VERY BRIEF (AND AWKWARD) SPSS INTRO by HC120609004238

VIEWS: 0 PAGES: 5

									VERY BRIEF (AND AWKWARD) SPSS INTRO.

FILE: OPEN:
>..SPSS\TUTORIAL\SAMPLE FILES\DEMO.SAW




5274, 5275,... are your observations (you’ll have to make about 50).

At the bottom there are two bookmarks: “data view” (current page) and “variable view”. Click on
“variable view”.




Here are described your variables (in rows). On previous screen you have them in columns.
On previous screen you have data, inserted into the cells, corresponding to the variables.
Age is numeric, 4 digits wide, 0 decimals, described as «age in years», values are entered
freely – not restricted as e.g. marital status, no missing values, 8 digits wide in «data view»,
right-aligned and (!) it is scale (or metric = interval or ratio).

Let’s correlate age and income.
Go to: analyze:correlate:bivariate




Add age and income. Select Pearson, Kendall’s, Spearman’s, one-tail (I test on the chance to
find a “bigger” correlation than this), flag significant correlations. Press “OK”.

Correlations
                                 Cor relations

                                                                   Income
                                                                 category in
                                                  Age in years thousands
  Age in years          Pearson Correlation                  1          ,351**
                        Sig. (1-tailed)                       ,         ,000
                        N                                6400          6400
  Income category       Pearson Correlation               ,351**           1
  in thous ands         Sig. (1-tailed)                   ,000              ,
                        N                                6400          6400
     **. Correlation is s ignif icant at the 0.01 level (1-tailed).
Nonparametric Correlations
                                             Cor relations

                                                                                           Income
                                                                                         category in
                                                                          Age in years thousands
  Kendall's tau_b      Age in years          Correlation Coef f ic ient          1,000          ,314**
                                             Sig. (1-tailed)                          ,         ,000
                                             N                                   6400          6400
                       Income category       Correlation Coef f ic ient           ,314**       1,000
                       in thous ands         Sig. (1-tailed)                      ,000              ,
                                             N                                   6400          6400
  Spearman's rho       Age in years          Correlation Coef f ic ient          1,000          ,387**
                                             Sig. (1-tailed)                          ,         ,000
                                             N                                   6400          6400
                       Income category       Correlation Coef f ic ient           ,387**       1,000
                       in thous ands         Sig. (1-tailed)                      ,000              ,
                                             N                                   6400          6400
    **. Correlation is s ignif icant at the .01 lev el (1-tailed).



Not surprisingly, the correlations are very close in values to each other.


Let’s regress job satisfaction against “years with current employer”, “level of education”, “price
of primary vehicle”, “household income in thousands”, “age in years”.
                          Model Sum m ary

                                          Adjusted       Std. Error of
  Model         R         R Square        R Square       the Estimate
  1              ,491 a       ,241             ,240             1,193
    a. Predictors: (Constant), Age in years, Lev el of
       educ ation, Household income in thousands , Years
       w ith current employer, Price of primary v ehicle


R Square is 0.241. Means, 24% of variance in dependent variable is explained by the model.

                                               ANOVAb

                             Sum of
  Model                     Squares             df        Mean Square           F           Sig.
  1        Regression       2886,252                5         577,250         405,661         ,000 a
           Residual         9098,588             6394           1,423
           Total           11984,840             6399
    a. Predictors: (Constant), Age in years, Lev el of educ ation, Household income in
       thousands , Years w ith c urrent employ er, Pric e of primary vehic le
    b. Dependent Variable: Job satisf ac tion


Anova table says that a model is better than simple guessing. A model is statistically strong
(high significance indicator).
                                                                           a
                                                               Coe fficients

                                                         Unstandardiz ed         Standardized
                                                           Coef f icients        Coef f icients
  Model                                                   B         Std. Error       Beta          t       Sig.
  1             (Cons tant)                               2,197           ,068                    32,491     ,000
                Y ears w ith c urrent
                                                   5,687E-02             ,003             ,404    22,413     ,000
                employ er
                Level of education                 -3,49E-02             ,014            -,031    -2,575     ,010
                Price of primary v ehicle          8,170E-03             ,001             ,131     6,710     ,000
                Hous ehold inc ome in
                                                   -1,00E-03             ,000            -,058    -3,168     ,002
                thousands
                A ge in years                      4,116E-03             ,002             ,037     2,658     ,008
     a. Dependent V ariable: Job satisf action


Coefficients. All are significant (the least strong is “level of education”, but still significant at 1%
level). Only “years with current employer” has big beta; other variables are much less effective.

(Initially, the experimental design is a mess; job satisfaction and years with current employer
may be in reverse causality or mutual relationship; the results are not trustworthy. You should
create logical experimental designs – this would be at least a good start for a regression).

Run SPSS tutorial (in “help menu” – “topics”; and type there anything you need). Regression is
better described in the car_sales.saw example.


Let’s make a distribution graph. (Menu “graphs”: “histogram”)
  1200



  1000



  800



  600



  400



  200                                                 Std. Dev = 12,29
                                                      Mean = 42,1

    0                                                 N = 6400,00
         20,0   30,0   40,0   50,0   60,0   70,0
            25,0   35,0   45,0   55,0   65,0   75,0


         Age in years

Age in years has approximately normal distribution (somewhat right-skewed). (“Good” data if
“normality” requirement must be met).

One-way ANOVA.
Let’s check how the parameter “years with current employer” changes (not changes) analyzing
different income categories. Go to: analyze:compare means: one way ANOVA.
In menu, factor variable is income categories, dependent – years with current employer.
(In “options,” check “descriptive”).
Oneway
                                                                                                    Des criptives

  Years w ith current employ er
                                                                                                                             95% Conf idence Interval f or
                                                                                                                                        Mean
                         N                                      Mean        Std. Deviation           Std. Error             Low er Bound Upper Bound         Minimum    Max imum
  Under $25              1174                                     5,24              8,040                  ,235                     4,78            5,70            0          54
  $25 - $49              2388                                     6,44              5,886                  ,120                     6,20            6,67            0          46
  $50 - $74              1120                                    11,07              7,276                  ,217                    10,64           11,49            0          44
  $75+                   1718                                    19,62            10,011                   ,242                    19,15           20,10            0          57
  Total                  6400                                    10,57              9,724                  ,122                    10,33           10,80            0          57


          Tes t of Homogene ity of Variance s

  Years w ith c urrent employ er
   Levene
   Statistic           df 1                                        df 2             Sig.
   186,182                                               3           6396             ,000

                                                                             ANOVA

  Years w ith current employ er
                                                          Sum of
                                                         Squares            df         Mean Square                     F                        Sig.
  Betw een Groups                                        215274,6              3         71758,194                  1177,295                      ,000
  Within Groups                                          389847,3           6396            60,952
  Total                                                  605121,9           6399



     30
                                                                                         70


                                                                                         60


                                                                                         50
     20
                                                                                         40
                           Years with current employer




                                                                                         30


                                                                                         20
     10
                                                                                         10


                                                                                             0

                                                                                         -10
      0                                                                                      N=        1174       2388         1120      1718

    Under $25       $25 - $49                                $50 - $74       $75+                   Under $25   $25 - $49    $50 - $74   $75+


          Income category in thousands                                                           Income category in thousands



The output and the mean’s plot is rather straightforward.

Multi-way ANOVA is addressed as “General Linear Model”:”Univariate”.
Go to “index” and review “how to” do GLM:Univariate.

Graphs are simple to obtain. Study how to build them and create a few interesting.

Descriptive statistics (means and standard deviations are in the “analyze”:”descriptive
statistics”; btw, cross-tabulation is also there).

T-tests are in “analyze”:”compare means”.

								
To top