Statnote 4: what if the data are not normal?

When testing the difference between two groups, if previous data indicate non-normality, then either transform the data if they comprise percentages, integers or scores or use a non-parametric test. If there is uncertainty whether the data are normally distributed, then deviations from normality are likely to be small if the data are measurements to three significant figures. Unless there is clear evidence that the distribution is non-normal, it is more efficient to use the conventional t-tests. It is poor statistical practice to carry out both the parametric and non-parametric tests on a set of data and then choose the result that is most convenient to the investigator!

Statnote 5: Is one set of data more variable than another?

There may be circumstances where it is necessary for microbiologists to compare variances rather than means, e,g., in analysing data from experiments to determine whether a particular treatment alters the degree of variability or testing the assumption of homogeneity of variance prior to other statistical tests. All of the tests described in this Statnote have their limitations. Bartlett’s test may be too sensitive but Levene’s and the Brown-Forsythe tests also have problems. We would recommend the use of the variance-ratio test to compare two variances and the careful application of Bartlett’s test if there are more than two groups. Considering that these tests are not particularly robust, it should be remembered that the homogeneity of variance assumption is usually the least important of those considered when carrying out an ANOVA. If there is concern about this assumption and especially if the other assumptions of the analysis are also not likely to be met, e.g., lack of normality or non additivity of treatment effects then it may be better either to transform the data or to carry out a non-parametric test on the data.

Statnote 14: the correlation of two variables (Pearson's 'r')

Pearson's correlation coefficient (‘r’) is one of the most widely used of all statistics. Nevertheless, care needs to be used in interpreting the results because with large numbers of observations, quite small values of ‘r’ become significant and the X variable may only account for a small proportion of the variance in Y. Hence, ‘r squared’ should always be calculated and included in a discussion of the significance of ‘r’. The use of ‘r’ also assumes that the data follow a bivariate normal distribution (see Statnote 17) and this assumption should be examined prior to the study. If the data do not conform to such a distribution, the use of a non-parametric correlation coefficient should be considered. A significant correlation should not be interpreted as indicating ‘causation’ especially in observational studies, in which the two variables may be correlated because of their mutual correlations with other confounding variables.

Data analysis methods in optometry. Part 5: correlation

1. Pearson's correlation coefficient only tests whether the data fit a linear model. With large numbers of observations, quite small values of r become significant and the X variable may only account for a minute proportion of the variance in Y. Hence, the value of r squared should always be calculated and included in a discussion of the significance of r. 2. The use of r assumes that a bivariate normal distribution is present and this assumption should be examined prior to the study. If Pearson's r is not appropriate, then a non-parametric correlation coefficient such as Spearman's rs may be used. 3. A significant correlation should not be interpreted as indicating causation especially in observational studies in which there is a high probability that the two variables are correlated because of their mutual correlations with other variables. 4. In studies of measurement error, there are problems in using r as a test of reliability and the ‘intra-class correlation coefficient’ should be used as an alternative. A correlation test provides only limited information as to the relationship between two variables. Fitting a regression line to the data using the method known as ‘least square’ provides much more information and the methods of regression and their application in optometry will be discussed in the next article.

Statistical analysis in microbiology:statnotes

This book is aimed primarily at microbiologists who are undertaking research and who require a basic knowledge of statistics to analyse their experimental data. Computer software employing a wide range of data analysis methods is widely available to experimental scientists. The availability of this software, however, makes it essential that investigators understand the basic principles of statistics. Statistical analysis of data can be complex with many different methods of approach, each of which applies in a particular experimental circumstance. Hence, it is possible to apply an incorrect statistical method to data and to draw the wrong conclusions from an experiment. The purpose of this book, which has its origin in a series of articles published in the Society for Applied Microbiology journal ‘The Microbiologist’, is an attempt to present the basic logic of statistics as clearly as possible and therefore, to dispel some of the myths that often surround the subject. The 28 ‘Statnotes’ deal with various topics that are likely to be encountered, including the nature of variables, the comparison of means of two or more groups, non-parametric statistics, analysis of variance, correlating variables, and more complex methods such as multiple linear regression and principal components analysis. In each case, the relevant statistical method is illustrated with examples drawn from experiments in microbiological research. The text incorporates a glossary of the most commonly used statistical terms and there are two appendices designed to aid the investigator in the selection of the most appropriate test.

The accuracy of parameter estimation from noisy data, with application to resonance peak estimation in distributed Brillouin sensing

Distributed Brillouin sensing of strain and temperature works by making spatially resolved measurements of the position of the measurand-dependent extremum of the resonance curve associated with the scattering process in the weakly nonlinear regime. Typically, measurements of backscattered Stokes intensity (the dependent variable) are made at a number of predetermined fixed frequencies covering the design measurand range of the apparatus and combined to yield an estimate of the position of the extremum. The measurand can then be found because its relationship to the position of the extremum is assumed known. We present analytical expressions relating the relative error in the extremum position to experimental errors in the dependent variable. This is done for two cases: (i) a simple non-parametric estimate of the mean based on moments and (ii) the case in which a least squares technique is used to fit a Lorentzian to the data. The question of statistical bias in the estimates is discussed and in the second case we go further and present for the first time a general method by which the probability density function (PDF) of errors in the fitted parameters can be obtained in closed form in terms of the PDFs of the errors in the noisy data.

Statistical guidelines for clinical studies of human vision

Citation information: Armstrong RA, Davies LN, Dunne MCM & Gilmartin B. Statistical guidelines for clinical studies of human vision. Ophthalmic Physiol Opt 2011, 31, 123-136. doi: 10.1111/j.1475-1313.2010.00815.x ABSTRACT: Statistical analysis of data can be complex and different statisticians may disagree as to the correct approach leading to conflict between authors, editors, and reviewers. The objective of this article is to provide some statistical advice for contributors to optometric and ophthalmic journals, to provide advice specifically relevant to clinical studies of human vision, and to recommend statistical analyses that could be used in a variety of circumstances. In submitting an article, in which quantitative data are reported, authors should describe clearly the statistical procedures that they have used and to justify each stage of the analysis. This is especially important if more complex or 'non-standard' analyses have been carried out. The article begins with some general comments relating to data analysis concerning sample size and 'power', hypothesis testing, parametric and non-parametric variables, 'bootstrap methods', one and two-tail testing, and the Bonferroni correction. More specific advice is then given with reference to particular statistical procedures that can be used on a variety of types of data. Where relevant, examples of correct statistical practice are given with reference to recently published articles in the optometric and ophthalmic literature.

Environmental quality in residential areas : a social survey approach to improvement priorities

The research is concerned with the measurement of residents' evaluations of the environmental quality of residential areas. The research reflects the increased attention being given to residents' values in planning decisions affecting the residential environment. The work was undertaken in co-operation with a local authority which was in the process of revising its housing strategy, and in particular the priorities for improvement action. The study critically examines the existing evidence on environmental values and their relationship to the environment and points to a number of methodological and conceptual deficiencies. The research strategy developed on the basis of the research review was constrained by the need to keep any survey methods simple so that they could easily be repeated, when necessary, by the sponsoring authority. A basic perception model was assumed, and a social survey carried out to measure residents' responses to different environmental conditions. The data was only assumed to have ordinal properties, necessitating the extensive use of non-parametric statistics. Residents' expressions of satisfaction with the component elements of the environment (ranging from convenience to upkeep and privacy) were successfully related to 'objective' measures of the environment. However the survey evidence did not justify the use of the 'objective' variables as environmental standards. A method of using the social survey data directly as an aid to decision-making is discussed. Alternative models of the derivation of overall satisfaction with the environment are tested, and the values implied by the additive model compared with residents' preferences as measured directly in the survey. Residents' overall satisfactions with the residential environment were most closely related to their satisfactions with the "Appearance" and the "Reputation" of their areas. By contrast the most important directly measured preference was "Friendliness of area". The differences point to the need to define concepts used in social research clearly in operational terms, and to take care in the use of values 'measured' by different methods.

Nonlinear speech analysis algorithms mapped to a standard metric achieve clinically useful quantification of average Parkinson's disease symptom severity

The standard reference clinical score quantifying average Parkinson's disease (PD) symptom severity is the Unified Parkinson's Disease Rating Scale (UPDRS). At present, UPDRS is determined by the subjective clinical evaluation of the patient's ability to adequately cope with a range of tasks. In this study, we extend recent findings that UPDRS can be objectively assessed to clinically useful accuracy using simple, self-administered speech tests, without requiring the patient's physical presence in the clinic. We apply a wide range of known speech signal processing algorithms to a large database (approx. 6000 recordings from 42 PD patients, recruited to a six-month, multi-centre trial) and propose a number of novel, nonlinear signal processing algorithms which reveal pathological characteristics in PD more accurately than existing approaches. Robust feature selection algorithms select the optimal subset of these algorithms, which is fed into non-parametric regression and classification algorithms, mapping the signal processing algorithm outputs to UPDRS. We demonstrate rapid, accurate replication of the UPDRS assessment with clinically useful accuracy (about 2 UPDRS points difference from the clinicians' estimates, p < 0.001). This study supports the viability of frequent, remote, cost-effective, objective, accurate UPDRS telemonitoring based on self-administered speech tests. This technology could facilitate large-scale clinical trials into novel PD treatments.

Absorptive capacity and the benefits from global reservoirs of knowledge:evidence from a linked China-OECD dataset

This paper investigates the role of absorptive capacity in the diffusion of global technology with sector and firm heterogeneity. We construct the FDI-intensity weighted global R&D stock for each industry and link it to Chinese firm-level panel data relating to 53,981 firms over the period 2001-2005. Non-parametric frontier analysis is employed to explore how absorptive capacity affects technical change and catch-up in the presence of global knowledge spillovers. We find that R&D activities and training at individual firms serve as an effective source of absorptive capability. The contribution of absorptive capacity varies according to the type of FDI and the extent of openness.

Is a little sunshine all we need? On the impact of sunshine regulation on profits, productivity and prices in the Dutch drinking water sector

This paper analyzes the performance of Dutch drinking water utilities before and after the introduction of sunshine regulation, which involves publication of the performance of utilities but no formal price regulation. By decomposing profit change into its economic drivers, our results suggest that, in the Dutch political and institutional context, sunshine regulation was effective in improving the productivity of publicly organised services. Nevertheless, while sunshine regulation did bring about a moderate reduction in water prices, sustained and substantial economic profits suggest that it may not have the potential to fully align output prices with economic costs in the long run. In methodological terms, the DEA based profit decomposition is extended to robust and conditional non-parametric efficiency measures, so as to account better for both uncertainty and differences in operating environment between utilities.

Selectivity and timing performance of UK investment trusts

This study examines the selectivity and timing performance of 218 UK investment trusts over the period July 1981 to June 2009. We estimate the Treynor and Mazuy (1966) and Henriksson and Merton (1981) models augmented with the size, value, and momentum factors, either under the OLS method adjusted with the Newey-West procedure or under the GARCH(1,1)-in-mean method following the specification of Glosten et al. (1993; hereafter GJR-GARCH-M). We find that the OLS method provides little evidence in favour of the selectivity and timing ability, consistent with previous studies. Interestingly, the GJR-GARCH-M method reverses this result, showing some relatively strong evidence on favourable selectivity ability, particularly for international funds, as well as favourable timing ability, particularly for domestic funds. We conclude that the GJR-GARCH-M method performs better in evaluating fund performance compared with the OLS method and the non-parametric approach, as it essentially accounts for the time-varying characteristics of factor loadings and hence obtains more reliable results, in particular, when the high frequency data, such as the daily returns, are used in the analysis. Our results are robust to various in-sample and out-of-sample tests and have valuable implications for practitioners in making their asset allocation decisions across different fund styles. © 2012 Elsevier B.V.

Statnote 34 : the Kolmogorov-Smirnov (KS) test

The Kolmogorov-Smirnov (KS) test is a non-parametric test which can be used in two different circumstances. First, it can be used as an alternative to chi-square (?2) as a ‘goodness-of-fit’ test to compare whether a given ‘observed’ sample of observations conforms to an ‘expected’ distribution of results (KS, one-sample test). An example of the use of the one-sample test to determine whether a sample of observations was normally distributed was described previously. Second, it can be used as an alternative to the Mann-Whitney test to compare two independent samples of observations (KS, two-sample test). Hence, this statnote describes the use of the KS test with reference to two scenarios: (1) to compare the observed frequency (Fo) of soil samples containing cysts of the protozoan Naegleria collected each month for a year with an expected equal frequency (Fe) across months (one-sample test), and (2) to compare the abundance of bacteria on cloths and sponges sampled in a domestic kitchen environment (two-sample test).

A framework for performance evaluation of employment offices:a case of Tunisia

This paper proposes a new framework for evaluating the performance of employment offices based on non-parametric technique of data envelopment analysis. This framework is explained using the assessment of technical efficiency of 82 employment offices in Tunisia which are under the direction of the National Agency for Employment and Independent Work. We further investigated the exogenous factors that may explain part of the variation in efficiency scores using a bootstrapping approach in period January 2006 to December 2008. Given the specialisation of employment offices, we used the proposed approach for the efficiency evaluation of graduate employment offices and multi-services employment offices, separately.

An aggregate measure of financial ratios using a multiplicative DEA model

This paper examines the problems in the definition of the General Non-Parametric Corporate Performance (GNCP) and introduces a multiplicative linear programming as an alternative model for corporate performance. We verified and tested a statistically significant difference between the two models based on the application of 27 UK industries using six performance ratios. Our new model is found to be a more robust performance model than the previous standard Data Envelopment Analysis (DEA) model.