Exam 3 Practice Questions Psych , Fall 9

Transcription

1 Vocabular Eam 3 Practice Questions Psch , Fall 9 Rather than choosing some practice terms at random, I suggest ou go through all the terms in the vocabular lists. The real eam will ask for definitions of about terms taken from these lists. Conceptual questions 1. Research has shown that people in Western (individualist) societies tend to think more analticall compared to people in Eastern (collectivist) societies, who tend to think more holisticall. Because this is a cultural difference, ou suspect the effect is stronger for adults than for children. To test this prediction, ou recruit children and adults from both this countr and China. Some subjects are tested on an analtic reasoning task, and others are tested on a holistic reasoning task. List the factors for this eperiment. Then list all the possible interactions ou could test. 2. What is the epected value of MS treatment, according to the null hpothesis that the group means are all equal? (You can answer in words or in smbols.) 3. Repeated-measures ANOVA differs from regular ANOVA in that we remove the variabilit due to before calculating. 4. The five scatterplots below show correlations of , 0,., and 1, in a scrambled order. Write the correlation under each plot Draw a picture of an F distribution being used for a hpothesis test (such as regression or ANOVA). Include the following: F statistic obtained from the data, p-value (as a shaded region), 0 (location on the horizontal ais).

2 Math questions You don t need to show our work, but I will give partial credit for partial answers. 1. Assume ou have data from three groups and want to know if the group means are reliabl different from each other. The scores from the three groups are [4,8,2,,], [3,9,,7], and [4,,2,]. Calculate the total sum of squares, the treatment sum of squares, and the residual sum of squares (in whatever wa ou want). 2. Use our answers from Question 1 to calculate F. The degrees of freedom are df treatment = 2 and df residual = 10. (Even if ou were to have the wrong SS answers for Question 1, ou would get full credit for Question 2 if ou did all the steps correctl going from SSs to F.) 3. Si subjects are measured on two variables, and the results are converted to z-scores. The z-scores for the two variables are [-.8,.8, 1.3,.34, -1.18, -.8] and [.18,.2, 1.29,.31, -.7, -1.3]. Calculate the correlation. 4. Researchers studing childhood language acquisition measure several variables for a sample of children, including age, number of siblings, and ears of education for both the mother and father. The use these variables as predictors in a regression to tr to eplain the children s vocabular size (in number of words known). The regression coefficient for age (in months) is 2. M son is currentl 8½. How man new words would we predict him to learn between now and age 9 (i.e., in months)?. A new educational program designed to teach kindergarteners the alphabet is tested in a controlled eperiment. Half of the children in the stud receive the new curriculum and half do not. This is done in both urban and rural schools. Therefore there are two factors with two levels each (eperimental vs. control and urban vs. rural). At the conclusion of the stud, the children are tested for how man letters the can correctl write. The urban-eperimental children score [17, 13, 21, 9, 20, 1]. The urban-control children score [9, 1, 1, 10, 14, 20]. The rural-eperimental children score [19, 1, 20, 9, 11, 1]. What would the mean for the rural-control children need to be for there to be no interaction in the sample? R questions 1. What is the result of the following command? > pf(0,1,2) 2. Write a number that could not be the result of the following command. > cor(,) 3. Find MS regression using the output of the regression analsis below. Remember that the residual standard error is the square root of. (Hint: Use the formula for F.) > summar(lm( ~ 1+2)) Call: lm(formula = ~ 1 + 2) Residuals: Min 1Q Median 3Q Ma Coefficients: Estimate Std. Error t value Pr(> t ) (Intercept) *** ** --- Signif. codes: 0 *** ** 0.01 * Residual standard error: on 47 degrees of freedom Multiple R-squared: , Adjusted R-squared: F-statistic: on 2 and 47 DF, p-value: 1.90e-0

3 4. Use the regression output below to find the predicted depression score for a 47-ear-old mother of three. Call: lm(formula = depression ~ age.ears + children) Coefficients: (Intercept) age.ears children Answers Conceptual 1. The factors are Countr (with levels US and China), Age (with levels Child and Adult), and Task (with levels Analtic and Holistic). The possible interactions are Countr:Age, Countr:Task, Age:Task, and Countr:Age:Task. 2. The epected value of MS treatment is the population variance, σ Repeated-measures ANOVA differs from regular ANOVA in that we remove the variabilit due to individual differences before calculating the residual variabilit or SS residual. 4. The order of correlations is.7, -1, -.7, 1, 0.. Math 1. M = M 1 = = M 2 = = 4 M 3 = = 4 4 = SS total = ( 4 " ) 2 + ( 8 " ) 2 + ( 2 " ) 2 + ( " ) 2 + ( " ) 2 + ( 3 " ) 2 + ( 9 " ) 2 + ( " ) 2 + ( 7 " ) 2 + ( 4 " ) 2 + ( " ) 2 + ( 2 " ) 2 + ( " ) 2 = = 4

4 SS residual = ( 4 " ) 2 + ( 8 " ) 2 + ( 2 " ) 2 + ( " ) 2 + ( " ) 2 + ( 3 " ) 2 + ( 9 " ) 2 + ( " ) 2 + ( 7 " ) 2 + ( 4 " 4) 2 + ( " 4) 2 + ( 2 " 4) 2 + ( " 4) 2 = = 4 SS treatment = ( " ) 2 + 4( " ) 2 + 4( 4 " ) 2 = = 8 You can also calculate SS residual = SS total " SS treatment = 4 " 8 = 4 or SS treatment = SS total " SS residual = 4 " 4 = 8 2. MS treatment = SS treatment df treatment = 8 2 = 4 = SS residual df residual = 4 10 = 4. F = MS treatment = 4 4. = r = = ".8# # # #.31+ ("1.18)(".7) + (".8)("1.3) "1 " = 4.11 = The regression coefficient for age tells how much the prediction increases for ever additional month of age. Therefore the prediction increases b " 2 =10 words M urban, eperimental = = M urban, control = = M rural, eperimental = =1 There is no interaction if the effect of the program is the same for children in both locations. That is, M urban, eperimental " M urban, control = M rural, eperimental " M rural, control 1 "14 =1 " M rural, control M rural, control =13

5 R 1. The command asks for the probabilit that F is less than 0. F can never be negative, so the probabilit is (or an other number bigger than 1 or less than -1) 3. = =.719 F = MS regression MS regression = F" =13.82".719 = " 47 #.170" 3 =1.19