|
Household Budget Continuous Survey. 1999 |
Analysis of the lack of response:
Introduction
Analysis of the data
Annex of Tables (In Spanish. XLS File 188 kb.)
Quality indicators (In Spanish. PDF 993 Kb.)
Among the errors that affect any survey the errors outside the sample are found,which occur in the different phases of the statistical process, appearing before the collection of information (framework deficiencies, insufficiencies in the definitions or in the questionnaires), during its collection (defects in the interviewers´ work, incorrect declarations or lack of response on the part of the informants) and, finally, in the operations subsequent to field work (errors in the coding, recording, etc.).
The evaluation of these errors presents many difficulties, among other reasons because of the large variety of things which can cause them. Among these causes the lack of response of the informant units is worthy of note. This can be due to a refusal to answer the questionnaire, to an absence of the informant units or the housing being inaccessible at the moment of the interview. In order to analyse the lack of response of the survey an evaluation questionnaire has been designed with which information will be obtained on the basic characteristics of the units that have not collaborated in the survey as well as quantifying another type of incidence, as is the case of the non-surveyable units.The incidences that can come up during the field work are the following:
All housing in agreement with the situation in which it is found, is classified in: Empty: housing which remains unoccupied all, or for the majority, of the year, due to being vacant, in ruins or seasonal housing. Non-locatable: the address that is on the Work Order cannot be located in the territory. For other uses : building destined entirely for different purposes from those of a family residence (for example commercial premises, storage, private garage). Within surveyable housing the households may be: Surveyed: when the household collaborates in the survey. Refusal: when the household refuses to collaborate. Absence: when the interviewer does not find any member of the household in the consecutive visits made to the household. If the housing is surveyable it is investigated. Non-surveyable housing is replaced and inaccessible housing receives the same treatment as absent houses. Whether they are replaced or not depends on the value of the effective collaboration key (ECK). If the ECK is equal to 1, in other words, if they have not collaborated previously, they have simply been captured, replacement is made if after repeated visits contact is not made with persons. If the ECK is greater than 1, in other words, if there has already been previous collaboration, if after repeated visits contact is not made with persons replacement is not made, returning to visit the following quarter. Replacement is made if after repeated visits contact is not made with persons. The evaluation questionnaire is structured in four sections. In the first data identifying the housing is recorded. In the second the type of incidence which has taken place when visiting the house is recorded and the visit when this occurred is specified; it is also indicated if the housing has been substituted or not. In the affirmative case, the order number of the replaced house is recorded. In the third section the number of members of the household and the following basic characteristics of their main breadwinner are collected: age, civil status, highest level of studies completed and situation in relation to activity. It is also indicated if the information has been obtained from a household member directly or whether it has been obtained indirectly by means of other sources. Finally, in the fourth section the interviewers subjective appreciation of the housings type and area of location is taken up. The questionnaire is only filled for sample households in which a positive interview has not been attained, but not for the reserve households that have been visited without achieving their collaboration (intermediate reserves).
A series of tables has been obtained from the use of the information contained in the evaluation questionnaires which shall be commented on below.
In table 1 distribution of the theoretical sample by Autonomous Communities is presented, which remains fixed over time, and the effective sample for the years four quarters. The effective sample is expressed in number of households, since in each house selected the household or households resident in the same (one or more households can coexist in the same house) are investigated. It can be observed that on a national level the effective sample represents, on average, 85 per cent of the theoretical sample, which means that there are 15 per cent of households that do not collaborate in the sample for various reasons and which are not substituted. It is also noted that while Madrid and the País Vasco are the autonomous communities with the lowest percentage of effective sample, Galicia and Castilla y León are the ones with the highest said percentage.
With respect to the municipalities distribution by size (table 2), it is observed that the smallest percentage of effective sample is obtained in the municipalities of Madrid and Barcelona, this being the general behaviour in all surveys.
Table 3 shows the distribution of the incidences according to type. It can be seen that the most numerous are the refusals with a percentage close to 50%, with absences as the second in importance since they account for 40% of the total. It can also be seen that while the number of refusals remains fairly stable throughout the year, the number of absences increases considerably in the third quarter with respect to the other three. This is due to the third quarter of the year coinciding with the population's most common vacation period.
If we look at the lack of response (table 3bis) we can observe that on average the percentage of refusals reaches a value of 55%, ten points higher than for absences.
If these figures are compared with those obtained in 1998 it can be demonstrated that there has been a decrease in refusals of almost nine percentage points, the same by which absences has increased.
The greatest percentage of lack of response according to number of members (table 4) corresponds to 2 person households, with the percentages of refusals and absences practically equal.
Table 5 studies the distribution of the refusals and absences according to the age and sex of the households main breadwinner. The highest percentage of refusals, on average, corresponds to the main breadwinner male over 65 years old, with the same occurring for absences. It can also be seen that when the main breadwinner is a female the highest percentage both for refusals and absences corresponds equally to the over 65 year olds group.
As far as marital status and the main breadwinner's sex (table 6), the highest percentage of refusals clearly corresponds to male and married main breadwinner, as also occurs for absences. This result is logical if we take into account, first, that the majority of the main breadwinners are males and second, that according to the general distribution of the population around 60% of males are married.
When the main breadwinner is female the highest percentage of refusals and absences corresponds to widows, an equally relatively logical result, since, according to the population distribution among women 70 years old and over, the highest percentage corresponds to widows.
Table 7 studies the distribution of refusals and absences according to the main breadwinner activity situation. It can be observed that the highest percentage of refusals corresponds to the employed main breadwinner, reaching an average value 50%, with the second place for retired persons or pensioners with 35%. This behaviour is the same in the case of absences.
With respect to the level of studies of the main breadwinner, it is observed that the greatest percentage of refusals corresponds to the case where the main breadwinner has Primary education as the highest level of studies completed . The result is the same for absences.
In table 9 the distribution of incidences is studied according to the residence area and type of housing, which are obtained through the interviewers valuation. It can be seen in the same that both for the refusals and for the absences and non-surveyables, the greatest percentages correspond to the case of an average house situated in an average urban area.
In the rest of the tables the percentage distributions are compared, according to various characteristics of the main breadwinner, of the households with lack of response, of the replacement households for these and of the households in the total effective sample each quarter. The main difference observed between the number of sample households with lack of response and the number of replaced households, in the case of level of studies, is due to the fact that households in collection are included among the former but not among the latter, since in collection the main breadwinners level of studies is not asked.
The discrepancies that are observed between the three distributions are small, with the most outstanding ones as far as level of studies (table 10) those existing between selected households with a lack of response and the replacement households with respect to the Pre-First Grade and to the First Grade Education.
As far as the number of household members (table 11), the discrepancies between the three distributions are also small. Only those existing in 1 person households are noteworthy, especially between selected households with lack of response and effective sample. The difference is close to 7 percentage points in 4 member households as an annual average. The same occurs between selected households and effective sample, where the difference is somewhat more than five percentage points as an annual average, although of an opposite sign.
Finally, table 12 shows the distribution of refusals according to the households effective collaboration code. The highest percentage of refusals occurs in collaboration code 1, in other words, in the first interview, reaching up to 35%. The second highest number of refusals occurs in the selection of households (collaboration code zero), at about 25%. It can be concluded from these results that the families demonstrate a certain doubt to collaborate already in the selection phase, which increases in the first interview presumably due to the increase in requested information. Generally after the first interview the families are ever more willing to continue collaborating.
Significant differences between the four quarters cannot be observed in the analysis carried out. With the exception of the already mentioned third quarter due to the summer holidays, the results generally coincide with those obtained in previous years.
Nothing relevant which affects the quality of the data published is concluded from the analysis carried out.
