Excel Analyse-it Introduction: Analyse-it is an add-in tool for Microsoft Excel that performs various parametric and nonparametric statistical analyses. Along with allowing its users to access some important standard and sophisticated statistical procedures, Analyse-it lets the user feel comfortable with the familiar and easy to use Microsoft Excel environment. This tool is compatible with all versions of Excel from 1997 and later. Some applications include analysis for descriptive and summary statistics and plots, ANOVA and regression, hypothesis testing, correlation, and Wilcoxon and Mann-Whitney tests. Analyse-it also contains another edition known as the Method Evaluation edition, which includes all of the applications of the standard package mentioned above along with other helpful analyses that can be used in clinical research. Some of these additional uses include method evaluation, validation, and demonstration for compliance, as well as weighted linear regression and CLSI Linearity. The extra topics found in Method Evaluation will not be discussed in this primer but can be further researched on Analyse-it’s homepage. As Analyse-it allows researchers to remain in the user-friendly and ever versatile Excel, users will be happy to know that Analyse-it prevents the necessity of exporting data to more complex statistical packages, since it is quite common for many of the social sciences to initialize research within the Excel framework. This benefit drastically reduces time in front of the computer as the user does not need to learn multiple software packages. Below we will see many practical uses for Analyse-it along with some demonstrations and print screens for using this package. These uses are summarized below: Descriptive and Summary Statistics Correlation Agreement Hypothesis Tests Regression Descriptive and Summary Statistics: Some of the many statistics included in Analyse-it are mean, median, standard deviation, confidence intervals, quartiles, and percentiles. Along with numerical summaries, graphical summaries are also included in this wonderful package. Some of these include box-whisker plots, mean-plots, percentile and standard deviation plots, normal probability plots, and frequency histograms with normal curve overlays. Analyse-it also allows the user to test and visualize normality with the Shapiro-Wilk test, Anderson Darling test, and Kolmogorov Smirnov tests. Figure 1a below shows an example of a histogram and boxplot. Figure 1b shows an example of a Normality Probability Plot. Figure 1a – Histogram and Boxplot Figure 1b – Normal Probability Plot Correlation: Analyse-it allows its users to explore correlations between variables with parametric and nonparametric tests and scatter plots. Such correlations include Pearson correlation coefficients, Spearman rank, and Kendall rank correlation statistics. Beyond computing the coefficients of correlation, Analyseit allows its users to test correlation degree and strength to determine if correlation exists between variables and how strong the correlation truly is. Analyse-it also provides plots to visualize potential correlation that exists between variables. Figure 2 below shows an example of a scatter plot between two highly, positively correlated variables, Age and Height. Figure 2 Agreement: For agreement, Analyse-it provides a rather complete implementation of Bland-Altman plots, along with Kappa and weighted Kappa for examining inter-rater agreement. Analyse-it shows bias and limits of agreement and allows the user to visualize agreement with difference and scatter plots. Figure 3a below shows an example of a Brand-Altman difference plot with bias and limits of agreement. Figure 3b shows an example of a Repeatability plot. Figure 3a Figure 3b More detailed plots for analytic method comparison can be seen in the Method Evaluation edition of Analyse-it Hypothesis Tests: A variety of tests to compare means, medians, errors, etc. are provided in Analyse-it. Some parametric methods include t-tests, F-tests, One-Way ANOVA, and Two-Way ANOVA. Non-parametric tests include Wilcoxon Signed Ranks test, Kruskal Wallis test, Friedman test, and Mann-Whitney test. Along with these global tests of differences, Analyse-it provides LSD, Bonferroni, Tukey, and Scheffe post-hoc multiple comparisons for their ANOVA tests. Beyond tests for differences in means, medians, and ranks, there are also tests for difference in proportion. These include Fisher exact test, McNemar test, and Chi-Square test. These differences can be given as relative-risk or odds-ratio where desired. Regression: When you purchase Analyse-it you are getting hooked up with single, multiple linear, and up to sixth order polynomial regression. You can also test which predictors significantly contribute to the response in order to determine which should be included in the model, and which should be removed. Beyond this, you can also use parameter estimates for prediction on future observations. Figure 4a below shows a scatterplot with statistics for a simple linear regression model. These statistics include coefficient values with confidence intervals and standard errors. Figure 4b shows an example of a Scatter plot which show fit along with confidence and prediction intervals. Finally, Figure 4c shows an example of a standardized residual plot. Finally, the user can perform an ANOVA test for fit to determine whether the predictors are a significant improvement over the mean for the model of the response and predictor variables. Figure 4a Figure 4b Figure 4c Conclusion: Performing the aforementioned function should be as simple as running any other add-in tool within the Excel Environment. Although I did not have access to this program, it appears to be as simple as any other Excel function with some point and click menus similar to SPSS. Anybody who is even a novice with Microsoft Excel should readily be able to extend their knowledge quickly and easily to Analyse-it. As we have seen, Analyse-it provides a convenient and practical way to explore and statistically analyze data in a convenient and user-friendly environment such as Excel. Although Analyse-it does not provide the same in depth analysis as a more sophisticated software package such as SAS or even SPSS, it provides the tools essential to explore a moderately sized set of data. The speed and accuracy of the procedures and results would need to be tested by acquiring the actual program and running a test. It would be useful to compare these results to a more well-known data package like SAS or R to ensure accuracy and compare speeds. Assuming everything runs properly, the advantages of not having to export and modify the data set to run standard statistical packages could be very useful, especially if the researcher already has several Excel files with data already in it. Should the standard Analyse-it package not include all the analysis needed to conduct research, it would be wise to check out the Method Evaluation edition, especially if clinical trial research is being performed and analyzed. Like the standard package of Analyse-it, the functions and analyses performed in the Method Evaluation edition should be a mere extension of everyday Microsoft Excel usage. Sources Used: http://www.analyse-it.com/products/standard/ http://en.wikipedia.org/wiki/Analyse-it