---
title: Instruction besoins Metallica
tags: Metallica
---
# Instruction besoins Metallica

Camme
| | | |
| -------- | -------- | -------- |

| **Couverture** |
| thématique | Conjoncture Economie | Text |
| Spatial | Text | Text |
| temporelle | Text | Text |
| **Champ** | Text | Text |
| UnitType | Text | Text |
| Universe | Text | Text |
| **Collecte** |
| Périodicité | Mensuel | Text |
| | Text | Text |
| Mode | Text | Text |
| **Questionnaire** | Text | Text |
| Text | Text | Text |
| Text | Text | Text |
| Text | Text | Text |
| **Protocole d'enquête** | Text | Text |

|Métier Métallica| Métier RMES| DDI | GSIM |
|----- | -------- | -------- | -------- |
| Text | Serie | Group | StatisticalProgram |
| Text | Operation | Group / StudyUnit | StatisticalProgramCycle |
| Text | Operation | Group / StudyUnit | StatisticalProgramCycle |
| | | DataCollection | |
| Campagne | | StudyUnit | |
| Lot | | CollectionEvent | |

|Métier Métallica| Métier RMES| DDI | GSIM |
|----- | -------- | -------- | -------- |
| Text | Serie | Group | StatisticalProgram |
| Text | Operation | Group / StudyUnit | StatisticalProgramCycle |
| Text | Operation | Group / StudyUnit | StatisticalProgramCycle |
| Campagne | | StudyUnit | StatisticalProgramCycle |
| Lot | | CollectionEvent | StatisticalProgramCycle |

# proposal

This mail to the DDI community adresses two general questions related to the data collection phase. The first one is how to describe sub steps of data collection in a survey. The second is about the possibility to trigger specific events (for instance send a mail, select unit to contact again) based on DDI metadata.We hope some advices or even discussions about the better way to handle these questions with DDI standard.

## 1a) Describing sub step of data collection in a survey

At the moment, in our metadata management repository, we describe, among others, objects that correspond to statistical program (in a GSIM sense). In our DDI repository, we use a Group to describe this object and in our own terminology this object corresponds to a « statistical series ».

These statistical program generally contains several statistical program cycles, where each statistical program cycle is an iteration of the statistical program for a given year for instance. A statistical program cycle produce and disseminate statistical products. In our terminology we call these a « statistical operation ». In DDI we use the StudyUnit object for describing this level.
In our metadata repository, a StudyUnit contains datacollection that has a reference to an instrument.

These statistical program cycles are defined following an annual rythm. This rythm is convenient to send to Eurostat the annual quality report that furnishes reference metadata about the statistical program cycle. To store reference metadata and create Eurostat report, this annual organisation is sufficient. But, now we would like to use DDI metadata, in a active manner, to drive finer steps of the process this modelisation need to be refined.

In particular in our terminology for each statistical operation we need to take into account at least one object called « campaign » . More precisely a campaign is a partition of statistical operation with the objective to produce results according to a calendar defined beforehand by the project owner. An operation can have one or several campaigns.

In each campaign we have on object called a « batch ». A batch is a partition of the data collection (partition of the sample or of the questionnaire * mode). It constitutes a division of a campaign to meet operational objectives. A collection unit belongs to one and only one batch at a given time. It does not give rise to statistical dissemination.

One important point is that we make the difference between the sub step of data collection depending on they create or not statistical products. An hindsight that would to be confirm at this stage, is to use a StudyUnit for each campaign since a campaign produce a statistical output. And the batch could be described with the ddi object CollectionEvent.

We will present two use cases, with the survey camme and the household survey to illustrate our questions and the solution we see to take into account the sub tasks of data collection and maybe how the introduction of campaign and batch could change the ddi object associated to the statistical operation.

##### Figure 1 : Tableau des définitions ou correspondances entre objets GSIM, DDI, RMéS

* statistical program
* statistical program cycle / statistical operation
* statistical program Cycle / campaign
* statistical operation
* datacollection
* campaign
* batch

*comment : Because of the new levels (campaign, and batch) there is a question concerning where is the statistical program cycle, statistical operation or the campaign ?*

##### Figure 2 : présentation de notre modélisation DDI actuelle

• group : statistical program (we call this serie. A serie contains one or several operations)
• StudyUnit: statistical program cycle (we call this operation. )
• datacollection
• contains a collection instrument (a questionnaire)

## 1b) Deux cas d’utilisation
### Camme
* enquête de conjoncture auprès des ménages par panel rotatif
* un échantillon global
* 12 échantillons
* 3 échantillons par mois
* une enquête mensuelle avec donnant lieu à une diffusion mensuelle de statistiques
* un questionnaire avec deux séquences : une séquence conjoncture et une séquence sociodémo
* Les répondants sont interrogés trois fois théoriquement
* Les répondants en S1 sont passés en S2 / répondant S1 termine le questionnaire
* Les non répondant S1 sortent
* Tous ceux qui ont répondu en S1 vont en S3 / sauf si refus explicite en S2
* Après S3 tout le monde sort
* Camme transporte ponctuellement (un mois donné) une enquête externe avec à un questionnaire disjoint.

### Logement
* One sample
* A questionnaire composed of three sequences in three different periodes of collecte
* 4 phases of collect with a sequential mixmode : respondants are interviewed
* 24 janvier 2022 – 13 février 2022 : sequence 1 by internet
* 28 février 2022 – 20 mars 2022 : sequence 1 par telephone et sequence 2 by internet
* 4 avril 2022 – 24 avril 2022 : sequence 2 by telephone et sequence 3 by internet
* 9 mai 2022 – 23 mai 2022 : sequence 3 by telephone
* One statistical dissemination when all the sequences and the data process are completed.

## 1c) modélisations envisagées

### Enquête Camme

#### solution 1
Group : Statistical program
Group : statistical operation (with the annual rythm)
the last group references 12 StudyUnit : one StudyUnit for each monthly survey (statistical program cycle that produces statistical output). These monthly survey could be seen as waves of the statistical operation. Each studyUnit correspond to a campaign in our terminology.
Each StudyUnit references one DataCollection (that contains capabilities to realise the collect).
Datacollection references one CollectionEvent. One collectionEvent can make reference to several instruments for each mode of collection. CollectionEvent corresponds to the batch in our terminology.
A CollectionEvent has date information (StartDate, EndDate, ie).
A CollectionEvent references three samples

#### Solution 2
Group : Statistical program
StudyUnit: statistical operation (with the annual rythm)
StudyUnit references 12 DataCollection: one DataCollection for each monthly survey (statistical program cycle that produces statistical output) that could be seen as waves of the statistical operation. This would correspond to the campaign.
DataCollection references one CollectionEvent. One collectionEvent can make reference to several instruments for mode of collection . CollectionEvent corresponds to the batch in our terminology.
A CollectionEvent has date information (StartDate, EndDate, ie).
A CollectionEvent references three samples

### Enquête Logement

#### Solution a

Group : Statistical program
Group: statistical operation
Group has one StudyUnit : that corresponds to statistical program cycle. This corresponds to a campaign.
DataCollection: that contains capabilities for the collection.
DataCollection references 4 CollectionEvents, one collection can make reference to several instruments with several mode of collection. CollectionEvent corresponds to the batch in our terminology.
the CollectionEvent has date information (StartDate, EndDate, ie).
A CollectionEvent may reference several samples

Comment : In this first solution a, the second Group is an object that contains general information about the current survey and is in a 1-1 relation to a campaign when there is only one campaign in the statistical operation. Otherwise, in a case of a longitudinal study this solution a allows to group several waves under the same object. It could provide a stable solution for all types of surveys, whether they have one or more waves and one or more batches per wave.

#### Solution b

Group : Statistical program
StudyUnit: statistical operation
StudyUnit has one datacollection that corresponds to the campaign.
Datacollection references 4 CollectionEvents, one CollectionEvent can make reference to several instruments with several mode of collection. CollectionEvent corresponds to the batch in our terminology.
the CollectionEvent has date information (StartDate, EndDate, ie).
A CollectionEvent may reference several samples

## 1d) Questions
dans DDI doc, lorsqu’une période de collecte donne lieu a une restitution stat on a recours un studyunit. Ainsi, dans une enquête longitudinale les vagues sont des studyUnits.
De ce fait, est ce que DataCollection, peut être utilisé comme une étape de collecte intermédiaire pour décrire une collecte qui donne lieu à une restitution ?

CollectionEvent : est ce que l’on peut les utiliser pour des évènements autour de la collecte proprement dite, par exemple les envois de courrier pour informer les futures personnes enquêtées qu’une enquête aura lieu ?

Par rapport aux solutions 1 à d quelles seraient vos préconisations ?

comment créer un calendrier, le mode et son calendrier
plusieurs calendrier par lot ??

---