반응형

Lab 4: Association and Regression

In this lab, we will investigate how total time spent on homework is associated with a student’s grades among the national sample of 10th grade students given in the NELS data collected in 1990. Follow the directions below to complete the analysis with SPSS.

Print out the outputs and submit in print. Also submit this worksheet with your answers to the questions.

  1. Open the SPSS dataset of “short NELS” that you created in Lab 1. If you don’t have the SPSS dataset ready to use, download the original data file “short_nels.csv” from Blackboard and follow Lab 1 Problem Set 2 to import it into SPSS.

  2. Download “nels_coding_system.pdf” from Blackboard. Follow the prompts below to locate the two variables of interest (i.e., time on homework, grades) in the dataset and learn about their coding systems.

    1. Open “nels_coding_system.pdf”. Search the keyword “homework” to learn about all variables related to homework available in this dataset.

    2. Find out the variable that is total time spent on homework out of school in 10th grade. Write down its variable name here.

      (Note: NELS is a longitudinal study that followed the sampled students over 12 years and collected data in 4 waves. The first wave was done in 1988 when students were in their 8th grade, which is also called the “base year”. The variables from this base year have their names starting with “by”. The 10th grade is the first follow-up wave and the variable names start with “f1”.)

    3. The dependent variable, student grades, is named ffugrad in the SPSS dataset. This is a composite score created by averaging each student’s math (f1s39a), English (f1s39b), history (f1s39c), and science (f1s39d) grades at the 10th grade.

    4. Use the variable names mentioned above as the keywords for searching the coding system PDF. This will lead you to coding systems of these variables, that is, what are the possible values this variable can assume, and what does each value represent. Write down the coding systems for the two variables of interest in the space below.

      (Note: ffugrad does not have its own coding system explained in the file. But because ffugrad is a composite score created using subject-specific grade variables, we can use one of the original variables, such as f1s39a, to find the coding system.

  3. Go to the SPSS. Follow the directions below to create a new, simpler dataset that includes only the variables of interest above.

    1. Open a new blank dataset: FileNewData

    2. Follow the directions below two times to copy and paste each of the two variables above

      from the short NELS dataset to the new dataset.

      Go to the short NELS SPSS data, Data ViewEditGo to VariableEnter the variable name in the text boxGoSelect the entire column, right-click and CopyGo to the new blank dataset, Data ViewSelect the first (blank) column, right-click and Paste.

    3. After both variables have been pasted into the new data file, save the new dataset.

    1. As the coding systems found above show, neither of these two variables of interest is a perfect continuous measure. For this lab’s purposes, we will treat ffugrad as if it’s a continuous measure. But for time on homework, follow the directions below to convert it into a more reasonably continuous measure --- hours.

      Open the new (simpler) SPSS dataset, Data ViewTransformRecode into Different VariablesSelect the homework variable (f1s36a2) as the Input VariableEnter homework as the Output Variable Name, click ChangeClick Old and New ValuesFor each pair of old (current) and new values below, enter the old and the corresponding new values in the text boxes then click AddAfter all pairs are defined, click ContinueOK


    1. Follow the directions below to create a scatter plot of the two variables. On the side of the printed graph, interpret the relation between the two variables.

      GraphsLegacy DialogsScatter/DotSimple ScatterSelect ffugrad for the Y Axis, select homework for the X Axis, create a title for the graphOK.

    2. Follow the directions below to find the Pearson’s correlation coefficient between the two variables. On the printed output, highlight/circle the Pearson’s correlation coefficient value, and then interpret this quantity on the side in terms of the direction and strength of the linear association between time on homework out of school and grades.

      AnalyzeCorrelateBivariateSelect ffugrad and homework as the VariablesOK.

    3. Follow the directions below to build a simple linear regression model to predict grades using

    the total time spent on homework out of school.

    1. Add a fitted linear regression line on the scatter plot.
      Go to the
      output window, double-click the scatter plotElementsFit Line at Total

      Apply.

    2. Run the linear regression model analysis.

      AnalyzeRegressionLinearDependent: ffugrad; Independent: homeworkOK.

    3. Write out the fitted model using the coefficients given in the outputs.

    4. Interpret the intercept.

    5. Interpret the slope.

    6. Find out the R2 value from the outputs and interpret it. 

Lab 4.sav


반응형

+ Recent posts