benam@csh.ac.at
Sign Up Login

Discussion Room:

This is the room for our more detailed discussions.

Let's jump into the confidence/dispute issue. I have used the examples that Jim sent me and modified them a bit to include some more potential (imaginary) cases.

Case 1: Numeric Values

Example 1: (Two Numeric values are disputed)

What we had so far:
Polity Year(s) Coded Value Disputed?
IqAbbs1 900CE 300000 disputed
IqAbbs1 900CE 900000 disputed
What we can have now:
Polity Year(s) Coded Value Confidence1 Disputed?2
IqAbbs1 900CE 300000 Inferred True
IqAbbs1 900CE 900000 Evidenced True
1. The potential values for Confidence are: (Inferred/Evidenced/Suspected).
2. The potential values for Disputed are: (True/False).

Example 2: (More than two numeric values are disputed)

What we had so far:
Polity Year(s) Coded Value Disputed?
IrPart1 1CE 7500000 disputed
IrPart1 1CE 15000000 disputed
IrPart1 1CE 25000000 disputed
What we can have now:
Polity Year(s) Coded Value Confidence1 Disputed?2
IrPart1 1CE 7500000 Evidenced True
IrPart1 1CE 15000000 Evidenced True
IrPart1 1CE 25000000 Suspected True
1. The potential values for Confidence are: (Inferred/Evidenced/Suspected).
2. The potential values for Disputed are: (True/False).

Example 3.a: (Two or more dates* are disputed)

* Remember that Dates are simple numeric values (integers) in the new databse. Date ranges will be a set of numeric values. We actually use years instead of dates. So we can have values such as yaer_from or year_to and peak_year_from or peak_year_to, which can represent a year range together.
What we had so far:
Polity Coded Value (Peak Date) Disputed?
IrSeleu 281 BCE disputed
IrSeleu 190 BCE disputed
What we can have now:
Polity Coded Value (peak_year) Confidence1 Disputed?2
IrSeleu -281 Inferred True
IrSeleu -190 Evidenced True
1. The potential values for Confidence are: (Inferred/Evidenced/Suspected).
2. The potential values for Disputed are: (True/False).

Example 3.b: (Two or more date ranges* are disputed)

* Remember that Dates are simple numeric values (integers) in the new databse. Date ranges will be a set of numeric values. We actually use years instead of dates. So we can have values such as yaer_from or year_to and peak_year_from or peak_year_to, which can represent a year range together.
What we had so far:
Polity Coded Value (Peak Date) Disputed?
IrSeleuB 281-271 BCE disputed
IrSeleuB 190-180 BCE disputed
What we can have now:
Polity Coded Value (peak_year_from) Coded Value (peak_year_to) Confidence1 Disputed?2
IrSeleuB -281 -271 Inferred True
IrSeleuB -190 -180 Evidenced True
1. The potential values for Confidence are: (Inferred/Evidenced/Suspected).
2. The potential values for Disputed are: (True/False).
Note 1:

- We can report facts as Disputed as soon as there is the slightest mismatch between two coded sets of values.

Note 2:

- In the column that is supposed to be a numeric value, we cannot store anything that is not a number. In particular, the value 'unknown' cannot be put into the database. For such cases, we simply use the concept of a null value, or no value at all (This can be done with None in Python, or null in SQL, etc.)
- When inserting the data manually through the front-end interface, we just need to leave the column for that coded value empty. That's it.
- How we then save it in .csv files for further use, depends totally on what we would like to do with the data. For example, if we are using R, we can simply save those values as NULL (as accepted by R) or as a string "unknown value" or "None", or simply anything we prefer.
- Therefore, we can have such rows in the SQL database:


Polity year_from year_to Coded Value (Population) Confidence Disputed?
LbAcPho1 -1200 -332 (null)3 Evidenced False
LbAcPhoB 2 -700 -652 (null) Suspected False
1. This row means we have evidence that the population is unknown.
2. This row means that we have a suspicion that the population might be unknown. The confidence can later be changed to Evidenced if a Seshat Expert decides so.
3. As mentioned above, (null) here means: no value was entered for population. This is equivalent to the cases when we used to call 'unknwon'. This is different from 'uncoded', as for 'uncoded' values, we don't have a corresponding confidence and disputed tag and they simply don't exist in the database.

Case 2: Multiple-choice options

Example 4.a: (Two or more multiple-choice options are disputed)

What we had so far:
Polity Years Coded Value (HS in-group) Disputed?
LbAcPho - * present disputed
LbAcPho - unknown disputed
* Note that whenever Years are not given in the old database, the polity year_range is assumed.
What we can have now:
Polity year_from year_to Coded Value (hs_in_group)1 Confidence2 Disputed?3
LbAcPho -1200 -332 present Evidenced True
LbAcPho -1200 -332 unknown Evidenced True
1. The potential coded values for Human Sacrifice are: (Absent/Present/transitional period (a->p)/ transitional period (p->a)). Note that "Scholarly Disagreement" will NOT be a potenntial value as it is represented in the 'Disputed?' column.
2. The potential values for Confidence are: (Inferred/Evidenced/Suspected).
3. The potential values for Disputed are: (True/False).
QUESTION 1: Is the confidence level of Suspected only used for cases when coded value is unknown? In other words, are suspected present or suspected transitional period (a->p) also valid combinations of coded value and confidence or not?

Example 4.b: (Two or more multiple-choice options are disputed, with different confidence levels)

What we had so far:
Polity Years Coded Value (HS relative) Disputed?
LbAcPho - * inferred present disputed
LbAcPho - unknown disputed
* Note that whenever Years are not given in the old database, the polity year_range is assumed.
What we can have now:
Polity year_from year_to Coded Value (hs_relative)1 Confidence2 Disputed?3
LbAcPho -1200 -332 present Inferred True
LbAcPho -1200 -332 unknown Evidenced True
1. The potential coded values for Human Sacrifice are: (Absent/Present/transitional period (a->p)/ transitional period (p->a)). Note that "Scholarly Disagreement" will NOT be a potenntial value as it is represented in the 'Disputed?' column.
2. The potential values for Confidence are: (Inferred/Evidenced/Suspected).
3. The potential values for Disputed are: (True/False).
QUESTION 2: I assume we won't have a dispute if the cases are only different in confidence level (and the coded values in the new database are the same). Right? Let's say if in the case above, the value coded as unknown was present. Even if the tag was suspected instead of inferred. In other words, are any combinations of present and inferred present and suspected present in dispute?

Example 4.c: (More multiple-choice options are disputed: another imaginary example)

What we had so far:
Polity Years Coded Value (HS in-group) Disputed?
LbAcPho - * present disputed
LbAcPho - unknown disputed
LbAcPho - inferred absent disputed
LbAcPho - suspected transitional period (a->p) disputed
* Note that whenever Years are not given in the old database, the polity year_range is assumed.
What we can have now:
Polity year_from year_to Coded Value (hs_in_group)1 Confidence2 Disputed?3
LbAcPho -1200 -332 present Evidenced True
LbAcPho -1200 -332 unknown Evidenced True
LbAcPho -1200 -332 absent inferred True
LbAcPho -1200 -332 transitional period (a->p) Suspected True
1. The potential coded values for Human Sacrifice are: (Absent/Present/transitional period (a->p)/ transitional period (p->a)). Note that "Scholarly Disagreement" will NOT be a potenntial value as it is represented in the 'Disputed?' column.
2. The potential values for Confidence are: (Inferred/Evidenced/Suspected).
3. The potential values for Disputed are: (True/False).

Changed On Seshat-db.com to Disputed?

512: EgNaqa2 (Naqada II): Scholarly Disagreement,
39: KhChenl (Chenla),
48: IdMedng (Medang Kingdom)