Presentation Slides 23 & 24

Goodness Fit Measures

The Model Construct closely resembles Figure 2.2 on page 88.

(Appendix 23)????Predictors?????????????????????ESTATEFAC, FFBPROC, LOGEFF

???????????????????????????Dependent Variable???? Oil Extraction Rate (OER)

Correlations coefficient R?=?0.953?????? Coefficient of Determination R(sq) =?0.908

?????????????? ?Adj. R(sq) = 0.898

??????????????The adjusted R(sq) is a modified version of R(sq) that has been adjusted for the number of predictors in the model.

??????????????It decreases when a predictor improves the model by less than expected by chance.

Q(sq) Predictive Relevance

Q(sq) represents measures of how well observed values are reconstructed by the model and its parameter (Chin, 2010).

Since blind folding is a sample re-use technique which systemically deletes data points and provides a prognosis of the original values, the procedure requires an omission distance D between 5 and 12 (Wold, 1982).

In this study, D is 7 and is obtained the Cross-Validated Redundancy (Table 4.24, page 254)

Cross Validated Redundancy, small 0.02, medium 0.15, large 0.35, > 0.5 good ?(Chin, 2010).

?Estate Factors is 0 and lacks predictive relevance in redundancy because this measures the capacity of the path model to predict the endogenous indirectly.

An alternative is Construct Cross Validated Communality. This measures the capacity of the model to predict directly and is 0.444 for estate factors (Table 4.24, page 254).

Goodness Fit Measures

Goodness of Fit measure refers to the geometric mean of the average communality and average R(sq) (Tenenhaus et al., 2005).

It tells if the sample data can represent the data expected to find in the actual population. GoF 0.10 small, 0.13 medium, 0.36 large.

GoF was found to be 0.663 and allow to conclude that this model performs well compared to the baseline values and adequately supported that the model was globally validated (Table 4.25, page 255).

Goodness fit measures, hypothesis H1 next slide 23-27

H1 Estate Factors have a direct influence on FFB Processing

Hypothesis testing is run to determine whether a claim is true or not, given a population parameter.

?

(Fig. 2.2, page 88)???? Predictor??????????????????????Estate Factors

????????????????????????????????????Dependent Variable??? FFB Processing

?

Appendix 24a Model Summary?

????????????????????????R = 0.470????R(sq) = 0.221

The square of the correlation tells about the amount of variability in y that is explained by the model (0.26, 0.13, 0.02)

Coefficients correlation measures the strength and direction of a linear relationship (+1 and -1)

???????????????????????????????

Appendix 24b ANOVA??????Analysis of Variation, to analyze the differences among group means in a sample (Ronald Fisher)

Sum of squares measures how far individual measurements are from the mean, also known as variation

Mean squares are estimates of variance across groups, (sum of squares/degree of freedom)

‘f’ is (variation between sample means/variation within the samples)

Significance probability or p-value

???????????????????????????sum of squares????????? df?????????????????????? ms????????????????????f???????????????????Sig.

Regression?????????1.885????????????????????????? df1???????1??????????? 1.885???????????????35.001 ?????????0.000

Residual??????????????6.660?????????????????????????df2 ??124.00???????0.054

df indicates the?number?of independent values that can vary?in an analysis without breaking any constraints

df1 how the cell means to relate to the?grand mean or marginal means

df2 how the single observations in the cells relate to the cell means

?

Appendix 24c Coefficients

std error is a measure of?the accuracy of predictors,

se is the square root of the average squared deviation

t = (coefficient/standard error)

unstandardized coefficient???????????se??????????stdcoefficient???t?????????????Sig.????????Tolerance???????????VIF

0.616???????????????? 0.104????0.470????????????????5.925?????0.000???? 1.000?????????????????1.000

?

Therefore, hypothesis H1, Estate factors have a direct influence on FFB Processing is supported.

Estate Factors influenced directly FFB processing is supported,?FFB Processing influenced directly OER next slide 24-27