Skip to main content

Table 3 Correlates of Obesity Perception. Non-pregnant participants ages 16 and up in NHANES 1999–2013, complete data on BMI and weight perception; BMI ≥ 30 or above 95th percentile

From: Self-estimated BMI, but not self-perceived body size, accurately identifies unhealthy weight in US adults

  Accurate
(self-perceives as ‘too heavy’)
Inaccurate
(self-perceives as ‘about right’ or ‘too thin’)
Odds ratio predicting accuracy1 P
Point estimate 95% confidence interval
Bottom Top
Percent of total obese 9.1 91
Years since 1999 8.0
0, 13
8.5
0, 13
0.979 0.962 0.996 0.02
Age, years 46
20, 73
43
17, 76
1.01 1.00 1.01 0.003
Male 45 68 0.326 0.277 0.385 ***
Ethnicity ***
 African-American 14 26 0.356 0.299 0.425 ***
 European-American (ref) 69 50 1.0 1.0 1.0 (ref)
 Hispanic 14 20 0.730 0.599 0.890 0.002
 Other 3.8 4.2 0.791 0.529 1.18 > 0.10
Household income as % of FMPI
  < 100 14 24 0.608 0.472 0.783 0.0001
 100–200 22 28 0.734 0.574 0.940 0.01
 200–300 16 17 1.0 1.0 1.0 (ref)
 300–400 15 12 1.31 0.956 1.79 0.09
 400–500 11 5.3 2.21 1.48 3.31 0.0001
  > 500 22 14 1.83 1.34 2.49 0.0001
Obesity class
 1 (ref) 56 83 1.0 1.0 1.0 (ref)
 2 26 14 2.86 2.33 3.52 ***
 3+ 18 3.5 7.92 5.35 11.7 ***
Self-estimated BMI as % of measured 104
95, 115
105
95, 120
n/a n/a n/a n/a
Category of self-estimated BMI     
 Underweight 0.002 0.060     
 Normal weight 0.17 0.99 n/a n/a n/a n/a
 Overweight 15 33     
 Obese 85 66     
  1. Continuous measures expressed as mean; 5th, 95th percentiles. Binary measures expressed as %
  2. P-value for top row of ethnicity, BMI category, and income is type 3 test of null hypothesis (all categories equal.) P-value for each row is that for pairwise comparison with the reference group. *** if p < 0.0001, n/a if not tested
  3. 1) Survey-weighted logistic regression used to fit multivariable corrected models
\