Regression Inferences

Recall that the simple regression model equation is 

Frequently, the investigator wants to know whether the slope, /3i, is significant, that is, not equal to zero (/3i 0). If /3i=0, then regression analysis should not be used, for /3o is a good estimate of y, that is, /3o = y. The significance test for /3i is a hypothesis test  

Returning to the /3i test, to evaluate whether /3i is significant (fii 0), the researchers set up a two-tail hypothesis, using the six-step procedure. 

Step 4 State the decision rule for /tabled = (a/2, 2) from Table B. 

Step 6 State the conclusion when comparing /calculated with /tabled- 

---

Robust and resistant isochrons can have very different characteristics than traditional least-squares or error-weighted least squares regressions. Some methods ignore analytical errors entirely, others infer them from the observed scatter of the data, and still others make use of analytical errors only to the extent that they are validated by their observed scatter. 

Once we have estimated the unknown parameter values in a linear regression model and the underlying assumptions appear to be reasonable, we can proceed and make statistical inferences about the parameter estimates and the response variables. 

A valuable inference that can be made to infer the quality of the model predictions is the (l-a)I00% confidence interval of the predicted mean response at x0. It should be noted that the predicted mean response of the linear regression model at x0 is y0 = F(x0)k or simply y0 = X0k. Although the error term e0 is not included, there is some uncertainty in the predicted mean response due to the uncertainty in k. Under the usual assumptions of normality and independence, the covariance matrix of the predicted mean response is given by... 

Both assumptions are mainly needed for constructing confidence intervals and tests for the regression parameters, as well as for prediction intervals for new observations in x. The assumption of normal distribution additionally helps avoid skewness and outliers, mean 0 guarantees a linear relationship. The constant variance, also called homoscedasticity, is also needed for inference (confidence intervals and tests). This assumption would be violated if the variance of y (which is equal to the residual variance a2, see below) is dependent on the value of x, a situation called heteroscedasticity, see Figure 4.8. 

An advantage of LR in comparison to LDA is the fact that statistical inference in the form of tests and confidence intervals for the regression parameters can be derived (compare Section 4.3). It is thus possible to test whether the /th regression coefficient bj = 0. If the hypothesis can be rejected, the jth regressor variable xj... 

In a paper that addresses both these topics, Gordon et al.11 explain how they followed a com mixture fermented by Fusarium moniliforme spores. They followed the concentrations of starch, lipids, and protein throughout the reaction. The amounts of Fusarium and even com were also measured. A multiple linear regression (MLR) method was satisfactory, with standard errors of prediction (SEP) for the constituents being 0.37% for starch, 4.57% for lipid, 4.62% for protein, 2.38% for Fusarium, and 0.16% for com. It may be inferred from the data that PLS or PCA (principal components analysis) may have given more accurate results. 

Differences in calibration graph results were found in amount and amount interval estimations in the use of three common data sets of the chemical pesticide fenvalerate by the individual methods of three researchers. Differences in the methods included constant variance treatments by weighting or transforming response values. Linear single and multiple curve functions and cubic spline functions were used to fit the data. Amount differences were found between three hand plotted methods and between the hand plotted and three different statistical regression line methods. Significant differences in the calculated amount interval estimates were found with the cubic spline function due to its limited scope of inference. Smaller differences were produced by the use of local versus global variance estimators and a simple Bonferroni adjustment. 

The substantial differences, when comparing response dispersion alone, may be explained in two additional ways. The first involves the size of the uncertainty when using the Bonferroni inference. Kurtz, et al., and Mitchell use a Bonferroni inference which splits the uncertainty in the regression between the regression and the response portions. Instead of a 95% probability there is a 97.5% probability for each portion. The contribution to the uncertainty for a higher probability thus has increased the width of the response uncertainty band. 

The types of problems capable of being solved with the following regression methods are defined by the manner in which each of the regression methods works. The basic premise is that a method is given input variables (bioactivities and descriptors) and in turn the method produces output variables (coefficients and a constant). These methods work best when information about the system of interest is known and inferences can be made about the problem being solved. This can only be done if there is confidence that a relationship exists between the known input data and the unknown output data before these methods are utilized if there is no relationship, then the model will be useless. 

There is a growing literature that addresses the transferability of a study s pooled results to subgroups. Approaches include evaluation of the homogeneity of different centers and countries results use of random effects models to borrow information from the pooled results when deriving center-specific or country-specific estimates direct statistical inference by use of net monetary benefit regression and use of decision analysis. 

The potential usefulness of the equation is indicated by the strength of the correlation between observed and inferred values characterized by the coefficient of determination (r2), as well as the standard error and the 95% confidence intervals associated with the regression. The overriding value of the relationship is that it can be used to infer past lake-water chemistry characteristics, with quantitative error estimates (e.g., The lake-water pH value, inferred from the sediment deposited at the 5.0-cm interval, is 6.3 with an estimated standard error of 0.3 pH units. ). To base inferred values only on the percent abundance of a limited number of categories is wasteful... 

Weighted Averaging. WA regression and calibration is a robust, computationally simple, and straightforward method for reconstructing environmental variables. It provides a more accurate and precise inference... 

Figure 10. Observed salinity versus diatom-inferred salinity (derived from WA regression) for 55 calibration set lakes in North and South Dakota and Saskatchewan (r2 = 0.83 standard error — 0.481 In salinity, ref. 11). Equations used for the WA calculations are given in the text (eq 3). (Reproduced with permission from reference 48. Copyright 1991 McMillan Publishing Co.)...

Figure 12. Graphs of observed versus diatom-inferred total phosphorus concentrations (TP) and observed minus diatom-inferred TP (i.e., a residual analysis) are based on weighted averaging regression and calibration models and classical deshrinking. The large circles indicate two coincident values. This analysis is discussed in detail in reference 46.

An index of toxicity is intended to be a simple tool that allows integrating and summarizing several variables into a single value. Realistically, this cannot be inferred without a judgement by environmental protection experts who consider all parameters available for their classification. PLS regression helped calculate an index fitted to expert judgement. The loss of information owing to the transformation of a multivariate situation to a univariate one was thus minimized since it is an inherent characteristic of multivariate analytical tools. 

---