Summary Demographic Table

Table 1 . summary Distribution of all variables ,Groupwise

Dependent: all all
age Mean (SD) 25.6 (4.1)
GA 37 59 (11.8)
38 185 (37.0)
39 233 (46.6)
40 23 (4.6)
Birth.Weight Mean (SD) 2.9 (0.4)
SFH Mean (SD) 32.6 (2.4)
Station -1 328 (65.6)
0 58 (11.6)
1 114 (22.8)
johnson Mean (SD) 3.1 (0.3)
Gravidity Mean (SD) 2.0 (1.0)
parity Mean (SD) 0.6 (0.7)
Sex_Of_neonate BOY 260 (52.0)
GIRL 240 (48.0)
error Mean (SD) -0.2 (0.2)
absolute_error Mean (SD) 0.3 (0.1)
percentage_error Mean (SD) 9.8 (6.2)
category < 2 2 (0.4)
2 - 2.5 74 (14.8)
2.5 - 3.5 223 (44.6)
3 - 3.5 174 (34.8)
> 3.5 27 (5.4)
estimation Overestimation 439 (87.8)
Underestimation 61 (12.2)

Demographic Variables

Distribution of Demographic Variables in Our Population

Age Distribution

The plot above shows a flipped bar plot of Counts(X axis) and percentages(annotated within bar) of various categories. The top 2 age categories are as follows : 200/500(40%) patients are in group 20 - 25 . 194/500(38.8%) patients are in group 25 - 30 . The Full details of distribution is in table below.

age categories n percentage
< 20 18 3.6
20 - 25 200 40.0
25 - 30 194 38.8
30 - 35 76 15.2
>35 12 2.4

Gestational Age Distribution

he Flipped Bar- plot above shows Counts(X axis) and percentages(annotated within bar) of various categories. The top 2 sub-groups are as follows : 233/500(46.6 %) patients are in sub-group 39 185/500(37 %) patients are in sub-group 38 The Full details of distribution is in table below.

Group n total percentage Confidence_Interval
39 233 500 46.6 42.26% - 50.98%
38 185 500 37.0 32.85% - 41.3%
37 59 500 11.8 9.19% - 14.85%
40 23 500 4.6 3.02% - 6.7%

Parity Distribution

The Flipped Bar- plot above shows Counts(X axis) and percentages(annotated within bar) of various categories. The top 2 sub-groups are as follows : 240/500(48 %) patients are in sub-group 0 212/500(42.4 %) patients are in sub-group 1 The Full details of distribution is in table below.

Group n total percentage Confidence_Interval
0 240 500 48.0 43.64% - 52.38%
1 212 500 42.4 38.12% - 46.77%
2 39 500 7.8 5.69% - 10.4%
3 5 500 1.0 0.38% - 2.18%
4 4 500 0.8 0.27% - 1.89%

Gravidity Distribution

The Flipped Bar- plot above shows Counts(X axis) and percentages(annotated within bar) of various categories. The top 2 sub-groups are as follows : 199/500(39.8 %) patients are in sub-group 1 167/500(33.4 %) patients are in sub-group 2 The Full details of distribution is in table below.

Group n total percentage Confidence_Interval
1 199 500 39.8 35.58% - 44.14%
2 167 500 33.4 29.37% - 37.62%
3 94 500 18.8 15.56% - 22.4%
4 29 500 5.8 4% - 8.11%
5 9 500 1.8 0.89% - 3.26%
6 2 500 0.4 0.08% - 1.28%

Actual Birth Weight Distribution

The Flipped Bar- plot above shows Counts(X axis) and percentages(annotated within bar) of various categories. The top 2 sub-groups are as follows : 223/500(44.6 %) patients are in sub-group 2.5 - 3.5 174/500(34.8 %) patients are in sub-group 3 - 3.5 The Full details of distribution is in table below.

Group n total percentage Confidence_Interval
2.5 - 3.5 223 500 44.6 40.28% - 48.98%
3 - 3.5 174 500 34.8 30.72% - 39.05%
2 - 2.5 74 500 14.8 11.89% - 18.11%
> 3.5 27 500 5.4 3.67% - 7.64%
< 2 2 500 0.4 0.08% - 1.28%

Estimated Birth Weight Distribution

The Flipped Bar- plot above shows Counts(X axis) and percentages(annotated within bar) of various categories. The top 2 sub-groups are as follows : 224/500(44.8 %) patients are in sub-group [2.50,3.00) 180/500(36 %) patients are in sub-group [3.00,3.50) The Full details of distribution is in table below.

Group n total percentage Confidence_Interval
[2.50,3.00) 224 500 44.8 40.48% - 49.18%
[3.00,3.50) 180 500 36.0 31.88% - 40.28%
[3.50,4.34] 93 500 18.6 15.38% - 22.19%
[2.00,2.50) 3 500 0.6 0.17% - 1.59%

Symphysio-Fundal Height Distribution

The Flipped Bar- plot above shows Counts(X axis) and percentages(annotated within bar) of various categories. The top 2 sub-groups are as follows : 444/500(88.8 %) patients are in sub-group [30.0,38.0) 41/500(8.2 %) patients are in sub-group [38.0,41.0] The Full details of distribution is in table below.

OUTCOMES Of INTEREST

Figure 8 Boxplot Of Distribution Of Actual and Predicted in our Population

In this Figure we see Box plot of Weight_Kg in 2 sub-groups of Type : johnson and Actual Birth.Weight respectively .The individual jittered data points of Weight_Kg are overlaid over transparent Boxplot for better visualisation. We see distribution of data in individual sub-groups of Type based on these box-plots. The lower edge of box plot represents -first quartile (Q1), Horizontal bar represents the median, Upper edge represnts third quartile (Q3), Two black lines (whiskers) emanating from box-plots signify range of non-outlier data for the particular sub-group. Lower whisker represents minimum(Q1- 1.5 interquartile range) non-outlier limit of Weight_Kg and upper whisker represnts maximum(Q1+1.5interquartile range) of Weight_Kg .Any data beyond whiskers of box-plots represents outliers in the sub-groups The big brown point in the box-plots represents mean Weight_Kg of 2 groups and it has been annotated in the figure itself. Summary Statistics of the groups is presented in table below

Table Summary Table Of Birth Weight within Groups

Group n Mean SD Median Minimum Maximum
Birth.Weight 500 2.895 0.380 2.895 1.80 4.04
johnson 500 3.129 0.344 3.100 2.48 4.34

The mean Actual Birth Weight [ 2.9 ± 0.38 ] was significantly lower than predicted by johnson formula [ 3.13 ± 0.34 ] . The mean difference was -0.23 and 95 % confidence interval of the difference was ( -0.28 - -0.19 ) . The p value was <0.001 . The t statistic was -10.18 and degree of freedom of the Welch unpaired two-sample t test was 988.11 .In Formal statistical notation this result is expressed as : t(988.11) = -10.18, p= <0.001.

Correlation between johnson and actual birth weight

The scatter plots above show relationship between Birth.Weight on X axis and johnson on Y axis. Graphically, we see that as Birth.Weight increases, johnson also increases . On a formal statistical linear regression analysis, we that line of best fit (blue line signifying line with least square difference) also has a positive slope implying a positive correlation. The gray shaded error around blue line signifies 95% confidence interval of linear regression line of best fit. The correlation between two variables is Significant . The Pearson’s correlation between Birth.Weight and johnson is 0.54 with 95% Confidence Interval of 0.48 to 0.6. the t statistic is 14.37 The p value is <0.001 .The degree of freedom is 498. In formal statistical notation this expressed as t(498)= 14.37, P= <0.001. r(Pearson) = 0.54 95% C.I. [0.48-0.6]. n= 500. The correlation is summmarised in table below

variable n Mean SD Median Minimum Maximum
Birth.Weight 500 2.895 0.380 2.895 1.80 4.04
johnson 500 3.129 0.344 3.100 2.48 4.34
Group 1 Group 2 Degree of Freedom T statistic Correlation 95 % C.I. P value
Birth.Weight johnson 498 38.15 0.86 0.84-0.88 <0.001
Group var1 var2 cor statistic conf.low conf.high parameter significance pvalue Confidence_Interval
2 - 2.5 Birth.Weight johnson 0.36 3.26 0.14 0.55 69 Significant <0.001 0.14-0.55
2.5 - 3.5 Birth.Weight johnson 0.56 9.96 0.46 0.64 221 Significant <0.001 0.46-0.64
3 - 3.5 Birth.Weight johnson 0.66 11.44 0.56 0.73 172 Significant <0.001 0.56-0.73
> 3.5 Birth.Weight johnson 0.36 1.95 -0.02 0.65 25 Non-Significant 0.06 -0.02-0.65

absolute error

he scatter plots above show relationship between Birth.Weight on X axis and absolute_error on Y axis. Graphically, we see that as Birth.Weight increases, absolute_error initially decreases and then increases (more at extremes). On a formal statistical linear regression analysis, we that line of best fit (blue line signifying line with least square difference) also has a negative slope implying a negative correlation. The gray shaded error around blue line signifies 95% confidence interval of linear regression line of best fit. The correlation between two variables is Significant . The Pearson’s correlation between Birth.Weight and absolute_error is -0.4 with 95% Confidence Interval of -0.47 to -0.33. the t statistic is -9.84 The p value is <0.001 .The degree of freedom is 498. In formal statistical notation this expressed as t(498)= -9.84, P= <0.001. r(Pearson) = -0.4 95% C.I. [-0.47–0.33]. n= 500. The correlation is summmarised in table below

variable n Mean SD Median Minimum Maximum
absolute_error 500 0.269 0.139 0.255 0.0 0.82
Birth.Weight 500 2.895 0.380 2.895 1.8 4.04

Box plot of Absolute error variation with categories of Birth weight

In this Figure we see Box plot of absolute_error in 5 sub-groups of category : 3 - 3.5,> 3.5,2 - 2.5,2.5 - 3.5 and < 2 respectively .The individual jittered data points of absolute_error are overlaid over transparent Boxplot for better visualisation. We see distribution of data in individual sub-groups of category based on these box-plots. The lower edge of box plot represents -first quartile (Q1), Horizontal bar represents the median, Upper edge represnts third quartile (Q3), Two black lines (whiskers) emanating from box-plots signify range of non-outlier data for the particular sub-group. Lower whisker represents minimum(Q1- 1.5 interquartile range) non-outlier limit of absolute_error and upper whisker represnts maximum(Q1+1.5interquartile range) of absolute_error .Any data beyond whiskers of box-plots represents outliers in the sub-groups The big brown point in the box-plots represents mean absolute_error of 5 groups and it has been annotated in the figure itself Summary Statistics of the groups is presented in table below

Group n Mean SD Median Minimum Maximum
< 2 2 0.642 0.053 0.642 0.605 0.680
2 - 2.5 74 0.436 0.142 0.423 0.155 0.790
2.5 - 3.5 223 0.236 0.108 0.240 0.000 0.515
3 - 3.5 174 0.235 0.106 0.235 0.000 0.555
> 3.5 27 0.269 0.188 0.250 0.000 0.820

We find that One-way ANOVA was significant for Group effect of category on absolute_error. In statistical notation it is expressed as F(4,495)=<0.01. The Effect size(Omega -Squared) of this One-way ANOVA test was 0.288 .

We find that One-way ANOVA was significant for Group effect of category on absolute_error. In statistical notation it is expressed as F(4,495)=<0.01. The  Effect size(Omega -Squared) of this One-way ANOVA  test was 0.288 .

Since Overall One-Way ANOVA was signifcant indicating an overall difference in groups, we undertook 10 unpaired t-test to look for inter-group differences The mean absolute_error in Group 2 was non-significantly lower than Group 2.5 . The difference was -0.21 and 95 % confidence interval was ( -0.44 - 0.03 ) . The adjusted p value was 0.11 . The mean absolute_error in Group 2.5 was significantly lower than Group 3.5 . The difference was -0.41 and 95 % confidence interval was ( -0.64 - -0.18 ) . The adjusted p value was <0.001 . The mean absolute_error in Group 3 was significantly lower than Group 3.5 . The difference was -0.41 and 95 % confidence interval was ( -0.64 - -0.18 ) . The adjusted p value was <0.001 . The mean absolute_error in Group > 3.5 was significantly lower than Group < 2 . The difference was -0.37 and 95 % confidence interval was ( -0.61 - -0.14 ) . The adjusted p value was <0.001 . The mean absolute_error in Group 2.5 was significantly lower than Group 3.5 . The difference was -0.2 and 95 % confidence interval was ( -0.24 - -0.16 ) . The adjusted p value was <0.001 . The mean absolute_error in Group 3 was significantly lower than Group 3.5 . The difference was -0.2 and 95 % confidence interval was ( -0.25 - -0.16 ) . The adjusted p value was <0.001 . The mean absolute_error in Group > 3.5 was significantly lower than Group 2 . The difference was -0.17 and 95 % confidence interval was ( -0.24 - -0.09 ) . The adjusted p value was <0.001 . The mean absolute_error in Group 3 was non-significantly lower than Group 3.5 . The difference was 0 and 95 % confidence interval was ( -0.03 - 0.03 ) . The adjusted p value was 1 . The mean absolute_error in Group > 3.5 was non-significantly higher than Group 2.5 . The difference was 0.03 and 95 % confidence interval was ( -0.03 - 0.1 ) . The adjusted p value was 0.63 . The mean absolute_error in Group > 3.5 was non-significantly higher than Group 3 . The difference was 0.03 and 95 % confidence interval was ( -0.03 - 0.1 ) . The adjusted p value was 0.63 . Table describing these tests with Tukey’s Post-Hoc correction is described below

Comparison Difference 95% Confidence Interval P value Significance
2 - 2.5 -0.21 -0.44 - 0.03 0.11 Non-significant
2.5 - 3.5 -0.41 -0.64 - -0.18 <0.001 Significant
3 - 3.5 -0.41 -0.64 - -0.18 <0.001 Significant
> 3.5 - < 2 -0.37 -0.61 - -0.14 <0.001 Significant
2.5 - 3.5 -0.20 -0.24 - -0.16 <0.001 Significant
3 - 3.5 -0.20 -0.25 - -0.16 <0.001 Significant
> 3.5 - 2 -0.17 -0.24 - -0.09 <0.001 Significant
3 - 3.5 0.00 -0.03 - 0.03 1 Non-significant
> 3.5 - 2.5 0.03 -0.03 - 0.1 0.63 Non-significant
> 3.5 - 3 0.03 -0.03 - 0.1 0.63 Non-significant

Percentage error

The scatter plots above show relationship between Birth.Weight on X axis and percentage_error on Y axis. Graphically, we see that as Birth.Weight increases, percentage_error initially decreases and then increases (i.e. more at extremes) On a formal statistical linear regression analysis, we that line of best fit (blue line signifying line with least square difference) also has a negative slope implying a negative correlation. The gray shaded error around blue line signifies 95% confidence interval of linear regression line of best fit. The correlation between two variables is Significant . The Pearson’s correlation between Birth.Weight and percentage_error is -0.39 with 95% Confidence Interval of -0.46 to -0.31. the t statistic is -9.36 The p value is <0.001 .The degree of freedom is 498. In formal statistical notation this expressed as t(498)= -9.36, P= <0.001. r(Pearson) = -0.39 95% C.I. [-0.46–0.31]. n= 500. The correlation is summmarised in table below

variable n Mean SD Median Minimum Maximum
Birth.Weight 500 2.895 0.380 2.895 1.8 4.04
percentage_error 500 9.764 6.209 8.685 0.0 39.50

Box plot of Absolute error variation with categories of Birth weight

In this Figure we see Box plot of percentage_error in 5 sub-groups of category : 3 - 3.5,> 3.5,2 - 2.5,2.5 - 3.5 and < 2 respectively .The individual jittered data points of percentage_error are overlaid over transparent Boxplot for better visualisation. We see distribution of data in individual sub-groups of category based on these box-plots. The lower edge of box plot represents -first quartile (Q1), Horizontal bar represents the median, Upper edge represnts third quartile (Q3), Two black lines (whiskers) emanating from box-plots signify range of non-outlier data for the particular sub-group. Lower whisker represents minimum(Q1- 1.5 interquartile range) non-outlier limit of percentage_error and upper whisker represnts maximum(Q1+1.5interquartile range) of percentage_error .Any data beyond whiskers of box-plots represents outliers in the sub-groups The big brown point in the box-plots represents mean percentage_error of 5 groups and it has been annotated in the figure itself Summary Statistics of the groups is presented in table below

Group n Mean SD Median Minimum Maximum
< 2 2 35.025 3.896 35.025 32.27 37.78
2 - 2.5 74 19.075 7.174 18.160 6.25 39.50
2.5 - 3.5 223 8.609 4.069 8.770 0.00 19.92
3 - 3.5 174 7.354 3.338 7.225 0.00 18.44
> 3.5 27 7.451 5.257 6.900 0.00 22.71

We find that One-way ANOVA was significant for Group effect of category on percentage_error. In statistical notation it is expressed as F(4,495)=<0.01. The Effect size(Omega -Squared) of this One-way ANOVA test was 0.475 .

We find that One-way ANOVA was significant for Group effect of category on percentage_error. In statistical notation it is expressed as F(4,495)=<0.01. The  Effect size(Omega -Squared) of this One-way ANOVA  test was 0.475 .

Since Overall One-Way ANOVA was signifcant indicating an overall difference in groups, we undertook 10 unpaired t-test to look for inter-group differences The mean percentage_error in Group 2 was significantly lower than Group 2.5 . The difference was -15.95 and 95 % confidence interval was ( -24.81 - -7.09 ) . The adjusted p value was <0.001 . The mean percentage_error in Group 2.5 was significantly lower than Group 3.5 . The difference was -26.42 and 95 % confidence interval was ( -35.2 - -17.63 ) . The adjusted p value was <0.001 . The mean percentage_error in Group 3 was significantly lower than Group 3.5 . The difference was -27.67 and 95 % confidence interval was ( -36.46 - -18.88 ) . The adjusted p value was <0.001 . The mean percentage_error in Group > 3.5 was significantly lower than Group < 2 . The difference was -27.57 and 95 % confidence interval was ( -36.63 - -18.51 ) . The adjusted p value was <0.001 . The mean percentage_error in Group 2.5 was significantly lower than Group 3.5 . The difference was -10.47 and 95 % confidence interval was ( -12.12 - -8.81 ) . The adjusted p value was <0.001 . The mean percentage_error in Group 3 was significantly lower than Group 3.5 . The difference was -11.72 and 95 % confidence interval was ( -13.44 - -10.01 ) . The adjusted p value was <0.001 . The mean percentage_error in Group > 3.5 was significantly lower than Group 2 . The difference was -11.62 and 95 % confidence interval was ( -14.4 - -8.84 ) . The adjusted p value was <0.001 . The mean percentage_error in Group 3 was non-significantly lower than Group 3.5 . The difference was -1.26 and 95 % confidence interval was ( -2.51 - 0 ) . The adjusted p value was 0.05 . The mean percentage_error in Group > 3.5 was non-significantly lower than Group 2.5 . The difference was -1.16 and 95 % confidence interval was ( -3.68 - 1.36 ) . The adjusted p value was 0.72 . The mean percentage_error in Group > 3.5 was non-significantly higher than Group 3 . The difference was 0.1 and 95 % confidence interval was ( -2.46 - 2.65 ) . The adjusted p value was 1 . Table describing these tests with Tukey’s Post-Hoc correction is described below

Comparison Difference 95% Confidence Interval P value Significance
2 - 2.5 -15.95 -24.81 - -7.09 <0.001 Significant
2.5 - 3.5 -26.42 -35.2 - -17.63 <0.001 Significant
3 - 3.5 -27.67 -36.46 - -18.88 <0.001 Significant
> 3.5 - < 2 -27.57 -36.63 - -18.51 <0.001 Significant
2.5 - 3.5 -10.47 -12.12 - -8.81 <0.001 Significant
3 - 3.5 -11.72 -13.44 - -10.01 <0.001 Significant
> 3.5 - 2 -11.62 -14.4 - -8.84 <0.001 Significant
3 - 3.5 -1.26 -2.51 - 0 0.05 Non-significant
> 3.5 - 2.5 -1.16 -3.68 - 1.36 0.72 Non-significant
> 3.5 - 3 0.10 -2.46 - 2.65 1 Non-significant

percentage_error categories n percentage
< 5 97 19.4
10 - 15 120 24.0
15 - 20 42 8.4
20 - 25 19 3.8
25 - 30 5 1.0
30 - 40 10 2.0
> 5 207 41.4

absolute_error categories n percentage
< 0.15 87 17.4
0.15 - 0.25 148 29.6
0.25 - 0.35 146 29.2
0.35 - 0.50 89 17.8
>0.50 30 6.0

Regression Equation

Multiple linear regression was conducted to find best combination of Age,Gestational_age,parity & SFH for predicting Birth_weight . Dummy indicator(0/1) were used for categorical variables. The Forest plot above shows standardized regression coefficients of Age,Gestational_age,parity & SFH with their confidence intervals as horizontal error bars on X axis. An error bar which crosses vertical line of zero in this plot is non-significant. We centered the variables at their means for better interpretation.

We fitted a linear model (estimated using OLS) to predict Birth.Weight with age, GA, parity, Interval and SFH.in.Cms (formula = Birth.Weight ~ age + GA + parity + Interval + SFH.in.Cms). Standardized parameters were obtained by fitting the model on a standardized version of the dataset. Effect sizes were labelled following Funder’s (2019) recommendations.

We fitted a linear model (estimated using OLS) to predict Birth.Weight with age, GA, parity and SFH (formula = Birth.Weight ~ age + GA + parity + SFH). Standardized parameters were obtained by fitting the model on a standardized version of the dataset. Effect sizes were labelled following Funder’s (2019) recommendations.

The model explains a significant and substantial proportion of variance (R2 = 0.71, F(4, 495) = 308.30, p < .001, adj. R2 = 0.71). The model’s intercept, corresponding to Birth.Weight = 0, age = 0, GA = 0, parity = 0 and SFH = 0, is at -2.30 (SE = 0.47, 95% CI [-3.23, -1.38[, std. intercept = 0.00, p < .001). Within this model:

Bland Altman

Bland-Altman analysis was performed to compare Johnson’s Method with Actual Birth Weight. There were 500 observations in our study. The mean measurementby method 1 was 3.13 .The mean measurementby method 2 was 2.9. There was a mean bias of 0.23 (95% C.I. 0.22 - 0.25 ) and 95% confidence limits of Agreement from a lower limit of -0.14 to upper limit of 0.61 implying 95% of differences between two methods were found in this range. The mean proportional bias was 8.21 % with 95% Confidence Limit of 7.42 % to 8.66 %. The difference between methods as a function of mean of two methods can be described by the equation y(differences) = -0.11 x(means) + 0.56 implying difference between two measures changes by -0.11 unit with 1 unit increase in average of two methods

The Graph above shows average Birth weight (average of Birth weight by Johnson’s Formula and Actual Birth weight on X Axis) and Difference between Johnson’s Formula and Actual Birth weight on Y axis. The Limit of Agreement (between+0.35 to -0.35) is represented as Gray zone. A dashed line represents average Bias (+0.23 kg) implying Johnson’s Formula overestimated Actual Birth Weight on average. The Blue shaded line (regression line) has a negative slope is implying , with increasing Birth weight,Bias decreases.

The two outward-most dashed lines represent upper and lower Confidence limits of Bias.

Interpretation

Bland-Altman analysis was performed to compare Johnson’s Method with Actual Birth Weight. There were 500 observations in our study. The mean measurementby method 1 was 3.13 .The mean measurementby method 2 was 2.9. There was a mean bias of 0.23 (95% C.I. 0.22 - 0.25 ) and 95% confidence limits of Agreement from a lower limit of -0.14 to upper limit of 0.61 implying 95% of differences between two methods were found in this range. The mean proportional bias was 8.21 % with 95% Confidence Limit of 7.42 % to 8.66 %. The difference between methods as a function of mean of two methods can be described by the equation y(differences) = -0.11 x(means) + 0.56 implying difference between two measures changes by -0.11 unit with 1 unit increase in average of two methods

Estimation

We found a Significant association between estimation and category. The chi-square statistic was 17.1 . The degree of freedom was 4 and P value was <0.001 .Contingency and Proportion table are shown below

category estimation n value 95 % Confidence Interval
< 2 Overestimation 2 2/2 ( 100 %) 33.32% - 99.98%
2 - 2.5 Overestimation 74 74/74 ( 100 %) 96.67% - 100%
2.5 - 3.5 Overestimation 196 196/223 ( 87.89 %) 83.13% - 91.68%
2.5 - 3.5 Underestimation 27 27/223 ( 12.11 %) 8.32% - 16.87%
3 - 3.5 Overestimation 147 147/174 ( 84.48 %) 78.55% - 89.28%
3 - 3.5 Underestimation 27 27/174 ( 15.52 %) 10.72% - 21.45%
> 3.5 Overestimation 20 20/27 ( 74.07 %) 55.71% - 87.58%
> 3.5 Underestimation 7 7/27 ( 25.93 %) 12.42% - 44.29%