Does Oil Hinder Democracy?
World Politics, 2001 · paper · Political science
Uses data from 113 states from 1971 to 1997 to test the claim that oil wealth impedes democracy. Examines whether the claim is valid, whether it holds beyond the Middle East and for other minerals, and what might explain it.
Checked: PDF from pages.ucsd.edu, fetched 5 Oct 2026. Page numbers are those of that PDF.
8 findings
Substance Text on Government Consumption says "The data cover 104 states and are compiled by the World Bank", but Table 5 column 2 (the Government Consumption regression) reports States = 110; 104 is the Taxes count (column 1).
- p. 23 Rentier Effect, second paragraph
The data cover 104 states and are compiled by the World
- p. 24 Table 5, column 2, States row
110
- p. 23 spending effect paragraph
The data cover 104 states
- p. 23
The data cover 104 states and are compiled by the World Bank
Judge's note: Viewed p.23 and p.24. Table 5 States row reads 104 / 110 / 111. The Taxes paragraph just above says "covers 104 of the 113 states", matching column 1. A regression cannot use more states than the variable covers. Appendix 1 (p.36) gives no state count for Government Consumption, and I found no other explanation. The likely fix is 110 in the text.
Second check: another model agreed
Report a problem with this finding
Substance Ethnic Tensions: body text says "Scores are available for 102 states between 1982 and 1997" and Table 6 column 3 has 102 states, but Appendix 1 says "The data cover 97 states between 1982 and 1997".
- p. 26 Repression Effect, last paragraph
available for 102 states between 1982 and 1997
- p. 27 Table 6, column 3, States row
102
- p. 37 Appendix 1, Ethnic Tensions definition
The data cover 97 states between 1982 and 1997
- p. 26 Repression effect text
Scores are available for 102 states between 1982 and 1997
Judge's note: Viewed p.26 (102 states), p.27 (Table 6 column 3, States = 102) and p.37 (Appendix 1, 97 states). Same variable, same source, same period. The regression uses 102 states, so the appendix figure of 97 is the inconsistent one. No explanation found in the text.
Second check: another model agreed
Report a problem with this finding
Substance Life expectancy: "The underlying data are compiled by several UN agencies and cover ninety states." but Table 9 column 3 (Life Expectancy regression) reports States = 103.
- p. 28 Modernization Effect, paragraph on health
cover ninety
- p. 31 Table 9, column 3, States row
103
- p. 28 Modernization Effect, paragraph on health
underlying data are compiled by several UN agencies and cover ninety
- p. 28 Modernization effect, health
cover ninety states
Only the authors can say what is right here.
Judge's note: Viewed p.28 ("cover ninety states") and p.31 (Table 9 column 3: Observations 777, States 103). The estimation sample cannot exceed the variable's coverage. Appendix 1 (p.36 text) gives no state count for Life Expectancy. Which of the two figures is right cannot be told from the paper.
Second check: another model agreed
Report a problem with this finding
Substance Text: "All of the variables measuring occupational specialization are highly significant and positively associated with democracy", but Table 7 column 4 prints Women in Services as "–.0185***" (negative).
- p. 29 Table 7, column 4, Women in Services
–.0185***
- p. 30 body text, top of page (sentence begins on p. 29)
significant and positively associated with democracy
- p. 29 Results text, last lines of page (continues on p. 30)
All of the variables measuring occupational specialization are highly
Only the authors can say what is right here.
Judge's note: Viewed p.29 and p.30. Table 7 shows Men in Industry .0733***, Women in Industry .0814***, Men in Services .0685***, and Women in Services –.0185*** (.00512), with the minus sign clearly printed. The sentence spanning pp.29-30 says all are positive and names only telephones as negatively correlated. Either the table sign or the sentence is wrong; only the authors can say which.
Second check: another model agreed
Report a problem with this finding
Reference or label Table 2 is out of descending order: "8. Bolivia 5.53" is ranked above "9. Congo, Dem. Rep. (1983) 7.00".
- p. 3 Table 2, rows 8-9
8. Bolivia 5.53
- p. 3 Table 2, rows 8-9
9. Congo, Dem. Rep. (1983) 7.00
- p. 3 Table 2 rows 8–9
9. Congo, Dem. Rep. (1983)
- p. 3 Table 2, rows 7-9
8. Bolivia 5.53 9. Congo, Dem. Rep. (1983) 7.00
Only the authors can say what is right here.
Judge's note: Table 2 read on the page image. Every other row in Tables 1 and 2 is non-increasing; row 9 (7.00) exceeds row 8 (5.53) and falls below row 7 (Togo, 7.79). The table note does not explain a different ordering rule. Either the rank order or one of the values is wrong, and only the author can say which.
Second check: another model agreed
Report a problem with this finding
Wording Footnote 5 defines the Middle East used "throughout this article" with 19 states including "Libya, Malta, Morocco, Oman", but the Appendix 1 Mideast dummy lists 18 states, "Lebanon, Libya, Morocco, Oman, Qatar ...", without Malta.
- p. 5 footnote 5
Libya, Malta, Morocco, Oman
- p. 35 Appendix 1, Mideast definition
Lebanon, Libya, Morocco, Oman, Qatar
- p. 5 Footnote 5
Libya, Malta, Morocco, Oman, Qatar, Saudi Arabia, Syria,Tunisia, United Arab Emirates, and Yemen
- p. 35 Appendix 1, Mideast
Lebanon, Libya, Morocco, Oman, Qatar, Saudi Arabia, Syria, Tunisia
Only the authors can say what is right here.
Judge's note: Counted both lists on the page images: footnote 5 has 19 states, Appendix 1 has 18, and the only difference is Malta. PDF page 21 says the Mideast dummy follows the World Bank classification. No statement in the text explains dropping Malta.
Second check: another model agreed
Report a problem with this finding
Wording The regression equation prints "b7(Year1) . . . + b33(Year26)"; with b7 on Year1, the 26th year dummy should carry b32, not b33.
- p. 13 basic regression model equation
b7(Year1) . . . + b33(Year26)
- p. 16 paragraph on year dummies
a set of twenty-six dummy variables
- p. 13 basic regression model
+ b7(Year1) . . . + b33(Year26)
- p. 13 basic regression model
b33(Year26 )
Judge's note: Equation read on the page image of PDF page 13: coefficients b1 to b6 precede the year dummies, and b7 is attached to Year1. PDF page 16 says "a set of twenty-six dummy variables". The range b7 to b33 spans 27 coefficients, so the last index is off by one.
Second check: another model agreed
Report a problem with this finding
Wording Footnote 64: "a government's reliance on personal and corporate tax revenues is strongly and negatively influenced by per capita income: poor states tend to rely on trade taxes, rich ones on personal and corporate taxes" — the clause after the colon describes a positive relationship.
- p. 25 footnote 64
strongly and negatively influenced by per capita income: poor states tend to rely on trade taxes, rich ones on personal and corporate taxes
- p. 25 footnote 64
is strongly and negatively influenced by per capita income: poor states tend to rely on trade taxes, rich ones on personal and corporate taxes
- p. 25 footnote 64
the actual effect of Taxes on regime types is probably larger
Judge's note: Viewed p.25, footnote 64. If rich states rely on personal and corporate taxes, reliance on those taxes rises with income, which is a positive relationship. The footnote's own closing inference (controlling for income means the Taxes effect is probably larger) also needs a positive link. So "negatively" contradicts the same sentence. The elaboration makes the intended meaning clear, hence S3.
Second check: another model agreed
Report a problem with this finding
18 minor (typos and the like; most checked by the judge only)
Wording The text says "at the 1995 level of Angola, Nigeria, or Kuwait", but Table 1 gives "4. Nigeria (1991)" and "6. Angola (1996)" and its note says no 1995 figure is available for Angola.
- p. 18 Results, first paragraph on page
1995 level of Angola, Nigeria, or Kuwait
- p. 2 Table 1 rows 4 and 6 and note
4. Nigeria (1991)
Judge's note: Both pages read on the images. Only Kuwait's Table 1 figure (46.14) is a 1995 value; Nigeria's is 1991 and Angola's 1996. The "1995 level" wording is loose. The 1.5-point and 2.1-point figures do match about 45 percent times the column 1 coefficients.
Second check: another model agreed
Report a problem with this finding
Wording Conclusion names Cambodia among 'major mineral exporters' ("including Angola, Chile, the Democratic Republic of Congo, Cambodia, and Peru"), but Cambodia is not in Table 2's 25 mineral-reliant states and is mentioned nowhere else in the article; the other four named states are all in Table 2.
- p. 32 Conclusion, third finding
Democratic Republic of Congo, Cambodia, and Peru
- p. 3 Table 2
Central African Republic
Only the authors can say what is right here.
Judge's note: Viewed the page 32 crop and the full page 3 image of Table 2. Angola (5), Chile (4), Congo Dem. Rep. (9) and Peru (11) are listed; Cambodia is not. A text search finds Cambodia only in this one sentence of the conclusion. Table 2 is only a top-25 list for (mostly) 1995, so Cambodia's inclusion is unsupported rather than demonstrably wrong; it may be a slip for another state, which only the author can say.
Second check: another model agreed
Report a problem with this finding
Wording The text says the OECD dummy covers OECD members "(excluding newer members Mexico and South Korea)", but the Appendix 1 list of 23 states also omits Turkey.
- p. 15 Third control variable paragraph
(excluding newer members Mexico and South
- p. 35 Appendix 1, OECD
Netherlands, New Zealand, Norway, Portugal, Spain, Sweden
Only the authors can say what is right here.
Judge's note: Rests on outside knowledge: that Turkey (and some 1990s entrants) were OECD members in the period. On the page images the text names only Mexico and South Korea as excluded, and the Appendix 1 list has 23 states with no Turkey. If the outside fact holds, the stated exclusion rule is incomplete; the author would need to confirm the coding.
Second check: another model agreed
Report a problem with this finding
Wording "a set of twenty-six dummy variables, one for each year covered by the data (1971–97), less one to mitigate autocorrelation": the stated reason for omitting one year dummy looks misworded.
- p. 16 paragraph on year dummies
less one to mitigate auto-
Judge's note: Text confirmed on the page image of PDF page 16. The same page says autocorrelation is handled by the lagged dependent variable and a panel-specific AR(1) correction. Omitting one dummy from a full set is normally done to avoid perfect collinearity with the constant. That rests on general statistical knowledge rather than a contradiction inside the paper, so this is an observation only.
Second check: another model agreed
Report a problem with this finding
Wording The worked example "losing about .93 on the 0–10 democracy scale" for the $800 country is not reproduced by the Table 3 column 1 coefficients that reproduce the $1,720 case (.54).
- p. 19 Paragraph beginning 'The test also implies'
losing about .93 on the 0–10 democracy scale
Only the authors can say what is right here.
Judge's note: I reworked the scenario (20 million people, $10bn of new oil exports, Oil coefficient –.0346, Income(log) coefficient .922), taking ln(26/16)=.4855, ln(44.4/34.4)=.2552 and ln(170/160)=.0606. The $1,720 case gives about .544, matching the printed .54. The $800 case gives about .883 against .93, and the $8,000 case about .148 against .16. The paper does not state the exact calculation, so this is a query to the author rather than a demonstrated error.
Second check: another model agreed
Report a problem with this finding
Wording Table 4 column 1 prints Minerals as "–.0439** (.00802)"; the two-star marker and the coefficient/standard-error ratio do not correspond under the declared test, and the text says the inclusion "leaves them highly significant".
- p. 20 Third test paragraph, last sentence
leaves them highly significant
- p. 21 Table 4, column 1, Minerals
–.0439**
- p. 21 Table 4, Minerals, column 1
–.0439** (.00802)
Only the authors can say what is right here.
Judge's note: On the page image of Table 4, column 1 Minerals has two stars while Oil in that column and Minerals in columns 2 and 3 have three. The ratio is about 5.47. Table 3 column 3 has a near-identical cell (–.0438, SE .0081, ratio 5.41) printed with three stars, and Table 4 column 3 Oil (–.0138, SE .00557, ratio 2.48) gets one star. A dropped star is likely, but the author should confirm.
Second check: another model agreed
Report a problem with this finding
Wording Table 7 column 2 prints Oil as "–.116" with SE "(.0202)" and no significance marker; the coefficient, standard error and absence of stars do not correspond under the table's star convention, and footnote 70 says Oil is not significant in these samples.
- p. 29 Table 7, column 2, Oil
–.116
- p. 30 footnote 70
Neither Oil nor Minerals is significantly correlated with democracy in these reduced samples
Only the authors can say what is right here.
Judge's note: Viewed p.29 and p.30. The cell reads –.116 (.0202) with no star, against –.0182, –.0187 and –.0315 in the other columns. The ratio is about 5.7, yet in the same table ratios of about 2.1 (.565/.271) get * and about 2.8 (.0154/.00545) get **. The test is not explicitly declared, so this is a query about whether the coefficient (for example –.0116) or the marker is misprinted.
Second check: another model agreed
Report a problem with this finding
Wording Table 7 column 4 prints Income (log) as "3.8" with SE "(.344)" and no significance marker; coefficient, standard error and absence of stars do not correspond under the table's star convention.
- p. 29 Table 7, column 4, Income (log)
3.8
Only the authors can say what is right here.
Judge's note: Viewed p.29. The cell reads 3.8 (.344) with no stars and is aligned differently from the other cells in the column. The other Income (log) entries in Table 7 are –.251, .565* and –.408, and the largest in any other table viewed is 1.24. The ratio is about 11, which would carry *** if the stars follow coefficient/SE as elsewhere in the table. The test is not explicitly declared, so this is a query about whether the value (for example .38) or the marker is misprinted.
Second check: another model agreed
Report a problem with this finding
Wording Table 7 column 1 OECD is printed ".752*" with SE "(.419)"; the coefficient/SE ratio and the 0.05-level star do not correspond under the convention evident elsewhere in the tables.
- p. 29 Table 7, OECD row, column 1
.752*
Only the authors can say what is right here.
Judge's note: Viewed p.29 and p.27. The ratio .752/.419 is about 1.79, below the 1.96 needed at the 0.05 level under a normal reference. In Table 6, Military/GNP –.0366 (.0197), ratio about 1.86, is unstarred and described in the text as "marginally significant at the 0.10 level", so a ratio near 1.8 does not earn a star elsewhere. The test distribution is not explicitly declared, hence a query only.
Second check: another model agreed
Report a problem with this finding
Wording Footnote 70: "Neither Oil nor Minerals is significantly correlated with democracy in these reduced samples", while Table 7 column 1 prints Minerals as ".146*" (significant at 0.05, positive).
- p. 30 footnote 70
Neither Oil nor Minerals is significantly correlated with democracy in these reduced samples
- p. 29 Table 7, column 1, Minerals
.146*
- p. 30 Table 8 col 3 Oil
–.033***
Judge's note: Viewed p.29 and p.30. Minerals in Table 7 column 1 is .146* (.0666). The footnote may refer to the untabulated comparison regressions on the reduced sample without the intervening variable (described on pp.22-23), in which case there is no conflict; the paper does not say which. Table 8 column 3 (Oil –.033***) is the college-enrolment sample, not an occupational-specialization sample, so that part of the finding does not apply. Ambiguity only.
Second check: another model agreed
Report a problem with this finding
Typo The same author's surname is spelled two ways on one page: body text "Kessler and Bazdresch and Levy" versus footnote 26 "Carlos Bazresch and Santiago Levy".
- p. 9 body text
Kessler and Bazdresch and Levy
- p. 9 footnote 26
Carlos Bazresch and Santiago Levy
- p. 9
Carlos Bazresch and Santiago Levy, “Populism and Economic Policy in Mexico
Judge's note: Both spellings are visible on the page image of PDF page 9. The inconsistency is internal to the page. Which spelling is correct rests on outside knowledge (probably Bazdresch).
Second check: none (judge only)
Report a problem with this finding
Typo The Results section opens "For the basic model described below", although the model is specified in the preceding section (above).
- p. 16 Results, first sentence
For the basic model described below
Judge's note: On PDF page 16 the sentence opens Results. The equation and variable definitions are on PDF pages 13 to 16, before it, and only results follow. "below" should read "above". Trivial directional slip.
Second check: none (judge only)
Report a problem with this finding
Typo Figure 1 note reads "per capita income of $1,720 dollars a year", doubling the currency marker.
- p. 18 Figure 1 note a
per capita income of $1,720 dollars a year
Judge's note: Visible on the page image of PDF page 18. Cosmetic redundancy of "$" and "dollars".
Second check: none (judge only)
Report a problem with this finding
Typo Footnote 61 URL is printed "http://cansim.epas.utoronto.ca;5680/pwt/pwt.htm/" with a semicolon where a host:port colon belongs.
- p. 23 footnote 61
utoronto.ca;5680/pwt/pwt.htm/
Judge's note: Viewed p.23. The semicolon before 5680 is visible in the footnote. A port number is separated from the host by a colon, so this is a punctuation slip.
Second check: none (judge only)
Report a problem with this finding
Typo Table 5 row label reads "Mineral" where the variable is "Minerals" in the text, footnote 63 and Tables 6-9.
- p. 24 Table 5, third row label
Mineral
Judge's note: Viewed p.24: the row label is "Mineral". Tables 6, 7, 8 and 9 (pp.27, 29, 30, 31) all print "Minerals". Cosmetic.
Second check: none (judge only)
Report a problem with this finding
Typo Conclusion: "it may well have the same affect on the oil-rich states of Central Asia" — "affect" should be "effect".
- p. 32 Conclusion, second paragraph
same affect on the oil-rich states of Central Asia
- p. 32 Conclusion, second finding
it may well have the same affect on the oil-rich states of Central Asia
- p. 32 Conclusion
same affect on the oil-rich states
Judge's note: Viewed p.32, Conclusion, second paragraph. "same affect" is printed; the noun required is "effect".
Second check: none (judge only)
Report a problem with this finding
Typo Appendix 1 defines the Arabian Peninsula dummy as "coded 1 for the states on the Saudi Arabian peninsula", whereas the main text (p. 22) and the variable name use 'Arabian Peninsula'.
- p. 35 Appendix 1, Arabian Peninsula definition
coded 1 for the states on the Saudi Arabian peninsula
Judge's note: Viewed the page 35 crop: the text reads 'Saudi Arabian peninsula (Bahrain, Kuwait, Oman, Saudi Arabia, Qatar, the United Arab Emirates, and Yemen)'. The main text on PDF page 22 says 'the seven states of the Arabian Peninsula' with the same list. This is a nonstandard and internally inconsistent name for the same region; the meaning is unambiguous, so it is a minor wording point.
Second check: none (judge only)
Report a problem with this finding
Typo Duplicated word in Appendix 1, Urban definition: "The data are from from the United Nations, World Urbanization Prospects: The 1996 Revision."
- p. 36 Appendix 1, Urban definition
The data are from from the United Nations
- p. 36
The data are from from the United Nations, World
- p. 36
otal population. The data are from from the United Nations,
Judge's note: Viewed the page 36 crop: 'from from' is printed twice on the same line in the page image, so it is not a line-break or extraction artefact. The fix is to delete one 'from'.
Second check: code check