Multiple Regression Laboratory

The Last psyc350 Lab

Correlation & Multiple Regression Exercise

We will be looking at several analyses from the Interpersonal Relationships dataset you collected earlier this semester.

Walk-Through à Criterion variable is the Liking People Scale

Predictor variables are gender, number of times moved as a child, hometown population, Greek affiliation, number of close friends and total score on the Trust Scale

Univariate Analysis

a. Get the univariate statistics for each variable. For quantitative variables get the mean and the standard deviation. For binary variables (2-category qualitative variables) get the n and % for each category. Collect that information in the space below.

Variable / Mean / Std / Grp1 name / n / % / Grp2 name / n / %
Liking People Scale
Gender
Number of moves
Hometown population
Greek affiliation
Number of close friends
Trust Scale score

Bivariate Analyses

b. Get the correlations (and p-values) between the criterion and each predictor. Collect that information below and answer the question to complete the right-most column.

Predictor / Correlation (p) with Criterion / Is this predictor correlated with the criterion?
Gender
Number of moves
Hometown population
Greek affiliation
Number of close friends
Trust Scale score

Multivariate Analysis

c. Perform the multiple regression analysis and fill in the following

R = ______R² = ______F = ______df = ______, ______p = ______

Predictor / b (p) / β / Does this predictor contribute to the multiple regression model?
Gender
Number of moves
Hometown population
Greek affiliation
Number of close friends
Trust Scale score

d. What is the best single predictor of Liking People ? ______

e. Which is a better predictor of Liking People, the

best single predictor or the multiple regression model? ______why?

f. Name any variables that are correlated with Liking People and contribute to the multiple regression model?

g. Name any variables that are correlated with Liking People but do not contribute to the multiple regression model?

h. Pick one variable from “f” and interpret its correlation and its multiple regression weight.

i. Pick one variable from “g” and interpret its correlation and its multiple regression weight.

j. Considering “f”, “g”, “h”, and “i” – What did we learn from the multiple regression that we didn’t learn from the set of correlations?

On Your Own à Criterion variable is the Miller Social Intimacy Scale (MSIS)

Predictor variables are gender, number of siblings, hometown population, Greek affiliation, Liking People Scale and total score on the Trust Scale

Bivariate Analyses

a. Get the correlations (and p-values) between the criterion and each predictor. Collect that information below and answer the question to complete the right-most column.

Predictor / Correlation (p) with Criterion / Is this predictor correlated with the criterion?
Gender
Number of siblings
Hometown population
Greek affiliation
Liking People Scale
Trust Scale score

Multivariate Analysis

b. Perform the multiple regression analysis. Collect that information below and answer the question to complete the right-most column.

R = ______R² = ______F = ______df = ______, ______p = ______

Predictor / b (p) / β / Does this predictor contribute to the multiple regression model?
Gender
Number of siblings
Hometown population
Greek affiliation
Liking People Scale
Trust Scale score

c. What is the best single predictor of MSIS ? ______

d. Which is a better predictor of MSIS, the best single predictor or the multiple regression model? ______why?

e. Name any variables that are correlated with MSIS and contribute to the multiple regression model?

f. Name any variables that are correlated with MSIS and contribute to the multiple regression model?

g. Name any variables that are correlated with MSIS but do not contribute to the multiple regression model

h. Pick one variable from “f” and interpret its correlation and its multiple regression weight.

i. Pick one variable from “g” and interpret its correlation and its multiple regression weight.

j. Considering “f”, “g”, “h”, and “i” – What did we learn from the multiple regression that we didn’t learn from the set of correlations?

ANOVA & Factorial ANOVA Exercise

We will be looking at several analyses from the factab.sav dataset

Walk Through à The study looks at mothers who have been identified as being depressed. The mothers and their youngest child were invited to come to the lab and complete a 30-minute structured play activity

The dependent variable is “Praise” (how many time the mother-participant praises her child during the data collection protocol)

The IVs are the Child’s Age (2-3 years old vs. 6-7 years old) and the Treatment Status of the Mother (received treatment or not)

Bivarite Analysis

a. Get the ANOVA with Praise as the DV and Treatment Status of the Mother as the IV

Mean praises for untreated mothers ______mean praises for treated mothers ______

F = ______df = ____, ______p = ______MSe

Interpret the results of the ANOVA

Should we give this analysis a causal interpretation? Why or why not?

Multivariate Analysis

b. Complete the factorial analysis and compose a table (don’t forget the marginal means), a line graph and a bar graph of the data.

Note: We’ll refer to this as “Table 1” below

c. Find the results of the test of the interaction:

F = ______df = ____, ______p = ______MSe ______Is there an interaction ???

Find the components for the LSDmmd computation, if necessary:

# conditions = ______n = ______df error ______MSe = ______LSDmmd = ______

Show the pattern of the interaction - use the simple effects of Mother’s Treatment Status for each Age in Table 1

· Use <, >, or = to show the pattern of the simple effect of Treatment Status for 2-3 year olds

· Use <, >, or = to show the pattern of the simple effect of Treatment Status for 6-7 year olds

· Describe the interaction below:

d. Find the results for the test of the Main effect of Treatment Status

F = ______df = ____, _____ p = ______MSe ______Is there a main effect ?

Show the pattern of this main effect using <, > or = between the corresponding marginal means in Table 1.

We want to check if the main effect of Treatment Status is descriptive or misleading …

· Use <, >, or = to show the pattern of the simple effect of Treatment Status for 2-3 year olds

· Use <, >, or = to show the pattern of the simple effect of Treatment Status for 6-7 year olds

· So, is the main effect of Treatment Status descriptive or misleading?

e. Find the results for the test of the Main effect of Child’s Age

F = ______df = ____, _____ p = ______MSe ______Is there a main effect of Child’s AGe ?

Show the pattern of this main effect using <, > or = between the corresponding marginal means in Table 1.

We want to check if the main effect of Child’s Age is descriptive or misleading …

· Use <, >, or = to show the pattern of the simple effect of Child’s age when the Mother has not been treated

· So, is the main effect of Child’s Age descriptive or misleading?

f. Considering the “a’, “c”, and “d” – What did we learn from the factorial ANOVA that we didn’t learn from the ANOVA

BG Factorial Analysis Your Turn

This analysis uses Negations as the DV.

Bivarite Analysis

a. Get the ANOVA with Negations as the DV and Treatment Status of the Mother as the IV

Mean negations for untreated mothers ______mean negations for treated mothers ______

F = ______df = ____, ______p = ______MSe

Interpret the results of the ANOVA

Should we give this analysis a causal interpretation? Why or why not?

Multivariate Analysis

b. Complete the factorial analysis and compose a table (don’t forget the marginal means), a line graph and a bar graph of the data.

Note: We’ll refer to this as “Table 1” below

c. Find the results of the test of the interaction:

F = ______df = ____, ______p = ______MSe ______Is there an interaction ???

Find the components for the LSDmmd computation, if necessary:

# conditions = ______n = ______df error ______MSe = ______LSDmmd = ______

Show the pattern of the interaction - use the simple effects of Mother’s Treatment Status for each Age in Table 1