Time Sereis Analysis Using Stata

7/26/2019 Time Sereis Analysis Using Stata

1/26

Welcome to meo school of research

WWW.SAEEDMEO.BLOGSPOT.COM

Muhammad saeed AAS Khan Meo, Superior university Lahore Pakistan

PRAY FOR MY TEACHERS AND FAMILY VISIT MY BLOG FOR RESEAR TIPS AND ECONOMETRIC

TECHNIQUES [email protected]

TIME SEREIS ANALYSIS USING STATA

How to run regression using stata

Step #1:

Import data into STATA

Setp#2:

At first step, always set time otherwise u may get error, set time with the help of following

command

tsset years, yearly

step#3:

If u need to see summary of variables type in stata command bar

summarize CO2 GDP OIL fdi PP or simpley write sum (CO2 GDP OIL fdi PP

these are my variables)

For detail of data give this command

describe or list or br (these are three different commands)

step #4:

If u wishes to run correlation test then u may run by typing following command

correlate CO2 GDP OIL fdi

step#5:

If u wishes to run regression then u can with the help of following command

regress CO2 GDP OIL fdi {note: CO2 GDP OIL fdi are my variables first I

wrote my dependent variable then all Independent variables}

step#6

If u wants to check normality then u has to perform two steps after regression means runtwo commands consecutively

predict myResiduals, r sktest myResiduals

PLEASE DONT EDIT THIS FILE

OR MY NAME , THANKS. if you want tdownload this file search following link.

https://drive.google.com/open?

id=0B5lNKqneWZwhYWlUQy04NFRKNXc


2/26






step#7

If u have run regression now, if u wish to check serial correlation then apply following

command

dwstat or estat bgodfrey

step#7

Suppose now u want to test heteroskedasticity

estat hettest, fstat or estat hettest

step#8

Suppose u now want to test multicollenearity

estat vif

Setep#8

Suppose now u want to see either model is miss specified or not /either we have omitted

variables or not/Ramsey RESET test

estat ovtest

Note all diagnostic tests can be run from post estimation option (statistics-----post

estimation)

How to test about structural breaks in data?

Statistics > Postestimation

Step#1

At first step run simple regression, normally we check structural break individually in each

variable, so run one by one regression like this, suppose I want to check structural breaks

in my dependent variable co2. So first I should run simple regression with only co2

regress co2

step#2

Now set time with following command

tsset year (if u have monthly data then write month )

step#3

Run following command to know about structural breaks.


3/26






estat sbsingle

Few old commands with new names

Out-of-date commands

These commands continue to work but are out-of-date as of Stata 9. Their replacements

are

Old command New command

------------------------------

hettest estat hottest for HSK(HETEROSCADESTICITY)

imtest estat imtest (meron & Trivedi's decomposition of IM-test)ovtest estat ovtest (Ramsey RESET test using powers of the fitted values of CO2

Ho: model has no omitted variables)

szroeter estat szroeter

vif estat vif

------------------------------

See regress postestimation.

Old command New command

------------------------------

archlm estat archlm

bgodfrey estat bgodfreydurbina estat durbinalt (FOR SERIAL CORRELATION ALTERNATIVE TO

DURBINWATSON TEST)

dwstat estat dwatson (DURBINWATSON TEST FOR S.C)

------------------------------

How to run time series ARDL MODEL?

Step#1

Import data into stata

Step#2 set times first otherwise u will get error message for time write the following

command.(if u have annually data otherwise u can change frequency like monthly etc.

tsset years, yearly


4/26






Step#2 First install following package

net install ardl, from(http://www.kripfganz.de/stata/)

Write following command into command barardl P YU EX HE, lag(1 1 2 3) ec

(not p,yu ex and he I have my variable first p is dependent variable while remaining are

independent variables , further I have space between all variables, and after comma I have

also space and after bracket close I have also space good luck,,, lags 1,1,2,3 indicating for

dependent variable there must be one lag and after dependendent variable for the first

independent variable also must be lag 1 and so one )

Step#4

If u wants to conform long run relationship to the help of bound test then write following

command in command box.

estat btest

.

_cons 3314510 169730.1 19.53 0.000 2960459 3668561

L2D. .0003703 .0015138 0.24 0.809 -.0027874 .003528

LD. -.0027493 .002675 -1.03 0.316 -.0083294 .0028307

D1. -.0011703 .0065507 -0.18 0.860 -.0148348 .0124942 HE

LD. -.0002417 .0000773 -3.13 0.005 -.000403 -.0000804

D1. -.0001549 .0000722 -2.14 0.045 -.0003056 -4.17e-06

EX

D1. 30.83755 26.00734 1.19 0.250 -23.41281 85.08792

YU

SR

HE .6086697 1.082744 0.56 0.580 -1.649894 2.867234

EX .0646155 .0122439 5.28 0.000 .0390751 .0901558

YU 1797.459 14701.89 0.12 0.904 -28870.14 32465.06

LR

L1. -.0059958 .0017579 -3.41 0.003 -.0096627 -.0023289

P

ADJ

D.P Coef. Std. Err. t P>|t| [95% Conf. Interval]

Root MSE = 83070.046

Adj R-squared = .87452691

R-squared = .91635127

Log likelihood = -388.34477

Number of obs = 31

Sample: 1983 - 2013

Model: ec

ARDL regression

. ardl P YU EX HE, lag(1 1 2 3) ec

This is error

correction term

Long run results

Short run results


5/26






second method of running ardl model

step#1 import data into STATA

Step#2ardl CO2 GDP OIL fdi , lags(. . . 3) maxlag(3 3 3 3) (note: here co2 is my dependent

variable while other are independent variables, while lags(. . . 3) is showing that for the

first three variables means one dependent and other two independent variables I am

saying to stata that ,its all up to stata ,program itself can select optimal lags but 3

indicating that for last independent variable Im limiting program that there must be lag 3

for last variable,maxlag (3 3 3 3) showing we can add maximum lags 3333 for all variables

but it is ignorable

Step#3

If want to see how stata chose optimal lags then run following command

matrix list e(lags)

step#4

Suppose now you want to see error correction term, long run as well as short run results

then apply follow owing command

ardl CO2 GDP OIL fdi , ec

now you want to see bound test,

estat btest

Third Method of running ARDL in STATA

Step#1 first of all install package again command is here

net install ardl, from(http://www.kripfganz.de/stata/)

Step#2 or search ARDL package through stata command box using

help ardl or findit ardl

Setp#3 here we are going to run simple ardl like in eviews we get ardl results before bounds

tests and long run and short run , run following command in comamd bar first write yourdependent variable then all independent variables

ardl co2 he pop , aic

Step#3 As before going to long run and short run we go for bound tests valuesto conform long run cointegration.
http://www.kripfganz.de/stata/http://www.kripfganz.de/stata/http://www.kripfganz.de/stata/http://www.kripfganz.de/stata/


6/26






ardl, noctable btest

Step#4 as in step 3 we conform about long run cointegration now we aregoing to run long run and short run results with error correction term(ADJ)

here first I wrote my dependent variables then all independent .

ardl co2 he pop , aic ec regstore(ecreg)

Step#5 as now we have generate all results but we have need now ofdiagnostic test for the store your step#4 results with this command estimates

restore ecreg

Step#6 after restoring your results in step 5 ,,, now run regres commandyou will see your step 4 results will appear and after this you may runfollowing diagnostic test,

Step#7 Frequently ask question about ARDL USING STATA , it is acknowledge that i havecopied this post from Aymen Ammaritime line

estat dwatson (Durbin Watson statistics, at 1st order autocorrelation).

estat archlm (ARCH LM test for higher order autocorrelation)

estat bgodfrey (Breusch Godfrey LM test for higher order autocorrelation)

estat hottest (Breusch Pagan Heteroscedasticity test)

estat ovtest (Ramsey RESET test)

estat vif (Test for the Multicollinearity)

And finally run after ARDL for the parameters stability .CUSUM TEST

Now If you want to run cusum test (parameters stability test) then run following command

first install this package ssc install cusum6 (note: internet is necessary forinstallation)

now type this command cusum6 variable1 variable2 variable3,cs(cusum) lw(lower)uw(upper)

How to select optimal lags

Statistics > Multivariate time series > VAR diagnostics and tests > Lag-order selection

statistics (preestimation)

Or select optimal lags through following command

varsoc LOGFDI LOGGDP LOGDD LOGINF LOGEXCHRT, maxlag(8)


7/26






OrStep#1

Step#2

Exogenous: _cons

Endogenous: LOGFDIGDP LOGGDP LOGDD LOGINF LOGEXCHRT

8 3354.19 113.04* 25 0.000 . -294.927* -293.642* -289.471*

7 3297.68 -78.245 25 . . -289.789 -288.504 -284.333

6 3336.8 200.88 25 0.000 . -293.345 -292.06 -287.89

5 3236.36 3065.7 25 0.000 . -284.214 -282.929 -278.759

4 1703.49 3269.3 25 0.000 5.6e-66* -145.318 -144.091 -140.11

3 68.8475 169.66 25 0.000 .000013 1.01387 1.94847 4.98129

2 -15.9819 75.413 25 0.000 .000715 6.4529 7.09544 9.1805

1 -53.6883 224.04 25 0.000 .001488 7.60803 7.95851 9.09582

0 -165.706 3.77983 15.5187 15.5772 15.7667

lag LL LR df p FPE AIC HQIC SBIC

Sample: 1992 - 2013 Number of obs = 22

Selection-order criteria

. varsoc LOGFDIGDP LOGGDP LOGDD LOGINF LOGEXCHRT, maxlag(8)

delta: 1 year

time variable: year, 1984 to 2013

. tsset year, yearly


8/26






How to test cointegration

Statistics > Multivariate time series > Cointegrating rank of a VECM

Step#1

Set time first otherwise u

may get error

Write ur all variables

Chose maximum lags

Normally use in between

5-10 and keep all thing

unchan ed


9/26






Step#2

Write ur variables, like first

dependent then all indep

Chose optimal lags, which u

deicide form lag length

criteria and ok


10/26






How to run VECM MODEL?

Statistics > Multivariate time series > Vector error-correction model (VECM)

Step#1

Step#2

And ok


11/26






Step#3

Step#4

_cons -164710.4 57151.86 -2.88 0.004 -276726 -52694.85

L3D. .0005969 .0003585 1.67 0.096 -.0001057 .0012996

L2D. .0011455 .0005591 2.05 0.040 .0000496 .0022414

LD. .0032082 .0011028 2.91 0.004 .0010467 .0053698

HE

L3D. .0000111 9.78e-06 1.13 0.257 -8.09e-06 .0000303

L2D. -1.80e-06 9.17e-06 -0.20 0.845 -.0000198 .0000162

LD. -4.32e-06 9.15e-06 -0.47 0.637 -.0000222 .0000136

EX

L3D. -2.692897 1.817526 -1.48 0.138 -6.255183 .8693891

L2D. -10.78266 4.202071 -2.57 0.010 -19.01857 -2.546755

LD. -17.44365 6.310434 -2.76 0.006 -29.81187 -5.075422

YU

L3D. 1.00195 .1026645 9.76 0.000 .8007314 1.203169

L2D. -2.631597 .1808343 -14.55 0.000 -2.986025 -2.277168

LD. 2.641314 .0914989 28.87 0.000 2.46198 2.820649

P

L1. -.0002116 .0000693 -3.05 0.002 -.0003474 -.0000758

_ce1

D_P

Coef. Std. Err. z P>|z| [95% Conf. Interval]

Write your variabels,firstdependent then all

independent variables

Write here number of

cointegration

equations which u finds

from Johansson test

but I would like to

suggest u add all the

time 1 for simplisticity

Add here

maximum

lags or

optimal

Long run causality value must be

negative and in between 0-

1..which indicate error

correction term ,speed of

adjustment

Short run causality


12/26






Wald test for short run causalities if you want to see jointly impact of lags variabels on

dependent variables

Go to statistics----post estimation---test, contrast, and comparison of parameters,---linear

test of parameters

How to run IMPULSE RESPONSE FUNTION

If u want to run through MANU,, follow these steps

Statistics > Multivariate time series > IRF and FEVD analysis > Graphs by impulse or

response

Step#1 (actually impulse response functions used after VAR models)

Run VECM model

Step#2

Then use irf create to estimate the IRFs and FEVDs and save them in a file, and finally use irf

graph or any of the other irf analysis commands to examine results:, like run following command

irf create order1, step(10) set(myirf1)

Step#3 now I want to see impulse response function, the following function will show over all

impulse response function results

irf graph irf, irf(order1)

step#4

suppose you are not interest in all variables response function ,I mean to say I just want to see

only independent variables shocks effect on dependent then apply following command.

irf graph irf, irf(order1) impulse(GDP OIL fdi) response(CO2) (note here GDP,OIL and fdi

are my independent variables and co2 dependent .

How to run var model?

Statistics > Multivariate time series > Vector autoregression (VAR)

Step#1


13/26






Step#2


14/26






Step#3

Step#4 for short run granger causality/wald test

Statistics > Multivariate time series > VAR diagnostics and tests > Granger causality tests

_cons 477355 214606.1 2.22 0.026 56734.76 897975.3

L2. -.0078436 .0040739 -1.93 0.054 -.0158283 .000141

L1. -.0013774 .0007436 -1.85 0.064 -.0028348 .00008

HE

L2. -13216.11 7908.145 -1.67 0.095 -28715.79 2283.57

L1. 10483.9 7579.004 1.38 0.167 -4370.678 25338.47

HDI

L2. .0000507 .0000366 1.39 0.166 -.000021 .0001223

L1. .0000743 .0000315 2.36 0.018 .0000125 .000136

EX

L2. .6010501 6.50086 0.09 0.926 -12.1404 13.3425

L1. 4.900434 11.82905 0.41 0.679 -18.28407 28.08494

YU

L2. -.8601668 .0746806 -11.52 0.000 -1.006538 -.7137955

L1. 1.859308 .0743769 25.00 0.000 1.713532 2.005084

P

P

Coef. Std. Err. z P>|z| [95% Conf. Interval]

HE 11 9.9e+06 0.3616 18.12447 0.0529

HDI 11 1.01628 0.9989 6665.462 0.0000

EX 11 2.3e+08 0.9691 1003.692 0.0000

YU 11 580.764 0.7792 112.9032 0.0000

P 11 43425.6 1.0000 2.14e+07 0.0000

Equation Parms RMSE R-sq chi2 P>chi2


15/26






Setp#

U have no need to

change anything just

click ok


16/26






Step#6 and finally granger causality test

Panel data ModelsHow to take panel unit root

Statistics > Longitudinal/panel data > Li near models > Linear regression (F E, RE, PA, BE)

Setep#1

Give fi rst id with following command..

egen countr y1=group( country) (note: i f you have countr ies data)

egen Company1=group( Company) (note: i f you have companies data) means declare data

panel

HDI ALL 1346.5 8 0.000

HDI HE 1042.9 2 0.000

HDI EX 7.3963 2 0.025

HDI YU 3.519 2 0.172

HDI P 12.807 2 0.002

EX ALL 15.527 8 0.050

EX HE 5.1576 2 0.076

EX HDI 4.4102 2 0.110

EX YU 1.4657 2 0.481

EX P 8.7329 2 0.013

YU ALL 93.069 8 0.000

YU HE 51.299 2 0.000

YU HDI 3.3333 2 0.189

YU EX 8.8705 2 0.012

YU P 9.9949 2 0.007

P ALL 40.749 8 0.000

P HE 5.7897 2 0.055

P HDI 5.4149 2 0.067

P EX 21.011 2 0.000

P YU .17274 2 0.917

Equation Excluded chi2 df Prob > chi2

Granger causality Wald tests

. vargranger

Here results showing that P is

dependent variables while YU,EX, HDI

ETC INDEPENDNET VARIAELS,

IN THE SECOND ROW OF RIGTHT SIDE

FIRST COLUM, SHOWING THT EX

JOINTLY GRANGER CUSE P IN SHORT

RUN . AS NULL HYPOTHESIS WAS NO

GRANGER CASUE AS PROBABLITY VALUE

IS LEST THAN 5% SO I CAN SAY HERE

THAT I HAVE TO ACCEPT ALTERNATIVE

HYPOTHESIS


17/26






Step#2

Click on unit root test

Select test type

Select variables to which u

want to take unit root

Set time and give panel

id to cross sections

Select optimal lags ,suppose

one

Select if u

need

suppose u

want to add

time trend

the check

first option


18/26






Panel data analysis from star t to end..

Poll ed, random ,f ixed effect ,hausman test.

1.import your data fi le into stata

2.now create a pool or simple stata give codes to each cross section or enti ty l ike if you have

dif ferent countr ies data or companies the u have to give specif ic code all countr ies or

companies, fu rther i f you have assign code by your self suppose u did not wr i te company name

like nestles but you indicated nestle with 111 now u see you have already given the code but

i f you have simple right the name of company then u need to give also code

egen countr y1=group( country) (note: i f you have countr ies data)

egen Company1=group( Company) (note: if you have companies data)

3.now set time which is most important

xtset Company1 year, yearly (note: hear I have yearl y data and company1 is new variable

which I genrate in step 2)

4.now look at descri ptive statistics

Xtsum variable1 variabel2

xtsum ENVC EPS ROA ROE ROC

4.1 suppose u want to make a graph

xtline CO2 energy gdp gi

4.2 for description of data

xtdescribe

5. Now run fixed effect model

Xtreg dependent variable1 independent variable 1234456,fe

xtreg ENVC EPS ROA ROE,fe

Now store result of fixed effect from this command

6.estimate store fe ( if u want to run by Manu Statistics > Postestimation )

7. Now run random effect model

xtreg dep indep1 indep2 indep3, re (replace with your variables name)


19/26






now store result of fixed effect from this command

8. estimate store re

9. now the last thing what model is suitable random effect or fixed effect for this run Housman

test(note if you do not restore results of random effect and fixed effect may u face error prob)

hausman fe re

how to run pooled regression in stata

reg dep indep indep

10. Further u can double check either random /fixed effect /or polled appropriate. But note,

suppose hausaman verified random effect is appropriate, so we can double check for step 9,

suppose hausman test verified random effect model is appropriates so in step ten we willconform either really random effect is appropriate or not so we will run test and verify

hypothesis between polled model and random effect actually we have already done with fixedeffect using hausman test.stepsstatisticslongitudinal /panel data---linear model---

langrangian multiplier and null hypothesis is polled is appropriate. And alternative hypothesis is

random effect is appropriate.

How to run 2sls two stage least squareStatistics > Endogenous covariates > Single-equation instrumental-variables regression

Step#1

ivregress 2sls consumtion remetence (income = investment)

(note here income is my endogenous and investment instrumental is my instrumental

variables)

Step#2

As I have run 2sls model but now I have to conform that either in reality really endogeniety

problem was exist or not

estat endog

if probability value comes more than 5% then we say there is no endogeniety but if prob valuecomes less than in this case we say there is endogeniety prob,, which is desirable;

setp#3

Now I have I have conform either endogeniety problem exist or not now I want to know either

my instruments are weak or strong

estat firststage


20/26






step#4

Now I want to know either my instruments are over identified or not?

estat overid{ sargan and basman test is used to know about the over identification if probability value

comes of thesis test more than5% we say model is correct specified

Null hypothesis for over identified instruments: instrument set is valid and the model is correct

specified}

PANEL ARDL USING STATA

1) First of all install this package to run PANEL ARDL ssc install xtpmg, replace

2) Suppose you think you have installed this package but still you are not sure then type incommand bar type xtpmg

3) If u see message of no found then install otherwise you have already install it. here we shall Run MG (average):

xtpmg d.CO2 d.energy d.gdp , lr(l.CO2 energy gdp ) ec(ECT) replace mg

here we shall Run MG (individual):

(It allows for all coefficients to vary and be heterogeneous in the long-run and short-run.However, the necessary condition for the consistency and validity of this approach is to

have a sufficiently large time-series dimension of the data.)

xtpmg d.CO2 d.energy d.gdp , lr(l.CO2 energy gdp ) ec(ECT) replace full mg

here we shall Run PMG (average):

xtpmg d.CO2 d.energy d.gdp , lr(l.CO2 energy gdp ) ec(ECT) replace pmg here we shall Run PMG (individual):

(The main characteristic of PMG is that it allows short-run coefficients, including theintercepts, the speed of adjustment to the long-run equilibrium values, and error

variances to be heterogeneous country by country, while the long-run slope coefficients

are restricted to be homogeneous across countries.)

xtpmg d.CO2 d.energy d.gdp , lr(l.CO2 energy gdp ) ec(ECT) replace full pmg

here we shall Run Hausman test to choose between MG and PMG:hausman mg pmg, sigmamorenow if our probability value comes more than 5% we run PMG

if our probability value comes less than 5% we run MG

Running DFE:xtpmg d.CO2 d.energy d.gdp , lr(l.CO2 energy gdp ) ec(ECT) replace dfe

* Running Hausman test to choose between MG and DFE:

hausman mg DFE, sigmamore

Note:


21/26






Suppose you want to run all these tests on your data, so simple just import your data intostata and copy command from here into stata command bar and replace my variables

name with yours.

Good luck.

PANEL ARDL

Pooled Mean Group (PMG) model

The main characteristic of PMG is that it allows short-run coefficients, including the intercepts, the

speed of adjustment to the long-run equilibrium values, and error variances to be heterogeneous

country by country, while the long-run slope coefficients are restricted to be homogeneous across

countries. This is particularly useful when there are reasons to expect that the long-run equilibrium

relationship between the variables is similar across countries or, at least, a sub-set of them. The

shortrun adjustment is allowed to be country-specific, due to the widely different impact of thevulnerability to financial crises and external shocks, stabilization policies, monetary policy and so on.

However, there are several requirements for the validity, consistency and efficiency of this

methodology. First, the existence of a long-run relationship among the variables of interest requires the

coefficient on the errorcorrection term to be negative and not lower than -2. Second, an important

assumption for the consistency of the ARDL model is that the resulting residual of the error-correction

model be serially uncorrelated and the explanatory variables can be treated as exogenous. Such

conditions can be fulfilled by including the ARDL (p,q) lags for the dependent (p) and independent

variables (q) in error correction form. Third, the relative size of T and N is crucial, since when both of

them are large this allows us to use the dynamic panel technique, which helps to avoid the bias in the

average estimators and resolves the issue of heterogeneity. Eberhardt and Teal (2010) argue that the

treatment of heterogeneity is central to understanding the growth process. Therefore, failing to fulfil

these conditions will produce inconsistent estimation in PMG.

The PMG estimator constrains the long term coefficients to be the same across countries and allows

only the short-term coefficients to vary.

Mean Group (MG) estimator

The second technique (MG) introduced by Pesaran and Smith, (1995) calls for estimating separate

regressions for each country and calculating the coefficients as unweight means of the estimated

coefficients for the individual countries. This does not impose any restrictions. It allows for all

coefficients to vary and be heterogeneous in the long-run and short-run. However, the necessary

condition for the consistency and validity of this approach is to have a sufficiently large time-series

dimension of the data. The cross-country dimension should also be large (to include about 20 to 30

countries). Additionally, for small N the average estimators (MG) in this approach are quite sensitive to

outliers and small model permutations (see Favara, 2003).


22/26






Dynamic Fixed Effects (DFE) model

Finally, the dynamic fixed effects estimator (DFE) is very similar to the PMG estimator and imposes

restrictions on the slope coefficient and error variances to be equal across all countries in the long run.

The DFE model further restricts the speed of adjustment coefficient and the short-run coefficient to be

equal too. However, the model features country-specific intercepts. DFE has cluster option to estimate

intra-group correlation with the standard error (Blackburne and Frank, 2007). Nevertheless, Baltagi, Gri,

and Xiong (2000) point out that this model is subject to a simultaneous equation bias due to the

endogeneity between the error term and the lagged dependent variable in case of small sample size.

How to run DOLS model

Setpe# import data

Step#2 install following package

ssc install ltimbimata, replace

Step#3 Before beginning the estimations, we use the set more off instruction to tell Stata not to

pause when displaying the output.

set more off

Step#4 now run dols model, we regress iskr (dependent variable) on the regressors (gdskr

irxmap1 defigd2 ltinflcd opins2 totwdct ltdgdpd).

xtdolshm iskr gdskr irxmap1 defigd2 ltinflcd opins2 totwdct ltdgdpd

Step#5 if you want to increase lags and leads

xtdolshm iskr gdskr irxmap1 defigd2 ltinflcd opins2 totwdct, nla(3) nle(4)

step#6 if you want estimation at 10 % level of significance

xtdolshm iskr gdskr irxmap1 defigd2 ltinflcd opins2 totwdct, nla(3) nle(4) level(90)


23/26






Primary data analysis

Step#1

Suppose I want to see descriptive statistics of my variables

summarize cs1 cs2 cs3 cs4 (cs1,2,3 and 4 are my variables)

step#2

*suppose you want to know about correlation among variables

correlate cs1 cs2 cs3 cs4 (cs1,2,3 and 4 are my variables)

step#3

*now you want to check reliability(cronbach alpha values) of items ,so first write alpha then all

items with space

alpha cs1 cs2 cs3 cs4 ( cs1 cs2 cs3 cs4 are my items)

Step#4

*now we are going to run PCA and want to see egen values /component means from the items

how much component we can create

pca cs1 cs2 cs3 cs4 ( cs1 cs2 cs3 cs4 are my items)

step#5

Now i also want to know about the KMO value of PCA

estat kmo, novar

step#6

*now i want to make a construct/variables from (4 items)cs1 cs2 cs3 cs4 and suppose i give

name to this new single variable like saeed1


24/26






predict saeed1, score

Step#6

Actually I have converted my all items into variables now I want to run regressions between

these variables

regress cs1 cs2 cs3 cs4 (suppose cs1,2,3 and 4 are my variables which are made after

PCA)

Last tips and tricks

Finally how to generate new variables

1. Suppose you want to generate a series of square of any variable

gen cs1sqrt=sqrt( cs1) (note cs1 is my variabel)

2. Suppose you want to take log

gen cs1log=log( cs1)

3. Suppose u want to add two variables

gen cs1pluscs3 = cs1+cs3

4. Suppose you want to generate series with first difference5. generate fdpop = d.pop

How to get help from stataSuppose you are running any test but at some points you got confused/ stuck so how u canprecede now, suppose I was running panel unit root but I was not sure about null hypothesis

of different test how I can precede now? So to get rid of this problem see following pics and

enjoy.


25/26






In this you can see description about the null hypothesis of all tests so u can get rid of any

problem just click on help button


26/26





Time Sereis Analysis Using Stata

Documents