In the following, we will explain several important matters:

- 7-minute lecture
- Use of i-Regand
- Topic outline

At a surface level, it’s possible to simply skim all the topics in this user guide in a fairly short time. But in practice it won’t be of much use. We recommend to read topics 1 to 3 first, and then try out the application as suggested in topic 2. After you are familiar with the basic operations of i-Regand, you can pick a topic that’s relevant to your needs and make use of i-Regand accordingly.

## 7-minute lecture

You can read a topic in this user guide in about seven minutes. We only provide a short explanation of what you need to prepare and what to do. Afterwards, you can try out all the menus in this Apps.

We assume that you have basic knowledge about regression analysis, or at the very least you understand what you need from regression analysis. We do not provide any notes about the various regression analysis methods, because such knowledge is already available in textbooks, journal articles, and other reference materials. In every topic we provide a bibliography listing the sources we referenced in developing this app. Of course you can also make use of other reference materials.

## What i-Regand can do for you

The various methods of regression analysis is used in nearly all occupation and fields of science. Regression analysis as applied in the field of economy and business is termed econometrics. It is also used in the life sciences, especially in computational biology. Nearly every field of science, from psychology, medicine, to engineering, makes use of regression analysis under various names. Considering the different terms used we adopted a standard name accepted by all fields, regression method. The name ‘i-Regand’ was taken from the term ‘regression analysis’ (regan). The final letter D was added as ‘Regan’ is a common personal name while the letter I was added at the beginning to impress the fact that this app is a mobile application that is usable anywhere and anytime.

Of course the field of regression method is very wide. It is difficult to create an app that can provide all the standard methods used in every field. As our background is in economics, there is an impression that the terminology and methods used in this app tends more towards econometrics. This is unavoidable, but in later versions we will add more regression methods used outside the field of econometrics.

In this version i-Regand provides computation capabilities for processing of cross-section, time series, panel (pooled), and binary data. The regression methods we provide are chosen from the most popular and commonly used methods among all fields, including OLS, WLS, 2SLS, W2SLS, VAR, VECM, ARCH-GARCH, Fixed and Random Effects, Probit, and Logit. We’ve even provided several methods that aren’t available in desktop software menus, such as threshold regression and FAVAR. At the end of this topic we provide a highlights section for each analysis method.

You might be surprised to find out that the computation accuracy and speed of i-Regand rivals that of specialized statistics and econometric software on a desktop platform. This is because we developed special algorithms that place less burden on a smart-phone CPU. While the speed of computation is reliant on smartphone processor speed and RAM size, the accuracy remains the same.

Our application can also handle datasets of various size, from small to large. An analysis with thousands of observations and tens of variables can still be handled by our software. In comparison, for desktop software increasing the size of the dataset also increases the time needed for computation.

Along with speed and accuracy i-Regand also makes inputing and reading output very easy. The menu interface is compact and finger-friendly, not just user-friendly. For those who have issues in reading from a small smartphone display, we also provide a rotate mode and zoom. This application is very ‘eyes friendly’.

## Outline

Topic 2: Must read, project sheet and OLS. For those of you who are already familiar with statistics or econometric software, you might only need to skim this topic. Here you will be guided to perform a regression analysis from data retrieval, model estimation, displaying the output, performing diagnostics, and robust standard error. OLS estimation is used as example, and you can follow the guide step-by-step. Once you are familiar with the OLS procedure it will be easy to perform other procedures, as the menus are very similar.

Topic 3: Data preparation. As i-Regand is focused on regression analysis estimations, we do not provide the capability to input data manually. Preparing data must be done through spreadsheet software such as Excel, either on a desktop or on a mobile device. I-Regand can read csv (comma delimited), text (txt) and xlsx file formats.

Topic 4: WLS. Weighted least square is commonly used in overcoming heteroskedasticity, and i-Regand provides a simple menu to perform WLS. All you need to do is define the appropriate weight variable and run it.

Topic 5: IV/2SLS. In i-Regand you can perform two-stage least square (2SLS) and instrumental variable (IV). As 2SLS is simply a special case of IVLS, the estimations are treated the same. We use the 2SLS procedure in performing IVLS estimations.

Topic 6: W2SLS. Weighted least squares can be used along with IV or 2SLS. Therefore, W2SLS is actually 3SLS.

Topic 7: Unit root test. Unit root test can be performed on level, first difference and second difference. We provide 2 of the most popular tests, Augmented Dickey Fuller (ADF) and Phillips-Perron(PP).

Topic 8: Cointegration test. For the cointegration test, we provide two types of tests: residual-based (ADF and PP) and rank test(Johansen). It should be noted that before performing the rank test you should determine the optimum lag length on VAR.

Topic 9: VAR. Vector autoregression (VAR) is very suited for analyses involving stationary time-series data. In i-Regand you can perform a VAR estimation and continue with impulse response function (IRF) and forecast error variance decomposition (FEVD).

Topic 10: VECM. Vector Error Correction Model (VECM) is the ‘twin’ of VAR for non-stationary data. In VECM you need to determine the number of cointegrating vectors (CV) or the rank of the cointegrating relations. As in VAR you can also perform IRF and FEVD analysis.

Topic 11: Binary regression. Sometimes we need to perform regression analysis with binary dependent variables. Two procedures are provided by i-Regand, Probit and Logit