**Sample Exercises for Dataset pwltsp.por**

James Gerber, San Diego State University

*© The Author, 1998; Last Modified 16 August 1998*

1. Descriptive statisticsHow does productivity (rgw

xx) vary by continent?

- (a) Run Statistics, Custom Tables, Basic Tables. In the Summary box, select rgw90 (real GDP, 1990), and in the Down box select ccode.

(b) Choose the Statistics button and click on Mean, then Add to cell statistics. Click on Maximum and Add to cell statistics. Click Continue, then click Okay. The table in the output window will show the mean and maximum values, by continent.

2. Compare meansFrom the above output, it appears that Asia has higher productivity than South America. Let's test this by comparing the averages.

- (a) Run Statistics, Compare Means, Independent Samples t Test. Choose rgw90 as the Test Variable, and ccode(3,4) as the grouping variable.

3. CorrelationInvestment is one of the key determinants of living standards and economic growth. But not all investment is created equal. For example, another strip mall (non-residential construction) in Southern California probably will have less impact on economic growth than investment in factory machines (producer durables). Which forms of investment are associated with greater increases in percapita income?

- (a) Run Transform, Compute. Create a new variable "kdurg" which is the difference between kdur90 and kdur80. Do the same for real GDP (rgdp
xx), nonresidential construction (knrxx) and residential consturction (krxx).

(b) Run Statistics, Correlate, Bivariate. Include all four newly constructed variables.

4. RegressionAnother way to think about the link between investment and GDP is to consider its impact relative to the impacts of consumption and government spending.

- (a) Run Statistics, Regression, Linear. Select rgdp90 as the dependent variable and cc90, cg90, and ci90 as the independents.

(b) Compare the results with a bivariate regression. Run Statistics, Correlate, Bivariate. Use all four of the same dependent and indpendent variables. Compare the correlation coefficients of each of the variables with rgdp90.