In statistical evaluation, residuals signify the distinction between noticed values and values predicted by a regression mannequin. These values are essential for assessing the match and validity of a mannequin. The TI-84 calculator offers functionalities to compute these residuals immediately, simplifying the evaluation course of. As an example, if an information level has an noticed y-value of 5 and the regression line predicts a y-value of 4, the residual for that time is 1 (5 – 4 = 1).
Calculating and analyzing these variations is vital for a number of causes. Residuals can reveal patterns within the knowledge which may point out the regression mannequin shouldn’t be acceptable. Analyzing the distribution of residuals can spotlight non-linearity, heteroscedasticity (unequal variance), or the presence of outliers, all of which might compromise the reliability of statistical inferences. Understanding these options is important for guaranteeing the validity of the mannequin.
The next sections element the steps concerned in computing residuals on the TI-84, decoding the outcomes, and using them to evaluate the adequacy of a linear regression mannequin. This course of begins with coming into knowledge into the calculator, performing a regression evaluation, after which accessing the saved residual values for additional examination.
1. Knowledge Entry
The method of calculating residuals on a TI-84 calculator essentially depends on correct knowledge entry. Inputting paired knowledge units (x-values and corresponding y-values) into the calculator’s lists (usually L1 and L2) constitutes the foundational step. Any error throughout this section propagates via subsequent calculations, resulting in incorrect residual values and a doubtlessly flawed interpretation of the regression mannequin’s match. For instance, if an information level (3, 7) is mistakenly entered as (3, 17), the calculated residuals will probably be considerably skewed, distorting any visible illustration and doubtlessly resulting in an inaccurate evaluation of the mannequin’s validity.
The right affiliation of x and y values is equally essential. A transposed entry (e.g., coming into x as y and vice-versa for a single knowledge level) will invalidate the whole residual evaluation. The TI-84 operates on the premise that knowledge in corresponding positions throughout the designated lists signify paired observations. Moreover, the variety of knowledge factors entered should be constant between the x and y lists. Discrepancies will both stop the regression evaluation from working or produce misguided outcomes. Scrupulous consideration to element throughout this section is crucial; many errors in regression evaluation might be traced again to incorrect knowledge enter.
In abstract, meticulous knowledge entry shouldn’t be merely a preliminary step however an integral part of the whole residual calculation course of on a TI-84. The accuracy of the enter immediately influences the validity of the calculated residuals and, consequently, the conclusions drawn concerning the appropriateness of the chosen regression mannequin. Challenges on this stage spotlight the necessity for verification procedures, corresponding to double-checking knowledge towards the unique supply and using the calculator’s modifying options to appropriate any errors earlier than continuing with the evaluation.
2. Regression Command
The regression command executed on a TI-84 calculator is instrumental within the course of. This command initiates the calculation of the regression equation, which then serves as the premise for figuring out the anticipated y-values. Consequently, the accuracy and appropriateness of the chosen regression command (e.g., LinReg(ax+b), QuadReg) immediately impression the ensuing residuals. The selection of an incorrect command, corresponding to utilizing a linear regression when the info displays a curvilinear relationship, will produce residuals that don’t precisely replicate the mannequin’s match. These residuals would possible show a sample, indicating a violation of the linearity assumption of linear regression and rendering the calculated residuals deceptive.
Contemplate a state of affairs the place knowledge factors representing the connection between time and distance traveled by a decelerating object are entered. If a linear regression command is used, the ensuing residuals will possible present a curved sample, demonstrating that the linear mannequin doesn’t adequately seize the true relationship. Conversely, choosing an acceptable command, corresponding to a quadratic regression, would produce residuals which can be randomly distributed round zero, suggesting a greater mannequin match. The TI-84 shops the coefficients derived from the chosen regression command. These coefficients are then used to calculate the anticipated y-values for every corresponding x-value, that are then subtracted from the noticed y-values to find out the residuals. The “RESID” record on the TI-84 holds these residual values.
In abstract, the regression command is a essential determinant of the validity of subsequent residual calculations. The command should align with the underlying relationship current within the knowledge; in any other case, the residuals will probably be systematically biased, resulting in misguided conclusions concerning the mannequin’s appropriateness. Due to this fact, choosing the proper regression command is a basic step when looking for to compute significant residuals on a TI-84 calculator, or when participating in any statistical evaluation involving residuals.
3. Residual Storage
Residual storage is an integral part within the technique of calculating and analyzing residuals on a TI-84 calculator. With out the capability to retailer these calculated values, additional evaluation and interpretation can be considerably hindered. Residual storage allows examination of the distribution, patterns, and total magnitude of the variations between noticed and predicted values, important for mannequin validation.
-
The RESID Checklist
Following the execution of a regression command on the TI-84, the calculated residuals are routinely saved in a delegated record named “RESID.” This record serves as the first repository for residual values. The power to entry and manipulate knowledge throughout the RESID record is key to subsequent analytical steps. With out this computerized storage, residuals would should be manually calculated and entered, a course of each time-consuming and susceptible to error.
-
Facilitating Residual Plots
Probably the most informative strategies for assessing the adequacy of a regression mannequin entails producing residual plots. Residual plots graphically show the residuals towards the corresponding x-values or predicted y-values. To create these plots on the TI-84, the calculator accesses the saved residuals from the RESID record. This direct entry permits for speedy technology of visible representations, which might reveal patterns corresponding to non-linearity, heteroscedasticity, or the presence of outliers, indicators of mannequin inadequacy.
-
Enabling Statistical Calculations
Past visible inspection, the saved residuals within the RESID record might be subjected to varied statistical calculations. As an example, one may calculate the sum of squared residuals (SSR), a measure of the general error within the mannequin. Moreover, the usual deviation of the residuals might be computed to evaluate the standard magnitude of the prediction errors. These calculations, facilitated by the storage of residuals, present quantitative measures of mannequin match, complementing the qualitative insights gained from residual plots.
-
Diagnostic Capabilities
The RESID record shouldn’t be solely a storage location but in addition a key component in diagnostic assessments. If patterns are noticed within the residual plots or if statistical calculations reveal uncommon values, the RESID record permits for the identification of particular knowledge factors contributing to those anomalies. By cross-referencing the residual values with the unique knowledge, potential outliers or influential observations might be pinpointed and additional investigated. This diagnostic functionality is essential for refining the mannequin and enhancing its predictive accuracy.
In conclusion, the efficient calculation and use of residuals on a TI-84 are immediately linked to the calculator’s residual storage capability. The automated storage throughout the RESID record not solely streamlines the analytical course of but in addition allows a complete evaluation of mannequin match via each visible and statistical strategies. With out this storage performance, most of the diagnostic advantages derived from residual evaluation can be considerably diminished.
4. Checklist Entry
Checklist entry on the TI-84 calculator is crucial for interacting with and decoding the residuals generated after regression evaluation. Accessing these lists shouldn’t be merely a technical step however an integral a part of understanding the mannequin’s efficiency and figuring out potential points with the info or the mannequin itself.
-
Accessing the RESID Checklist
The “RESID” record, the place residuals are routinely saved after a regression calculation, should be immediately accessed to view the residual values. The TI-84 offers a number of pathways to perform this. For instance, the “STAT” menu, adopted by “EDIT,” permits the consumer to view the contents of any record, together with “RESID.” This direct entry is essential as a result of it permits the consumer to look at particular person residual values, establish outliers, and customarily assess the unfold of the residuals. If, as an illustration, a residual worth is considerably bigger than the others, it could point out an outlier requiring additional investigation.
-
Creating Residual Plots
Visible illustration of residuals is commonly extra informative than inspecting uncooked numerical values. The TI-84 permits the creation of scatter plots with the x-values (L1) on the x-axis and the residuals (RESID) on the y-axis. This requires accessing each the L1 record and the RESID record via the “STAT PLOT” menu. By plotting the residuals towards the unbiased variable, patterns corresponding to non-linearity or heteroscedasticity might be readily recognized. The absence of a sample suggests the linear mannequin is suitable, whereas the presence of a discernible sample signifies the necessity for mannequin refinement or transformation.
-
Performing Calculations on Residuals
Past visible inspection, record entry allows the efficiency of statistical calculations on the residuals themselves. As an example, the sum of squared residuals (SSR) might be computed by accessing the RESID record and making use of the suitable statistical features (obtainable beneath the “LIST” menu, then “MATH”). A decrease SSR signifies a greater match of the mannequin to the info. Equally, calculating the usual deviation of the residuals offers a measure of the standard dimension of the prediction errors. These calculations require the power to specify the RESID record because the enter variable for the statistical features.
-
Utilizing Residuals in Additional Modeling
In some superior statistical methods, residuals from an preliminary mannequin could also be used as enter variables in a subsequent mannequin. This strategy is perhaps employed when addressing points like autocorrelation in time collection knowledge. Accessing the RESID record permits these residual values to be copied into a brand new record to be used as a predictor variable in a refined mannequin. The power to simply switch residual values is crucial for iterative mannequin constructing and refinement.
The capability to entry and manipulate lists is subsequently central to successfully making use of the ideas related to “how one can calculate residuals on ti 84”. Checklist entry is required not solely to look at the residuals themselves but in addition to create visualizations, carry out statistical calculations, and doubtlessly use the residuals as inputs in additional modeling phases, underlining its position in regression evaluation.
5. Residual Plotting
Residual plotting is intrinsically linked to the method of computing residuals, serving as a visible diagnostic device that permits evaluation of the appropriateness of a fitted regression mannequin. The creation of a residual plot follows immediately from the calculation of residuals; it entails plotting these residuals towards both the unbiased variable or the anticipated values. The patterns noticed within the residual plot immediately inform the validity of the assumptions underlying the regression mannequin. If the residuals are randomly scattered round zero, this means that the linear mannequin is an effective match for the info. Nevertheless, systematic patterns, corresponding to a curved form or growing variance, point out that the linear mannequin is insufficient and that different fashions or knowledge transformations must be thought of.
Contemplate a state of affairs the place a linear regression mannequin is fitted to knowledge exhibiting a curvilinear relationship. The ensuing residual plot would possible show a U-shaped or inverted U-shaped sample. This sample demonstrates that the linear mannequin systematically overestimates some values and underestimates others, a transparent violation of the linearity assumption. Equally, if the variance of the residuals will increase as the anticipated values enhance (heteroscedasticity), the residual plot would exhibit a funnel form. This means that the mannequin’s prediction accuracy varies relying on the magnitude of the unbiased variable. The identification of such patterns is essential for guaranteeing the reliability of statistical inferences drawn from the regression mannequin. With out residual plotting, these potential points could stay undetected, resulting in deceptive conclusions.
In abstract, residual plotting is an indispensable step within the “how one can calculate residuals on ti 84” course of. It transforms numerical residuals into a visible format, facilitating the detection of patterns and anomalies which will compromise the validity of the regression evaluation. The power to interpret residual plots is, subsequently, a core ability for anybody using regression fashions for knowledge evaluation. Challenges on this space spotlight the necessity for improved understanding of statistical assumptions and diagnostic methods to make sure the correct and dependable utility of regression strategies.
6. Statistical Validity
Statistical validity, within the context of regression evaluation, hinges upon the success of sure assumptions concerning the knowledge and the error time period. The method of calculating residuals is key to assessing whether or not these assumptions are met, and subsequently, in figuring out the statistical validity of the ensuing mannequin. The calculated residuals function diagnostics to disclose deviations from these assumptions.
-
Linearity Evaluation
One key assumption for linear regression is that the connection between the unbiased and dependent variables is linear. Residuals present a way to check this assumption. If a plot of residuals towards the unbiased variable displays a discernible sample (e.g., a curve or a funnel form), it means that the linearity assumption is violated. Failure to fulfill this assumption compromises the statistical validity of the mannequin, doubtlessly resulting in biased coefficient estimates and unreliable predictions. Correct residual calculation permits for the visible and statistical detection of non-linearity.
-
Homoscedasticity Analysis
Homoscedasticity, or fixed variance of the error time period, is one other essential assumption. A violation of this assumption (heteroscedasticity) can result in inefficient estimates and inaccurate commonplace errors. Residuals are used to evaluate whether or not the variance of the errors is fixed throughout all ranges of the unbiased variable. A residual plot displaying a non-constant unfold of residuals signifies heteroscedasticity. Figuring out this challenge via residual evaluation is essential for guaranteeing the statistical validity of the regression mannequin and for choosing acceptable remedial measures, corresponding to weighted least squares.
-
Independence of Errors
The belief of unbiased errors stipulates that the error phrases for various observations are uncorrelated. Serial correlation, a violation of this assumption, can come up in time collection knowledge or when observations are clustered. Residuals are used to detect serial correlation via methods such because the Durbin-Watson check or by inspecting patterns in a time-ordered plot of the residuals. Failure to account for serial correlation can inflate the importance of regression coefficients and result in deceptive inferences. Correct calculation and evaluation of residuals are, subsequently, important for validating the independence assumption.
-
Normality of Errors
Whereas not strictly crucial for unbiased coefficient estimates, the belief of usually distributed errors is vital for speculation testing and confidence interval development. Residuals present a way to evaluate the normality assumption via histograms, regular likelihood plots, or statistical exams just like the Shapiro-Wilk check. Vital deviations from normality can have an effect on the accuracy of p-values and confidence intervals, thus impacting the statistical validity of inferences drawn from the regression mannequin. Analyzing the distribution of residuals helps to find out whether or not the normality assumption is affordable.
In abstract, the method of calculating residuals shouldn’t be merely a mathematical train, however a essential step in assessing the statistical validity of a regression mannequin. Correct calculation and cautious evaluation of residuals present insights into the extent to which the underlying assumptions of the mannequin are met. Failure to deal with violations of those assumptions can result in flawed conclusions and unreliable predictions, underscoring the significance of residual evaluation in guaranteeing the integrity of statistical inference.
7. Error Identification
Error identification is a direct consequence of calculating residuals and inspecting their properties. Residuals, representing the distinction between noticed and predicted values, function indicators of the mannequin’s predictive accuracy for every knowledge level. When residuals are giant or exhibit systematic patterns, they sign potential errors both within the mannequin specification, the info itself, or violations of the assumptions underlying the regression evaluation. For instance, a single giant residual may level to an outliera knowledge level that deviates considerably from the general development. Systematic patterns, corresponding to a curved relationship in a residual plot, point out that the linear mannequin is inappropriate and a non-linear mannequin is perhaps extra appropriate.
The sensible significance of error identification via residual evaluation is appreciable. In a monetary modeling context, giant residuals may reveal fraudulent transactions or inaccurate knowledge entries. In scientific analysis, figuring out patterns in residuals can result in the invention of beforehand unacknowledged elements influencing the phenomenon beneath examine. Contemplate a examine modeling crop yield based mostly on rainfall; a sample within the residuals associated to soil kind may point out the necessity to embrace soil traits as further predictors within the mannequin. The power to pinpoint such errors and omissions facilitates extra strong and correct modeling, resulting in extra dependable predictions and better-informed choices.
The reliance on correct residual calculations is essential. If residuals are computed incorrectly, error identification efforts change into futile and doubtlessly deceptive. Challenges in precisely calculating residuals can stem from knowledge entry errors, incorrect utility of the regression command, or improper dealing with of information transformations. Guaranteeing the proper implementation of those procedures is, subsequently, a prerequisite for efficient error identification and for guaranteeing the general validity of the regression evaluation. Understanding the connection between error identification and the way residuals are calculated on a TI-84 is central to reaching correct and dependable statistical inference.
8. Mannequin Evaluation
Mannequin evaluation represents an important stage in statistical evaluation, figuring out the diploma to which a developed mannequin precisely represents the underlying knowledge. Residual calculation kinds an indispensable a part of this evaluation, permitting for a rigorous examination of a mannequin’s match and predictive functionality.
-
General Match Analysis
The sum of squared residuals (SSR) is a basic metric for gauging a mannequin’s total match. Decrease SSR values point out a greater match, reflecting a smaller discrepancy between noticed and predicted values. Within the context of “how one can calculate residuals on TI-84”, precisely figuring out the residuals is a prerequisite for computing a significant SSR. As an example, if two competing fashions are evaluated, the one with the decrease SSR, derived from accurately calculated residuals, would usually be most well-liked. An inaccurate calculation would result in a flawed evaluation and doubtlessly the number of a suboptimal mannequin.
-
Assumption Validation
Mannequin evaluation additionally entails verifying that the underlying assumptions of the chosen statistical methodology are met. In linear regression, these embrace linearity, homoscedasticity, and independence of errors. Residuals play a pivotal position on this validation course of. A residual plot, constructed utilizing the calculated residuals, permits for visible evaluation of those assumptions. A curved sample within the residual plot, for instance, signifies a violation of the linearity assumption. Due to this fact, the correct calculation and subsequent plotting of residuals are important for guaranteeing the validity of the mannequin’s assumptions, and for deciding whether or not the mannequin might be reliably used.
-
Outlier Detection
Residual evaluation can spotlight the presence of outliers, knowledge factors that deviate considerably from the general sample. Giant residuals typically flag these outliers, which can signify knowledge entry errors, uncommon occasions, or observations from a special inhabitants. Figuring out and addressing outliers is essential for enhancing mannequin accuracy and stability. In econometrics, for instance, a big residual may point out an uncommon market occasion that requires separate consideration or the removing of the info level to stop undue affect on the mannequin. An misguided calculation may masks or misidentify these influential factors, compromising the integrity of the mannequin evaluation course of.
-
Predictive Accuracy Measurement
The foundation imply squared error (RMSE), computed utilizing the residuals, offers a measure of the mannequin’s predictive accuracy. It represents the standard magnitude of the prediction errors. A decrease RMSE signifies higher predictive accuracy. The accuracy of the RMSE calculation immediately will depend on the correct dedication of the residuals. In advertising and marketing analytics, for instance, a mannequin predicting buyer churn with a decrease RMSE, based mostly on accurately calculated residuals, can be extra worthwhile for figuring out at-risk prospects than a mannequin with a better RMSE. Any errors in calculating residuals will immediately translate to inaccuracies within the RMSE and subsequently within the evaluation of the mannequin’s predictive capabilities.
Every part underscores the basic position of exact residual calculation in mannequin evaluation. The act of assessing hinges on dependable knowledge from how residuals are derived. Correct evaluation helps choosing and refining statistical fashions for enhanced analytical accuracy.
9. Diagnostic Instrument
The potential to calculate residuals on a TI-84 calculator transforms the gadget right into a diagnostic device for evaluating the adequacy of regression fashions. By quantifying the distinction between noticed and predicted values, the residuals provide a lens via which the validity of mannequin assumptions and the presence of outliers might be examined. This diagnostic perform is paramount in guaranteeing the reliability of statistical inferences derived from regression evaluation. With out the power to readily calculate and analyze residuals, assessing the appropriateness of a selected mannequin can be considerably compromised. As an example, in environmental science, a regression mannequin is perhaps used to foretell air pollution ranges based mostly on industrial output. The calculated residuals present insights into the mannequin’s accuracy and might reveal the presence of surprising occasions, corresponding to gear malfunctions, that considerably impression air pollution ranges and deviate from the anticipated development.
Residuals present a diagnostic instrument for evaluating the assumptions basic to regression evaluation. The homoscedasticity assumption, for instance, requires that the variance of the errors is fixed throughout all ranges of the unbiased variable. By plotting the residuals towards the anticipated values, a researcher can visually assess whether or not this assumption is violated. A funnel form within the residual plot signifies heteroscedasticity, implying that the mannequin’s prediction accuracy varies relying on the magnitude of the unbiased variable. The correct dedication of residuals is thus important for figuring out violations of those assumptions and guiding the number of acceptable remedial measures, corresponding to knowledge transformations or the usage of strong commonplace errors. This diagnostic perform enhances the statistical rigor of the evaluation. A flawed residual calculation undermines this diagnostic functionality, resulting in doubtlessly incorrect conclusions concerning the mannequin’s validity.
In abstract, the capability to calculate residuals on a TI-84 is indispensable for rendering the calculator an efficient diagnostic device in regression evaluation. This diagnostic functionality allows evaluation of mannequin match, validation of underlying assumptions, and detection of outliers, thereby contributing to the reliability and accuracy of statistical inferences. Any errors within the residual calculation course of will immediately undermine the worth of this diagnostic device. The power to evaluate these is core to how efficient that diagnostic device is.
Continuously Requested Questions
This part addresses frequent inquiries and misconceptions concerning the calculation of residuals utilizing the TI-84 calculator, offering readability on this vital statistical course of.
Query 1: What constitutes a residual within the context of regression evaluation?
A residual represents the distinction between an noticed worth and the worth predicted by the regression mannequin for that remark. It quantifies the error within the mannequin’s prediction for a selected knowledge level.
Query 2: How does one guarantee knowledge is precisely entered into the TI-84 for residual calculation?
Knowledge accuracy is paramount. Paired knowledge units (x and y values) should be entered meticulously into the TI-84’s lists (usually L1 and L2). Double-checking entries towards the unique knowledge supply is suggested.
Query 3: Which regression command on the TI-84 is suitable for linear knowledge?
For linear knowledge, the “LinReg(ax+b)” command is usually used. This command calculates the linear regression equation (y = ax + b) and shops the coefficients (a and b) for subsequent residual calculation.
Query 4: The place are the residuals saved on the TI-84 after working a regression command?
Following the execution of a regression command, the calculated residuals are routinely saved in a delegated record named “RESID.” This record might be accessed for additional evaluation and plotting.
Query 5: What does a residual plot reveal concerning the adequacy of a linear regression mannequin?
A residual plot, plotting residuals towards both the unbiased variable or predicted values, offers a visible evaluation of the mannequin’s assumptions. A random scatter of residuals round zero signifies an acceptable mannequin, whereas patterns recommend violations of the assumptions.
Query 6: What actions must be taken if a residual plot displays a non-random sample?
A non-random sample within the residual plot means that the linear mannequin shouldn’t be acceptable for the info. Contemplate different fashions, corresponding to polynomial regression, or knowledge transformations to deal with the non-linearity.
Correct residual calculation, coupled with knowledgeable interpretation, is essential for validating regression fashions and guaranteeing the reliability of statistical conclusions. Residuals are important for refining fashions.
The next part will discover frequent errors encountered throughout the “how one can calculate residuals on ti 84” course of and supply troubleshooting suggestions.
Important Ideas for Exact Residual Calculation on the TI-84
The correct dedication of residuals on a TI-84 calculator is paramount for legitimate statistical inference. Errors throughout the calculation course of compromise the reliability of the evaluation. The next suggestions define finest practices to make sure precision and accuracy.
Tip 1: Confirm Knowledge Entry: Scrutinize all knowledge entered into the TI-84’s lists. Errors in knowledge entry immediately translate into inaccurate residuals. Double-check every entry towards the unique knowledge supply earlier than continuing.
Tip 2: Choose the Acceptable Regression Command: The number of the proper regression command is crucial. Make sure the command aligns with the character of the connection between the variables. As an example, make the most of LinReg(ax+b) for linear relationships, and QuadReg for quadratic relationships.
Tip 3: Clear Earlier Residuals: Earlier than working a brand new regression evaluation, clear the “RESID” record. Retaining earlier residual values will contaminate subsequent calculations, resulting in misguided outcomes. This record might be cleared by way of the STAT Edit menu.
Tip 4: Retailer the Regression Equation: Storing the regression equation to Y1 permits for direct calculation of predicted values and facilitates residual plotting. Entry VARS, then Y-VARS, and choose Operate to retailer the equation.
Tip 5: Interpret Residual Plots Critically: Give attention to figuring out patterns or tendencies in residual plots. A random scatter of residuals signifies an excellent mannequin match, whereas patterns recommend violations of mannequin assumptions. Right interpretation facilitates mannequin refinement or transformation.
Tip 6: Calculate the Sum of Squared Residuals (SSR): Computing the SSR offers a quantitative measure of mannequin match. Decrease SSR values recommend a greater match. This calculation requires accessing the “RESID” record and making use of the sum and sq. features.
Tip 7: Verify for Outliers: Look at particular person residual values to establish potential outliers. A residual considerably bigger than others could point out an outlier requiring additional investigation or attainable removing from the info set.
By adhering to those pointers, practitioners can improve the accuracy and reliability of residual calculation on the TI-84, resulting in extra legitimate statistical conclusions.
The next part will deal with troubleshooting methods for frequent points encountered throughout residual evaluation.
Conclusion
The previous exploration has illuminated the methodology for calculating residuals on a TI-84 calculator. The method, whereas seemingly easy, necessitates a meticulous strategy encompassing correct knowledge entry, acceptable command choice, diligent storage, and insightful interpretation. The worth of accurately computing residuals transcends mere calculation, as it’s a essential component in assessing the validity and reliability of regression fashions.
The insights gained via residual analysisoutlier detection, validation of statistical assumptions, and analysis of total mannequin fitare very important for knowledgeable decision-making throughout various fields. Continued refinement of analytical expertise and a focus to element are important to leverage the total potential of the TI-84 as a diagnostic device in statistical evaluation.
