Download presentation

1
**Chapter 13 (Ch. 11 in 2nd Can. Ed.)**

Association Between Variables Measured at the Nominal Level: Phi, Cramer’s V, and Lambda

2
**Nominal Level Measures of Association**

It is always useful to compute column percentages for bivariate tables. But, it is also useful to have a summary measure – a single number – to indicate the strength of the relationship.

3
**Nominal Level Measures of Association**

For nominal level variables, there are two commonly used types of measures of association: 1. Chi-square based statistics: Phi or Cramer’s V 2. A PRE (Proportional Reduction in Error) measure known as Lambda

4
**Chi-square Based Measure: Phi ( )**

Phi is used for 2x2 tables. The formula for Phi:

5
**Example of Nominal Measure: Calculating Phi for #12.1 (#11.1 in 2nd)**

Step 1. Using the 5 step method explained in Chapter 11, calculate for the table. = (critical value = 3.841) There is a significant relationship between Authoritarianism and Worker Efficiency ( )

6
**Example: Calculating Phi for 12. 1(11**

Example: Calculating Phi for 12.1(11.1) Step 2: Calculate Phi… What does this mean?* Phi = There is a strong association between authoritarianism and efficiency *Keep this table handy! Value Strength 0.0 > 0.10 Weak 0.10 > 0.30 Moderate > 0.30 Strong

7
**Nominal Measures: Cramer’s V (V)**

Cramer’s V is used for tables larger than 2x2. Formula for Cramer’s V: Where (min r-1, c-1) is the minimum value of either the # rows-1 or the # columns-1.

8
**Limitations of Phi and V**

Phi is used for 2x2 tables only. For larger tables, use Cramer’s V. Phi (or V) is an index of the strength of the relationship only. It does not identify the pattern. Phi or V cannot be used to compare the strength of one relationship to another. To analyze the pattern of the relationship, see the column % in the bivariate table.

9
**Nominal Measures: Lambda (λ)**

Like Phi and V, Lambda is used to measure the strength of the relationship between nominal variables in bivariate tables. It can also be used to compare one relationship to another. Unlike Phi and V, Lambda is a PRE measure and its value has a more direct interpretation. PRE = Proportional Reduction in Error = how much less error do you make in your prediction of the dependent variable (y) when you take into account the values of the independent variable (x) Lambda tells us the improvement in predicting Y while taking X into account.

10
**Lambda (λ) cont. Formula for Lambda:**

Where E1 = (N – the largest row total) and E2 = (For each column, subtract the largest cell frequency from its column total and then add the differences together)

11
**Calculating Lambda (λ)- #12.1 (11.1 in 2nd):**

To compute λ, we must first find E1 and E2: E1 = N – largest row total = 44 – 22 = 22 E2 = For each column, subtract the largest cell frequency from the col. total and add together = (27 – 17) + (17 – 12) = = 15 Authoritarianism (x) Efficiency (y) Low High Total 10 12 22 17 5 27 44

12
**Calculating Lambda (cont.)**

Lambda is a PRE measure. A Lambda of .32 means that knowing something about authoritarianism (X) reduces our error in prediction of efficiency by 32%. We can also say that knowing about X increases our ability to predict efficiency (Y) by 32%.

13
More about Lambda (λ) The values of Lambda vary from 0.0 – +1.00, where 0.0 is very weak and 1.00 is very strong. Lambda can be used to compare the strength of two or more bivariate relationships. Suppose that you calculate: 1. Prestige X Happiness, λ = .55 and 2. Income X Happiness, λ = .20, …then you could say that Relationship 1 is stronger, and that Prestige has a greater effect on Happiness than Income.

14
More about Lambda (λ) Lambda, like Phi and V, tells you about the strength of the relationship only. It does not give information about pattern. To analyze the pattern of the relationship, use the column % in the bivariate table. Note that when row totals are very unequal, lambda can be zero even when there is an association between the variables.

15
**Answering the “Three Questions” when working with Nominal Level Variables.**

1. “Is there an association?” can be answered by looking at the % and by doing a test of significance like Х2. 2. “How strong is the association?” can be answered by using a measure of strength like Phi, Cramer’s V, or Lambda. 3. “What is the pattern of the association?” can be assessed by looking at the percentages.

16
**Practice Question: Healey #13.1 (11.2 in 2nd Ed.)**

Work with a partner on question #13.1 (a,b,c) and find an answer to all three of the above questions Answers are on the last slide. Compare the three relationships, and use Lambda to evaluate which relationship is the strongest. Write a brief summary of your findings.

17
**Answers to Healey #13.1 (11.2 in 2nd Ed.)**

Phi = .00 Lambda = .00 #13.1 (11.2) b Phi = .09 #13.1 (11.2) c Phi = .25 Lambda = .14

Similar presentations

OK

Chapter 14 in 1e Ch. 12 in 2/3 Can. Ed. Association Between Variables Measured at the Ordinal Level Using the Statistic Gamma and Conducting a Z-test for.

Chapter 14 in 1e Ch. 12 in 2/3 Can. Ed. Association Between Variables Measured at the Ordinal Level Using the Statistic Gamma and Conducting a Z-test for.

© 2017 SlidePlayer.com Inc.

All rights reserved.

Ads by Google

Ppt on agriculture insurance in india Upload and view ppt online form Ppt on forward rate agreement ppt Ppt on dubai shopping festival Ppt on indian equity market Ppt on 5g wireless system Ppt on series and parallel circuits Ppt on panel discussion Ppt on you can win the race Ppt on world population day