Transcript ISAE - OECD

What Do Citizens Know about Statistics:
Results of an OECD- ISAE Survey on Italian
Consumers
F. Fullone (*) M.Gamba (**) E. Giovannini (**) M. Malgarini (*)
__________________________
(*) ISTITUTO DI STUDI E ANALISI ECONOMICA
Piazza dell’ Indipendenza, 4 00185 – ROMA (Italy)
(**) ORGANISATION FOR ECONOMIC CO-OPERATION AND DEVELOPMENT
2 Rue André-Pascal 75016- Paris (France)
_____________________________
Measuring the Progress of Societies
World Forum on Statistics, Knowledge and Politics
Istanbul, June 27th 2007
World Forum on Statistics, Knowledge and Politics - Istanbul, June 27th 2007
An introduction to the ISAE Survey on
Italian Consumers


Every month a 2-step representative sample of 2,000 consumers is selected

It also provides some insights on the consumers, among which:
The survey includes qualitative questions on the personal situation of
consumers and general questions on the country
 Number of persons in the household
 Type of household, including information on mortgages, rent etc.
 Area of residence and number of people living in the municipality
 Age, gender, professional qualification and education of the respondent
 Income quartile
World Forum on Statistics, Knowledge and Politics - Istanbul, June 27th 2007
The OECD/ ISAE Survey on Statistical
Knowledge

In March and April 2007, a number of questions have been added
concerning the statistical knowledge of Italian citizens:

MARCH 07: GDP growth, Inflation rate, Unemployment rate, Deficit/
GDP ratio and the Euro/ Dollar exchange rate

APRIL 07: (on top of March’s questions) size of Italian population,
Italian life expectancy at birth, R&D expenditure, literacy of Italian
students and CO2 emissions

MARCH & APRIL: two questions on the importance and desire of
being informed and on the media used to gather relevant
information are added.
World Forum on Statistics, Knowledge and Politics - Istanbul, June 27th 2007
Quantitative questions


LOW Response rates to quantitative questions


UNCERTANTY among respondents was rather HIGH - as indicated by Std. Dev
On average, Italian citizens tend to be OVEROPTIMISTIC about GDP growth and
PESSIMISTIC for the remaining variables
The MEDIAN is generally closer to true values than the mean, indicating that
answers distributions are RIGHT- SKEWED
GDP
March
Inflation
April
Average
March
April
Average
Unempl
Deficit/
GDP
March
March
Response rate (%)
27
19.2
23.1
32.6
23.6
28.1
32.1
12.7
Don't Know (%)
70.5
72.8
71.7
65.1
68.8
66.9
65.5
84.5
Refuse to answer (%)
2.5
8
5.2
2.3
7.6
5
2.5
2.8
Average
2.7
2.1
2.5
4.5
2.9
3.8
14.5
8.5
Median
2
1.9
2
2.4
2.1
2.2
10
3.4
P25
1.5
1.3
1.5
2
1.9
2
7
2
P75
2.4
2.2
2.2
3
3
3
16
4.4
Std. Dev.
3.7
1.8
3.1
8.7
2.9
6.9
13
14.5
True value
1.9*
1.9*
1.8**
1.7***
6.8****
4.4*
(*) 2006, (**) February, (***) March, (****) III Quarter 2006, seasonally adjusted.
Qualitative questions

Higher response rates for qualitative
questions ( between 60 and 87%)
HOWEVER…

Correct answers are uncommon, especially
concerning population, life expectancy and
CO2 emissions
World Forum on Statistics, Knowledge and Politics - Istanbul, June 27th 2007
Results on qualitative questions
Response rate
Don't Know
Refuse to
answer
Weighted Survey answers (true values in RED)
Increased
Euro/ Dollar (%)
R&D Expenditure (%)
73.1
71.4
25.3
26.6
1.7
2
Stable
19.8
Decreased
13.2
Higher
5.1
Equal
17.7
Lower
Population (%)
Life expectancy (%)
63.4
77.4
34.5
20.4
2.1
2.2
89.7
8.5
1.9
39.5
Between 58ml and 59ml
4.5
More than 59ml
19.4
Less than 76 years
23.5
Between 76 and 80 years
22.3
Increased a lot
61.5
19.5
2
16.4
Stable
8.7
Decreased
0.4
Decreased a lot
2.7
More qualified
78.5
31.6
More than 80 years
Much more qualified
Literacy (%)
48.6
Between 50ml and 58ml
Increased
CO2 emissions (%)
40.1
5
6.7
Equally qualified
26.9
Less qualified
31.8
Much less qualified
8.1
The Importance of statistical information:
results
How important is to be informed
on the issues we have discussed
so far?
Extremely important
Average
(%)
•
8.2
Very important
24.4
Important
46.1
Not very important
12.4
Not important
3.2
Don't know
•
More than 15% believe that
information is not very important or
not important at all
•
More than 40% of the respondents
do not want to be more informed
about these issues
4
Refuse to answer
1.7
Would you like to be more informed on these
issues?
Yes
54.8
No
40.5
Don’t Know/ Refuse to Answer
4.7
Italian citizens believing that
information is “extremely” or
“very” important are only a 33%
of the population
World Forum on Statistics, Knowledge and Politics - Istanbul, June 27th 2007
Results on Channels of Information
•
TELEVISION is the most often cited
channel used to acquire statistical
information
•
NEWSPAPERS & INTERNET are cited
by respectively 52 and 23% of the
respondents
•
Opinions shared among friends and
relatives
(cited
by
11,2%
of
respondents) and political and civic
leaders (7%) lag behind the radio
(almost 19%) as other relevant
sources of statistical information
Information channels %
Television
82
Radio
18.6
Newspapers, periodicals
51.9
Internet
23.3
Political, opinion leaders
Friends, relative
Refuse to answer
7
11.2
1.8
World Forum on Statistics, Knowledge and Politics - Istanbul, June 27th 2007
Building a Knowledge Score
• What are the determinants of statistical knowledge?
Method employed:
 Construct
a “Knowledge Score” aggregating the answers to the 5
five questions included in March’s questionnaire
 Assign
a score to each respondent based on the accuracy of
individual answers to each specific knowledge question
 Compute
the ABSOLUTE VALUE of the individuals errors (i.e. the
difference between individual answer and official data)
 SORT
Individual errors and calculate a “RAW SCORE” for each
question
 Compute a PERCENTILE RANK SCORE (PRS)
 AGGREGATE
by averaging each PRS and get the “Knowledge
Score”
World Forum on Statistics, Knowledge and Politics - Istanbul, June 27th 2007
A Model for Knowledge - CONCEPT
•
Model estimation using the Knowledge score as dependent variable
allowing the residuals to be heteroskedastic and use robust
methods in the OLS estimates :

Ki = f (SDi, Di, Si) + ui
 Ki = knowledge score
 SDi = socio-demographic characteristics of the respondent
 Di = desire of being informed
 Si = source used for being informed
 ui = unobserved error term, assumed to be uncorrelated with
the regressors
World Forum on Statistics, Knowledge and Politics - Istanbul, June 27th 2007
A Model for Knowledge - VARIABLES
•
Control variables






•
GENDER (M/F)
AGE ( <30 years, 30- 49, 50-64, and 65+)
ZONE OF RESIDENCE (North West, North East, Center and
South of Italy)
EMPLOYMENT STATUS (self-employed, employed,
unemployed and out of LF)
LEVEL OF EDUCATION (Primary school, Secondary school and
University level and higher), and
INCOME (divided into quartiles).
Desire to be informed and the channels used to acquire
information are measured on the basis of the answers provided to
the survey questions
World Forum on Statistics, Knowledge and Politics - Istanbul, June 27th 2007
OUTCOMES- Socio-demographic factors
VARIABLES
BETA
STD DEV
T-STAT
•
Professional category (baseline: Employed)
4.1
1.7
2.3
1
1.1
0.9
1.4
0.4
3.5
30-50 years
4.01
2.1
1.91
*
50-65 years
9.49
2.15
4.41
***
> 65 years
1.88
2.35
0.8
-12.08
1.12
-10.75
Self employed
Unemployed
Out of labour force
**
 For men with respect
to women
***
Age (baseline: up to 30 years)
 For people between
30 and 65 years of age
 For better educated
people
Gender (baseline: Male)
Female
***
 For self-employed
and inactive people
(including students) with
respect to dependent
workers
Zone of residence (baseline: North-West)
0.01
1.62
0.01
-0.94
1.6
-0.59
-5.7
1.46
-3.91
***
High School or less
5.05
1.27
3.99
***
University degree
8.72
1.89
4.62
***
-0.25
1.56
-0.16
0.7
1.57
0.44
2.03
1.51
1.34
North-East
Center
South
Knowledge is HIGHER:
Education (baseline: primary school)
Income (baseline: 1st quartile)
Second Quartile
Third Quartile
Fourth Quartile
Statistically Significance : * = 10%
** = 5%
*** = 1%
•
It is LOWER for people
leaving in the South of Italy
•
It is NOT INFLUENCED by
income
OUTCOMES- Importance and desire to be
informed
VARIABLES
BETA
STD DEV
T-STAT
•
Knowledge is NOT
STATISTICALLY
SIGNIFICANT for
people thinking it is
VERY important to be
informed
•
Knowledge is NOT
SIGNIFICANTLY
influenced
by
the
desired to be more
informed about these
subjects
Importance of being informed
(baseline: Extremely important)
Very Important
Important
Not very important
Absolutely not
important
-1.8
1.99
-0.9
-7.42
1.86
-3.98
***
-11.45
2.47
-4.64
***
-19.29
4.15
-4.65
Desire to be more informed
(baseline: No Desire)
Yes
0.71
***= Statistically Significant at 1%
1.26
0.56
***
World Forum on Statistics, Knowledge and Politics - Istanbul, June 27th 2007
OUTCOMES- Information Channels
•
VARIABLES
BETA
Std Dev
Television
0.43
1.58
0.27
Radio
1.39
1.37
1.01
Newspapers
5.15
1.15
4.46
Internet
2.95
1.47
2
Political, civic leaders
-2.08
1.73
-1.2
Friends, relatives
-0.19
1.74
-0.11
**= 5%
***= 1%
T-STAT
Knowledge
SIGNIFICANTLY
for:
HIGHER
 Italians reading
NEWSPAPERS to
achieve economic
information
***
 Italians using INTERNET
**
•
NO significant EFFECTS of
watching television,
listening to the radio and
discussing economic issues
among friends and relatives
or political and civic leaders
World Forum on Statistics, Knowledge and Politics - Istanbul, June 27th 2007
Questions?
…..Thank you!
[email protected]
[email protected]
[email protected]
[email protected]
World Forum on Statistics, Knowledge and Politics - Istanbul, June 27th 2007