This preview shows page 1. Sign up to view the full content.
Unformatted text preview: Name: _____________________________________ Lab #3 Spring08 Lab Partners: ________________________ Lab Day/Time: __________ Part I. Social support & recovery
Researchers have found significant relationships between social support and mortality rates in women suffering from breast cancer, such that women with a great deal of social support experience better courses and outcome of their disease, and women in social support groups have a longer lifespan. From Blackboard > Course Documents > Lab Information, Lab #3 open the file Lab3_BreastCancer.sav. The current data consist of information from a longitudinal study of women with early stage breast cancer in which participants were asked about their levels of social support and then time to recovery in months was measured. The data consist of patient ID, age, a rating of their current social support (0-to-10 point Likert scale with 0 being very poor and 10 being strong social support) and time in months until recovery. We are interested in learning about the relationship between levels of social support and recovery time. 1. Looking at the dataset, what is/are the independent variables? There are no IVs. 2. What is/are the dependent variables? What is/are the measurement scales associated with each variable? DV#1: Levels of support on a 0 10 Likert Scale with 0 being poor support and 10 being strong social support. Interval Data. DV #2: Recovery Time in months. Ratio Data. Next, we are going to find the measures of central tendency and variability for the sample. On the toolbar, click on ANALYZE > DESCRIPTIVE STATISTICS > FREQUENCIES. Move TIME over to the variable(s) box. Uncheck the Display Frequency Tables box. Click on STATISTICS and select STD DEVIATION, VARIANCE, RANGE, MINIMUM, MAXIMUM, MEAN, MEDIAN, MODE, SUM, SKEWNESS & KURTOSIS Click CONTINUE Click on CHARTS and select HISTOGRAMS, WITH NORMAL CURVE. Click on CONTINUE Click on OK 1. Looking at your output, fill in the following information: Mode Median s2 46.3627 30 39.5 996.482 S Range N 31.56711 125.13 85 Page 1 of 5 Name: _____________________________________ Lab #3 Spring08 Statistics Time (months) N Mean Median Mode Std. Deviation Variance Skewness Std. Error of Skewness Kurtosis Std. Error of Kurtosis Range Minimum Maximum Sum Valid Missing 85 0 46.3627 39.5000 30.00 31.56711 996.482 .716 .261 -.299 .517 125.13 7.10 132.23 3940.83 20 15 Frequency 10 5 0 0.00 25.00 50.00 75.00 100.00 125.00 Time (months) 2. Describe the shape your histogram? Talk about skew and normality in terms of skewness and kurtosis measures (go back to Lab#2 if you forgot how to). Looking at the graph, one can see that the distribution has a positive skew. The skewness value was 0.716. To determine the magnitude of the skew, you take that value and divide it by its standard error (2.74). This number is greater than 2 so the distribution has a large skew. For normality, you take the kurtosis value and divide it by its standard error (-0.578). This number is less than 1 which indicates that although the distribution does have a bit too few values in the tail (particularly on the lower end) and in the middle (see graph), it is not different enough to consider it vastly different than a normal distribution. Correlation & Regression.
Next we will examine whether there is an association between social support and time to recovery, and whether time to recovery could have been predicted by the social support scores. On the toolbar, click on ANALYZE > CORRELATE > BIVARIATE. Move TIME and SUPPORT over to the variable(s) box. Click on OK
Correlations 1) r = -0.544, p < 0.0001 If we set alpha to 0.05, is there a statistically significant correlation between the two variables? Explain the nature of this relationship. Make sure to mention both direction and magnitude.
support Pearson Correlation Sig. (2-tailed) N Pearson Correlation Sig. (2-tailed) N support Time (months) Time (months) 1 -.544** .000 85 85 -.544** 1 .000 85 85 **. Correlation is significant at the 0.01 level (2-tailed). Page 2 of 5 Name: _____________________________________ Lab #3 Spring08 Yes, there is a statistically significant correlation because our p-value (observed value) falls beyond our alpha value/critical value (0.05). This suggests that there is a strong, negative relationship between level of support and recovery time in months such that as level of support increases, recovery time is likely to decrease (and vice-versa). We cannot say anything about causation. Now, we will investigate whether or not time to recovery can be predicted from level of social support. On the toolbar, click on ANALYZE > REGRESSION > LINEAR REGRESSION. Move TIME to the "Dependent" box and SUPPORT over to the "Independents" box. Click on OK You can find the results of your regression in the box labeled Coefficients. (Constant) refers to the yintercept of your regression equation, and beneath that the Support number is your slope. 2) What is your regression equation?
Coefficientsa Unstandardized Coefficients B Std. Error 82.663 6.794 -7.417 1.256 Standardized Coefficients Beta -.544 ^ Y = a + b( X ) = 82.663 + (-7.417) X Model 1 (Constant) support t 12.168 -5.904 Sig. .000 .000 a. Dependent Variable: Time (months) 3) On the provided scatter plot (see page 4), find two points and graph the regression line. Make sure to label your two points. 4) What percentage of variability in time to recovery can be explained by support?
R 2 = r 2 This means that roughly 29% of the variability can be explained in terms of social support. = .2959 29.59% 5) What is the proportion increase in prediction (PIP)? Calculate by hand and explain what this value represents. PIP = 1 - 1 - r 2 = .1608 = 16.08% Our estimate of time to recovery is 16.08% more accurate when we know the person's level of social support. Page 3 of 5 Name: _____________________________________ Lab #3 Spring08 6) What is the standard error of the estimate (from SPSS, under Model Summary)? 26.64945 Model Summary Model 1 R R Square .544a .296 Adjusted R Square .287 Std. Error of the Estimate 26.64945 a. Predictors: (Constant), support 7) Complete the following: Rachel just joined this ongoing study. Not knowing anything about her level of social support, what is your best estimate of her time to recovery (assume that demographically and medically she is typical of the rest of your sample). The mean for our sample of time in months to recovery, which is 46.3627 months. Obviously it is unlikely that she will recover in exactly these many months. What is your best estimate of variability around this prediction (conceptually, and the actual value)? The standard deviation, which is 31.56711 Now consider her level of social support. You know that her level of social support is higher than the average for your sample. Does this change what your estimate of her recovery time? Why? What is your measure of variability (or accuracy) in this estimate? We would adjust our prediction and guess that she will recover in less than the average number of months, since we know that social support is negatively correlated with recovery time and that she scored higher than average on the social support scale. This is using the regression line to predict scores based on this relationship. Our measure of variability or accuracy in this prediction is our standard error of the estimate, which is 26.64945. How much smaller is this value than your original estimate of variability? 31.56711 26.64945 = 4.91766 What percentage of your original variability estimate is this difference between estimates? What is this number (conceptually) HINT- look back earlier in the lab... what number is this very close to? 4.91766/31.56711= .1558 .... Around 16%.... PIP! Page 4 of 5 Name: _____________________________________ Lab #3 Spring08 140 120 100 80 60 Time (months) 40 20 0 -2 0 2 4 6 8 10 SUPPORT Page 5 of 5 ...
View Full Document
This lab report was uploaded on 04/07/2008 for the course PSY 031 taught by Professor Dicorcia during the Spring '08 term at Tufts.
- Spring '08