPA – Secretaría de Agricultura Ganadería, Desarollo Rural, Pesca y Alimentación – for information about the production of ginger in Mexico).

Finally, in some cases, FOI officers responded to questions not related to their agency or operations, such as the case where the Secretariat for the Environment and Natural Resources (SEMARNAT, Secretaría de Medio Ambiente y Recursos Naturales) provided information in response to a request for the latest weather conditions in Finland and Paris (ID 57).

The results of this study show that the majority of responses issued by federal government bodies to citizens using Mexico's access to information system satisfied the original request. The results also show that Mexico's dependencias performed well in fulfilling their requirements as mandated by the Transparency Law during the period in question (from June 12, 2003 to April 30, 2006). In the first three years of Mexico's electronic freedom of information system, public officials performed at a "satisfactory" level in 67% of the cases, 9% percent of agency responses exceeded the expectations of the requester, and only 24% of the responses were demonstrably unsatisfactory.

There was a large discrepancy between different agencies in the way they responded to requests. Some agencies, such as the Instituto Mexicano del Seguro Social (IMSS), were consistently professional. The IMSS is the agency that by far receives the most requests (almost 30,000 to date), (note 3) and in our sample we found that their answers always included a scanned document with the agency header, signatures, stamps, etc. (e.g. ID 107).

Another agency that was notably professional in its responses was Secretarìa de Salud (SSA),which is ranked 5th in the list of agencies with the most number of requests (8,048 as of February 2008). (note 4) Salud, like IMSS, consistently responded with a scanned document with proper headers and signatures (ID 799). Another notable example was the Procuraduría General de la República (ID 64), which include scanned copies of oficios with proper headers and signatures.

Other agencies included in our random sample did not perform as well. Secretaría de la Defensa Nacional's (SEDENA's) responses were always provided as unsigned Word documents without any letterhead (ID 25). In cases where SEDENA did provide information on important issues such as drug eradication efforts (ID 525) or crimes committed by military officials (ID 521), it was proportioned in such a manner that the source of the information could not be determined and the data could not be verified. Presidencia de la República responses varied from Word documents without letterheads (ID 112) to using stationary with some type of official header.

When Mexico's law was first being discussed, there were many concerns with regards to protecting the identity of citizens filing information requests. In response, the federal law and the newly approved constitutional reforms established that users of the FOI law were not required to provide a reason for petitioning information. (note 5) Nevertheless, we would like to point out an example in which a requester asked for the date of graduation of several Mexican generals, to which the Secretaría de la Defensa Nacional responded, "When division, brigade and group generals were consulted about this request, they responded that in order for them to provide the required information they need to know why the information is being requested" (ID 22). This response was in clear violation of the FOI law.

In terms of the responses classified as "not satisfactory," we found several recurrent examples of inefficient government practices. In some cases the agency's response had nothing to do with the FOI request (e.g. requester asked for the requirements to apply for a job in SEDENA and their response was that they were not hiring at the moment - ID 23). Other times a requester asked for documents and the agency did not attach them, but limited its response to a set of data instead. There were also cases in which the agency official replied that the information was publicly available on their Web Page, but failed to provide specific instructions on where to locate the information (ID 231). (note 6)

Out of the 1,000 requests we analyzed, we found that government agencies denied information because it was "reserved" or "confidential" in 17 cases. We read each case very carefully. The justifications given by the agencies were the following:

Despite our careful examination of the 17 requests, it was impossible to verify the government's classification claims independently and with absolute certainty. However, since agencies have the legal obligation to withhold properly reserved information, and the denials in these 17 instances all cited specific articles of the transparency law (reasonably well-justified in the agencies' letters to requesters), we catalogued these 17 responses as "satisfactory."

Regarding responses that we classified according to our methodology as "non-existent information," we found 29 cases in which the government failed to produce information on the grounds that it did not exist. Agencies justified their responses for different reasons. For example, in 14 of the cases, requesters were told that the agency did not generate the information sought (e.g. a citizen asked for the salary of First Lady Martha Sahagún, to which Presidencia stated that it was "non-existent" because the wives of Mexican presidents do not receive remuneration for their activities – ID 110). In 12 instances, agencies were unable to find the requested information (e.g. PROFEPA – Procuraduría Federal de Protección al Ambiente – was asked for a copy of a specific complaint filed in its Michoacán office in 2002, but was unable to locate the relevant documents – ID 714). Finally, in three of the cases, agencies claimed that information did not exist in the form sought by the requester (e.g. a requester sought the amount of money spent by the National Institute for Senior Citizens [Instituto Nacional de las Personas Adultas Mayores] in the municipality of Zapopan, state of Jalisco, for a certain year; the agency replied that it did not aggregate its budget data by municipality – ID 998).

We carefully reviewed all of the "non-existent" responses that surfaced in our random sample and found that the majority were explained in a clear and reasonable manner; accordingly, we classified them as "satisfactory" responses. However, it was not possible to verify with absolute certainty whether every assertion of "non-existence" was justified, or if the agency was using the response for other purposes – such as a desire to hide the information or because of the difficulty in locating it. It is worth noting that, according to the IFAI's figures, "non-existent information" responses increased from 2.8% of all cases filed through SISI in 2003, to 4.6% in 2006. (note 7) One possible solution to the problem posed in trying to analyze this type of response would be to look at whether or not appeals were filed with IFAI in these cases and what their outcomes were – an exercise that beyond the scope of our study.

A third kind of response that we evaluated was when a requester sought information from one agency that was generated by another. This response was classified according to our methodology as "no es competencia de la entidad," or, the agency was not the correct entity from which to solicit the information. In some of those cases, agencies told requesters that they had asked the wrong office, but failed to direct them to the proper authority. Those responses were rated "non-satisfactory."

4.5% of the requests examined in our study are classified as "no es competencia de la entidad" and were categorized as "satisfactory" responses in accordance with Article 69 of the Transparency Law, which states that "those Liaison Units receiving requests for access to information that are not in the possession of the corresponding department or entity must assist and give orientation to the petitioners… concerning the departments or entities that could have such information."

Five of the requests analyzed were related to state or municipal information. In their corresponding responses, federal agencies referred the users to a local entity (e.g. a request asking for a proposal concerning water and sewer services provided by states was sent to the Comisión Nacional del Agua, which informed the requester that the municipal governments were the corresponding entities that managed these services - ID 709).

In eight cases, the agencies informed the users that the data sought were not generated by their offices, but could be found publicly on the Web sites of other entities, to which the agencies referred them (e.g. a request asking about regulations controlling logging in Mexico was sent to the Secretaría de Medio Ambiente y Recursos Naturales, which referred the user to the Web page of the Comisión Federal de Mejora Regulatoria – ID 565).

In the majority of the cases – 32 of these – it was clear that the users sent the request to the wrong agency. In many of the cases the text of the request itself indicates the correct agency to which the user should have sent the petition (e.g. one request asks for information related to a private entity "registered in the Registro Público de Comercio," sent to the Secretaría de Economia; the agency responded that the information was produced by the Registro Público de Comercio – ID 536).

The fact that the greatest number of cases in our study where information was denied fell into this category – "no es competencia" – demonstrates that while requesters are slowly becoming familiar with the Law, they are still unfamiliar with the nature of the agencies from which they are soliciting information. As the requests become more sophisticated and users understand the Law better, the volume of cases in this category is likely to decrease.

Most of the requests examined in our study sought information concerning current events. In the few cases when FOI users asked for data prior to the year 2000, agencies often answered that they did not hold information dated earlier than 1999 (ID 734); or only cited information generated after the year 2000 (ID 409). This type of response appears to reflect a problem frequently raised by government officials, IFAI and civil society representatives during public discussions on the right to information – the condition of government archives. The lack of specific norms regulating archives and records management makes it difficult for government officials to systematically organize their files in a manner that allows them to efficiently respond to information requests.

Other examples demonstrate that federal agencies may lack the necessary resources necessary to respond adequately to certain information requests. In one case, public officials from Mexico's national archives (Archivo General de la Nación—AGN) took digital photographs of the responsive documents instead of scanning them – possibly due to the lack of an available scanner? – which produced illegible images that were then provided to the requester (ID 374).

Finally, responses that exceeded the expectations of the user were generally those in which the user asked for data and the government responded attaching official documentation such as in example number 3 in the section below, in which an FOI user sent 61 questions to the airport authorities in Mexico City. The response was outstanding, clear and ordered. Airport officials included official documents in response to several of the questions posed by the user even though no official documents were requested (ID 666).

While it was impossible for us to determine the background of all users, in some cases we were able to determine user profiles from the information contained in their requests. While our study did not include a category for the user's background, we believe it is important to look at the individuals requesting information in order to develop a better understanding of the type of demographic using Mexico's FOI law. This can help freedom of information advocates take into account these groups when developing educational resources and tools for FOI users.

There are several cases in which people identified themselves as government officials using SISI to access information. In one somewhat stunning case (ID 728), public officials in Nuevo León sent several requests to government agencies to get contact information for their local offices in the state. In a focus group organized by the Annenberg School of Communication, federal government officials from the Unidades de Enlace raised the ironic fact that it is sometimes easier to get information using the Transparency Law, rather than directly asking their government colleagues for the information. (note 8)

In other cases, members of the private sector sought various types of information such as information on government suppliers (ID 132), government contracts (ID 422), companies that win bids for government contracts (ID 146), and government practices.

Many cases demonstrated that people were inexperienced and uncertain as to how to use the FOI law, and were using SISI for other purposes. We examined several requests from private companies, for example, offering services to government agencies, a clear abuse of the information system (ID 487). We also found several instances where people asked for a job through the SISI (ID 470).

In other cases, users identified themselves as students (ID 459) and stated that they needed information to complete a specific project. According to figures cited by IFAI Commissioner Juan Pablo Guerrero, 33.4% of users in 2004 identified themselves as members of the academic community, constituting the largest group of information requesters. (note 9)

Another set of requests worth mentioning involved regular citizens describing problems they faced in the hope that government information could help them. One request involved a tenant refusing to leave the requester's property (ID 757). In another case, the requester was a relative of a senior citizen seeking information about government-provided care for the elderly in the state of Chihuahua (ID 742). Other interesting examples include that of a citizen requesting information to help calculate his IMSS pension (ID 93), a requester asking for data regarding a medical malpractice case (ID 94), and a request from a concerned citizen asking Luz y Fuerza del Centro the reasons for a recent blackout (ID 980). These cases demonstrate that everyday citizens see Mexico's Transparency Law as a tool that can be utilized to gain closer access to their government representatives.

We did not encounter any examples where the information in the request identified the user as a representative of the news media. These findings coincide with data cited by Commissioner Guerrero Amparán, who cited that in 2004 only 9.0% of the users identified themselves as representatives of the news media. (note 10) In a focus group with Mexican reporters organized by the Annenberg School of Communication, they mentioned that they were careful to not reveal their connection to the media, believing officials would be less willing to disclose the information if they thought it could be exposed by news outlets. (note 11)

In response to a FOI request for information on "the amount spent by the federal government from 1998 to 2003, separated by state," Secretaría de Hacienda provided the citizen with seventeen pages of responsive documents. The proportioned information came from a 2003 government report on public expenditures and contained charts on the allocation of federal resources in various public sectors at the state and municipal level, from 1994-2003.

In this case the requester asked for information about travel expenses of IFAI Commissioner José Octavio López Presa during his 2004 business trips, along with justification for the expenditures. In the response, IFAI officials provided a spreadsheet with the total amount spent by the Commissioner on transportation, hotels and food. The response included proof of the expenses, including scanned copies of credit card statements and receipts for rental cars and hotels.

In this request the FOI user sent 61 questions to the Mexico City airport authorities regarding the airport's operations. Officials responded to this request with 21 pages of agency information related to the number of passengers that passed through the airport, statistics on departures and arrivals, expenditures for airline equipment and various services, names of the providers of the airport's information systems, security information, etc. This is an outstanding response to a very broad request. Although the requester did not ask for government records, airport officials included official documents among the 21 pages sent to the requester.

This response from Secretaría de Salud (SSA) was for a request seeking information regarding expenditures on medicines in 2003 and 2004. The 19-page response included information from six agency components in response to the questions forwarded to them by the Unidad de Enlace. Each entity responded by sending a memo with proper headers and signatures and in some instances even included copies of official documents. For example, officials from Juarez attached copies of documents with information on the Juarez hospital contracts from 2003 and 2004. It is important to note that the user had not asked for documents but only for data, thus this request was classified as "exceeding" the user's expectations.

In response to a citizens' request for their social security number, the liaison official informed the requester that they could locate the information on the agency's Web page. Instead of merely providing the address of the website containing the information, however, the response included images of the IMSS site with arrows directing the user exactly where to locate the information. The Unidad de Enlace went so far as to provide a phone number at the end of the request in case the user needed further assistance locating the requested information.

IMSS responded to a request for information on an agreement reached between the social security agency and its union syndicate, following a workers' call to strike. The response provided by the liaison official included copies of pages of the contract agreement signed by IMSS functionaries, as well as representatives from their syndicate – Trabajadores del Seguro Social. The documents were clearly unaltered scanned copies of original files pulled from the government archives. They included the agency heading, as well as official stamps and signatures.

The questions that initially influenced this study were answered as a result of our examination of 1,000 information requests and government responses taken from our statistically representative sample of requests filed between June 12, 2003, and April 30, 2006. These questions were: what was the level of satisfaction of those using Mexico's transparency law, and what percentage of the users were soliciting government documents?

We determined that 76% of the cases examined received satisfactory responses or responses that exceeded expectations during the first three years of the law, according to our methodology. We were able to draw upon our conclusions about the type of information that was being requested. The results of the study demonstrate that the majority of SISI users asked for information related to the structure of government agencies, salaries of its employees, the way an agency distributes its resources, and personnel information.

The results also demonstrate that agencies were more adept to responding to requests relating to relatively permanent characteristics of the agency. However, when requests became much more complex and sought information relating to the operations and results of an agency's programs, the percentage of satisfactory responses dropped, from an average 81% satisfaction rate for low and medium-complexity requests, to 57% level of satisfaction for very high-complexity requests. These results reflect the difficulty agencies have complying with the demand of more sophisticated FOI requests.

Our initial hypothesis that a majority of users of the Transparency Law sought more data than official government records was confirmed. Only 18% of the examined requests solicited documents, while 71% asked for general information. (The remaining 11% is explained by requests for personal information, which we could not review.) It is important to emphasize that even when a requester asks for data, the law clearly states in Article 41 that agencies are obliged to release documents from their archives and that "these files can only be provided in the format in which they are found, and delivered in whole or in part to the requester." Nevertheless, we observed repeated instances when officials would provide the requester with a simple Worddocument containing the information sought – without letterhead, date or signature – in violation of the law.

As a result, although we were impressed by the rates of satisfaction provided by agency responses during the period in question, the study raises two troubling issues. First, the most complex freedom of information requests filed by citizens received the lowest rate of satisfaction. The study should serve as a warning for future trends in the SISI system. As citizens become more experienced in using the Mexican freedom of information law over time, it is highly likely that requests overall will become more sophisticated and complex. The government will need to address agency capacity for responding to these more complex requests if it hopes to satisfy a new generation of requesters.

Second, while the majority of requesters in our sample did not seek the release of official records, the demand for documents will inevitably rise over time. Agencies are poorly equipped to deal with such a demand. Although Article 41 of the LFTAIPG clearly states they are required to supply government records in response to information requests, there has been no effort to enforce this obligation. Furthermore, some agencies still know little about the law, or do not understand the scope of its obligations on their information practice. As a consequence, agencies have tended to provide information in a variety of improvised forms – incorporated inside their letters to requesters, for example, or conveyed in simple and unofficial documents created ad hoc by the agency. The government should ensure that agencies provide official records in response to requests. Doing so would reinforce the legitimacy and credibility of the SISI process in the future, as requesters increasingly seek records through more complex requests for information.

Finally, for those interested in studying the quality of government responses further, we recommend that three additional variables be analyzed. First, we think it would be valuable to add the date category. Our current qualitative method cannot measure changes over time. Are requests getting measurably more complex? Were more documents (as opposed to data) requested in 2007 than in 2003? These questions are not answered in our study. Second, it would be interesting to add the agency category to look for any trends within each dependencia. And lastly, we believe that in order to evaluate government denials more accurately, a subsequent study should consider analyzing which requests were appealed to the IFAI, and what resolutions did the IFAI reach.

Diario Oficial de la Federación. (2007). "Decreto por el que se adiciona un segundo párrafo con siete fracciones al Artículo 6o. de la Constitución Política de los Estados Unidos Mexicanos."

Instituto Federal de Acceso a la Información Pública. (2007). "Estadísticas del SISI." http://www.sisi.org.mx

Instituto Federal de Acceso a la Información Pública. (2006). "Ley Federal de Acceso a la Información Pública Gubernamental." http://www.ifai.org.mx/AcercaIfai/Marco

Instituto Federal de Acceso a la Información Pública. (2006). "Reglamento de la Ley Federal de Acceso a la Información Pública Gubernamental." http://www.ifai.org.mx/AcercaIfai/Marco

Instituto Federal de Acceso a la Información Pública. (2004). Transparency, Access to Information and Personal Data. Mexico: IFAI.

1. In a 2006 report, Human Rights Watch described the passage of the LFTAIPG as the "most unambiguous achievement in the area of human rights during the Fox presidency."

2. The confidence interval (or margin of error) is conservative for two reasons: first because it is calculated as if each reported percentage were 50 percent, which is the point at which the standard error estimator is at its maximum. Many of the reported percentages are less or greater than 50 percent, and their errors are correspondingly smaller. Second, we stratified the sample proportionally by institution. If the stratification information were included in the error calculation, the error would be smaller. For these two reasons, the true error is likely less than the 3.1 percent reported here.

3. According to IFAI, IMSS has received 28,192 requests as of February 14, 2008. The agency that came in second in number of requests is Secretaría de Hacienda y Crédito Público. It is important to note the substantial difference in the number of requests between the agencies. SHCP received 13,523 requests during the same time period, less than half of those received by IMSS.

4. IFAI. Estadísticas del SISI.

5. According to IFAI, "the Federal Law of Transparency offers fast and clear channels with which to exercise the Right of Acces to Information without restrictions… for any desired purposes. Neither the Law nor the IFAI can prejudge the juridical status of the individuals or their intentions as petitioners…" Fraction three of the constitutional reform establishes that "it is not necessary for any person to justify their use of the law, which guarantees them free access to public information, and to their personal data."

6. In our study, these responses are classified as non-satisfactory, following the LFTAIPG regulations that establish that government agencies are required to provide information on "the source, place and manner in which the information can be obtained or reproduced" (Fraction 2, Article 70 of LFTAIPG).

7. IFAI. Estadísticas del SISI.

8. Information obtained during focus group sessions organized by the Annenberg School for Communication (University of Pennsylvania), along with officials from Mexico's federal information offices (Mexico City, June 2007).

9. Guerrero Amparán, Juan Pablo. (2005). "Instituto Federal de Acceso a la Información Pública". (Panel discussion at the Center for Security and International Studies in Washington DC February 2005).

10. Ibid.

11. Information obtained during focus group sessions organized by the Annenberg School for Communication (University of Pennsylvania), along with officials from Mexico's federal information offices (Mexico City, June 2007)